2020.5

PAKBuilder/main.cpp

@@ -39,10 +39,8 @@ std::queue<FileEntry> g_fileEntries;
 
 void TraverseDirectory(CWSZ parentDir, CWSZ foundDir, IO::RFile pakFile);
 
-int main(VERUS_MAIN_DEFAULT_ARGS)
+void Run()
 {
-	try
-	{
 	std::wcout << _T("PACK Builder 1.2.1") << std::endl;
 	std::wcout << _T("Copyright (c) 2006-2020 Dmitry Maluev") << std::endl;
 	std::wcout << _T("Arguments: [PAK/folder name] [input directory] [PAK output directory]") << std::endl;
@@ -126,13 +124,6 @@ int main(VERUS_MAIN_DEFAULT_ARGS)
 	pakFile.Close();
 	if (argCount < 2)
 		system("pause");
-	}
-	catch (const std::exception&)
-	{
-		system("pause");
-		return -1;
-	}
-	return 0;
 }
 
 INT64 WriteData(CWSZ name, RcString password, IO::RFile pakFile, const BYTE* p, INT64 size)
@@ -332,3 +323,17 @@ void TraverseDirectory(CWSZ parentDir, CWSZ foundDir, IO::RFile pakFile)
 	}
 	g_depth--;
 }
+
+int main(VERUS_MAIN_DEFAULT_ARGS)
+{
+	try
+	{
+		Run();
+	}
+	catch (const std::exception& e)
+	{
+		std::wcerr << _T("EXCEPTION: ") << e.what() << std::endl;
+		return EXIT_FAILURE;
+	}
+	return EXIT_SUCCESS;
+}

README.md

@@ -1,2 +1,8 @@
 # verus
 Game Engine
+
+# Release
+undef VERUS_RELEASE_DEBUG
+UseFullPaths=false
+DebugInformationFormat=None
+GenerateDebugInformation=false (Link)

RendererDirect3D12/src/CGI/CommandBufferD3D12.cpp

@@ -48,7 +48,7 @@ void CommandBufferD3D12::End()
 		throw VERUS_RUNTIME_ERROR << "Close(), hr=" << VERUS_HR(hr);
 }
 
-void CommandBufferD3D12::BeginRenderPass(int renderPassHandle, int framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor)
+void CommandBufferD3D12::BeginRenderPass(RPHandle renderPassHandle, FBHandle framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor)
 {
 	VERUS_QREF_RENDERER_D3D12;
 
@@ -99,7 +99,7 @@ void CommandBufferD3D12::EndRenderPass()
 		if (_vAttachmentStates[index] != attachment._finalState)
 		{
 			const auto& resources = _pFramebuffer->_vResources[index];
-			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(resources, _vAttachmentStates[index], attachment._finalState);
+			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(resources, _vAttachmentStates[index], attachment._finalState, 0);
 		}
 		index++;
 		if (VERUS_MAX_RT == barrierCount)
@@ -157,6 +157,13 @@ void CommandBufferD3D12::BindPipeline(PipelinePtr pipe)
 
 void CommandBufferD3D12::SetViewport(std::initializer_list<Vector4> il, float minDepth, float maxDepth)
 {
+	if (il.size() > 0)
+	{
+		const float w = il.begin()->Width();
+		const float h = il.begin()->Height();
+		_viewportSize = Vector4(w, h, 1 / w, 1 / h);
+	}
+
 	CD3DX12_VIEWPORT vpD3D12[VERUS_MAX_RT];
 	int count = 0;
 	for (const auto& rc : il)
@@ -182,7 +189,7 @@ void CommandBufferD3D12::SetBlendConstants(const float* p)
 	GetD3DGraphicsCommandList()->OMSetBlendFactor(p);
 }
 
-bool CommandBufferD3D12::BindDescriptors(ShaderPtr shader, int setNumber, int complexSetHandle)
+bool CommandBufferD3D12::BindDescriptors(ShaderPtr shader, int setNumber, CSHandle complexSetHandle)
 {
 	bool copyDescOnly = false;
 	if (setNumber < 0)
@@ -197,7 +204,7 @@ bool CommandBufferD3D12::BindDescriptors(ShaderPtr shader, int setNumber, int co
 	if (shaderD3D12.TryRootConstants(setNumber, *this))
 		return true;
 
-	const D3D12_GPU_DESCRIPTOR_HANDLE hGPU = shaderD3D12.UpdateUniformBuffer(setNumber, complexSetHandle, copyDescOnly);
+	const D3D12_GPU_DESCRIPTOR_HANDLE hGPU = shaderD3D12.UpdateUniformBuffer(setNumber, complexSetHandle.Get(), copyDescOnly);
 	if (!hGPU.ptr)
 		return false;
 
@@ -208,9 +215,9 @@ bool CommandBufferD3D12::BindDescriptors(ShaderPtr shader, int setNumber, int co
 	else
 		pCmdList->SetGraphicsRootDescriptorTable(rootParameterIndex, hGPU);
 
-	if (complexSetHandle >= 0 && !shaderD3D12.IsCompute())
+	if (complexSetHandle.IsSet() && !shaderD3D12.IsCompute())
 	{
-		const D3D12_GPU_DESCRIPTOR_HANDLE hGPU = shaderD3D12.UpdateSamplers(setNumber, complexSetHandle);
+		const D3D12_GPU_DESCRIPTOR_HANDLE hGPU = shaderD3D12.UpdateSamplers(setNumber, complexSetHandle.Get());
 		if (hGPU.ptr)
 		{
 			const UINT rootParameterIndex = shaderD3D12.GetDescriptorSetCount();
@@ -283,7 +290,7 @@ void CommandBufferD3D12::PrepareSubpass()
 		if (_vAttachmentStates[ref._index] != ref._state)
 		{
 			const auto& resources = fs._vResources[resIndex];
-			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(resources, _vAttachmentStates[ref._index], ref._state);
+			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(resources, _vAttachmentStates[ref._index], ref._state, 0);
 			_vAttachmentStates[ref._index] = ref._state;
 		}
 		resIndex++;
@@ -294,7 +301,7 @@ void CommandBufferD3D12::PrepareSubpass()
 		if (_vAttachmentStates[ref._index] != ref._state)
 		{
 			const auto& resources = fs._vResources[resIndex];
-			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(resources, _vAttachmentStates[ref._index], ref._state);
+			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(resources, _vAttachmentStates[ref._index], ref._state, 0);
 			_vAttachmentStates[ref._index] = ref._state;
 		}
 		resIndex++;
@@ -304,7 +311,7 @@ void CommandBufferD3D12::PrepareSubpass()
 		const auto& ref = subpass._depthStencil;
 		if (_vAttachmentStates[ref._index] != ref._state)
 		{
-			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(fs._vResources.back(), _vAttachmentStates[ref._index], ref._state);
+			barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(fs._vResources.back(), _vAttachmentStates[ref._index], ref._state, 0);
 			_vAttachmentStates[ref._index] = ref._state;
 		}
 	}

RendererDirect3D12/src/CGI/CommandBufferD3D12.h

@@ -23,7 +23,7 @@ namespace verus
 			virtual void Begin() override;
 			virtual void End() override;
 
-			virtual void BeginRenderPass(int renderPassHandle, int framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor) override;
+			virtual void BeginRenderPass(RPHandle renderPassHandle, FBHandle framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor) override;
 			virtual void NextSubpass() override;
 			virtual void EndRenderPass() override;
 
@@ -35,7 +35,7 @@ namespace verus
 			virtual void SetScissor(std::initializer_list<Vector4> il) override;
 			virtual void SetBlendConstants(const float* p) override;
 
-			virtual bool BindDescriptors(ShaderPtr shader, int setNumber, int complexSetHandle) override;
+			virtual bool BindDescriptors(ShaderPtr shader, int setNumber, CSHandle complexSetHandle) override;
 			virtual void PushConstants(ShaderPtr shader, int offset, int size, const void* p, ShaderStageFlags stageFlags) override;
 
 			virtual void PipelineImageMemoryBarrier(TexturePtr tex, ImageLayout oldLayout, ImageLayout newLayout, Range<int> mipLevels, int arrayLayer) override;

RendererDirect3D12/src/CGI/GeometryD3D12.cpp

@@ -20,11 +20,11 @@ void GeometryD3D12::Init(RcGeometryDesc desc)
 	_dynBindingsMask = desc._dynBindingsMask;
 	_32BitIndices = desc._32BitIndices;
 
-	_vInputElementDesc.reserve(GetInputElementDescCount(desc._pInputElementDesc));
+	_vInputElementDesc.reserve(GetVertexInputAttrDescCount(desc._pVertexInputAttrDesc));
 	int i = 0;
-	while (desc._pInputElementDesc[i]._offset >= 0)
+	while (desc._pVertexInputAttrDesc[i]._offset >= 0)
 	{
-		int binding = desc._pInputElementDesc[i]._binding;
+		int binding = desc._pVertexInputAttrDesc[i]._binding;
 		D3D12_INPUT_CLASSIFICATION inputClassification = D3D12_INPUT_CLASSIFICATION_PER_VERTEX_DATA;
 		UINT instanceDataStepRate = 0;
 		if (binding < 0)
@@ -36,11 +36,11 @@ void GeometryD3D12::Init(RcGeometryDesc desc)
 		}
 		D3D12_INPUT_ELEMENT_DESC ieDesc =
 		{
-			ToNativeSemanticName(desc._pInputElementDesc[i]._usage),
-			static_cast<UINT>(desc._pInputElementDesc[i]._usageIndex),
-			ToNativeFormat(desc._pInputElementDesc[i]._usage, desc._pInputElementDesc[i]._type, desc._pInputElementDesc[i]._components),
+			ToNativeSemanticName(desc._pVertexInputAttrDesc[i]._usage),
+			static_cast<UINT>(desc._pVertexInputAttrDesc[i]._usageIndex),
+			ToNativeFormat(desc._pVertexInputAttrDesc[i]._usage, desc._pVertexInputAttrDesc[i]._type, desc._pVertexInputAttrDesc[i]._components),
 			static_cast<UINT>(binding),
-			static_cast<UINT>(desc._pInputElementDesc[i]._offset),
+			static_cast<UINT>(desc._pVertexInputAttrDesc[i]._offset),
 			inputClassification,
 			instanceDataStepRate
 		};
@@ -48,7 +48,7 @@ void GeometryD3D12::Init(RcGeometryDesc desc)
 		i++;
 	}
 
-	_vStrides.reserve(GetBindingCount(desc._pInputElementDesc));
+	_vStrides.reserve(GetBindingCount(desc._pVertexInputAttrDesc));
 	i = 0;
 	while (desc._pStrides[i] > 0)
 	{

RendererDirect3D12/src/CGI/Native.cpp

@@ -103,6 +103,8 @@ D3D_PRIMITIVE_TOPOLOGY CGI::ToNativePrimitiveTopology(PrimitiveTopology primitiv
 	case PrimitiveTopology::lineStrip:     return D3D_PRIMITIVE_TOPOLOGY_LINESTRIP;
 	case PrimitiveTopology::triangleList:  return D3D_PRIMITIVE_TOPOLOGY_TRIANGLELIST;
 	case PrimitiveTopology::triangleStrip: return D3D_PRIMITIVE_TOPOLOGY_TRIANGLESTRIP;
+	case PrimitiveTopology::patchList3:    return D3D_PRIMITIVE_TOPOLOGY_3_CONTROL_POINT_PATCHLIST;
+	case PrimitiveTopology::patchList4:    return D3D_PRIMITIVE_TOPOLOGY_4_CONTROL_POINT_PATCHLIST;
 	default: throw VERUS_RECOVERABLE << "ToNativePrimitiveTopology()";
 	}
 }
@@ -116,6 +118,8 @@ D3D12_PRIMITIVE_TOPOLOGY_TYPE CGI::ToNativePrimitiveTopologyType(PrimitiveTopolo
 	case PrimitiveTopology::lineStrip:     return D3D12_PRIMITIVE_TOPOLOGY_TYPE_LINE;
 	case PrimitiveTopology::triangleList:  return D3D12_PRIMITIVE_TOPOLOGY_TYPE_TRIANGLE;
 	case PrimitiveTopology::triangleStrip: return D3D12_PRIMITIVE_TOPOLOGY_TYPE_TRIANGLE;
+	case PrimitiveTopology::patchList3:    return D3D12_PRIMITIVE_TOPOLOGY_TYPE_PATCH;
+	case PrimitiveTopology::patchList4:    return D3D12_PRIMITIVE_TOPOLOGY_TYPE_PATCH;
 	default: throw VERUS_RECOVERABLE << "ToNativePrimitiveTopologyType()";
 	}
 }
@@ -141,14 +145,16 @@ CSZ CGI::ToNativeSemanticName(IeUsage usage)
 	static const CSZ names[] =
 	{
 		"POSITION",     // IeUsage::position
-		"BLENDWEIGHT",  // IeUsage::blendWeight
+		"BLENDWEIGHTS", // IeUsage::blendWeights
 		"BLENDINDICES", // IeUsage::blendIndices
 		"NORMAL",       // IeUsage::normal
+		"TANGENT",      // IeUsage::tangent
+		"BINORMAL",     // IeUsage::binormal
+		"COLOR",        // IeUsage::color
 		"PSIZE",        // IeUsage::psize
 		"TEXCOORD",     // IeUsage::texCoord
-		"TEXCOORD",     // IeUsage::tangent
-		"TEXCOORD",     // IeUsage::binormal
-		"COLOR"         // IeUsage::color
+		"INSTDATA",     // IeUsage::instData
+		"ATTR"          // IeUsage::attr
 	};
 	return names[+usage];
 }

RendererDirect3D12/src/CGI/PipelineD3D12.cpp

@@ -51,11 +51,12 @@ void PipelineD3D12::Init(RcPipelineDesc desc)
 
 	D3D12_GRAPHICS_PIPELINE_STATE_DESC gpsDesc = {};
 	gpsDesc.pRootSignature = _pRootSignature;
-	gpsDesc.VS = ToBytecode(shader.GetD3DBlob(desc._shaderBranch, BaseShader::Stage::vs));
-	gpsDesc.PS = ToBytecode(shader.GetD3DBlob(desc._shaderBranch, BaseShader::Stage::fs));
-	gpsDesc.DS = ToBytecode(shader.GetD3DBlob(desc._shaderBranch, BaseShader::Stage::ds));
-	gpsDesc.HS = ToBytecode(shader.GetD3DBlob(desc._shaderBranch, BaseShader::Stage::hs));
-	gpsDesc.GS = ToBytecode(shader.GetD3DBlob(desc._shaderBranch, BaseShader::Stage::gs));
+	ShaderD3D12::RcCompiled compiled = shader.GetCompiled(desc._shaderBranch);
+	gpsDesc.VS = ToBytecode(compiled._pBlobs[+BaseShader::Stage::vs].Get());
+	gpsDesc.PS = ToBytecode(compiled._pBlobs[+BaseShader::Stage::fs].Get());
+	gpsDesc.DS = ToBytecode(compiled._pBlobs[+BaseShader::Stage::ds].Get());
+	gpsDesc.HS = ToBytecode(compiled._pBlobs[+BaseShader::Stage::hs].Get());
+	gpsDesc.GS = ToBytecode(compiled._pBlobs[+BaseShader::Stage::gs].Get());
 	gpsDesc.StreamOutput = {};
 
 	gpsDesc.BlendState = {};
@@ -141,7 +142,7 @@ void PipelineD3D12::Init(RcPipelineDesc desc)
 	gpsDesc.RasterizerState.SlopeScaledDepthBias = desc._rasterizationState._depthBiasSlopeFactor;
 	gpsDesc.RasterizerState.DepthClipEnable = desc._rasterizationState._depthClampEnable;
 	gpsDesc.RasterizerState.MultisampleEnable = FALSE;
-	gpsDesc.RasterizerState.AntialiasedLineEnable = TRUE;
+	gpsDesc.RasterizerState.AntialiasedLineEnable = (desc._colorAttachBlendEqs[0] == VERUS_COLOR_BLEND_ALPHA) ? TRUE : FALSE;
 	gpsDesc.RasterizerState.ForcedSampleCount = 0;
 	gpsDesc.RasterizerState.ConservativeRaster = D3D12_CONSERVATIVE_RASTERIZATION_MODE_OFF;
 
@@ -200,7 +201,8 @@ void PipelineD3D12::InitCompute(RcPipelineDesc desc)
 
 	D3D12_COMPUTE_PIPELINE_STATE_DESC cpsDesc = {};
 	cpsDesc.pRootSignature = _pRootSignature;
-	cpsDesc.CS = ToBytecode(shader.GetD3DBlob(desc._shaderBranch, BaseShader::Stage::cs));
+	ShaderD3D12::RcCompiled compiled = shader.GetCompiled(desc._shaderBranch);
+	cpsDesc.CS = ToBytecode(compiled._pBlobs[+BaseShader::Stage::cs].Get());
 	cpsDesc.NodeMask = 0;
 	cpsDesc.CachedPSO = { nullptr, 0 };
 	cpsDesc.Flags = D3D12_PIPELINE_STATE_FLAG_NONE;

RendererDirect3D12/src/CGI/RendererD3D12.cpp

@@ -40,13 +40,13 @@ void RendererD3D12::Done()
 		Renderer::I().ImGuiSetCurrentContext(nullptr);
 	}
 
-	DeleteFramebuffer(-1);
-	DeleteRenderPass(-1);
+	DeleteFramebuffer(FBHandle::Make(-2));
+	DeleteRenderPass(RPHandle::Make(-2));
 
 	_vSamplers.clear();
 
 	_dhSamplers.Reset();
-	_dhCbvSrvUav.Reset();
+	_dhViews.Reset();
 
 	if (_hFence != INVALID_HANDLE_VALUE)
 	{
@@ -210,7 +210,7 @@ void RendererD3D12::InitD3D()
 	_pFence = CreateFence();
 	_hFence = CreateEvent(nullptr, FALSE, FALSE, nullptr);
 
-	_dhCbvSrvUav.Create(_pDevice.Get(), D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV, settings.GetLimits()._d3d12_dhCbvSrvUavCapacity, 16, true);
+	_dhViews.Create(_pDevice.Get(), D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV, settings.GetLimits()._d3d12_dhViewsCapacity, 16, true);
 	_dhSamplers.Create(_pDevice.Get(), D3D12_DESCRIPTOR_HEAP_TYPE_SAMPLER, settings.GetLimits()._d3d12_dhSamplersCapacity, 0, true);
 
 	CreateSamplers();
@@ -326,40 +326,40 @@ void RendererD3D12::CreateSamplers()
 
 	// <Wrap>
 	desc = init;
-	_vSamplers[+Sampler::linear3D] = desc;
+	_vSamplers[+Sampler::linearMipL] = desc;
 
 	desc = init;
 	desc.Filter = ApplyTrilinearFilter(D3D12_FILTER_MIN_MAG_MIP_POINT);
-	_vSamplers[+Sampler::nearest3D] = desc;
+	_vSamplers[+Sampler::nearestMipL] = desc;
 
 	desc = init;
 	desc.Filter = D3D12_FILTER_MIN_MAG_LINEAR_MIP_POINT;
-	_vSamplers[+Sampler::linear2D] = desc;
+	_vSamplers[+Sampler::linearMipN] = desc;
 
 	desc = init;
 	desc.Filter = D3D12_FILTER_MIN_MAG_MIP_POINT;
-	_vSamplers[+Sampler::nearest2D] = desc;
+	_vSamplers[+Sampler::nearestMipN] = desc;
 	// </Wrap>
 
 	// <Clamp>
 	desc = init;
 	desc.AddressU = desc.AddressV = desc.AddressW = D3D12_TEXTURE_ADDRESS_MODE_CLAMP;
-	_vSamplers[+Sampler::linearClamp3D] = desc;
+	_vSamplers[+Sampler::linearClampMipL] = desc;
 
 	desc = init;
 	desc.Filter = ApplyTrilinearFilter(D3D12_FILTER_MIN_MAG_MIP_POINT);
 	desc.AddressU = desc.AddressV = desc.AddressW = D3D12_TEXTURE_ADDRESS_MODE_CLAMP;
-	_vSamplers[+Sampler::nearestClamp3D] = desc;
+	_vSamplers[+Sampler::nearestClampMipL] = desc;
 
 	desc = init;
 	desc.Filter = D3D12_FILTER_MIN_MAG_LINEAR_MIP_POINT;
 	desc.AddressU = desc.AddressV = desc.AddressW = D3D12_TEXTURE_ADDRESS_MODE_CLAMP;
-	_vSamplers[+Sampler::linearClamp2D] = desc;
+	_vSamplers[+Sampler::linearClampMipN] = desc;
 
 	desc = init;
 	desc.Filter = D3D12_FILTER_MIN_MAG_MIP_POINT;
 	desc.AddressU = desc.AddressV = desc.AddressW = D3D12_TEXTURE_ADDRESS_MODE_CLAMP;
-	_vSamplers[+Sampler::nearestClamp2D] = desc;
+	_vSamplers[+Sampler::nearestClampMipN] = desc;
 	// </Clamp>
 }
 
@@ -368,7 +368,7 @@ D3D12_STATIC_SAMPLER_DESC RendererD3D12::GetStaticSamplerDesc(Sampler s) const
 	return _vSamplers[+s];
 }
 
-void RendererD3D12::ImGuiInit(int renderPassHandle)
+void RendererD3D12::ImGuiInit(RPHandle renderPassHandle)
 {
 	VERUS_QREF_RENDERER;
 	VERUS_QREF_CONST_SETTINGS;
@@ -392,12 +392,12 @@ void RendererD3D12::ImGuiInit(int renderPassHandle)
 	ImGui::StyleColorsDark();
 
 	ImGui_ImplSDL2_InitForD3D(renderer.GetMainWindow()->GetSDL());
-	auto hp = _dhCbvSrvUav.GetStaticHandlePair(0);
+	auto hp = _dhViews.GetStaticHandlePair(0);
 	ImGui_ImplDX12_Init(
 		_pDevice.Get(),
 		s_ringBufferSize,
 		DXGI_FORMAT_B8G8R8A8_UNORM_SRGB,
-		_dhCbvSrvUav.GetD3DDescriptorHeap(),
+		_dhViews.GetD3DDescriptorHeap(),
 		hp._hCPU,
 		hp._hGPU);
 }
@@ -450,7 +450,7 @@ void RendererD3D12::BeginFrame(bool present)
 
 	renderer.GetCommandBuffer()->Begin();
 
-	ID3D12DescriptorHeap* ppHeaps[] = { _dhCbvSrvUav.GetD3DDescriptorHeap(), _dhSamplers.GetD3DDescriptorHeap() };
+	ID3D12DescriptorHeap* ppHeaps[] = { _dhViews.GetD3DDescriptorHeap(), _dhSamplers.GetD3DDescriptorHeap() };
 	static_cast<CommandBufferD3D12*>(&(*renderer.GetCommandBuffer()))->GetD3DGraphicsCommandList()->SetDescriptorHeaps(2, ppHeaps);
 }
 
@@ -552,7 +552,7 @@ void RendererD3D12::DeleteTexture(PBaseTexture p)
 	TStoreTextures::Delete(static_cast<PTextureD3D12>(p));
 }
 
-int RendererD3D12::CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD)
+RPHandle RendererD3D12::CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD)
 {
 	RP::D3DRenderPass renderPass;
 
@@ -627,16 +627,16 @@ int RendererD3D12::CreateRenderPass(std::initializer_list<RP::Attachment> ilA, s
 		renderPass._vSubpasses.push_back(std::move(d3dSubpass));
 	}
 
-	const int handle = GetNextRenderPassHandle();
-	if (handle >= _vRenderPasses.size())
+	const int nextIndex = GetNextRenderPassIndex();
+	if (nextIndex >= _vRenderPasses.size())
 		_vRenderPasses.push_back(std::move(renderPass));
 	else
-		_vRenderPasses[handle] = renderPass;
+		_vRenderPasses[nextIndex] = renderPass;
 
-	return handle;
+	return RPHandle::Make(nextIndex);
 }
 
-int RendererD3D12::CreateFramebuffer(int renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex)
+FBHandle RendererD3D12::CreateFramebuffer(RPHandle renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex)
 {
 	RP::RcD3DRenderPass renderPass = GetRenderPass(renderPassHandle);
 	RP::D3DFramebuffer framebuffer;
@@ -746,32 +746,32 @@ int RendererD3D12::CreateFramebuffer(int renderPassHandle, std::initializer_list
 		framebuffer._vSubpasses.push_back(std::move(fs));
 	}
 
-	const int handle = GetNextFramebufferHandle();
-	if (handle >= _vFramebuffers.size())
+	const int nextIndex = GetNextFramebufferIndex();
+	if (nextIndex >= _vFramebuffers.size())
 		_vFramebuffers.push_back(std::move(framebuffer));
 	else
-		_vFramebuffers[handle] = framebuffer;
+		_vFramebuffers[nextIndex] = framebuffer;
 
-	return handle;
+	return FBHandle::Make(nextIndex);
 }
 
-void RendererD3D12::DeleteRenderPass(int handle)
+void RendererD3D12::DeleteRenderPass(RPHandle handle)
 {
-	if (handle >= 0)
-		_vRenderPasses[handle] = RP::D3DRenderPass();
-	else
+	if (handle.IsSet())
+		_vRenderPasses[handle.Get()] = RP::D3DRenderPass();
+	else if (-2 == handle.Get())
 		_vRenderPasses.clear();
 }
 
-void RendererD3D12::DeleteFramebuffer(int handle)
+void RendererD3D12::DeleteFramebuffer(FBHandle handle)
 {
-	if (handle >= 0)
-		_vFramebuffers[handle] = RP::D3DFramebuffer();
-	else
+	if (handle.IsSet())
+		_vFramebuffers[handle.Get()] = RP::D3DFramebuffer();
+	else if (-2 == handle.Get())
 		_vFramebuffers.clear();
 }
 
-int RendererD3D12::GetNextRenderPassHandle() const
+int RendererD3D12::GetNextRenderPassIndex() const
 {
 	const int count = Utils::Cast32(_vRenderPasses.size());
 	VERUS_FOR(i, count)
@@ -782,7 +782,7 @@ int RendererD3D12::GetNextRenderPassHandle() const
 	return count;
 }
 
-int RendererD3D12::GetNextFramebufferHandle() const
+int RendererD3D12::GetNextFramebufferIndex() const
 {
 	const int count = Utils::Cast32(_vFramebuffers.size());
 	VERUS_FOR(i, count)
@@ -793,12 +793,12 @@ int RendererD3D12::GetNextFramebufferHandle() const
 	return count;
 }
 
-RP::RcD3DRenderPass RendererD3D12::GetRenderPass(int handle) const
+RP::RcD3DRenderPass RendererD3D12::GetRenderPass(RPHandle handle) const
 {
-	return _vRenderPasses[handle];
+	return _vRenderPasses[handle.Get()];
 }
 
-RP::RcD3DFramebuffer RendererD3D12::GetFramebuffer(int handle) const
+RP::RcD3DFramebuffer RendererD3D12::GetFramebuffer(FBHandle handle) const
 {
-	return _vFramebuffers[handle];
+	return _vFramebuffers[handle.Get()];
 }

RendererDirect3D12/src/CGI/RendererD3D12.h

@@ -20,7 +20,7 @@ namespace verus
 			ComPtr<IDXGISwapChain4>           _pSwapChain;
 			Vector<ComPtr<ID3D12Resource>>    _vSwapChainBuffers;
 			DescriptorHeap                    _dhSwapChainBuffersRTVs;
-			DynamicDescriptorHeap             _dhCbvSrvUav;
+			DynamicDescriptorHeap             _dhViews;
 			DynamicDescriptorHeap             _dhSamplers;
 			TMapCommandAllocators             _mapCommandAllocators[s_ringBufferSize];
 			ComPtr<ID3D12Fence>               _pFence;
@@ -64,7 +64,7 @@ namespace verus
 			ID3D12CommandAllocator* GetD3DCommandAllocator(int ringBufferIndex) const { return _mapCommandAllocators[ringBufferIndex].at(std::this_thread::get_id()).Get(); }
 			D3D12_STATIC_SAMPLER_DESC GetStaticSamplerDesc(Sampler s) const;
 
-			virtual void ImGuiInit(int renderPassHandle) override;
+			virtual void ImGuiInit(RPHandle renderPassHandle) override;
 			virtual void ImGuiRenderDrawData() override;
 
 			virtual void ResizeSwapChain() override;
@@ -91,17 +91,17 @@ namespace verus
 			virtual void DeleteShader(PBaseShader p) override;
 			virtual void DeleteTexture(PBaseTexture p) override;
 
-			virtual int CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD) override;
-			virtual int CreateFramebuffer(int renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex = -1) override;
-			virtual void DeleteRenderPass(int handle) override;
-			virtual void DeleteFramebuffer(int handle) override;
-			int GetNextRenderPassHandle() const;
-			int GetNextFramebufferHandle() const;
-			RP::RcD3DRenderPass GetRenderPass(int handle) const;
-			RP::RcD3DFramebuffer GetFramebuffer(int handle) const;
+			virtual RPHandle CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD) override;
+			virtual FBHandle CreateFramebuffer(RPHandle renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex = -1) override;
+			virtual void DeleteRenderPass(RPHandle handle) override;
+			virtual void DeleteFramebuffer(FBHandle handle) override;
+			int GetNextRenderPassIndex() const;
+			int GetNextFramebufferIndex() const;
+			RP::RcD3DRenderPass GetRenderPass(RPHandle handle) const;
+			RP::RcD3DFramebuffer GetFramebuffer(FBHandle handle) const;
 
-			RDynamicDescriptorHeap GetHeapCbvSrvUav() { return _dhCbvSrvUav; }
-			RDynamicDescriptorHeap GetHeapSampler() { return _dhSamplers; }
+			RDynamicDescriptorHeap GetViewHeap() { return _dhViews; }
+			RDynamicDescriptorHeap GetSamplerHeap() { return _dhSamplers; }
 		};
 		VERUS_TYPEDEFS(RendererD3D12);
 	}

RendererDirect3D12/src/CGI/ShaderD3D12.cpp

@@ -66,12 +66,12 @@ void ShaderD3D12::Init(CSZ source, CSZ sourceName, CSZ* branches)
 
 	while (*branches)
 	{
-		String entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages;
+		String entry, stageEntries[+Stage::count], stages;
 		Vector<String> vMacroName;
 		Vector<String> vMacroValue;
-		const String entry = Parse(*branches, entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages, vMacroName, vMacroValue, "DEF_");
+		const String branch = Parse(*branches, entry, stageEntries, stages, vMacroName, vMacroValue, "DEF_");
 
-		if (IsInIgnoreList(_C(entry)))
+		if (IsInIgnoreList(_C(branch)))
 		{
 			branches++;
 			continue;
@@ -118,12 +118,13 @@ void ShaderD3D12::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		// </System defines>
 
 		Compiled compiled;
+		compiled._entry = entry;
 
 		if (strchr(_C(stages), 'V'))
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex].Name = "_VS";
-			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(entryVS), _C("vs_" + version), flags, 0, &compiled._pBlobs[+Stage::vs], &pErrorMsgs);
+			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::vs]), _C("vs_" + version), flags, 0, &compiled._pBlobs[+Stage::vs], &pErrorMsgs);
 			CheckErrorMsgs(pErrorMsgs);
 		}
 
@@ -131,7 +132,7 @@ void ShaderD3D12::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex].Name = "_HS";
-			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(entryHS), _C("hs_" + version), flags, 0, &compiled._pBlobs[+Stage::hs], &pErrorMsgs);
+			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::hs]), _C("hs_" + version), flags, 0, &compiled._pBlobs[+Stage::hs], &pErrorMsgs);
 			CheckErrorMsgs(pErrorMsgs);
 		}
 
@@ -139,7 +140,7 @@ void ShaderD3D12::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex].Name = "_DS";
-			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(entryDS), _C("ds_" + version), flags, 0, &compiled._pBlobs[+Stage::ds], &pErrorMsgs);
+			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::ds]), _C("ds_" + version), flags, 0, &compiled._pBlobs[+Stage::ds], &pErrorMsgs);
 			CheckErrorMsgs(pErrorMsgs);
 		}
 
@@ -147,7 +148,7 @@ void ShaderD3D12::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex].Name = "_GS";
-			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(entryGS), _C("gs_" + version), flags, 0, &compiled._pBlobs[+Stage::gs], &pErrorMsgs);
+			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::gs]), _C("gs_" + version), flags, 0, &compiled._pBlobs[+Stage::gs], &pErrorMsgs);
 			CheckErrorMsgs(pErrorMsgs);
 		}
 
@@ -155,7 +156,7 @@ void ShaderD3D12::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex].Name = "_FS";
-			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(entryFS), _C("ps_" + version), flags, 0, &compiled._pBlobs[+Stage::fs], &pErrorMsgs);
+			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::fs]), _C("ps_" + version), flags, 0, &compiled._pBlobs[+Stage::fs], &pErrorMsgs);
 			CheckErrorMsgs(pErrorMsgs);
 		}
 
@@ -163,12 +164,12 @@ void ShaderD3D12::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex].Name = "_CS";
-			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(entryCS), _C("cs_" + version), flags, 0, &compiled._pBlobs[+Stage::cs], &pErrorMsgs);
+			hr = D3DCompile(source, len, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::cs]), _C("cs_" + version), flags, 0, &compiled._pBlobs[+Stage::cs], &pErrorMsgs);
 			CheckErrorMsgs(pErrorMsgs);
 			_compute = true;
 		}
 
-		_mapCompiled[entry] = compiled;
+		_mapCompiled[branch] = compiled;
 
 		branches++;
 	}
@@ -284,6 +285,16 @@ void ShaderD3D12::CreatePipelineLayout()
 			stageFlags |= dsd._stageFlags;
 			if (dsd._capacity > 0)
 			{
+				D3D12_SHADER_VISIBILITY visibility = D3D12_SHADER_VISIBILITY_ALL;
+				switch (dsd._stageFlags)
+				{
+				case ShaderStageFlags::vs: visibility = D3D12_SHADER_VISIBILITY_VERTEX; break;
+				case ShaderStageFlags::hs: visibility = D3D12_SHADER_VISIBILITY_HULL; break;
+				case ShaderStageFlags::ds: visibility = D3D12_SHADER_VISIBILITY_DOMAIN; break;
+				case ShaderStageFlags::gs: visibility = D3D12_SHADER_VISIBILITY_GEOMETRY; break;
+				case ShaderStageFlags::fs: visibility = D3D12_SHADER_VISIBILITY_PIXEL; break;
+				}
+
 				CD3DX12_DESCRIPTOR_RANGE1 descRange;
 				descRange.Init(D3D12_DESCRIPTOR_RANGE_TYPE_CBV, 1, 0, space);
 				vDescRanges.push_back(descRange);
@@ -313,18 +324,15 @@ void ShaderD3D12::CreatePipelineLayout()
 						{
 							Sampler s = dsd._vSamplers[i];
 							if (Sampler::input == s)
-								s = Sampler::nearest2D;
+								s = Sampler::nearestMipN;
 							D3D12_STATIC_SAMPLER_DESC samplerDesc = pRendererD3D12->GetStaticSamplerDesc(s);
 							samplerDesc.ShaderRegister = i + 1;
 							samplerDesc.RegisterSpace = space;
-							samplerDesc.ShaderVisibility = D3D12_SHADER_VISIBILITY_ALL;
+							samplerDesc.ShaderVisibility = visibility;
 							vStaticSamplers.push_back(samplerDesc);
 						}
 					}
 				}
-				D3D12_SHADER_VISIBILITY visibility = D3D12_SHADER_VISIBILITY_ALL;
-				if (ShaderStageFlags::fs == dsd._stageFlags)
-					visibility = D3D12_SHADER_VISIBILITY_PIXEL;
 				CD3DX12_ROOT_PARAMETER1 rootParam;
 				rootParam.InitAsDescriptorTable(1 + textureCount, pDescriptorRanges, visibility);
 				vRootParams.push_back(rootParam);
@@ -359,6 +367,8 @@ void ShaderD3D12::CreatePipelineLayout()
 			rootSignatureFlags |= D3D12_ROOT_SIGNATURE_FLAG_DENY_PIXEL_SHADER_ROOT_ACCESS;
 	}
 
+	UpdateDebugInfo(vRootParams, vStaticSamplers, rootSignatureFlags);
+
 	CD3DX12_VERSIONED_ROOT_SIGNATURE_DESC rootSignatureDesc;
 	rootSignatureDesc.Init_1_1(
 		Utils::Cast32(vRootParams.size()), vRootParams.data(),
@@ -376,7 +386,7 @@ void ShaderD3D12::CreatePipelineLayout()
 		throw VERUS_RUNTIME_ERROR << "CreateRootSignature(), hr=" << VERUS_HR(hr);
 }
 
-int ShaderD3D12::BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips)
+CSHandle ShaderD3D12::BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips)
 {
 	VERUS_QREF_RENDERER_D3D12;
 
@@ -400,7 +410,7 @@ int ShaderD3D12::BindDescriptorSetTextures(int setNumber, std::initializer_list<
 
 	RComplexSet complexSet = _vComplexSets[complexSetHandle];
 	complexSet._vTextures.reserve(il.size());
-	complexSet._dhSrvUav.Create(pRendererD3D12->GetD3DDevice(), D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV, Utils::Cast32(il.size()));
+	complexSet._dhViews.Create(pRendererD3D12->GetD3DDevice(), D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV, Utils::Cast32(il.size()));
 	if (!dsd._staticSamplersOnly)
 		complexSet._dhSamplers.Create(pRendererD3D12->GetD3DDevice(), D3D12_DESCRIPTOR_HEAP_TYPE_SAMPLER, Utils::Cast32(il.size()));
 	int index = 0;
@@ -412,14 +422,14 @@ int ShaderD3D12::BindDescriptorSetTextures(int setNumber, std::initializer_list<
 		if (Sampler::storage == dsd._vSamplers[index])
 		{
 			pRendererD3D12->GetD3DDevice()->CopyDescriptorsSimple(1,
-				complexSet._dhSrvUav.AtCPU(index),
+				complexSet._dhViews.AtCPU(index),
 				texD3D12.GetDescriptorHeapUAV().AtCPU(mip),
 				D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV);
 		}
 		else
 		{
 			pRendererD3D12->GetD3DDevice()->CopyDescriptorsSimple(1,
-				complexSet._dhSrvUav.AtCPU(index),
+				complexSet._dhViews.AtCPU(index),
 				texD3D12.GetDescriptorHeapSRV().AtCPU(mip),
 				D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV);
 			if (Sampler::custom == dsd._vSamplers[index])
@@ -433,18 +443,18 @@ int ShaderD3D12::BindDescriptorSetTextures(int setNumber, std::initializer_list<
 		index++;
 	}
 
-	return complexSetHandle;
+	return CSHandle::Make(complexSetHandle);
 }
 
-void ShaderD3D12::FreeDescriptorSet(int& complexSetHandle)
+void ShaderD3D12::FreeDescriptorSet(CSHandle& complexSetHandle)
 {
-	if (complexSetHandle >= 0 && complexSetHandle < _vComplexSets.size())
+	if (complexSetHandle.IsSet() && complexSetHandle.Get() < _vComplexSets.size())
 	{
-		_vComplexSets[complexSetHandle]._vTextures.clear();
-		_vComplexSets[complexSetHandle]._dhSrvUav.Reset();
-		_vComplexSets[complexSetHandle]._dhSamplers.Reset();
+		_vComplexSets[complexSetHandle.Get()]._vTextures.clear();
+		_vComplexSets[complexSetHandle.Get()]._dhViews.Reset();
+		_vComplexSets[complexSetHandle.Get()]._dhSamplers.Reset();
 	}
-	complexSetHandle = -1;
+	complexSetHandle = CSHandle();
 }
 
 void ShaderD3D12::BeginBindDescriptors()
@@ -532,7 +542,7 @@ CD3DX12_GPU_DESCRIPTOR_HANDLE ShaderD3D12::UpdateUniformBuffer(int setNumber, in
 		dsd._index++;
 	}
 
-	auto hpBase = pRendererD3D12->GetHeapCbvSrvUav().GetNextHandlePair();
+	auto hpBase = pRendererD3D12->GetViewHeap().GetNextHandlePair();
 
 	// Copy CBV:
 	pRendererD3D12->GetD3DDevice()->CopyDescriptorsSimple(1,
@@ -545,8 +555,8 @@ CD3DX12_GPU_DESCRIPTOR_HANDLE ShaderD3D12::UpdateUniformBuffer(int setNumber, in
 		const auto& complexSet = _vComplexSets[complexSetHandle];
 		const int count = Utils::Cast32(complexSet._vTextures.size());
 		pRendererD3D12->GetD3DDevice()->CopyDescriptorsSimple(count,
-			pRendererD3D12->GetHeapCbvSrvUav().GetNextHandlePair(count)._hCPU,
-			complexSet._dhSrvUav.AtCPU(0),
+			pRendererD3D12->GetViewHeap().GetNextHandlePair(count)._hCPU,
+			complexSet._dhViews.AtCPU(0),
 			D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV);
 	}
 
@@ -562,15 +572,24 @@ CD3DX12_GPU_DESCRIPTOR_HANDLE ShaderD3D12::UpdateSamplers(int setNumber, int com
 	if (dsd._staticSamplersOnly)
 		return CD3DX12_GPU_DESCRIPTOR_HANDLE();
 
+	CD3DX12_GPU_DESCRIPTOR_HANDLE hOffset = CD3DX12_GPU_DESCRIPTOR_HANDLE();
 	const auto& complexSet = _vComplexSets[complexSetHandle];
-	const int count = Utils::Cast32(complexSet._vTextures.size());
-	auto hpSampler = pRendererD3D12->GetHeapSampler().GetNextHandlePair(count);
-	pRendererD3D12->GetD3DDevice()->CopyDescriptorsSimple(count,
+	const int maxCount = Utils::Cast32(complexSet._vTextures.size());
+	VERUS_FOR(i, maxCount)
+	{
+		if (Sampler::custom == dsd._vSamplers[i])
+		{
+			auto hpSampler = pRendererD3D12->GetSamplerHeap().GetNextHandlePair(1);
+			if (!hOffset.ptr)
+				hOffset = hpSampler._hGPU;
+			pRendererD3D12->GetD3DDevice()->CopyDescriptorsSimple(1,
 				hpSampler._hCPU,
-		complexSet._dhSamplers.AtCPU(0),
+				complexSet._dhSamplers.AtCPU(i),
 				D3D12_DESCRIPTOR_HEAP_TYPE_SAMPLER);
+		}
+	}
 
-	return hpSampler._hGPU;
+	return hOffset;
 }
 
 void ShaderD3D12::OnError(CSZ s)
@@ -587,3 +606,112 @@ void ShaderD3D12::OnError(CSZ s)
 	else
 		renderer.OnShaderWarning(s);
 }
+
+void ShaderD3D12::UpdateDebugInfo(
+	const Vector<CD3DX12_ROOT_PARAMETER1>& vRootParams,
+	const Vector<D3D12_STATIC_SAMPLER_DESC>& vStaticSamplers,
+	D3D12_ROOT_SIGNATURE_FLAGS rootSignatureFlags)
+{
+	StringStream ss;
+
+	auto PrintShaderVisibility = [&ss](D3D12_SHADER_VISIBILITY shaderVisibility)
+	{
+		switch (shaderVisibility)
+		{
+		case D3D12_SHADER_VISIBILITY_ALL: ss << "All"; break;
+		case D3D12_SHADER_VISIBILITY_VERTEX: ss << "V"; break;
+		case D3D12_SHADER_VISIBILITY_HULL: ss << "H"; break;
+		case D3D12_SHADER_VISIBILITY_DOMAIN: ss << "D"; break;
+		case D3D12_SHADER_VISIBILITY_GEOMETRY: ss << "G"; break;
+		case D3D12_SHADER_VISIBILITY_PIXEL: ss << "P"; break;
+		}
+	};
+
+	int totalSize = 0;
+	int index = 0;
+	for (const auto& rootParam : vRootParams)
+	{
+		ss << "Index=" << index;
+		ss << ", Type=";
+		switch (rootParam.ParameterType)
+		{
+		case D3D12_ROOT_PARAMETER_TYPE_DESCRIPTOR_TABLE: ss << "Table"; totalSize += 4; break;
+		case D3D12_ROOT_PARAMETER_TYPE_32BIT_CONSTANTS: ss << "Const"; totalSize += 4 * rootParam.Constants.Num32BitValues; break;
+		case D3D12_ROOT_PARAMETER_TYPE_CBV: ss << "CBV"; totalSize += 8; break;
+		case D3D12_ROOT_PARAMETER_TYPE_SRV: ss << "SRV"; totalSize += 8; break;
+		case D3D12_ROOT_PARAMETER_TYPE_UAV: ss << "UAV"; totalSize += 8; break;
+		}
+		ss << ", Visibility=";
+		PrintShaderVisibility(rootParam.ShaderVisibility);
+		if (totalSize > 64)
+			ss << " (spilled)";
+		ss << std::endl;
+		switch (rootParam.ParameterType)
+		{
+		case D3D12_ROOT_PARAMETER_TYPE_DESCRIPTOR_TABLE:
+		{
+			VERUS_FOR(i, rootParam.DescriptorTable.NumDescriptorRanges)
+			{
+				auto& descRange = rootParam.DescriptorTable.pDescriptorRanges[i];
+				ss << "    Type=";
+				switch (descRange.RangeType)
+				{
+				case D3D12_DESCRIPTOR_RANGE_TYPE_SRV: ss << "SRV"; break;
+				case D3D12_DESCRIPTOR_RANGE_TYPE_UAV: ss << "UAV"; break;
+				case D3D12_DESCRIPTOR_RANGE_TYPE_CBV: ss << "CBV"; break;
+				case D3D12_DESCRIPTOR_RANGE_TYPE_SAMPLER: ss << "Sampler"; break;
+				}
+				ss << ", Count=" << descRange.NumDescriptors;
+				ss << ", Register=" << descRange.BaseShaderRegister;
+				ss << ", Space=" << descRange.RegisterSpace;
+				ss << ", Flags=" << descRange.Flags;
+				ss << ", Offset=" << static_cast<int>(descRange.OffsetInDescriptorsFromTableStart);
+				ss << std::endl;
+			}
+		}
+		break;
+		case D3D12_ROOT_PARAMETER_TYPE_32BIT_CONSTANTS:
+		{
+			ss << "    ";
+			ss << "Register=" << rootParam.Constants.ShaderRegister;
+			ss << ", Space=" << rootParam.Constants.RegisterSpace;
+			ss << ", Count=" << rootParam.Constants.Num32BitValues << "x4 bytes";
+			ss << std::endl;
+		}
+		break;
+		default:
+		{
+		}
+		break;
+		}
+		index++;
+	}
+	for (const auto& sampler : vStaticSamplers)
+	{
+		ss << "StaticSampler Register=" << sampler.ShaderRegister;
+		ss << ", Space=" << sampler.RegisterSpace;
+		ss << ", Visibility=";
+		PrintShaderVisibility(sampler.ShaderVisibility);
+		ss << std::endl;
+	}
+	ss << "TotalSize=" << totalSize << " bytes (256 max)";
+	ss << ", Flags=";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_ALLOW_INPUT_ASSEMBLER_INPUT_LAYOUT)
+		ss << "[AIAIL]";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_DENY_VERTEX_SHADER_ROOT_ACCESS)
+		ss << "[DVSRA]";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_DENY_HULL_SHADER_ROOT_ACCESS)
+		ss << "[DHSRA]";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_DENY_DOMAIN_SHADER_ROOT_ACCESS)
+		ss << "[DDSRA]";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_DENY_GEOMETRY_SHADER_ROOT_ACCESS)
+		ss << "[DGSRA]";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_DENY_PIXEL_SHADER_ROOT_ACCESS)
+		ss << "[DPSRA]";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_ALLOW_STREAM_OUTPUT)
+		ss << "[ASO]";
+	if (rootSignatureFlags & D3D12_ROOT_SIGNATURE_FLAG_LOCAL_ROOT_SIGNATURE)
+		ss << "[LRS]";
+
+	_debugInfo = ss.str();
+}

RendererDirect3D12/src/CGI/ShaderD3D12.h

@@ -19,11 +19,16 @@ namespace verus
 
 		class ShaderD3D12 : public BaseShader
 		{
+		public:
 			struct Compiled
 			{
 				ComPtr<ID3DBlob> _pBlobs[+Stage::count];
+				String           _entry;
 				int              _stageCount = 0;
 			};
+			VERUS_TYPEDEFS(Compiled);
+
+		private:
 			typedef Map<String, Compiled> TMapCompiled;
 
 			struct DescriptorSetDesc
@@ -48,7 +53,7 @@ namespace verus
 			struct ComplexSet
 			{
 				Vector<TexturePtr> _vTextures;
-				DescriptorHeap     _dhSrvUav;
+				DescriptorHeap     _dhViews;
 				DescriptorHeap     _dhSamplers;
 			};
 			VERUS_TYPEDEFS(ComplexSet);
@@ -57,6 +62,7 @@ namespace verus
 			Vector<DescriptorSetDesc>   _vDescriptorSetDesc;
 			Vector<ComplexSet>          _vComplexSets;
 			ComPtr<ID3D12RootSignature> _pRootSignature;
+			String                      _debugInfo;
 			UINT64                      _currentFrame = UINT64_MAX;
 			bool                        _compute = false;
 
@@ -69,8 +75,8 @@ namespace verus
 
 			virtual void CreateDescriptorSet(int setNumber, const void* pSrc, int size, int capacity, std::initializer_list<Sampler> il, ShaderStageFlags stageFlags) override;
 			virtual void CreatePipelineLayout() override;
-			virtual int BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips) override;
-			virtual void FreeDescriptorSet(int& complexSetHandle) override;
+			virtual CSHandle BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips) override;
+			virtual void FreeDescriptorSet(CSHandle& complexSetHandle) override;
 
 			virtual void BeginBindDescriptors() override;
 			virtual void EndBindDescriptors() override;
@@ -79,8 +85,7 @@ namespace verus
 			// D3D12
 			//
 
-			ID3DBlob* GetD3DBlob(CSZ branch, Stage stage) const { return _mapCompiled.at(branch)._pBlobs[+stage].Get(); }
-			int GetStageCount(CSZ branch) const { return _mapCompiled.at(branch)._stageCount; }
+			RcCompiled GetCompiled(CSZ branch) const { return _mapCompiled.at(branch); }
 
 			ID3D12RootSignature* GetD3DRootSignature() const { return _pRootSignature.Get(); }
 
@@ -92,6 +97,11 @@ namespace verus
 			bool IsCompute() const { return _compute; }
 
 			void OnError(CSZ s);
+
+			void UpdateDebugInfo(
+				const Vector<CD3DX12_ROOT_PARAMETER1>& vRootParams,
+				const Vector<D3D12_STATIC_SAMPLER_DESC>& vStaticSamplers,
+				D3D12_ROOT_SIGNATURE_FLAGS rootSignatureFlags);
 		};
 		VERUS_TYPEDEFS(ShaderD3D12);
 	}

RendererDirect3D12/src/CGI/TextureD3D12.cpp

@@ -29,7 +29,7 @@ void TextureD3D12::Init(RcTextureDesc desc)
 	_bytesPerPixel = FormatToBytesPerPixel(desc._format);
 	const bool renderTarget = (_desc._flags & TextureDesc::Flags::colorAttachment);
 	const bool depthFormat = IsDepthFormat(desc._format);
-	const bool depthSampled = _desc._flags & TextureDesc::Flags::depthSampled;
+	const bool depthSampled = _desc._flags & (TextureDesc::Flags::depthSampledR | TextureDesc::Flags::depthSampledW);
 	if (_desc._flags & TextureDesc::Flags::anyShaderResource)
 		_mainLayout = ImageLayout::xsReadOnly;
 
@@ -50,17 +50,13 @@ void TextureD3D12::Init(RcTextureDesc desc)
 	D3D12_RESOURCE_STATES initialResourceState = ToNativeImageLayout(_mainLayout);
 	D3D12_CLEAR_VALUE clearValue = {};
 	clearValue.Format = ToNativeFormat(_desc._format, false);
-	if (renderTarget)
-	{
-		initialResourceState = D3D12_RESOURCE_STATE_PIXEL_SHADER_RESOURCE;
 	memcpy(clearValue.Color, _desc._clearValue.ToPointer(), sizeof(clearValue.Color));
-	}
 	if (depthFormat)
 	{
 		initialResourceState = D3D12_RESOURCE_STATE_DEPTH_WRITE;
 		clearValue.DepthStencil.Depth = _desc._clearValue.getX();
 		clearValue.DepthStencil.Stencil = static_cast<UINT8>(_desc._clearValue.getY());
-		if (depthSampled)
+		if (_desc._flags & TextureDesc::Flags::depthSampledR)
 			initialResourceState = D3D12_RESOURCE_STATE_DEPTH_READ | D3D12_RESOURCE_STATE_PIXEL_SHADER_RESOURCE;
 	}
 	D3D12MA::ALLOCATION_DESC allocDesc = {};
@@ -83,6 +79,12 @@ void TextureD3D12::Init(RcTextureDesc desc)
 		resDescUAV.Height = Math::Max(1, _desc._height >> 1);
 		resDescUAV.MipLevels = uavMipLevels;
 		resDescUAV.Flags |= D3D12_RESOURCE_FLAG_ALLOW_UNORDERED_ACCESS;
+		// sRGB cannot be used with UAV:
+		switch (resDescUAV.Format)
+		{
+		case DXGI_FORMAT_R8G8B8A8_UNORM_SRGB: resDescUAV.Format = DXGI_FORMAT_R8G8B8A8_UNORM; break;
+		case DXGI_FORMAT_B8G8R8A8_UNORM_SRGB: resDescUAV.Format = DXGI_FORMAT_B8G8R8A8_UNORM; break;
+		}
 
 		D3D12MA::ALLOCATION_DESC allocDesc = {};
 		allocDesc.HeapType = D3D12_HEAP_TYPE_DEFAULT;
@@ -91,18 +93,18 @@ void TextureD3D12::Init(RcTextureDesc desc)
 			&resDescUAV,
 			D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE,
 			nullptr,
-			&_uavResource._pMaAllocation,
-			IID_PPV_ARGS(&_uavResource._pResource))))
+			&_uaResource._pMaAllocation,
+			IID_PPV_ARGS(&_uaResource._pResource))))
 			throw VERUS_RUNTIME_ERROR << "CreateResource(D3D12_HEAP_TYPE_DEFAULT), hr=" << VERUS_HR(hr);
 
 		_dhUAV.Create(pRendererD3D12->GetD3DDevice(), D3D12_DESCRIPTOR_HEAP_TYPE_CBV_SRV_UAV, uavMipLevels);
 		VERUS_FOR(i, uavMipLevels)
 		{
 			D3D12_UNORDERED_ACCESS_VIEW_DESC uavDesc = {};
-			uavDesc.Format = resDesc.Format;
+			uavDesc.Format = resDescUAV.Format;
 			uavDesc.ViewDimension = D3D12_UAV_DIMENSION_TEXTURE2D;
 			uavDesc.Texture2D.MipSlice = i;
-			pRendererD3D12->GetD3DDevice()->CreateUnorderedAccessView(_uavResource._pResource.Get(), nullptr, &uavDesc, _dhUAV.AtCPU(i));
+			pRendererD3D12->GetD3DDevice()->CreateUnorderedAccessView(_uaResource._pResource.Get(), nullptr, &uavDesc, _dhUAV.AtCPU(i));
 		}
 	}
 
@@ -152,9 +154,9 @@ void TextureD3D12::Done()
 	_dhDSV.Reset();
 	_dhRTV.Reset();
 	_dhSRV.Reset();
-	VERUS_SMART_RELEASE(_uavResource._pMaAllocation);
-	VERUS_COM_RELEASE_CHECK(_uavResource._pResource.Get());
-	_uavResource._pResource.Reset();
+	VERUS_SMART_RELEASE(_uaResource._pMaAllocation);
+	VERUS_COM_RELEASE_CHECK(_uaResource._pResource.Get());
+	_uaResource._pResource.Reset();
 	VERUS_SMART_RELEASE(_resource._pMaAllocation);
 	VERUS_COM_RELEASE_CHECK(_resource._pResource.Get());
 	_resource._pResource.Reset();
@@ -258,7 +260,7 @@ void TextureD3D12::GenerateMips(PBaseCommandBuffer pCB)
 		ub._srcMipLevel = srcMip;
 		ub._mipLevelCount = dispatchMipCount;
 		ub._srcDimensionCase = (srcHeight & 1) << 1 | (srcWidth & 1);
-		ub._srgb = false;
+		ub._srgb = IsSRGBFormat(_desc._format);
 		ub._texelSize.x = 1.f / dstWidth;
 		ub._texelSize.y = 1.f / dstHeight;
 
@@ -267,7 +269,7 @@ void TextureD3D12::GenerateMips(PBaseCommandBuffer pCB)
 			int mips[5] = {}; // For input texture (always mip 0) and 4 UAV mips.
 			VERUS_FOR(mip, dispatchMipCount)
 				mips[mip + 1] = srcMip + mip;
-			const int complexSetHandle = shader->BindDescriptorSetTextures(0, { tex, tex, tex, tex, tex }, mips);
+			const CSHandle complexSetHandle = shader->BindDescriptorSetTextures(0, { tex, tex, tex, tex, tex }, mips);
 			_vCshGenerateMips.push_back(complexSetHandle);
 			pCB->BindDescriptors(shader, 0, complexSetHandle);
 		}
@@ -278,7 +280,7 @@ void TextureD3D12::GenerateMips(PBaseCommandBuffer pCB)
 
 		pCB->Dispatch(Math::DivideByMultiple(dstWidth, 8), Math::DivideByMultiple(dstHeight, 8));
 
-		auto rb = CD3DX12_RESOURCE_BARRIER::UAV(_uavResource._pResource.Get());
+		auto rb = CD3DX12_RESOURCE_BARRIER::UAV(_uaResource._pResource.Get());
 		pCmdList->ResourceBarrier(1, &rb);
 
 		// Transition state for upcoming CopyTextureRegion():
@@ -288,7 +290,8 @@ void TextureD3D12::GenerateMips(PBaseCommandBuffer pCB)
 			VERUS_FOR(mip, dispatchMipCount)
 			{
 				const int subUAV = srcMip + mip;
-				barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(_uavResource._pResource.Get(), D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE, D3D12_RESOURCE_STATE_COPY_SOURCE, subUAV);
+				barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(_uaResource._pResource.Get(),
+					D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE, D3D12_RESOURCE_STATE_COPY_SOURCE, subUAV);
 			}
 			pCmdList->ResourceBarrier(barrierCount, barriers);
 
@@ -301,7 +304,7 @@ void TextureD3D12::GenerateMips(PBaseCommandBuffer pCB)
 			pCmdList->CopyTextureRegion(
 				&CD3DX12_TEXTURE_COPY_LOCATION(_resource._pResource.Get(), subSRV),
 				0, 0, 0,
-				&CD3DX12_TEXTURE_COPY_LOCATION(_uavResource._pResource.Get(), subUAV),
+				&CD3DX12_TEXTURE_COPY_LOCATION(_uaResource._pResource.Get(), subUAV),
 				nullptr);
 		}
 		// Transition state for next Dispatch():
@@ -311,7 +314,8 @@ void TextureD3D12::GenerateMips(PBaseCommandBuffer pCB)
 			VERUS_FOR(mip, dispatchMipCount)
 			{
 				const int subUAV = srcMip + mip;
-				barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(_uavResource._pResource.Get(), D3D12_RESOURCE_STATE_COPY_SOURCE, D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE, subUAV);
+				barriers[barrierCount++] = CD3DX12_RESOURCE_BARRIER::Transition(_uaResource._pResource.Get(),
+					D3D12_RESOURCE_STATE_COPY_SOURCE, D3D12_RESOURCE_STATE_NON_PIXEL_SHADER_RESOURCE, subUAV);
 			}
 			pCmdList->ResourceBarrier(barrierCount, barriers);
 
@@ -348,7 +352,7 @@ Continue TextureD3D12::Scheduled_Update()
 	{
 		VERUS_QREF_RENDERER;
 		auto shader = renderer.GetShaderGenerateMips();
-		for (int csh : _vCshGenerateMips)
+		for (auto& csh : _vCshGenerateMips)
 			shader->FreeDescriptorSet(csh);
 		_vCshGenerateMips.clear();
 	}

RendererDirect3D12/src/CGI/TextureD3D12.h

@@ -13,9 +13,9 @@ namespace verus
 			};
 
 			ResourceEx         _resource;
-			ResourceEx         _uavResource;
+			ResourceEx         _uaResource;
 			Vector<ResourceEx> _vStagingBuffers;
-			Vector<int>        _vCshGenerateMips;
+			Vector<CSHandle>   _vCshGenerateMips;
 			DescriptorHeap     _dhSRV;
 			DescriptorHeap     _dhUAV;
 			DescriptorHeap     _dhRTV;

RendererVulkan/src/CGI/CommandBufferVulkan.cpp

@@ -44,7 +44,7 @@ void CommandBufferVulkan::End()
 		throw VERUS_RUNTIME_ERROR << "vkEndCommandBuffer(), res=" << res;
 }
 
-void CommandBufferVulkan::BeginRenderPass(int renderPassHandle, int framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor)
+void CommandBufferVulkan::BeginRenderPass(RPHandle renderPassHandle, FBHandle framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor)
 {
 	VERUS_QREF_RENDERER_VULKAN;
 	VERUS_QREF_CONST_SETTINGS;
@@ -120,6 +120,13 @@ void CommandBufferVulkan::BindPipeline(PipelinePtr pipe)
 
 void CommandBufferVulkan::SetViewport(std::initializer_list<Vector4> il, float minDepth, float maxDepth)
 {
+	if (il.size() > 0)
+	{
+		const float w = il.begin()->Width();
+		const float h = il.begin()->Height();
+		_viewportSize = Vector4(w, h, 1 / w, 1 / h);
+	}
+
 	VkViewport vpVulkan[VERUS_MAX_RT];
 	int count = 0;
 	for (const auto& rc : il)
@@ -155,7 +162,7 @@ void CommandBufferVulkan::SetBlendConstants(const float* p)
 	vkCmdSetBlendConstants(GetVkCommandBuffer(), p);
 }
 
-bool CommandBufferVulkan::BindDescriptors(ShaderPtr shader, int setNumber, int complexSetHandle)
+bool CommandBufferVulkan::BindDescriptors(ShaderPtr shader, int setNumber, CSHandle complexSetHandle)
 {
 	if (setNumber < 0)
 		return true;
@@ -167,7 +174,7 @@ bool CommandBufferVulkan::BindDescriptors(ShaderPtr shader, int setNumber, int c
 	if (offset < 0)
 		return false;
 
-	const VkDescriptorSet descriptorSet = (complexSetHandle >= 0) ?
+	const VkDescriptorSet descriptorSet = complexSetHandle.IsSet() ?
 		shaderVulkan.GetComplexVkDescriptorSet(complexSetHandle) : shaderVulkan.GetVkDescriptorSet(setNumber);
 	const uint32_t dynamicOffset = offset;
 	const uint32_t dynamicOffsetCount = 1;

RendererVulkan/src/CGI/CommandBufferVulkan.h

@@ -19,7 +19,7 @@ namespace verus
 			virtual void Begin() override;
 			virtual void End() override;
 
-			virtual void BeginRenderPass(int renderPassHandle, int framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor) override;
+			virtual void BeginRenderPass(RPHandle renderPassHandle, FBHandle framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor) override;
 			virtual void NextSubpass() override;
 			virtual void EndRenderPass() override;
 
@@ -31,7 +31,7 @@ namespace verus
 			virtual void SetScissor(std::initializer_list<Vector4> il) override;
 			virtual void SetBlendConstants(const float* p) override;
 
-			virtual bool BindDescriptors(ShaderPtr shader, int setNumber, int complexSetHandle) override;
+			virtual bool BindDescriptors(ShaderPtr shader, int setNumber, CSHandle complexSetHandle) override;
 			virtual void PushConstants(ShaderPtr shader, int offset, int size, const void* p, ShaderStageFlags stageFlags) override;
 
 			virtual void PipelineImageMemoryBarrier(TexturePtr tex, ImageLayout oldLayout, ImageLayout newLayout, Range<int> mipLevels, int arrayLayer) override;

RendererVulkan/src/CGI/GeometryVulkan.cpp

@@ -19,13 +19,13 @@ void GeometryVulkan::Init(RcGeometryDesc desc)
 	_dynBindingsMask = desc._dynBindingsMask;
 	_32BitIndices = desc._32BitIndices;
 
-	const int bindingCount = GetBindingCount(desc._pInputElementDesc);
+	const int bindingCount = GetBindingCount(desc._pVertexInputAttrDesc);
 	_vVertexInputBindingDesc.reserve(bindingCount);
-	_vVertexInputAttributeDesc.reserve(GetInputElementDescCount(desc._pInputElementDesc));
+	_vVertexInputAttributeDesc.reserve(GetVertexInputAttrDescCount(desc._pVertexInputAttrDesc));
 	int i = 0;
-	while (desc._pInputElementDesc[i]._offset >= 0)
+	while (desc._pVertexInputAttrDesc[i]._offset >= 0)
 	{
-		int binding = desc._pInputElementDesc[i]._binding;
+		int binding = desc._pVertexInputAttrDesc[i]._binding;
 		VkVertexInputRate inputRate = VK_VERTEX_INPUT_RATE_VERTEX;
 		if (binding < 0)
 		{
@@ -53,10 +53,10 @@ void GeometryVulkan::Init(RcGeometryDesc desc)
 		}
 
 		VkVertexInputAttributeDescription vkviad = {};
-		vkviad.location = ToNativeLocation(desc._pInputElementDesc[i]._usage, desc._pInputElementDesc[i]._usageIndex);
+		vkviad.location = ToNativeLocation(desc._pVertexInputAttrDesc[i]._usage, desc._pVertexInputAttrDesc[i]._usageIndex);
 		vkviad.binding = binding;
-		vkviad.format = ToNativeFormat(desc._pInputElementDesc[i]._usage, desc._pInputElementDesc[i]._type, desc._pInputElementDesc[i]._components);
-		vkviad.offset = desc._pInputElementDesc[i]._offset;
+		vkviad.format = ToNativeFormat(desc._pVertexInputAttrDesc[i]._usage, desc._pVertexInputAttrDesc[i]._type, desc._pVertexInputAttrDesc[i]._components);
+		vkviad.offset = desc._pVertexInputAttrDesc[i]._offset;
 		_vVertexInputAttributeDesc.push_back(vkviad);
 		i++;
 	}

RendererVulkan/src/CGI/Native.cpp

@@ -90,6 +90,8 @@ VkPrimitiveTopology CGI::ToNativePrimitiveTopology(PrimitiveTopology primitiveTo
 	case PrimitiveTopology::lineStrip:     return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP;
 	case PrimitiveTopology::triangleList:  return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
 	case PrimitiveTopology::triangleStrip: return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP;
+	case PrimitiveTopology::patchList3:    return VK_PRIMITIVE_TOPOLOGY_PATCH_LIST;
+	case PrimitiveTopology::patchList4:    return VK_PRIMITIVE_TOPOLOGY_PATCH_LIST;
 	default: throw VERUS_RECOVERABLE << "ToNativePrimitiveTopology()";
 	}
 }
@@ -148,14 +150,16 @@ int CGI::ToNativeLocation(IeUsage usage, int usageIndex)
 	switch (usage)
 	{
 	case IeUsage::position: return 0;
-	case IeUsage::blendWeight: return 1;
+	case IeUsage::blendWeights: return 1;
 	case IeUsage::blendIndices: return 6;
 	case IeUsage::normal: return 2;
-	case IeUsage::psize: return 7;
-	case IeUsage::texCoord: return 8 + usageIndex;
 	case IeUsage::tangent: return 14;
 	case IeUsage::binormal: return 15;
 	case IeUsage::color: return 3 + usageIndex;
+	case IeUsage::psize: return 7;
+	case IeUsage::texCoord: return 8 + usageIndex;
+	case IeUsage::instData: return 16 + usageIndex;
+	case IeUsage::attr: return usageIndex;
 	default: throw VERUS_RECOVERABLE << "ToNativeLocation()";
 	}
 }

RendererVulkan/src/CGI/PipelineVulkan.cpp

@@ -38,27 +38,31 @@ void PipelineVulkan::Init(RcPipelineDesc desc)
 		attachmentCount++;
 	}
 
+	const bool tess = desc._topology >= PrimitiveTopology::patchList3 && desc._topology <= PrimitiveTopology::patchList4;
+
+	ShaderVulkan::RcCompiled compiled = shader.GetCompiled(desc._shaderBranch);
+	Vector<String> entryNames(+BaseShader::Stage::count);
 	Vector<VkPipelineShaderStageCreateInfo> vShaderStages;
-	vShaderStages.reserve(shader.GetStageCount(desc._shaderBranch));
-	auto PushShaderStage = [&vShaderStages, &shader, &desc](CSZ name, BaseShader::Stage stage, VkShaderStageFlagBits shaderStageFlagBits)
+	vShaderStages.reserve(compiled._stageCount);
+	auto PushShaderStage = [&vShaderStages, &compiled, &entryNames](CSZ suffix, BaseShader::Stage stage, VkShaderStageFlagBits shaderStageFlagBits)
 	{
-		const VkShaderModule shaderModule = shader.GetVkShaderModule(desc._shaderBranch, stage);
+		const VkShaderModule shaderModule = compiled._shaderModules[+stage];
+		entryNames[+stage] = compiled._entry + suffix;
 		if (shaderModule != VK_NULL_HANDLE)
 		{
 			VkPipelineShaderStageCreateInfo vkpssci = {};
 			vkpssci.sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO;
 			vkpssci.stage = shaderStageFlagBits;
 			vkpssci.module = shaderModule;
-			vkpssci.pName = name;
+			vkpssci.pName = _C(entryNames[+stage]);
 			vShaderStages.push_back(vkpssci);
 		}
 	};
-
-	PushShaderStage("mainVS", BaseShader::Stage::vs, VK_SHADER_STAGE_VERTEX_BIT);
-	PushShaderStage("mainHS", BaseShader::Stage::hs, VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT);
-	PushShaderStage("mainDS", BaseShader::Stage::ds, VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT);
-	PushShaderStage("mainGS", BaseShader::Stage::gs, VK_SHADER_STAGE_GEOMETRY_BIT);
-	PushShaderStage("mainFS", BaseShader::Stage::fs, VK_SHADER_STAGE_FRAGMENT_BIT);
+	PushShaderStage("VS", BaseShader::Stage::vs, VK_SHADER_STAGE_VERTEX_BIT);
+	PushShaderStage("HS", BaseShader::Stage::hs, VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT);
+	PushShaderStage("DS", BaseShader::Stage::ds, VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT);
+	PushShaderStage("GS", BaseShader::Stage::gs, VK_SHADER_STAGE_GEOMETRY_BIT);
+	PushShaderStage("FS", BaseShader::Stage::fs, VK_SHADER_STAGE_FRAGMENT_BIT);
 
 	Vector<VkVertexInputBindingDescription> vVertexInputBindingDesc;
 	Vector<VkVertexInputAttributeDescription> vVertexInputAttributeDesc;
@@ -70,6 +74,14 @@ void PipelineVulkan::Init(RcPipelineDesc desc)
 	inputAssemblyState.topology = ToNativePrimitiveTopology(desc._topology);
 	inputAssemblyState.primitiveRestartEnable = desc._primitiveRestartEnable;
 
+	VkPipelineTessellationStateCreateInfo tessellationState = {};
+	tessellationState.sType = VK_STRUCTURE_TYPE_PIPELINE_TESSELLATION_STATE_CREATE_INFO;
+	tessellationState.patchControlPoints = 3;
+	switch (desc._topology)
+	{
+	case PrimitiveTopology::patchList4: tessellationState.patchControlPoints = 4; break;
+	}
+
 	VkPipelineViewportStateCreateInfo viewportState = {};
 	viewportState.sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO;
 	viewportState.viewportCount = desc._multiViewport;
@@ -77,6 +89,8 @@ void PipelineVulkan::Init(RcPipelineDesc desc)
 
 	VkPipelineRasterizationLineStateCreateInfoEXT rasterizationLineState = {};
 	rasterizationLineState.sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_LINE_STATE_CREATE_INFO_EXT;
+	rasterizationLineState.lineRasterizationMode = VK_LINE_RASTERIZATION_MODE_BRESENHAM_EXT;
+	if (desc._colorAttachBlendEqs[0] == VERUS_COLOR_BLEND_ALPHA)
 		rasterizationLineState.lineRasterizationMode = VK_LINE_RASTERIZATION_MODE_RECTANGULAR_SMOOTH_EXT;
 	VkPipelineRasterizationStateCreateInfo rasterizationState = {};
 	rasterizationState.sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO;
@@ -94,7 +108,6 @@ void PipelineVulkan::Init(RcPipelineDesc desc)
 
 	VkPipelineMultisampleStateCreateInfo multisampleState = {};
 	multisampleState.sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO;
-	multisampleState.sampleShadingEnable = VK_FALSE;
 	multisampleState.rasterizationSamples = ToNativeSampleCount(desc._sampleCount);
 
 	VkPipelineDepthStencilStateCreateInfo depthStencilState = {};
@@ -196,6 +209,7 @@ void PipelineVulkan::Init(RcPipelineDesc desc)
 	vkgpci.pStages = vShaderStages.data();
 	vkgpci.pVertexInputState = &vertexInputState;
 	vkgpci.pInputAssemblyState = &inputAssemblyState;
+	vkgpci.pTessellationState = tess ? &tessellationState : nullptr;
 	vkgpci.pViewportState = &viewportState;
 	vkgpci.pRasterizationState = &rasterizationState;
 	vkgpci.pMultisampleState = &multisampleState;
@@ -225,14 +239,17 @@ void PipelineVulkan::InitCompute(RcPipelineDesc desc)
 	VkResult res = VK_SUCCESS;
 
 	RcShaderVulkan shader = static_cast<RcShaderVulkan>(*desc._shader);
-	const VkShaderModule shaderModule = shader.GetVkShaderModule(desc._shaderBranch, BaseShader::Stage::cs);
+
+	ShaderVulkan::RcCompiled compiled = shader.GetCompiled(desc._shaderBranch);
+	const VkShaderModule shaderModule = compiled._shaderModules[+BaseShader::Stage::cs];
+	String entryName = compiled._entry + "CS";
 
 	VkComputePipelineCreateInfo vkcpci = {};
 	vkcpci.sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO;
 	vkcpci.stage.sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO;
 	vkcpci.stage.stage = VK_SHADER_STAGE_COMPUTE_BIT;
 	vkcpci.stage.module = shaderModule;
-	vkcpci.stage.pName = "mainCS";
+	vkcpci.stage.pName = _C(entryName);
 	vkcpci.layout = shader.GetVkPipelineLayout();
 	if (VK_SUCCESS != (res = vkCreateComputePipelines(pRendererVulkan->GetVkDevice(), VK_NULL_HANDLE, 1, &vkcpci, pRendererVulkan->GetAllocator(), &_pipeline)))
 		throw VERUS_RUNTIME_ERROR << "vkCreateComputePipelines(), res=" << res;

RendererVulkan/src/CGI/RendererVulkan.cpp

@@ -100,8 +100,8 @@ void RendererVulkan::Done()
 		Renderer::I().ImGuiSetCurrentContext(nullptr);
 	}
 
-	DeleteFramebuffer(-1);
-	DeleteRenderPass(-1);
+	DeleteFramebuffer(FBHandle::Make(-2));
+	DeleteRenderPass(RPHandle::Make(-2));
 
 	for (auto sampler : _vSamplers)
 		VERUS_VULKAN_DESTROY(sampler, vkDestroySampler(_device, sampler, GetAllocator()));
@@ -189,6 +189,7 @@ VKAPI_ATTR VkBool32 VKAPI_CALL RendererVulkan::DebugUtilsMessengerCallback(
 		severity = D::Log::Severity::warning;
 	if (messageSeverity & VK_DEBUG_UTILS_MESSAGE_SEVERITY_ERROR_BIT_EXT)
 		severity = D::Log::Severity::error;
+	if (!strstr(pCallbackData->pMessage, "ptr to input arr[1] of struct of"))
 		D::Log::I().Write(pCallbackData->pMessage, std::this_thread::get_id(), __FILE__, __LINE__, severity);
 	return VK_FALSE;
 }
@@ -394,10 +395,12 @@ void RendererVulkan::PickPhysicalDevice()
 
 void RendererVulkan::CreateDevice()
 {
+	VERUS_QREF_CONST_SETTINGS;
 	VkResult res = VK_SUCCESS;
 
 	VkPhysicalDeviceLineRasterizationFeaturesEXT lineRasterizationFeatures = {};
 	lineRasterizationFeatures.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_LINE_RASTERIZATION_FEATURES_EXT;
+	lineRasterizationFeatures.bresenhamLines = VK_TRUE;
 	lineRasterizationFeatures.smoothLines = VK_TRUE;
 
 	_queueFamilyIndices = FindQueueFamilyIndices(_physicalDevice);
@@ -416,6 +419,8 @@ void RendererVulkan::CreateDevice()
 		vDeviceQueueCreateInfos.push_back(vkdqci);
 	}
 	VkPhysicalDeviceFeatures physicalDeviceFeatures = {};
+	physicalDeviceFeatures.geometryShader = VK_TRUE;
+	physicalDeviceFeatures.tessellationShader = settings._gpuTessellation ? VK_TRUE : VK_FALSE;
 	physicalDeviceFeatures.fillModeNonSolid = VK_TRUE;
 	physicalDeviceFeatures.multiViewport = VK_TRUE;
 	physicalDeviceFeatures.samplerAnisotropy = VK_TRUE;
@@ -624,46 +629,46 @@ void RendererVulkan::CreateSamplers()
 
 	// <Repeat>
 	vksci = init;
-	_vSamplers[+Sampler::linear3D] = Create(vksci);
+	_vSamplers[+Sampler::linearMipL] = Create(vksci);
 
 	vksci = init;
 	vksci.magFilter = VK_FILTER_NEAREST;
 	vksci.minFilter = VK_FILTER_NEAREST;
-	_vSamplers[+Sampler::nearest3D] = Create(vksci);
+	_vSamplers[+Sampler::nearestMipL] = Create(vksci);
 
 	vksci = init;
 	vksci.mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST;
-	_vSamplers[+Sampler::linear2D] = Create(vksci);
+	_vSamplers[+Sampler::linearMipN] = Create(vksci);
 
 	vksci = init;
 	vksci.magFilter = VK_FILTER_NEAREST;
 	vksci.minFilter = VK_FILTER_NEAREST;
 	vksci.mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST;
-	_vSamplers[+Sampler::nearest2D] = Create(vksci);
+	_vSamplers[+Sampler::nearestMipN] = Create(vksci);
 	// </Repeat>
 
 	// <Clamp>
 	vksci = init;
 	vksci.addressModeU = vksci.addressModeV = vksci.addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE;
-	_vSamplers[+Sampler::linearClamp3D] = Create(vksci);
+	_vSamplers[+Sampler::linearClampMipL] = Create(vksci);
 
 	vksci = init;
 	vksci.magFilter = VK_FILTER_NEAREST;
 	vksci.minFilter = VK_FILTER_NEAREST;
 	vksci.addressModeU = vksci.addressModeV = vksci.addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE;
-	_vSamplers[+Sampler::nearestClamp3D] = Create(vksci);
+	_vSamplers[+Sampler::nearestClampMipL] = Create(vksci);
 
 	vksci = init;
 	vksci.mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST;
 	vksci.addressModeU = vksci.addressModeV = vksci.addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE;
-	_vSamplers[+Sampler::linearClamp2D] = Create(vksci);
+	_vSamplers[+Sampler::linearClampMipN] = Create(vksci);
 
 	vksci = init;
 	vksci.magFilter = VK_FILTER_NEAREST;
 	vksci.minFilter = VK_FILTER_NEAREST;
 	vksci.mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST;
 	vksci.addressModeU = vksci.addressModeV = vksci.addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE;
-	_vSamplers[+Sampler::nearestClamp2D] = Create(vksci);
+	_vSamplers[+Sampler::nearestClampMipN] = Create(vksci);
 	// </Clamp>
 }
 
@@ -692,7 +697,7 @@ void RendererVulkan::ImGuiCheckVkResultFn(VkResult res)
 		throw VERUS_RUNTIME_ERROR << "ImGuiCheckVkResultFn(), res=" << res;
 }
 
-void RendererVulkan::ImGuiInit(int renderPassHandle)
+void RendererVulkan::ImGuiInit(RPHandle renderPassHandle)
 {
 	VkResult res = VK_SUCCESS;
 	VkDescriptorPoolSize vkdps = { VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, 1 };
@@ -720,7 +725,7 @@ void RendererVulkan::ImGuiInit(int renderPassHandle)
 		IO::FileSystem::LoadResource(_C(settings._imguiFont), vData);
 		void* pFontData = IM_ALLOC(vData.size());
 		memcpy(pFontData, vData.data(), vData.size());
-		io.Fonts->AddFontFromMemoryTTF(pFontData, Utils::Cast32(vData.size()), settings.GetFontSize(), nullptr, io.Fonts->GetGlyphRangesCyrillic());
+		io.Fonts->AddFontFromMemoryTTF(pFontData, Utils::Cast32(vData.size()), static_cast<float>(settings.GetFontSize()), nullptr, io.Fonts->GetGlyphRangesCyrillic());
 	}
 
 	ImGui::StyleColorsDark();
@@ -738,7 +743,7 @@ void RendererVulkan::ImGuiInit(int renderPassHandle)
 	info.MinImageCount = settings._screenVSync ? 3 : 2;;
 	info.ImageCount = s_ringBufferSize;
 	info.CheckVkResultFn = ImGuiCheckVkResultFn;
-	ImGui_ImplVulkan_Init(&info, _vRenderPasses[renderPassHandle]);
+	ImGui_ImplVulkan_Init(&info, _vRenderPasses[renderPassHandle.Get()]);
 
 	CommandBufferVulkan commandBuffer;
 	commandBuffer.InitOneTimeSubmit();
@@ -912,7 +917,7 @@ void RendererVulkan::DeleteTexture(PBaseTexture p)
 	TStoreTextures::Delete(static_cast<PTextureVulkan>(p));
 }
 
-int RendererVulkan::CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD)
+RPHandle RendererVulkan::CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD)
 {
 	VkResult res = VK_SUCCESS;
 
@@ -1111,16 +1116,16 @@ int RendererVulkan::CreateRenderPass(std::initializer_list<RP::Attachment> ilA,
 	if (VK_SUCCESS != (res = vkCreateRenderPass(_device, &vkrpci, GetAllocator(), &renderPass)))
 		throw VERUS_RUNTIME_ERROR << "vkCreateRenderPass(), res=" << res;
 
-	const int handle = GetNextRenderPassHandle();
-	if (handle >= _vRenderPasses.size())
+	const int nextIndex = GetNextRenderPassIndex();
+	if (nextIndex >= _vRenderPasses.size())
 		_vRenderPasses.push_back(renderPass);
 	else
-		_vRenderPasses[handle] = renderPass;
+		_vRenderPasses[nextIndex] = renderPass;
 
-	return handle;
+	return RPHandle::Make(nextIndex);
 }
 
-int RendererVulkan::CreateFramebuffer(int renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex)
+FBHandle RendererVulkan::CreateFramebuffer(RPHandle renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex)
 {
 	VkResult res = VK_SUCCESS;
 
@@ -1138,7 +1143,7 @@ int RendererVulkan::CreateFramebuffer(int renderPassHandle, std::initializer_lis
 	for (const auto& x : il)
 	{
 		auto& texVulkan = static_cast<RTextureVulkan>(*x);
-		imageViews[count] = texVulkan.GetVkImageView();
+		imageViews[count] = texVulkan.GetVkImageViewForFramebuffer();
 		count++;
 	}
 	vkfci.attachmentCount = count;
@@ -1149,27 +1154,27 @@ int RendererVulkan::CreateFramebuffer(int renderPassHandle, std::initializer_lis
 	if (VK_SUCCESS != (res = vkCreateFramebuffer(_device, &vkfci, GetAllocator(), &framebuffer)))
 		throw VERUS_RUNTIME_ERROR << "vkCreateFramebuffer(), res=" << res;
 
-	const int handle = GetNextFramebufferHandle();
+	const int nextIndex = GetNextFramebufferIndex();
 	Framebuffer fb;
 	fb._framebuffer = framebuffer;
 	fb._width = w;
 	fb._height = h;
-	if (handle >= _vFramebuffers.size())
+	if (nextIndex >= _vFramebuffers.size())
 		_vFramebuffers.push_back(fb);
 	else
-		_vFramebuffers[handle] = fb;
+		_vFramebuffers[nextIndex] = fb;
 
-	return handle;
+	return FBHandle::Make(nextIndex);
 }
 
-void RendererVulkan::DeleteRenderPass(int handle)
+void RendererVulkan::DeleteRenderPass(RPHandle handle)
 {
-	if (handle >= 0)
+	if (handle.IsSet())
 	{
-		vkDestroyRenderPass(_device, _vRenderPasses[handle], GetAllocator());
-		_vRenderPasses[handle] = VK_NULL_HANDLE;
+		vkDestroyRenderPass(_device, _vRenderPasses[handle.Get()], GetAllocator());
+		_vRenderPasses[handle.Get()] = VK_NULL_HANDLE;
 	}
-	else
+	else if (-2 == handle.Get())
 	{
 		for (auto renderPass : _vRenderPasses)
 			vkDestroyRenderPass(_device, renderPass, GetAllocator());
@@ -1177,14 +1182,14 @@ void RendererVulkan::DeleteRenderPass(int handle)
 	}
 }
 
-void RendererVulkan::DeleteFramebuffer(int handle)
+void RendererVulkan::DeleteFramebuffer(FBHandle handle)
 {
-	if (handle >= 0)
+	if (handle.IsSet())
 	{
-		vkDestroyFramebuffer(_device, _vFramebuffers[handle]._framebuffer, GetAllocator());
-		_vFramebuffers[handle]._framebuffer = VK_NULL_HANDLE;
+		vkDestroyFramebuffer(_device, _vFramebuffers[handle.Get()]._framebuffer, GetAllocator());
+		_vFramebuffers[handle.Get()]._framebuffer = VK_NULL_HANDLE;
 	}
-	else
+	else if (-2 == handle.Get())
 	{
 		for (auto framebuffer : _vFramebuffers)
 			vkDestroyFramebuffer(_device, framebuffer._framebuffer, GetAllocator());
@@ -1192,7 +1197,7 @@ void RendererVulkan::DeleteFramebuffer(int handle)
 	}
 }
 
-int RendererVulkan::GetNextRenderPassHandle() const
+int RendererVulkan::GetNextRenderPassIndex() const
 {
 	const int count = Utils::Cast32(_vRenderPasses.size());
 	VERUS_FOR(i, count)
@@ -1203,7 +1208,7 @@ int RendererVulkan::GetNextRenderPassHandle() const
 	return count;
 }
 
-int RendererVulkan::GetNextFramebufferHandle() const
+int RendererVulkan::GetNextFramebufferIndex() const
 {
 	const int count = Utils::Cast32(_vFramebuffers.size());
 	VERUS_FOR(i, count)
@@ -1214,14 +1219,14 @@ int RendererVulkan::GetNextFramebufferHandle() const
 	return count;
 }
 
-VkRenderPass RendererVulkan::GetRenderPass(int handle) const
+VkRenderPass RendererVulkan::GetRenderPass(RPHandle handle) const
 {
-	return _vRenderPasses[handle];
+	return _vRenderPasses[handle.Get()];
 }
 
-RendererVulkan::RcFramebuffer RendererVulkan::GetFramebuffer(int handle) const
+RendererVulkan::RcFramebuffer RendererVulkan::GetFramebuffer(FBHandle handle) const
 {
-	return _vFramebuffers[handle];
+	return _vFramebuffers[handle.Get()];
 }
 
 void RendererVulkan::CreateBuffer(VkDeviceSize size, VkBufferUsageFlags usage, VmaMemoryUsage vmaUsage, VkBuffer& buffer, VmaAllocation& vmaAllocation)

RendererVulkan/src/CGI/RendererVulkan.h

@@ -127,7 +127,7 @@ namespace verus
 			const VkSampler* GetImmutableSampler(Sampler s) const;
 
 			static void ImGuiCheckVkResultFn(VkResult res);
-			virtual void ImGuiInit(int renderPassHandle) override;
+			virtual void ImGuiInit(RPHandle renderPassHandle) override;
 			virtual void ImGuiRenderDrawData() override;
 
 			virtual void ResizeSwapChain() override;
@@ -154,14 +154,14 @@ namespace verus
 			virtual void DeleteShader(PBaseShader p) override;
 			virtual void DeleteTexture(PBaseTexture p) override;
 
-			virtual int CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD) override;
-			virtual int CreateFramebuffer(int renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex) override;
-			virtual void DeleteRenderPass(int handle) override;
-			virtual void DeleteFramebuffer(int handle) override;
-			int GetNextRenderPassHandle() const;
-			int GetNextFramebufferHandle() const;
-			VkRenderPass GetRenderPass(int handle) const;
-			RcFramebuffer GetFramebuffer(int handle) const;
+			virtual RPHandle CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD) override;
+			virtual FBHandle CreateFramebuffer(RPHandle renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex) override;
+			virtual void DeleteRenderPass(RPHandle handle) override;
+			virtual void DeleteFramebuffer(FBHandle handle) override;
+			int GetNextRenderPassIndex() const;
+			int GetNextFramebufferIndex() const;
+			VkRenderPass GetRenderPass(RPHandle handle) const;
+			RcFramebuffer GetFramebuffer(FBHandle handle) const;
 
 			void CreateBuffer(VkDeviceSize size, VkBufferUsageFlags usage, VmaMemoryUsage vmaUsage, VkBuffer& buffer, VmaAllocation& vmaAllocation);
 			void CopyBuffer(VkBuffer srcBuffer, VkBuffer dstBuffer, VkDeviceSize size, PBaseCommandBuffer pCB = nullptr);

RendererVulkan/src/CGI/ShaderVulkan.cpp

@@ -70,12 +70,12 @@ void ShaderVulkan::Init(CSZ source, CSZ sourceName, CSZ* branches)
 
 	while (*branches)
 	{
-		String entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages;
+		String entry, stageEntries[+Stage::count], stages;
 		Vector<String> vMacroName;
 		Vector<String> vMacroValue;
-		const String entry = Parse(*branches, entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages, vMacroName, vMacroValue, "DEF_");
+		const String branch = Parse(*branches, entry, stageEntries, stages, vMacroName, vMacroValue, "DEF_");
 
-		if (IsInIgnoreList(_C(entry)))
+		if (IsInIgnoreList(_C(branch)))
 		{
 			branches++;
 			continue;
@@ -126,12 +126,13 @@ void ShaderVulkan::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		// </System defines>
 
 		Compiled compiled;
+		compiled._entry = entry;
 
 		if (strchr(_C(stages), 'V'))
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex] = "_VS";
-			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(entryVS), "vs", flags, &pCode, &size, &pErrorMsgs))
+			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::vs]), "vs", flags, &pCode, &size, &pErrorMsgs))
 				CheckErrorMsgs(pErrorMsgs);
 			CreateShaderModule(pCode, size, compiled._shaderModules[+Stage::vs]);
 		}
@@ -140,7 +141,7 @@ void ShaderVulkan::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex] = "_HS";
-			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(entryHS), "hs", flags, &pCode, &size, &pErrorMsgs))
+			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::hs]), "hs", flags, &pCode, &size, &pErrorMsgs))
 				CheckErrorMsgs(pErrorMsgs);
 			CreateShaderModule(pCode, size, compiled._shaderModules[+Stage::hs]);
 		}
@@ -149,7 +150,7 @@ void ShaderVulkan::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex] = "_DS";
-			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(entryDS), "ds", flags, &pCode, &size, &pErrorMsgs))
+			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::ds]), "ds", flags, &pCode, &size, &pErrorMsgs))
 				CheckErrorMsgs(pErrorMsgs);
 			CreateShaderModule(pCode, size, compiled._shaderModules[+Stage::ds]);
 		}
@@ -158,7 +159,7 @@ void ShaderVulkan::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex] = "_GS";
-			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(entryGS), "gs", flags, &pCode, &size, &pErrorMsgs))
+			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::gs]), "gs", flags, &pCode, &size, &pErrorMsgs))
 				CheckErrorMsgs(pErrorMsgs);
 			CreateShaderModule(pCode, size, compiled._shaderModules[+Stage::gs]);
 		}
@@ -167,7 +168,7 @@ void ShaderVulkan::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex] = "_FS";
-			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(entryFS), "fs", flags, &pCode, &size, &pErrorMsgs))
+			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::fs]), "fs", flags, &pCode, &size, &pErrorMsgs))
 				CheckErrorMsgs(pErrorMsgs);
 			CreateShaderModule(pCode, size, compiled._shaderModules[+Stage::fs]);
 		}
@@ -176,13 +177,13 @@ void ShaderVulkan::Init(CSZ source, CSZ sourceName, CSZ* branches)
 		{
 			compiled._stageCount++;
 			vDefines[typeIndex] = "_CS";
-			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(entryCS), "cs", flags, &pCode, &size, &pErrorMsgs))
+			if (!RendererVulkan::VulkanCompile(source, sourceName, vDefines.data(), &inc, _C(stageEntries[+Stage::cs]), "cs", flags, &pCode, &size, &pErrorMsgs))
 				CheckErrorMsgs(pErrorMsgs);
 			CreateShaderModule(pCode, size, compiled._shaderModules[+Stage::cs]);
 			_compute = true;
 		}
 
-		_mapCompiled[entry] = compiled;
+		_mapCompiled[branch] = compiled;
 
 		branches++;
 	}
@@ -248,6 +249,24 @@ void ShaderVulkan::CreatePipelineLayout()
 	VERUS_QREF_RENDERER_VULKAN;
 	VkResult res = VK_SUCCESS;
 
+	StringStream ss;
+
+	auto PrintStageFlags = [&ss](VkShaderStageFlags stageFlags)
+	{
+		if (stageFlags & VK_SHADER_STAGE_VERTEX_BIT)
+			ss << "[V]";
+		if (stageFlags & VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT)
+			ss << "[TC]";
+		if (stageFlags & VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT)
+			ss << "[TE]";
+		if (stageFlags & VK_SHADER_STAGE_GEOMETRY_BIT)
+			ss << "[G]";
+		if (stageFlags & VK_SHADER_STAGE_FRAGMENT_BIT)
+			ss << "[F]";
+		if (stageFlags & VK_SHADER_STAGE_COMPUTE_BIT)
+			ss << "[C]";
+	};
+
 	VkPushConstantRange vkpcr = {};
 	VkPipelineLayoutCreateInfo vkplci = {};
 	vkplci.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO;
@@ -307,6 +326,34 @@ void ShaderVulkan::CreatePipelineLayout()
 				vkdslci.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO;
 				vkdslci.bindingCount = Utils::Cast32(vBindings.size());
 				vkdslci.pBindings = vBindings.data();
+
+				ss << "Set=" << _vDescriptorSetLayouts.size();
+				ss << std::endl;
+				for (const auto& binding : vBindings)
+				{
+					ss << "    binding=" << binding.binding;
+					ss << ", type=";
+					switch (binding.descriptorType)
+					{
+					case VK_DESCRIPTOR_TYPE_SAMPLER: ss << "SAMPLER"; break;
+					case VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER: ss << "COMBINED_IMAGE_SAMPLER"; break;
+					case VK_DESCRIPTOR_TYPE_SAMPLED_IMAGE: ss << "SAMPLED_IMAGE"; break;
+					case VK_DESCRIPTOR_TYPE_STORAGE_IMAGE: ss << "STORAGE_IMAGE"; break;
+					case VK_DESCRIPTOR_TYPE_UNIFORM_TEXEL_BUFFER: ss << "UNIFORM_TEXEL_BUFFER"; break;
+					case VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER: ss << "STORAGE_TEXEL_BUFFER"; break;
+					case VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER: ss << "UNIFORM_BUFFER"; break;
+					case VK_DESCRIPTOR_TYPE_STORAGE_BUFFER: ss << "STORAGE_BUFFER"; break;
+					case VK_DESCRIPTOR_TYPE_UNIFORM_BUFFER_DYNAMIC: ss << "UNIFORM_BUFFER_DYNAMIC"; break;
+					case VK_DESCRIPTOR_TYPE_STORAGE_BUFFER_DYNAMIC: ss << "STORAGE_BUFFER_DYNAMIC"; break;
+					case VK_DESCRIPTOR_TYPE_INPUT_ATTACHMENT: ss << "INPUT_ATTACHMENT"; break;
+					}
+					ss << ", count=" << binding.descriptorCount;
+					ss << ", flags=";
+					PrintStageFlags(binding.stageFlags);
+					ss << ", immutableSampler=" << binding.pImmutableSamplers;
+					ss << std::endl;
+				}
+
 				VkDescriptorSetLayout descriptorSetLayout = VK_NULL_HANDLE;
 				if (VK_SUCCESS != (res = vkCreateDescriptorSetLayout(pRendererVulkan->GetVkDevice(), &vkdslci, pRendererVulkan->GetAllocator(), &descriptorSetLayout)))
 					throw VERUS_RUNTIME_ERROR << "vkCreateDescriptorSetLayout(), res=" << res;
@@ -321,6 +368,14 @@ void ShaderVulkan::CreatePipelineLayout()
 				vkpcr.size = dsd._size;
 				vkplci.pushConstantRangeCount = 1;
 				vkplci.pPushConstantRanges = &vkpcr;
+
+				ss << "Set=PushConst";
+				ss << std::endl;
+				ss << "    flags=";
+				PrintStageFlags(vkpcr.stageFlags);
+				ss << ", offset=" << vkpcr.offset;
+				ss << ", size=" << vkpcr.size;
+				ss << std::endl;
 			}
 		}
 
@@ -328,6 +383,8 @@ void ShaderVulkan::CreatePipelineLayout()
 		vkplci.pSetLayouts = _vDescriptorSetLayouts.data();
 	}
 
+	_debugInfo = ss.str();
+
 	if (VK_SUCCESS != (res = vkCreatePipelineLayout(pRendererVulkan->GetVkDevice(), &vkplci, pRendererVulkan->GetAllocator(), &_pipelineLayout)))
 		throw VERUS_RUNTIME_ERROR << "vkCreatePipelineLayout(), res=" << res;
 
@@ -340,7 +397,7 @@ void ShaderVulkan::CreatePipelineLayout()
 	}
 }
 
-int ShaderVulkan::BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips)
+CSHandle ShaderVulkan::BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips)
 {
 	VERUS_QREF_RENDERER_VULKAN;
 	VkResult res = VK_SUCCESS;
@@ -448,21 +505,21 @@ int ShaderVulkan::BindDescriptorSetTextures(int setNumber, std::initializer_list
 
 	vkUpdateDescriptorSets(pRendererVulkan->GetVkDevice(), Utils::Cast32(vWriteDescriptorSet.size()), vWriteDescriptorSet.data(), 0, nullptr);
 
-	return complexSetHandle;
+	return CSHandle::Make(complexSetHandle);
 }
 
-void ShaderVulkan::FreeDescriptorSet(int& complexSetHandle)
+void ShaderVulkan::FreeDescriptorSet(CSHandle& complexSetHandle)
 {
 	VERUS_QREF_RENDERER_VULKAN;
 	VkResult res = VK_SUCCESS;
 
-	if (complexSetHandle >= 0 && complexSetHandle < _vComplexDescriptorSets.size() && _vComplexDescriptorSets[complexSetHandle] != VK_NULL_HANDLE)
+	if (complexSetHandle.IsSet() && complexSetHandle.Get() < _vComplexDescriptorSets.size() && _vComplexDescriptorSets[complexSetHandle.Get()] != VK_NULL_HANDLE)
 	{
-		if (VK_SUCCESS != (res = vkFreeDescriptorSets(pRendererVulkan->GetVkDevice(), _descriptorPool, 1, &_vComplexDescriptorSets[complexSetHandle])))
+		if (VK_SUCCESS != (res = vkFreeDescriptorSets(pRendererVulkan->GetVkDevice(), _descriptorPool, 1, &_vComplexDescriptorSets[complexSetHandle.Get()])))
 			throw VERUS_RUNTIME_ERROR << "vkFreeDescriptorSets(), res=" << res;
-		_vComplexDescriptorSets[complexSetHandle] = VK_NULL_HANDLE;
+		_vComplexDescriptorSets[complexSetHandle.Get()] = VK_NULL_HANDLE;
 	}
-	complexSetHandle = -1;
+	complexSetHandle = CSHandle();
 }
 
 void ShaderVulkan::BeginBindDescriptors()
@@ -582,9 +639,9 @@ VkDescriptorSet ShaderVulkan::GetVkDescriptorSet(int setNumber)
 	return _vDescriptorSetDesc[setNumber]._descriptorSet;
 }
 
-VkDescriptorSet ShaderVulkan::GetComplexVkDescriptorSet(int descSetID)
+VkDescriptorSet ShaderVulkan::GetComplexVkDescriptorSet(CSHandle descSetID)
 {
-	return _vComplexDescriptorSets[descSetID];
+	return _vComplexDescriptorSets[descSetID.Get()];
 }
 
 bool ShaderVulkan::TryPushConstants(int setNumber, RBaseCommandBuffer cb)
@@ -621,7 +678,7 @@ int ShaderVulkan::UpdateUniformBuffer(int setNumber)
 void ShaderVulkan::OnError(CSZ s)
 {
 	VERUS_QREF_RENDERER;
-	if (strstr(s, ": error "))
+	if (strstr(s, " ERROR: "))
 		renderer.OnShaderError(s);
 	else
 		renderer.OnShaderWarning(s);

RendererVulkan/src/CGI/ShaderVulkan.h

@@ -6,11 +6,16 @@ namespace verus
 	{
 		class ShaderVulkan : public BaseShader
 		{
+		public:
 			struct Compiled
 			{
 				VkShaderModule _shaderModules[+Stage::count] = {};
+				String         _entry;
 				int            _stageCount = 0;
 			};
+			VERUS_TYPEDEFS(Compiled);
+
+		private:
 			typedef Map<String, Compiled> TMapCompiled;
 
 			struct DescriptorSetDesc
@@ -36,6 +41,7 @@ namespace verus
 			Vector<VkDescriptorSet>       _vComplexDescriptorSets;
 			VkDescriptorPool              _descriptorPool = VK_NULL_HANDLE;
 			VkPipelineLayout              _pipelineLayout = VK_NULL_HANDLE;
+			String                        _debugInfo;
 			UINT64                        _currentFrame = UINT64_MAX;
 			uint32_t                      _poolComplexUniformBuffers = 0;
 			uint32_t                      _poolComplexImageSamplers = 0;
@@ -52,8 +58,8 @@ namespace verus
 
 			virtual void CreateDescriptorSet(int setNumber, const void* pSrc, int size, int capacity, std::initializer_list<Sampler> il, ShaderStageFlags stageFlags) override;
 			virtual void CreatePipelineLayout() override;
-			virtual int BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips) override;
-			virtual void FreeDescriptorSet(int& complexSetHandle) override;
+			virtual CSHandle BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips) override;
+			virtual void FreeDescriptorSet(CSHandle& complexSetHandle) override;
 
 			virtual void BeginBindDescriptors() override;
 			virtual void EndBindDescriptors() override;
@@ -65,10 +71,9 @@ namespace verus
 			void CreateDescriptorPool();
 			void CreateDescriptorSets();
 			VkDescriptorSet GetVkDescriptorSet(int setNumber);
-			VkDescriptorSet GetComplexVkDescriptorSet(int descSetID);
+			VkDescriptorSet GetComplexVkDescriptorSet(CSHandle descSetID);
 
-			VkShaderModule GetVkShaderModule(CSZ branch, Stage stage) const { return _mapCompiled.at(branch)._shaderModules[+stage]; }
-			int GetStageCount(CSZ branch) const { return _mapCompiled.at(branch)._stageCount; }
+			RcCompiled GetCompiled(CSZ branch) const { return _mapCompiled.at(branch); }
 
 			VkPipelineLayout GetVkPipelineLayout() const { return _pipelineLayout; }
 

RendererVulkan/src/CGI/TextureVulkan.cpp

@@ -28,9 +28,8 @@ void TextureVulkan::Init(RcTextureDesc desc)
 	_desc._mipLevels = desc._mipLevels ? desc._mipLevels : Math::ComputeMipLevels(desc._width, desc._height, desc._depth);
 	_bytesPerPixel = FormatToBytesPerPixel(desc._format);
 	const bool renderTarget = (_desc._flags & TextureDesc::Flags::colorAttachment);
-	const bool inputAttach = (_desc._flags & TextureDesc::Flags::inputAttachment);
 	const bool depthFormat = IsDepthFormat(desc._format);
-	const bool depthSampled = _desc._flags & TextureDesc::Flags::depthSampled;
+	const bool depthSampled = _desc._flags & (TextureDesc::Flags::depthSampledR | TextureDesc::Flags::depthSampledW);
 	if (_desc._flags & TextureDesc::Flags::anyShaderResource)
 		_mainLayout = ImageLayout::xsReadOnly;
 
@@ -50,14 +49,16 @@ void TextureVulkan::Init(RcTextureDesc desc)
 	vkici.initialLayout = VK_IMAGE_LAYOUT_UNDEFINED;
 	if (renderTarget)
 		vkici.usage = VK_IMAGE_USAGE_SAMPLED_BIT | VK_IMAGE_USAGE_COLOR_ATTACHMENT_BIT;
-	if (inputAttach)
-		vkici.usage |= VK_IMAGE_USAGE_INPUT_ATTACHMENT_BIT;
+	if (_desc._flags & TextureDesc::Flags::generateMips)
+		vkici.usage |= VK_IMAGE_USAGE_TRANSFER_DST_BIT;
 	if (depthFormat)
 	{
 		vkici.usage = VK_IMAGE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT;
 		if (depthSampled)
 			vkici.usage |= VK_IMAGE_USAGE_SAMPLED_BIT;
 	}
+	if (_desc._flags & TextureDesc::Flags::inputAttachment)
+		vkici.usage |= VK_IMAGE_USAGE_INPUT_ATTACHMENT_BIT;
 	pRendererVulkan->CreateImage(&vkici, VMA_MEMORY_USAGE_GPU_ONLY, _image, _vmaAllocation);
 
 	if (_desc._flags & TextureDesc::Flags::generateMips)
@@ -68,6 +69,12 @@ void TextureVulkan::Init(RcTextureDesc desc)
 		vkiciStorage.extent.height = Math::Max(1, _desc._height >> 1);
 		vkiciStorage.mipLevels = Math::Max(1, _desc._mipLevels - 1);
 		vkiciStorage.usage = VK_IMAGE_USAGE_TRANSFER_SRC_BIT | VK_IMAGE_USAGE_STORAGE_BIT;
+		// sRGB cannot be used with storage image:
+		switch (vkiciStorage.format)
+		{
+		case VK_FORMAT_R8G8B8A8_SRGB: vkiciStorage.format = VK_FORMAT_R8G8B8A8_UNORM; break;
+		case VK_FORMAT_B8G8R8A8_SRGB: vkiciStorage.format = VK_FORMAT_B8G8R8A8_UNORM; break;
+		}
 		pRendererVulkan->CreateImage(&vkiciStorage, VMA_MEMORY_USAGE_GPU_ONLY, _storageImage, _storageVmaAllocation);
 
 		_vCshGenerateMips.reserve((_desc._mipLevels + 3) / 4);
@@ -78,7 +85,7 @@ void TextureVulkan::Init(RcTextureDesc desc)
 			vkivci.sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO;
 			vkivci.image = _storageImage;
 			vkivci.viewType = VK_IMAGE_VIEW_TYPE_2D;
-			vkivci.format = vkici.format;
+			vkivci.format = vkiciStorage.format;
 			vkivci.subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT;
 			vkivci.subresourceRange.baseMipLevel = mip;
 			vkivci.subresourceRange.levelCount = 1;
@@ -103,6 +110,12 @@ void TextureVulkan::Init(RcTextureDesc desc)
 	vkivci.subresourceRange.layerCount = vkici.arrayLayers;
 	if (VK_SUCCESS != (res = vkCreateImageView(pRendererVulkan->GetVkDevice(), &vkivci, pRendererVulkan->GetAllocator(), &_imageView)))
 		throw VERUS_RUNTIME_ERROR << "vkCreateImageView(), res=" << res;
+	if (renderTarget)
+	{
+		vkivci.subresourceRange.levelCount = 1;
+		if (VK_SUCCESS != (res = vkCreateImageView(pRendererVulkan->GetVkDevice(), &vkivci, pRendererVulkan->GetAllocator(), &_imageViewLevelZero)))
+			throw VERUS_RUNTIME_ERROR << "vkCreateImageView(LevelZero), res=" << res;
+	}
 
 	if (_desc._pSamplerDesc)
 		CreateSampler();
@@ -116,7 +129,7 @@ void TextureVulkan::Init(RcTextureDesc desc)
 	}
 	if (depthFormat)
 	{
-		const ImageLayout layout = depthSampled ? ImageLayout::depthStencilReadOnly : ImageLayout::depthStencilAttachment;
+		const ImageLayout layout = (_desc._flags & TextureDesc::Flags::depthSampledR) ? ImageLayout::depthStencilReadOnly : ImageLayout::depthStencilAttachment;
 		CommandBufferVulkan commandBuffer;
 		commandBuffer.InitOneTimeSubmit();
 		commandBuffer.PipelineImageMemoryBarrier(TexturePtr::From(this), ImageLayout::undefined, layout, 0, 0);
@@ -134,6 +147,7 @@ void TextureVulkan::Done()
 	ForceScheduled();
 
 	VERUS_VULKAN_DESTROY(_sampler, vkDestroySampler(pRendererVulkan->GetVkDevice(), _sampler, pRendererVulkan->GetAllocator()));
+	VERUS_VULKAN_DESTROY(_imageViewLevelZero, vkDestroyImageView(pRendererVulkan->GetVkDevice(), _imageViewLevelZero, pRendererVulkan->GetAllocator()));
 	VERUS_VULKAN_DESTROY(_imageView, vkDestroyImageView(pRendererVulkan->GetVkDevice(), _imageView, pRendererVulkan->GetAllocator()));
 	for (auto view : _vStorageImageViews)
 		VERUS_VULKAN_DESTROY(view, vkDestroyImageView(pRendererVulkan->GetVkDevice(), view, pRendererVulkan->GetAllocator()));
@@ -237,7 +251,7 @@ void TextureVulkan::GenerateMips(PBaseCommandBuffer pCB)
 		ub._srcMipLevel = srcMip;
 		ub._mipLevelCount = dispatchMipCount;
 		ub._srcDimensionCase = (srcHeight & 1) << 1 | (srcWidth & 1);
-		ub._srgb = false;
+		ub._srgb = IsSRGBFormat(_desc._format);
 		ub._texelSize.x = 1.f / dstWidth;
 		ub._texelSize.y = 1.f / dstHeight;
 
@@ -246,7 +260,7 @@ void TextureVulkan::GenerateMips(PBaseCommandBuffer pCB)
 			int mips[5] = {}; // For input texture (always mip 0) and 4 storage mips.
 			VERUS_FOR(mip, dispatchMipCount)
 				mips[mip + 1] = srcMip + mip;
-			const int complexSetHandle = shader->BindDescriptorSetTextures(0, { tex, tex, tex, tex, tex }, mips);
+			const CSHandle complexSetHandle = shader->BindDescriptorSetTextures(0, { tex, tex, tex, tex, tex }, mips);
 			_vCshGenerateMips.push_back(complexSetHandle);
 			pCB->BindDescriptors(shader, 0, complexSetHandle);
 		}
@@ -312,7 +326,7 @@ Continue TextureVulkan::Scheduled_Update()
 	{
 		VERUS_QREF_RENDERER;
 		auto shader = renderer.GetShaderGenerateMips();
-		for (int csh : _vCshGenerateMips)
+		for (auto& csh : _vCshGenerateMips)
 			shader->FreeDescriptorSet(csh);
 		_vCshGenerateMips.clear();
 	}

RendererVulkan/src/CGI/TextureVulkan.h

@@ -17,11 +17,12 @@ namespace verus
 			VkImage             _storageImage = VK_NULL_HANDLE;
 			VmaAllocation       _storageVmaAllocation = VK_NULL_HANDLE;
 			VkImageView         _imageView = VK_NULL_HANDLE;
+			VkImageView         _imageViewLevelZero = VK_NULL_HANDLE;
 			VkSampler           _sampler = VK_NULL_HANDLE;
 			Vector<UINT32>      _vDefinedSubresources;
 			Vector<VkImageView> _vStorageImageViews;
 			Vector<VkBufferEx>  _vStagingBuffers;
-			Vector<int>         _vCshGenerateMips;
+			Vector<CSHandle>    _vCshGenerateMips;
 			bool                _definedStorage = false;
 
 		public:
@@ -45,6 +46,7 @@ namespace verus
 
 			VkImage GetVkImage() const { return _image; }
 			VkImageView GetVkImageView() const { return _imageView; }
+			VkImageView GetVkImageViewForFramebuffer() const { return _imageViewLevelZero ? _imageViewLevelZero : _imageView; }
 			VkImageView GetStorageVkImageView(int mip) const { return _vStorageImageViews[mip]; }
 			VkSampler GetVkSampler() const { return _sampler; }
 			ImageLayout GetSubresourceMainLayout(int mipLevel, int arrayLayer) const;

Scripts/Compress (BC1).bat

@@ -1 +0,0 @@
-"C:\Home\nvidia-texture-tools-2.1.1-win64\bin64\nvcompress.exe" -bc1a %1

Scripts/Compress (BC3).bat

@@ -1 +0,0 @@
-"C:\Home\nvidia-texture-tools-2.1.1-win64\bin64\nvcompress.exe" -bc3 %1

Scripts/Compress (BC7).bat

@@ -1 +0,0 @@
-"C:\Home\nvidia-texture-tools-2.1.1-win64\bin64\nvcompress.exe" -bc7 %1

Scripts/Compress Colors (BC7).bat

@@ -0,0 +1 @@
+"C:\Home\Projects\Verus\verus\Bin\TextureTool.exe" %1

Scripts/Compress Non-Color Data (BC7).bat

@@ -0,0 +1 @@
+"C:\Home\Projects\Verus\verus\Bin\TextureTool.exe" --non-color-data %1

Scripts/Compress Normals (BC7).bat

@@ -0,0 +1 @@
+"C:\Home\Projects\Verus\verus\Bin\TextureTool.exe" --normal-map %1

TextureTool/TextureTool.vcxproj

@@ -0,0 +1,97 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <VCProjectVersion>16.0</VCProjectVersion>
+    <ProjectGuid>{5A1A3E76-7F69-48B6-B1E3-F6BB281B7E73}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>TextureTool</RootNamespace>
+    <WindowsTargetPlatformVersion>10.0</WindowsTargetPlatformVersion>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v142</PlatformToolset>
+    <CharacterSet>Unicode</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v142</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>Unicode</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+    <Import Project="..\Verus\Verus.props" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+    <Import Project="..\Verus\Verus.props" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <SDLCheck>true</SDLCheck>
+      <PreprocessorDefinitions>_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <ConformanceMode>true</ConformanceMode>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <SDLCheck>true</SDLCheck>
+      <PreprocessorDefinitions>NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <ConformanceMode>true</ConformanceMode>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemGroup>
+    <ClCompile Include="main.cpp" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="..\Verus\Verus.vcxproj">
+      <Project>{b154d670-e4b1-4d8a-885c-69546a5bd833}</Project>
+    </ProjectReference>
+  </ItemGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>

TextureTool/TextureTool.vcxproj.filters

@@ -0,0 +1,22 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;c++;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+    <Filter Include="Header Files">
+      <UniqueIdentifier>{93995380-89BD-4b04-88EB-625FBE52EBFB}</UniqueIdentifier>
+      <Extensions>h;hh;hpp;hxx;h++;hm;inl;inc;ipp;xsd</Extensions>
+    </Filter>
+    <Filter Include="Resource Files">
+      <UniqueIdentifier>{67DA6AB6-F800-4c08-8B7A-83BB121AAD01}</UniqueIdentifier>
+      <Extensions>rc;ico;cur;bmp;dlg;rc2;rct;bin;rgs;gif;jpg;jpeg;jpe;resx;tiff;tif;png;wav;mfcribbon-ms</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="main.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+</Project>

TextureTool/main.cpp

@@ -0,0 +1,363 @@
+#define VERUS_INCLUDE_COMPRESSONATOR
+#include <verus.h>
+
+using namespace verus;
+
+void DeleteMipmap(CWSZ pathname);
+
+CMP_BOOL CompressionCallback(CMP_FLOAT progress, CMP_DWORD_PTR pUser1, CMP_DWORD_PTR pUser2)
+{
+	std::wcout << _T("ProcessTexture progress = ") << static_cast<int>(progress) << _T("%") << std::endl;
+	return false;
+}
+
+void Run()
+{
+	int argCount = 0;
+	LPWSTR* argArray = CommandLineToArgvW(GetCommandLine(), &argCount);
+	WideString pathNameW;
+	std::wcout << _T("TextureTool") << std::endl;
+	std::wcout << _T("Copyright (c) 2016-2020 Dmitry Maluev") << std::endl;
+	if (argCount < 2)
+	{
+		std::wcout << _T("Enter file name: ");
+		std::wcin >> pathNameW;
+	}
+	else
+		pathNameW = argArray[argCount - 1];
+	String pathname = Str::WideToUtf8(pathNameW);
+	bool argDeleteMipmap = false;
+	bool argFade = false;
+	bool argNormalMap = false;
+	bool argColorData = true;
+	VERUS_FOR(i, argCount)
+	{
+		if (!_wcsicmp(argArray[i], L"--delete-mipmap"))
+			argDeleteMipmap = true;
+		if (!_wcsicmp(argArray[i], L"--fade"))
+			argFade = true;
+		if (!_wcsicmp(argArray[i], L"--normal-map"))
+			argNormalMap = true;
+		if (!_wcsicmp(argArray[i], L"--non-color-data"))
+			argColorData = false;
+	}
+	LocalFree(argArray);
+
+	if (argDeleteMipmap)
+	{
+		DeleteMipmap(_C(pathNameW));
+		return;
+	}
+
+	CMP_ERROR cmpError = CMP_OK;
+
+	CMP_MipSet srcMipSet = {};
+	cmpError = CMP_LoadTexture(_C(pathname), &srcMipSet);
+	if (CMP_OK != cmpError)
+	{
+		std::wcerr << _T("ERROR: CMP_LoadTexture(), ") << cmpError << std::endl;
+		throw std::exception();
+	}
+
+	// FX image with extra info:
+	Vector<UINT32> vMipFX[2];
+	String pathnameFX(pathname);
+	Str::ReplaceExtension(pathnameFX, ".FX.png");
+	if (argNormalMap && IO::FileSystem::FileExist(_C(pathnameFX)))
+	{
+		std::wcout << _T("FX file: ") << Str::Utf8ToWide(pathnameFX) << std::endl;
+		IO::Image imageFX;
+		imageFX.Init(_C(pathnameFX));
+		if (imageFX._width == srcMipSet.m_nWidth && imageFX._height == srcMipSet.m_nHeight)
+		{
+			const int pixelCount = imageFX._width * imageFX._height;
+			vMipFX[0].resize(pixelCount);
+			vMipFX[1].resize(pixelCount);
+			VERUS_FOR(i, pixelCount)
+				memcpy(&vMipFX[0][i], imageFX._p + imageFX._bytesPerPixel * i, imageFX._bytesPerPixel);
+		}
+		else
+		{
+			std::wcerr << _T("ERROR: FX image has different dimensions: ") << imageFX._width << _T("x") << imageFX._height << std::endl;
+			throw std::exception();
+		}
+	}
+
+	if (srcMipSet.m_nMipLevels <= 1)
+		CMP_GenerateMIPLevels(&srcMipSet, 0);
+
+	CMP_MipLevel* pMipLevel = srcMipSet.m_pMipLevelTable[0];
+	const BYTE* pData = pMipLevel->m_pbData;
+	const int pixelCount = srcMipSet.m_nWidth * srcMipSet.m_nHeight;
+
+	Vector<UINT32> vMip[2];
+	vMip[0].resize(pixelCount);
+	vMip[1].resize(pixelCount);
+
+	const float fadePerMip = pow(0.02f, 1.f / srcMipSet.m_nMipLevels);
+
+	int mip = 0;
+
+	bool finish = false;
+	do
+	{
+		std::wcout << _T("Processing mipmap ") << mip << std::endl;
+
+		const int mipW = Math::Max(1, srcMipSet.m_nWidth >> mip);
+		const int mipH = Math::Max(1, srcMipSet.m_nHeight >> mip);
+
+		// Switch between in and out:
+		const int inIndex = (mip & 0x1);
+		const int outIndex = ((mip + 1) & 0x1);
+
+		VERUS_FOR(i, mipH)
+		{
+			VERUS_FOR(j, mipW)
+			{
+				int offset = 0;
+				UINT32 color = 0;
+				UINT32 colorFX = 0;
+
+				if (mip) // Generate mipmap level 1+:
+				{
+					offset = i * mipW + j;
+
+					const int aboveMipW = mipW << 1;
+					const int aboveMipH = mipH << 1;
+					const int aboveMipEdgeW = aboveMipW - 1;
+					const int aboveMipEdgeH = aboveMipH - 1;
+
+					const int sampleA[2] = { Math::Clamp((j << 1) + 0, 0, aboveMipEdgeW), Math::Clamp((i << 1) + 0, 0, aboveMipEdgeH) };
+					const int sampleB[2] = { Math::Clamp((j << 1) + 1, 0, aboveMipEdgeW), Math::Clamp((i << 1) + 0, 0, aboveMipEdgeH) };
+					const int sampleC[2] = { Math::Clamp((j << 1) + 0, 0, aboveMipEdgeW), Math::Clamp((i << 1) + 1, 0, aboveMipEdgeH) };
+					const int sampleD[2] = { Math::Clamp((j << 1) + 1, 0, aboveMipEdgeW), Math::Clamp((i << 1) + 1, 0, aboveMipEdgeH) };
+
+					const int offsetA = sampleA[1] * aboveMipW + sampleA[0];
+					const int offsetB = sampleB[1] * aboveMipW + sampleB[0];
+					const int offsetC = sampleC[1] * aboveMipW + sampleC[0];
+					const int offsetD = sampleD[1] * aboveMipW + sampleD[0];
+
+					if (argNormalMap)
+					{
+						Vector4 data[4];
+						float raw[4];
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetA], raw, false); data[0] = Vector4::MakeFromPointer(raw);
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetB], raw, false); data[1] = Vector4::MakeFromPointer(raw);
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetC], raw, false); data[2] = Vector4::MakeFromPointer(raw);
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetD], raw, false); data[3] = Vector4::MakeFromPointer(raw);
+
+						// Generate FX mipmap:
+						if (!vMipFX[inIndex].empty())
+						{
+							const BYTE* rgbaA = reinterpret_cast<BYTE*>(&vMipFX[inIndex][offsetA]);
+							const BYTE* rgbaB = reinterpret_cast<BYTE*>(&vMipFX[inIndex][offsetB]);
+							const BYTE* rgbaC = reinterpret_cast<BYTE*>(&vMipFX[inIndex][offsetC]);
+							const BYTE* rgbaD = reinterpret_cast<BYTE*>(&vMipFX[inIndex][offsetD]);
+							BYTE rgbaOut[4] =
+							{
+								static_cast<BYTE>((rgbaA[0] + rgbaB[0] + rgbaC[0] + rgbaD[0]) >> 2),
+								static_cast<BYTE>((rgbaA[1] + rgbaB[1] + rgbaC[1] + rgbaD[1]) >> 2),
+								static_cast<BYTE>((rgbaA[2] + rgbaB[2] + rgbaC[2] + rgbaD[2]) >> 2),
+								static_cast<BYTE>((rgbaA[3] + rgbaB[3] + rgbaC[3] + rgbaD[3]) >> 2)
+							};
+							memcpy(&colorFX, rgbaOut, 4);
+						}
+
+						const float prevAvgLength[4] =
+						{
+							data[0].getW(),
+							data[1].getW(),
+							data[2].getW(),
+							data[3].getW()
+						};
+						data[0].setW(0);
+						data[1].setW(0);
+						data[2].setW(0);
+						data[3].setW(0);
+
+						float minAvgLength = 1;
+						VERUS_FOR(i, 4) // To [-1 to 1]:
+						{
+							data[i] = data[i] * 2.f - Vector4(1, 1, 1, 0);
+							minAvgLength = Math::Min(minAvgLength, prevAvgLength[i]);
+						}
+
+						Vector4 avg = (
+							data[0] * prevAvgLength[0] +
+							data[1] * prevAvgLength[1] +
+							data[2] * prevAvgLength[2] +
+							data[3] * prevAvgLength[3]) * 0.25f;
+						float len = VMath::length(avg);
+
+						if (len > minAvgLength) // Don't make it too smooth.
+							len = (len + minAvgLength) * 0.5f;
+
+						Vector4 dataOut = VMath::normalize(data[0] + data[1] + data[2] + data[3]);
+						if (argFade)
+							dataOut = VMath::normalize(VMath::lerp(fadePerMip, Vector4(0, 0, 1, 0), dataOut));
+						dataOut = dataOut * 0.5f + Vector4(0.5f, 0.5f, 0.5f, 0); // To [0 to 1]:
+						dataOut.setW(len);
+						color = Convert::ColorFloatToInt32(dataOut.ToPointer(), false);
+					}
+					else
+					{
+						Vector4 data[4];
+						float raw[4];
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetA], raw, argColorData); data[0] = Vector4::MakeFromPointer(raw);
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetB], raw, argColorData); data[1] = Vector4::MakeFromPointer(raw);
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetC], raw, argColorData); data[2] = Vector4::MakeFromPointer(raw);
+						Convert::ColorInt32ToFloat(vMip[inIndex][offsetD], raw, argColorData); data[3] = Vector4::MakeFromPointer(raw);
+						Vector4 dataOut = (data[0] + data[1] + data[2] + data[3]) * 0.25f;
+						color = Convert::ColorFloatToInt32(dataOut.ToPointer(), argColorData);
+					}
+				}
+				else // Handle topmost mipmap level:
+				{
+					if (argNormalMap)
+					{
+						const bool upsideDownBGRA = false;
+						if (upsideDownBGRA)
+						{
+							// Input image is upside down BGRA (?):
+							offset = (mipH - i - 1) * mipW + j;
+							UINT32 input = 0;
+							memcpy(&input, &pData[offset * 4], 4);
+							reinterpret_cast<BYTE*>(&input)[3] >>= 2; // Divide alpha by 4.
+							color = VERUS_COLOR_TO_D3D(input);
+							offset = i * mipW + j;
+						}
+						else
+						{
+							offset = i * mipW + j;
+							UINT32 input = 0;
+							memcpy(&input, &pData[offset * 4], 4);
+							reinterpret_cast<BYTE*>(&input)[3] = 0xFF;
+							color = input;
+
+							if (!vMipFX[inIndex].empty())
+							{
+								colorFX = vMipFX[inIndex][offset];
+								BYTE* rgbaFX = reinterpret_cast<BYTE*>(&colorFX);
+								rgbaFX[0] /= 4;
+								rgbaFX[1] /= 4;
+								rgbaFX[2] /= 4;
+								rgbaFX[3] /= 4;
+							}
+						}
+					}
+					else // Simple copy for color:
+					{
+						offset = i * mipW + j;
+						UINT32 input = 0;
+						memcpy(&input, &pData[offset * 4], 4);
+						color = input;
+					}
+				}
+				vMip[outIndex][offset] = color;
+				if (!vMipFX[outIndex].empty())
+					vMipFX[outIndex][offset] = colorFX;
+			}
+		}
+
+		if (argNormalMap)
+		{
+			// Shuffle channels, read from outIndex, write to inIndex:
+			const int pixelCount = mipW * mipH;
+			VERUS_FOR(i, pixelCount)
+			{
+				BYTE* rgbaFX = vMipFX[outIndex].empty() ? nullptr : reinterpret_cast<BYTE*>(&vMipFX[outIndex][i]);
+				BYTE* rgba = reinterpret_cast<BYTE*>(&vMip[outIndex][i]);
+				const int len = Math::Clamp(rgba[3] - 128, 0, 0xFF);
+				vMip[inIndex][i] = VERUS_COLOR_RGBA(rgbaFX ? rgbaFX[0] : 0, rgba[1], len, rgba[0]);
+			}
+
+			// Save shuffled mip level, read from  inIndex:
+			CMP_MipLevel* pDstMipLevel = srcMipSet.m_pMipLevelTable[mip];
+			memcpy(pDstMipLevel->m_pbData, vMip[inIndex].data(), mipW * mipH * sizeof(UINT32));
+		}
+		else
+		{
+			CMP_MipLevel* pDstMipLevel = srcMipSet.m_pMipLevelTable[mip];
+			memcpy(pDstMipLevel->m_pbData, vMip[outIndex].data(), mipW * mipH * sizeof(UINT32));
+		}
+
+		finish = (mipW == 1) && (mipH == 1);
+		mip++; // Next level, switch between in and out.
+	} while (!finish);
+
+	CMP_MipSet dstMipSet = {};
+
+	KernelOptions kernelOptions = {};
+	kernelOptions.format = CMP_FORMAT_BC7;
+	kernelOptions.fquality = 0.51f; // Tests show that 51% is production quality.
+
+	cmpError = CMP_ProcessTexture(&srcMipSet, &dstMipSet, kernelOptions, CompressionCallback);
+	if (CMP_OK != cmpError)
+	{
+		std::wcerr << _T("ERROR: CMP_ProcessTexture(), ") << cmpError << std::endl;
+		throw std::exception();
+	}
+
+	Str::ReplaceExtension(pathname, ".dds");
+	cmpError = CMP_SaveTexture(_C(pathname), &dstMipSet);
+	if (CMP_OK != cmpError)
+	{
+		std::wcerr << _T("ERROR: CMP_SaveTexture(), ") << cmpError << std::endl;
+		throw std::exception();
+	}
+
+	CMP_FreeMipSet(&srcMipSet);
+	CMP_FreeMipSet(&dstMipSet);
+}
+
+void DeleteMipmap(CWSZ pathname)
+{
+	const String pathnameIn = Str::WideToUtf8(pathname);
+	const String pathnameOut = pathnameIn + "-del-mip.dds";
+	IO::File file, fileOut;
+	if (file.Open(_C(pathnameIn)) && file.GetSize() > sizeof(IO::DDSHeader) && fileOut.Open(_C(pathnameOut), "wb"))
+	{
+		IO::DDSHeader header;
+		file >> header;
+		if (!header.Validate())
+		{
+			std::wcerr << _T("ERROR: Invalid DDS header: ") << Str::Utf8ToWide(pathnameIn) << std::endl;
+			throw std::exception();
+		}
+		if (header.IsBC1() && header._mipMapCount > 1)
+		{
+			const int levelSize = IO::DDSHeader::ComputeBcLevelSize(header._width, header._height, true);
+
+			header._mipMapCount--;
+			header._width >>= 1;
+			header._height >>= 1;
+			header._pitchOrLinearSize = IO::DDSHeader::ComputeBcLevelSize(header._width, header._height, true);
+
+			fileOut << header;
+			file.Seek(sizeof(IO::DDSHeader) + levelSize, SEEK_SET);
+			Vector<BYTE> vData;
+			vData.resize(file.GetSize() - levelSize - sizeof(IO::DDSHeader));
+			file.Read(vData.data(), vData.size());
+			fileOut.Write(vData.data(), vData.size());
+		}
+		else
+		{
+			std::wcerr << _T("ERROR: Must be DXT1 with mipmaps") << std::endl;
+			throw std::exception();
+		}
+	}
+}
+
+int main(VERUS_MAIN_DEFAULT_ARGS)
+{
+	try
+	{
+		Run();
+	}
+	catch (const std::exception& e)
+	{
+		std::wcerr << _T("EXCEPTION: ") << e.what() << std::endl;
+		return EXIT_FAILURE;
+	}
+	return EXIT_SUCCESS;
+}

Verus.sln

@@ -15,6 +15,8 @@ Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "VulkanShaderCompiler", "Vul
 EndProject
 Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "PAKBuilder", "PAKBuilder\PAKBuilder.vcxproj", "{EBF1E2F9-65AA-419D-A3D3-AD66EB086E57}"
 EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "TextureTool", "TextureTool\TextureTool.vcxproj", "{5A1A3E76-7F69-48B6-B1E3-F6BB281B7E73}"
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|x64 = Debug|x64
@@ -45,6 +47,10 @@ Global
 		{EBF1E2F9-65AA-419D-A3D3-AD66EB086E57}.Debug|x64.Build.0 = Debug|x64
 		{EBF1E2F9-65AA-419D-A3D3-AD66EB086E57}.Release|x64.ActiveCfg = Release|x64
 		{EBF1E2F9-65AA-419D-A3D3-AD66EB086E57}.Release|x64.Build.0 = Release|x64
+		{5A1A3E76-7F69-48B6-B1E3-F6BB281B7E73}.Debug|x64.ActiveCfg = Debug|x64
+		{5A1A3E76-7F69-48B6-B1E3-F6BB281B7E73}.Debug|x64.Build.0 = Debug|x64
+		{5A1A3E76-7F69-48B6-B1E3-F6BB281B7E73}.Release|x64.ActiveCfg = Release|x64
+		{5A1A3E76-7F69-48B6-B1E3-F6BB281B7E73}.Release|x64.Build.0 = Release|x64
 	EndGlobalSection
 	GlobalSection(SolutionProperties) = preSolution
 		HideSolutionNode = FALSE

Verus/Verus.props

@@ -3,8 +3,8 @@
   <ImportGroup Label="PropertySheets" />
   <PropertyGroup Label="UserMacros" />
   <PropertyGroup>
-    <IncludePath>C:\Home\Projects\Verus\verus\Verus\src;C:\Home\Middleware\AMD Tootle 2.3\include;C:\Home\Middleware\bullet3-2.88\src;C:\Home\Middleware\bullet3-2.88\Extras;C:\Home\Middleware\DevIL-Windows-SDK-1.8.0\include;C:\Home\Middleware\libogg-1.3.4\include;C:\Home\Middleware\libvorbis-1.3.6\include;C:\Home\Middleware\SDL2-2.0.10\include;C:\Program Files %28x86%29\OpenAL 1.1 SDK\include;C:\VulkanSDK\1.2.131.2\Include;C:\VulkanSDK\1.2.131.2\glslang;$(IncludePath)</IncludePath>
-    <LibraryPath>C:\Home\Middleware\bullet3-2.88\bin;C:\Home\Middleware\AMD Tootle 2.3\lib;C:\Home\Middleware\DevIL-Windows-SDK-1.8.0\lib\x64\Release;C:\Home\Middleware\libogg-1.3.4\lib;C:\Home\Middleware\libvorbis-1.3.6\lib2;C:\Home\Middleware\SDL2-2.0.10\lib\x64;C:\Program Files %28x86%29\OpenAL 1.1 SDK\libs\Win64;C:\VulkanSDK\1.2.131.2\Lib;$(LibraryPath)</LibraryPath>
+    <IncludePath>C:\Compressonator_3.2.4691\include;C:\Home\Projects\Verus\verus\Verus\src;C:\Home\Middleware\AMD Tootle 2.3\include;C:\Home\Middleware\bullet3-2.88\src;C:\Home\Middleware\bullet3-2.88\Extras;C:\Home\Middleware\DevIL-Windows-SDK-1.8.0\include;C:\Home\Middleware\libogg-1.3.4\include;C:\Home\Middleware\libvorbis-1.3.6\include;C:\Home\Middleware\SDL2-2.0.10\include;C:\Program Files %28x86%29\OpenAL 1.1 SDK\include;C:\VulkanSDK\1.2.131.2\Include;C:\VulkanSDK\1.2.131.2\glslang;$(IncludePath)</IncludePath>
+    <LibraryPath>C:\Compressonator_3.2.4691\lib\VS2017\x64;C:\Home\Middleware\bullet3-2.88\bin;C:\Home\Middleware\AMD Tootle 2.3\lib;C:\Home\Middleware\DevIL-Windows-SDK-1.8.0\lib\x64\Release;C:\Home\Middleware\libogg-1.3.4\lib;C:\Home\Middleware\libvorbis-1.3.6\lib2;C:\Home\Middleware\SDL2-2.0.10\lib\x64;C:\Program Files %28x86%29\OpenAL 1.1 SDK\libs\Win64;C:\VulkanSDK\1.2.131.2\Lib;$(LibraryPath)</LibraryPath>
   </PropertyGroup>
   <ItemDefinitionGroup />
   <ItemGroup />

Verus/Verus.vcxproj

@@ -117,6 +117,7 @@
     <ClInclude Include="src\CGI\DeferredShading.h" />
     <ClInclude Include="src\CGI\RendererParser.h" />
     <ClInclude Include="src\CGI\Scheduled.h" />
+    <ClInclude Include="src\CGI\TextureRAM.h" />
     <ClInclude Include="src\CGI\Types.h" />
     <ClInclude Include="src\CGI\Formats.h" />
     <ClInclude Include="src\CGI\Renderer.h" />
@@ -202,6 +203,7 @@
     <ClInclude Include="src\Scene\Camera.h" />
     <ClInclude Include="src\Scene\CameraOrbit.h" />
     <ClInclude Include="src\Scene\EditorTerrain.h" />
+    <ClInclude Include="src\Scene\Grass.h" />
     <ClInclude Include="src\Scene\Helpers.h" />
     <ClInclude Include="src\Scene\MaterialManager.h" />
     <ClInclude Include="src\Scene\Mesh.h" />
@@ -209,6 +211,7 @@
     <ClInclude Include="src\Scene\SceneManager.h" />
     <ClInclude Include="src\Scene\ShadowMap.h" />
     <ClInclude Include="src\Scene\Terrain.h" />
+    <ClInclude Include="src\Scene\Water.h" />
     <ClInclude Include="src\Security\CipherRC4.h" />
     <ClInclude Include="src\Security\Security.h" />
     <ClInclude Include="src\ThirdParty\imgui\imconfig.h" />
@@ -274,6 +277,7 @@
     <ClCompile Include="src\CGI\RendererParser.cpp" />
     <ClCompile Include="src\CGI\RenderPass.cpp" />
     <ClCompile Include="src\CGI\Scheduled.cpp" />
+    <ClCompile Include="src\CGI\TextureRAM.cpp" />
     <ClCompile Include="src\D\D.cpp" />
     <ClCompile Include="src\D\Log.cpp" />
     <ClCompile Include="src\Extra\Extra.cpp" />
@@ -333,6 +337,7 @@
     <ClCompile Include="src\Scene\Camera.cpp" />
     <ClCompile Include="src\Scene\CameraOrbit.cpp" />
     <ClCompile Include="src\Scene\EditorTerrain.cpp" />
+    <ClCompile Include="src\Scene\Grass.cpp" />
     <ClCompile Include="src\Scene\Helpers.cpp" />
     <ClCompile Include="src\Scene\MaterialManager.cpp" />
     <ClCompile Include="src\Scene\Mesh.cpp" />
@@ -340,6 +345,7 @@
     <ClCompile Include="src\Scene\SceneManager.cpp" />
     <ClCompile Include="src\Scene\ShadowMap.cpp" />
     <ClCompile Include="src\Scene\Terrain.cpp" />
+    <ClCompile Include="src\Scene\Water.cpp" />
     <ClCompile Include="src\ThirdParty\fast_atof.c">
       <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">NotUsing</PrecompiledHeader>
       <PrecompiledHeader Condition="'$(Configuration)|$(Platform)'=='Release|x64'">NotUsing</PrecompiledHeader>
@@ -578,6 +584,56 @@
       <FileType>Document</FileType>
     </None>
   </ItemGroup>
+  <ItemGroup>
+    <None Include="src\Shaders\LibSurface.hlsl">
+      <FileType>Document</FileType>
+    </None>
+  </ItemGroup>
+  <ItemGroup>
+    <None Include="src\Shaders\LibTessellation.hlsl">
+      <FileType>Document</FileType>
+    </None>
+  </ItemGroup>
+  <ItemGroup>
+    <None Include="src\Shaders\DS_Grass.hlsl">
+      <FileType>Document</FileType>
+    </None>
+    <None Include="src\Shaders\DS_Grass.inc.hlsl">
+      <FileType>Document</FileType>
+    </None>
+  </ItemGroup>
+  <ItemGroup>
+    <None Include="src\Shaders\Sky.hlsl">
+      <FileType>Document</FileType>
+    </None>
+    <None Include="src\Shaders\Sky.inc.hlsl">
+      <FileType>Document</FileType>
+    </None>
+  </ItemGroup>
+  <ItemGroup>
+    <None Include="src\Shaders\Water.hlsl">
+      <FileType>Document</FileType>
+    </None>
+    <None Include="src\Shaders\Water.inc.hlsl">
+      <FileType>Document</FileType>
+    </None>
+  </ItemGroup>
+  <ItemGroup>
+    <None Include="src\Shaders\WaterGen.hlsl">
+      <FileType>Document</FileType>
+    </None>
+    <None Include="src\Shaders\WaterGen.inc.hlsl">
+      <FileType>Document</FileType>
+    </None>
+  </ItemGroup>
+  <ItemGroup>
+    <None Include="src\Shaders\SimpleTerrain.hlsl">
+      <FileType>Document</FileType>
+    </None>
+    <None Include="src\Shaders\SimpleTerrain.inc.hlsl">
+      <FileType>Document</FileType>
+    </None>
+  </ItemGroup>
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
   <ImportGroup Label="ExtensionTargets">
   </ImportGroup>

Verus/Verus.vcxproj.filters

@@ -552,6 +552,15 @@
     <ClInclude Include="src\IO\Xxx.h">
       <Filter>src\IO</Filter>
     </ClInclude>
+    <ClInclude Include="src\Scene\Grass.h">
+      <Filter>src\Scene</Filter>
+    </ClInclude>
+    <ClInclude Include="src\CGI\TextureRAM.h">
+      <Filter>src\CGI</Filter>
+    </ClInclude>
+    <ClInclude Include="src\Scene\Water.h">
+      <Filter>src\Scene</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="src\CGI\BaseGeometry.cpp">
@@ -917,6 +926,15 @@
     <ClCompile Include="src\IO\Xxx.cpp">
       <Filter>src\IO</Filter>
     </ClCompile>
+    <ClCompile Include="src\Scene\Grass.cpp">
+      <Filter>src\Scene</Filter>
+    </ClCompile>
+    <ClCompile Include="src\CGI\TextureRAM.cpp">
+      <Filter>src\CGI</Filter>
+    </ClCompile>
+    <ClCompile Include="src\Scene\Water.cpp">
+      <Filter>src\Scene</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <None Include="src\Shaders\Lib.hlsl">
@@ -979,6 +997,42 @@
     <None Include="src\Shaders\DebugDraw.inc.hlsl">
       <Filter>src\Shaders</Filter>
     </None>
+    <None Include="src\Shaders\LibSurface.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\LibTessellation.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\Sky.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\Sky.inc.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\DS_Grass.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\DS_Grass.inc.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\Water.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\Water.inc.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\WaterGen.inc.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\WaterGen.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\SimpleTerrain.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
+    <None Include="src\Shaders\SimpleTerrain.inc.hlsl">
+      <Filter>src\Shaders</Filter>
+    </None>
   </ItemGroup>
   <ItemGroup>
     <Natvis Include="src\ThirdParty\pugixml-1.10\pugixml.natvis">

Verus/src/App/Settings.cpp

@@ -25,6 +25,7 @@ void Settings::Load()
 	_gapi = GetI("gapi", _gapi);
 	_gpuAnisotropyLevel = GetI("gpuAnisotropyLevel", _gpuAnisotropyLevel);
 	_gpuAntialiasingLevel = GetI("gpuAntialiasingLevel", _gpuAntialiasingLevel);
+	_gpuTessellation = GetB("gpuTessellation", _gpuTessellation);
 	_gpuTextureLodLevel = GetI("gpuTextureLodLevel", _gpuTextureLodLevel);
 	_gpuTrilinearFilter = GetB("gpuTrilinearFilter", _gpuTrilinearFilter);
 	_inputMouseSensitivity = GetF("inputMouseSensitivity", _inputMouseSensitivity);
@@ -67,6 +68,7 @@ void Settings::Save()
 	Set("gapi", _gapi);
 	Set("gpuAnisotropyLevel", _gpuAnisotropyLevel);
 	Set("gpuAntialiasingLevel", _gpuAntialiasingLevel);
+	Set("gpuTessellation", _gpuTessellation);
 	Set("gpuTextureLodLevel", _gpuTextureLodLevel);
 	Set("gpuTrilinearFilter", _gpuTrilinearFilter);
 	Set("inputMouseSensitivity", _inputMouseSensitivity);

Verus/src/App/Settings.h

@@ -7,7 +7,7 @@ namespace verus
 		class Limits
 		{
 		public:
-			int _d3d12_dhCbvSrvUavCapacity = 10000;
+			int _d3d12_dhViewsCapacity = 10000;
 			int _d3d12_dhSamplersCapacity = 500;
 			int _mesh_ubPerFrameCapacity = 100;
 			int _mesh_ubPerMaterialFSCapacity = 1000;
@@ -64,6 +64,7 @@ namespace verus
 			int           _gapi = 0;
 			int           _gpuAnisotropyLevel = 4;
 			int           _gpuAntialiasingLevel = 0;
+			bool          _gpuTessellation = true;
 			int	          _gpuTextureLodLevel = 0;
 			bool          _gpuTrilinearFilter = false;
 			float         _inputMouseSensitivity = 1;

Verus/src/CGI/BaseCommandBuffer.h

@@ -7,6 +7,8 @@ namespace verus
 		class BaseCommandBuffer : public Object, public Scheduled
 		{
 		protected:
+			Vector4 _viewportSize = Vector4(0);
+
 			BaseCommandBuffer() = default;
 			virtual ~BaseCommandBuffer() = default;
 
@@ -17,7 +19,7 @@ namespace verus
 			virtual void Begin() = 0;
 			virtual void End() = 0;
 
-			virtual void BeginRenderPass(int renderPassHandle, int framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor = true) = 0;
+			virtual void BeginRenderPass(RPHandle renderPassHandle, FBHandle framebufferHandle, std::initializer_list<Vector4> ilClearValues, bool setViewportAndScissor = true) = 0;
 			virtual void NextSubpass() = 0;
 			virtual void EndRenderPass() = 0;
 
@@ -29,7 +31,7 @@ namespace verus
 			virtual void SetScissor(std::initializer_list<Vector4> il) = 0;
 			virtual void SetBlendConstants(const float* p) = 0;
 
-			virtual bool BindDescriptors(ShaderPtr shader, int setNumber, int complexSetHandle = -1) = 0;
+			virtual bool BindDescriptors(ShaderPtr shader, int setNumber, CSHandle complexSetHandle = CSHandle()) = 0;
 			virtual void PushConstants(ShaderPtr shader, int offset, int size, const void* p, ShaderStageFlags stageFlags = ShaderStageFlags::vs_fs) = 0;
 
 			virtual void PipelineImageMemoryBarrier(TexturePtr tex, ImageLayout oldLayout, ImageLayout newLayout,
@@ -39,6 +41,8 @@ namespace verus
 			virtual void DrawIndexed(int indexCount, int instanceCount = 1, int firstIndex = 0, int vertexOffset = 0, int firstInstance = 0) = 0;
 
 			virtual void Dispatch(int groupCountX, int groupCountY, int groupCountZ = 1) = 0;
+
+			RcVector4 GetViewportSize() const { return _viewportSize; }
 		};
 		VERUS_TYPEDEFS(BaseCommandBuffer);
 

Verus/src/CGI/BaseGeometry.cpp

@@ -3,7 +3,7 @@
 using namespace verus;
 using namespace verus::CGI;
 
-int BaseGeometry::GetInputElementDescCount(PcInputElementDesc p)
+int BaseGeometry::GetVertexInputAttrDescCount(PcVertexInputAttrDesc p)
 {
 	int i = 0;
 	while (p[i]._offset >= 0)
@@ -11,7 +11,7 @@ int BaseGeometry::GetInputElementDescCount(PcInputElementDesc p)
 	return i;
 }
 
-int BaseGeometry::GetBindingCount(PcInputElementDesc p)
+int BaseGeometry::GetBindingCount(PcVertexInputAttrDesc p)
 {
 	int count = 0;
 	int i = 0;

Verus/src/CGI/BaseGeometry.h

@@ -8,7 +8,7 @@ namespace verus
 
 		struct GeometryDesc
 		{
-			PcInputElementDesc _pInputElementDesc;
+			PcVertexInputAttrDesc _pVertexInputAttrDesc;
 			const int* _pStrides = nullptr;
 			UINT32                _dynBindingsMask = 0;
 			bool                  _32BitIndices = false;
@@ -35,8 +35,8 @@ namespace verus
 			virtual void CreateIndexBuffer(int count) = 0;
 			virtual void UpdateIndexBuffer(const void* p, BaseCommandBuffer* pCB = nullptr) = 0;
 
-			static int GetInputElementDescCount(PcInputElementDesc p);
-			static int GetBindingCount(PcInputElementDesc p);
+			static int GetVertexInputAttrDescCount(PcVertexInputAttrDesc p);
+			static int GetBindingCount(PcVertexInputAttrDesc p);
 
 			bool Has32BitIndices() const { return _32BitIndices; }
 		};

Verus/src/CGI/BasePipeline.h

@@ -14,7 +14,7 @@ namespace verus
 			PipelineRasterizationState _rasterizationState;
 			PrimitiveTopology          _topology = PrimitiveTopology::triangleList;
 			int                        _sampleCount = 1;
-			int                        _renderPassHandle = -1;
+			RPHandle                   _renderPassHandle;
 			int                        _subpass = 0;
 			int                        _multiViewport = 1; // See SV_ViewportArrayIndex semantic output by a geometry shader.
 			UINT32                     _vertexInputBindingsFilter = UINT32_MAX;
@@ -26,7 +26,7 @@ namespace verus
 			bool                       _compute = false; // Compute pipeline, use PipelineDesc(ShaderPtr, CSZ) to set this to true.
 
 			// What to draw (geo)? How to draw (shader)? Where to draw (render pass)?
-			PipelineDesc(GeometryPtr geo, ShaderPtr shader, CSZ branch, int renderPassHandle, int subpass = 0) :
+			PipelineDesc(GeometryPtr geo, ShaderPtr shader, CSZ branch, RPHandle renderPassHandle, int subpass = 0) :
 				_geometry(geo), _shader(shader), _shaderBranch(branch), _renderPassHandle(renderPassHandle), _subpass(subpass)
 			{
 				_colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_OFF;

Verus/src/CGI/BaseRenderer.cpp

@@ -45,6 +45,30 @@ void BaseRenderer::UpdateScheduled()
 		}), _vScheduled.end());
 }
 
+RPHandle BaseRenderer::CreateSimpleRenderPass(Format format, ImageLayout layout)
+{
+	return CreateRenderPass(
+		{
+			RP::Attachment("Attach", format).LoadOpDontCare().Layout(layout),
+		},
+		{
+			RP::Subpass("Sp0").Color({RP::Ref("Attach", ImageLayout::colorAttachment)}),
+		},
+		{});
+}
+
+RPHandle BaseRenderer::CreateShadowRenderPass(Format format)
+{
+	return CreateRenderPass(
+		{
+			RP::Attachment("Depth", format).LoadOpClear().Layout(ImageLayout::depthStencilReadOnly),
+		},
+		{
+			RP::Subpass("Sp0").Color({}).DepthStencil(RP::Ref("Depth", ImageLayout::depthStencilAttachment)),
+		},
+		{});
+}
+
 void BaseRenderer::SetAlphaBlendHelper(
 	CSZ sz,
 	int& colorBlendOp,

Verus/src/CGI/BaseRenderer.h

@@ -50,7 +50,7 @@ namespace verus
 			void Unschedule(PScheduled p);
 			void UpdateScheduled();
 
-			virtual void ImGuiInit(int renderPassHandle) = 0;
+			virtual void ImGuiInit(RPHandle renderPassHandle) = 0;
 			virtual void ImGuiRenderDrawData() = 0;
 
 			virtual void ResizeSwapChain() = 0;
@@ -77,10 +77,12 @@ namespace verus
 			virtual void DeleteShader(PBaseShader p) = 0;
 			virtual void DeleteTexture(PBaseTexture p) = 0;
 
-			virtual int CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD) = 0;
-			virtual int CreateFramebuffer(int renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex = -1) = 0;
-			virtual void DeleteRenderPass(int handle) = 0;
-			virtual void DeleteFramebuffer(int handle) = 0;
+			RPHandle CreateSimpleRenderPass(Format format, ImageLayout layout = ImageLayout::fsReadOnly);
+			RPHandle CreateShadowRenderPass(Format format);
+			virtual RPHandle CreateRenderPass(std::initializer_list<RP::Attachment> ilA, std::initializer_list<RP::Subpass> ilS, std::initializer_list<RP::Dependency> ilD) = 0;
+			virtual FBHandle CreateFramebuffer(RPHandle renderPassHandle, std::initializer_list<TexturePtr> il, int w, int h, int swapChainBufferIndex = -1) = 0;
+			virtual void DeleteRenderPass(RPHandle handle) = 0;
+			virtual void DeleteFramebuffer(FBHandle handle) = 0;
 
 			static void SetAlphaBlendHelper(
 				CSZ sz,

Verus/src/CGI/BaseShader.cpp

@@ -36,32 +36,24 @@ void BaseShader::Load(CSZ url)
 
 String BaseShader::Parse(
 	CSZ branchDesc,
-	RString entryVS,
-	RString entryHS,
-	RString entryDS,
-	RString entryGS,
-	RString entryFS,
-	RString entryCS,
+	RString entry,
+	String stageEntries[],
 	RString stages,
 	Vector<String>& vMacroName,
 	Vector<String>& vMacroValue,
 	CSZ prefix)
 {
 	VERUS_RT_ASSERT(branchDesc);
-	entryVS.clear();
-	entryHS.clear();
-	entryDS.clear();
-	entryGS.clear();
-	entryFS.clear();
-	entryCS.clear();
+	VERUS_FOR(i, +Stage::count)
+		stageEntries[i].clear();
 	stages = "(VF)";
 	vMacroName.clear();
 	vMacroValue.clear();
 
 	CSZ space = strchr(branchDesc, ' ');
 
-	String entry;
-	space ? entry.assign(branchDesc, space) : entry.assign(branchDesc);
+	String entryBranch;
+	space ? entryBranch.assign(branchDesc, space) : entryBranch.assign(branchDesc);
 
 	if (space)
 	{
@@ -97,44 +89,46 @@ String BaseShader::Parse(
 		} while (branchDesc);
 	}
 
-	const size_t colon = entry.find(":");
+	const size_t colon = entryBranch.find(":");
 	if (String::npos == colon)
 	{
-		entryVS = entry + "VS";
-		entryHS = entry + "HS";
-		entryDS = entry + "DS";
-		entryGS = entry + "GS";
-		entryFS = entry + "FS";
-		entryCS = entry + "CS";
-		return entry;
+		entry = entryBranch;
+		stageEntries[+Stage::vs] = entryBranch + "VS";
+		stageEntries[+Stage::hs] = entryBranch + "HS";
+		stageEntries[+Stage::ds] = entryBranch + "DS";
+		stageEntries[+Stage::gs] = entryBranch + "GS";
+		stageEntries[+Stage::fs] = entryBranch + "FS";
+		stageEntries[+Stage::cs] = entryBranch + "CS";
+		return entryBranch;
 	}
 	else
 	{
-		const String func = entry.substr(0, colon);
-		entryVS = func + "VS";
-		entryHS = func + "HS";
-		entryDS = func + "DS";
-		entryGS = func + "GS";
-		entryFS = func + "FS";
-		entryCS = func + "CS";
-		return entry.substr(colon + 1);
+		const String func = entryBranch.substr(0, colon);
+		entry = func;
+		stageEntries[+Stage::vs] = func + "VS";
+		stageEntries[+Stage::hs] = func + "HS";
+		stageEntries[+Stage::ds] = func + "DS";
+		stageEntries[+Stage::gs] = func + "GS";
+		stageEntries[+Stage::fs] = func + "FS";
+		stageEntries[+Stage::cs] = func + "CS";
+		return entryBranch.substr(colon + 1);
 	}
 }
 
 void BaseShader::TestParse()
 {
-	String entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages;
+	String entry, stageEntries[+Stage::count], stages;
 	Vector<String> vMacroName;
 	Vector<String> vMacroValue;
 
 	Parse("",
-		entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages, vMacroName, vMacroValue, "DEF_");
+		entry, stageEntries, stages, vMacroName, vMacroValue, "DEF_");
 	Parse("Foo",
-		entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages, vMacroName, vMacroValue, "DEF_");
+		entry, stageEntries, stages, vMacroName, vMacroValue, "DEF_");
 	Parse("Bar FOO BAR XYZ (VGF)",
-		entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages, vMacroName, vMacroValue, "DEF_");
+		entry, stageEntries, stages, vMacroName, vMacroValue, "DEF_");
 	Parse("FooBar FOO=1 BAR=xyz XYZ= XYZ2=",
-		entryVS, entryHS, entryDS, entryGS, entryFS, entryCS, stages, vMacroName, vMacroValue, "DEF_");
+		entry, stageEntries, stages, vMacroName, vMacroValue, "DEF_");
 }
 
 bool BaseShader::IsInIgnoreList(CSZ name) const

Verus/src/CGI/BaseShader.h

@@ -46,12 +46,8 @@ namespace verus
 			void Load(CSZ url);
 			static String Parse(
 				CSZ branchDesc,
-				RString entryVS,
-				RString entryHS,
-				RString entryDS,
-				RString entryGS,
-				RString entryFS,
-				RString entryCS,
+				RString entry,
+				String stageEntries[],
 				RString stages,
 				Vector<String>& vMacroName,
 				Vector<String>& vMacroValue,
@@ -61,8 +57,8 @@ namespace verus
 			virtual void CreateDescriptorSet(int setNumber, const void* pSrc, int size,
 				int capacity = 1, std::initializer_list<Sampler> il = {}, ShaderStageFlags stageFlags = ShaderStageFlags::vs_fs) = 0;
 			virtual void CreatePipelineLayout() = 0;
-			virtual int BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips = nullptr) = 0;
-			virtual void FreeDescriptorSet(int& complexSetHandle) = 0;
+			virtual CSHandle BindDescriptorSetTextures(int setNumber, std::initializer_list<TexturePtr> il, const int* pMips = nullptr) = 0;
+			virtual void FreeDescriptorSet(CSHandle& complexSetHandle) = 0;
 
 			virtual void BeginBindDescriptors() = 0;
 			virtual void EndBindDescriptors() = 0;

Verus/src/CGI/BaseTexture.cpp

@@ -28,8 +28,10 @@ void BaseTexture::LoadDDS(CSZ url, int texturePart)
 	IO::Async::I().Load(url, this, IO::Async::TaskDesc(false, false, texturePart));
 }
 
-void BaseTexture::LoadDDS(RcBlob blob)
+void BaseTexture::LoadDDS(CSZ url, RcBlob blob)
 {
+	_name = url;
+
 	int offset = sizeof(IO::DDSHeader);
 	IO::DDSHeader header;
 	memcpy(&header, blob._p, sizeof(header));
@@ -70,6 +72,7 @@ void BaseTexture::LoadDDS(RcBlob blob)
 		desc._width = header._width >> lod;
 		desc._height = header._height >> lod;
 		desc._mipLevels = header._mipMapCount - lod;
+		desc._pSamplerDesc = _desc._pSamplerDesc;
 
 		Init(desc);
 
@@ -97,6 +100,7 @@ void BaseTexture::LoadDDS(RcBlob blob)
 			desc._width = header._width;
 			desc._height = header._height;
 			desc._mipLevels = header._mipMapCount;
+			desc._pSamplerDesc = _desc._pSamplerDesc;
 
 			Init(desc);
 
@@ -118,6 +122,7 @@ void BaseTexture::LoadDDS(RcBlob blob)
 			desc._width = header._width;
 			desc._height = header._height;
 			desc._mipLevels = header._mipMapCount;
+			desc._pSamplerDesc = _desc._pSamplerDesc;
 
 			Init(desc);
 
@@ -151,6 +156,7 @@ void BaseTexture::LoadDDS(RcBlob blob)
 			desc._width = header._width;
 			desc._height = header._height;
 			desc._mipLevels = header._mipMapCount;
+			desc._pSamplerDesc = _desc._pSamplerDesc;
 
 			Init(desc);
 
@@ -202,10 +208,17 @@ void BaseTexture::LoadDDSArray(CSZ* urls)
 		const BYTE* pData = vData.data();
 		const size_t size = vData.size();
 
+		int offset = sizeof(IO::DDSHeader);
 		IO::DDSHeader header;
 		memcpy(&header, pData, sizeof(header));
 		if (!header.Validate())
 			throw VERUS_RUNTIME_ERROR << "Invalid DDS header: " << _name;
+		IO::DDSHeaderDXT10 header10;
+		if (header.IsDXT10())
+		{
+			memcpy(&header10, pData + offset, sizeof(header10));
+			offset += sizeof(IO::DDSHeaderDXT10);
+		}
 
 		VERUS_QREF_CONST_SETTINGS;
 
@@ -240,9 +253,12 @@ void BaseTexture::LoadDDSArray(CSZ* urls)
 				desc._format = SelectFormat(Format::unormBC2, Format::srgbBC2);
 			else if (header.IsBC3())
 				desc._format = SelectFormat(Format::unormBC3, Format::srgbBC3);
+			else if (header10.IsBC7())
+				desc._format = SelectFormat(Format::unormBC7, Format::srgbBC7);
 			desc._width = w;
 			desc._height = h;
 			desc._mipLevels = header._mipMapCount - lod;
+			desc._pSamplerDesc = _desc._pSamplerDesc;
 
 			if (init)
 			{
@@ -250,7 +266,6 @@ void BaseTexture::LoadDDSArray(CSZ* urls)
 				Init(desc);
 			}
 
-			int offset = sizeof(header);
 			VERUS_U_FOR(i, header._mipMapCount)
 			{
 				UINT32 w = header._width >> i;
@@ -274,7 +289,7 @@ void BaseTexture::LoadDDSArray(CSZ* urls)
 
 void BaseTexture::Async_Run(CSZ url, RcBlob blob)
 {
-	LoadDDS(blob);
+	LoadDDS(url, blob);
 }
 
 int BaseTexture::FormatToBytesPerPixel(Format format)
@@ -307,6 +322,20 @@ int BaseTexture::FormatToBytesPerPixel(Format format)
 	return 0;
 }
 
+bool BaseTexture::IsSRGBFormat(Format format)
+{
+	switch (format)
+	{
+	case Format::srgbR8G8B8A8: return true;
+	case Format::srgbB8G8R8A8: return true;
+	case Format::srgbBC1:      return true;
+	case Format::srgbBC2:      return true;
+	case Format::srgbBC3:      return true;
+	case Format::srgbBC7:      return true;
+	}
+	return false;
+}
+
 bool BaseTexture::IsBC(Format format)
 {
 	return format >= Format::unormBC1 && format <= Format::srgbBC7;
@@ -329,11 +358,12 @@ void TexturePtr::Init(RcTextureDesc desc)
 	VERUS_QREF_RENDERER;
 	VERUS_RT_ASSERT(!_p);
 	_p = renderer->InsertTexture();
+	_p->SetSamplerDesc(desc._pSamplerDesc);
 	if (desc._flags & TextureDesc::Flags::sync)
 	{
-		Vector<BYTE> v;
-		IO::FileSystem::LoadResource(desc._url, v);
-		_p->LoadDDS(Blob(v.data(), v.size()));
+		Vector<BYTE> vData;
+		IO::FileSystem::LoadResource(desc._url, vData);
+		_p->LoadDDS(desc._url, Blob(vData.data(), vData.size()));
 	}
 	else
 	{

Verus/src/CGI/BaseTexture.h

@@ -32,11 +32,12 @@ namespace verus
 				none = 0,
 				colorAttachment = (1 << 0),
 				inputAttachment = (1 << 1),
-				depthSampled = (1 << 2),
-				anyShaderResource = (1 << 3), // Will use xsReadOnly as main layout.
-				generateMips = (1 << 4), // Allows GenerateMips calls.
-				forceArrayTexture = (1 << 5), // Create array texture even if arrayLayers=1.
-				sync = (1 << 6)
+				depthSampledR = (1 << 2),
+				depthSampledW = (1 << 3),
+				anyShaderResource = (1 << 4), // Will use xsReadOnly as main layout.
+				generateMips = (1 << 5), // Allows GenerateMips calls.
+				forceArrayTexture = (1 << 6), // Create array texture even if arrayLayers=1.
+				sync = (1 << 7)
 			};
 
 			Vector4       _clearValue = Vector4(0);
@@ -98,8 +99,10 @@ namespace verus
 			RcVector4 GetSize() const { return _size; }
 			bool IsSRGB() const;
 
+			void SetSamplerDesc(PcSamplerDesc pSamplerDesc) { _desc._pSamplerDesc = pSamplerDesc; }
+
 			void LoadDDS(CSZ url, int texturePart = 0);
-			void LoadDDS(RcBlob blob);
+			void LoadDDS(CSZ url, RcBlob blob);
 			void LoadDDSArray(CSZ* urls);
 
 			virtual void Async_Run(CSZ url, RcBlob blob) override;
@@ -109,6 +112,7 @@ namespace verus
 			virtual void GenerateMips(BaseCommandBuffer* pCB = nullptr) = 0;
 
 			static int FormatToBytesPerPixel(Format format);
+			static bool IsSRGBFormat(Format format);
 			static bool IsBC(Format format);
 			static bool Is4BitsBC(Format format);
 			static bool IsDepthFormat(Format format);

Verus/src/CGI/CGI.h

@@ -12,6 +12,7 @@
 #include "BaseCommandBuffer.h"
 #include "BaseRenderer.h"
 
+#include "TextureRAM.h"
 #include "DebugDraw.h"
 #include "DeferredShading.h"
 #include "Renderer.h"

Verus/src/CGI/DebugDraw.cpp

@@ -25,13 +25,13 @@ void DebugDraw::Init()
 	_shader->CreatePipelineLayout();
 
 	GeometryDesc geoDesc;
-	const InputElementDesc ied[] =
+	const VertexInputAttrDesc viaDesc[] =
 	{
 		{0, offsetof(Vertex, _pos),   IeType::floats, 3, IeUsage::position, 0},
 		{0, offsetof(Vertex, _color), IeType::ubytes, 4, IeUsage::color, 0},
-		InputElementDesc::End()
+		VertexInputAttrDesc::End()
 	};
-	geoDesc._pInputElementDesc = ied;
+	geoDesc._pVertexInputAttrDesc = viaDesc;
 	const int strides[] = { sizeof(Vertex), 0 };
 	geoDesc._pStrides = strides;
 	geoDesc._dynBindingsMask = 0x1;
@@ -43,7 +43,7 @@ void DebugDraw::Init()
 	_offset = 0;
 
 	{
-		PipelineDesc pipeDesc(_geo, _shader, "#", renderer.GetRenderPass_SwapChainDepth());
+		PipelineDesc pipeDesc(_geo, _shader, "#", renderer.GetRenderPassHandle_SwapChainDepth());
 		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ALPHA;
 		pipeDesc._topology = PrimitiveTopology::pointList;
 		_pipe[PIPE_POINTS].Init(pipeDesc);
@@ -51,7 +51,7 @@ void DebugDraw::Init()
 		_pipe[PIPE_POINTS_NO_Z].Init(pipeDesc);
 	}
 	{
-		PipelineDesc pipeDesc(_geo, _shader, "#", renderer.GetRenderPass_SwapChainDepth());
+		PipelineDesc pipeDesc(_geo, _shader, "#", renderer.GetRenderPassHandle_SwapChainDepth());
 		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ALPHA;
 		pipeDesc._topology = PrimitiveTopology::lineList;
 		_pipe[PIPE_LINES].Init(pipeDesc);
@@ -59,7 +59,7 @@ void DebugDraw::Init()
 		_pipe[PIPE_LINES_NO_Z].Init(pipeDesc);
 	}
 	{
-		PipelineDesc pipeDesc(_geo, _shader, "#", renderer.GetRenderPass_SwapChainDepth());
+		PipelineDesc pipeDesc(_geo, _shader, "#", renderer.GetRenderPassHandle_SwapChainDepth());
 		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ALPHA;
 		pipeDesc._rasterizationState._polygonMode = PolygonMode::line;
 		pipeDesc._topology = PrimitiveTopology::triangleList;

Verus/src/CGI/DebugDraw.h

@@ -16,7 +16,7 @@ namespace verus
 				PIPE_LINES_NO_Z,
 				PIPE_POLY,
 				PIPE_POLY_NO_Z,
-				PIPE_MAX
+				PIPE_COUNT
 			};
 
 		public:
@@ -38,7 +38,7 @@ namespace verus
 
 			GeometryPwn              _geo;
 			ShaderPwn                _shader;
-			PipelinePwns<PIPE_MAX> _pipe;
+			PipelinePwns<PIPE_COUNT> _pipe;
 			Vector<Vertex>           _vDynamicBuffer;
 			UINT64                   _currentFrame = UINT64_MAX;
 			const int                _maxVerts = 0x10000;

Verus/src/CGI/DeferredShading.cpp

@@ -27,12 +27,11 @@ void DeferredShading::Init()
 	VERUS_QREF_CONST_SETTINGS;
 	VERUS_QREF_RENDERER;
 
-	_rp = renderer->CreateRenderPass(
+	_rph = renderer->CreateRenderPass(
 		{
 			RP::Attachment("GBuffer0", Format::srgbR8G8B8A8).LoadOpClear().Layout(ImageLayout::fsReadOnly),
-			RP::Attachment("GBuffer1", Format::floatR32).LoadOpClear().Layout(ImageLayout::fsReadOnly),
-			RP::Attachment("GBuffer2", Format::unormR10G10B10A2).LoadOpClear().Layout(ImageLayout::fsReadOnly),
-			RP::Attachment("GBuffer3", Format::unormR8G8B8A8).LoadOpClear().Layout(ImageLayout::fsReadOnly),
+			RP::Attachment("GBuffer1", Format::unormR10G10B10A2).LoadOpClear().Layout(ImageLayout::fsReadOnly),
+			RP::Attachment("GBuffer2", Format::unormR8G8B8A8).LoadOpClear().Layout(ImageLayout::fsReadOnly),
 			RP::Attachment("LightAccDiff", Format::floatR11G11B10).LoadOpClear().Layout(ImageLayout::fsReadOnly),
 			RP::Attachment("LightAccSpec", Format::floatR11G11B10).LoadOpClear().Layout(ImageLayout::fsReadOnly),
 			RP::Attachment("Depth", Format::unormD24uintS8).LoadOpClear().Layout(ImageLayout::depthStencilAttachment),
@@ -42,62 +41,85 @@ void DeferredShading::Init()
 				{
 					RP::Ref("GBuffer0", ImageLayout::colorAttachment),
 					RP::Ref("GBuffer1", ImageLayout::colorAttachment),
-					RP::Ref("GBuffer2", ImageLayout::colorAttachment),
-					RP::Ref("GBuffer3", ImageLayout::colorAttachment)
+					RP::Ref("GBuffer2", ImageLayout::colorAttachment)
 				}).DepthStencil(RP::Ref("Depth", ImageLayout::depthStencilAttachment)),
 			RP::Subpass("Sp1").Input(
 				{
 					RP::Ref("GBuffer0", ImageLayout::fsReadOnly),
 					RP::Ref("GBuffer1", ImageLayout::fsReadOnly),
 					RP::Ref("GBuffer2", ImageLayout::fsReadOnly),
-					RP::Ref("GBuffer3", ImageLayout::fsReadOnly)
+					RP::Ref("Depth", ImageLayout::depthStencilReadOnly)
 				}).Color(
 				{
 					RP::Ref("LightAccDiff", ImageLayout::colorAttachment),
 					RP::Ref("LightAccSpec", ImageLayout::colorAttachment),
-				}).DepthStencil(RP::Ref("Depth", ImageLayout::depthStencilAttachment))
+				}).DepthStencil(RP::Ref("Depth", ImageLayout::depthStencilReadOnly))
+		},
+		{
+			RP::Dependency("Sp0", "Sp1").Mode(1)
+		});
+	_rphCompose = renderer->CreateRenderPass(
+		{
+			RP::Attachment("Composed", Format::floatR11G11B10).LoadOpDontCare().Layout(ImageLayout::fsReadOnly),
+			RP::Attachment("Depth", Format::unormD24uintS8).Layout(ImageLayout::depthStencilAttachment),
+		},
+		{
+			RP::Subpass("Sp0").Color(
+				{
+					RP::Ref("Composed", ImageLayout::colorAttachment)
+				}).DepthStencil(RP::Ref("Depth", ImageLayout::depthStencilReadOnly)),
+			RP::Subpass("Sp1").Color(
+				{
+					RP::Ref("Composed", ImageLayout::colorAttachment)
+				}).DepthStencil(RP::Ref("Depth", ImageLayout::depthStencilAttachment)),
 		},
 		{
 			RP::Dependency("Sp0", "Sp1").Mode(1)
 		});
 
-	_shader[S_LIGHT].Init("[Shaders]:DS.hlsl");
-	_shader[S_LIGHT]->CreateDescriptorSet(0, &s_ubPerFrame, sizeof(s_ubPerFrame));
-	_shader[S_LIGHT]->CreateDescriptorSet(1, &s_ubTexturesFS, sizeof(s_ubTexturesFS), 1,
+	_shader[SHADER_LIGHT].Init("[Shaders]:DS.hlsl");
+	_shader[SHADER_LIGHT]->CreateDescriptorSet(0, &s_ubPerFrame, sizeof(s_ubPerFrame));
+	_shader[SHADER_LIGHT]->CreateDescriptorSet(1, &s_ubTexturesFS, sizeof(s_ubTexturesFS), 1,
 		{
 			Sampler::input,
 			Sampler::input,
 			Sampler::input,
 			Sampler::input,
 			Sampler::shadow,
-			Sampler::nearest2D
+			Sampler::nearestMipN
 		}, CGI::ShaderStageFlags::fs);
-	_shader[S_LIGHT]->CreateDescriptorSet(2, &s_ubPerMeshVS, sizeof(s_ubPerMeshVS), 1000, {}, CGI::ShaderStageFlags::vs);
-	_shader[S_LIGHT]->CreateDescriptorSet(3, &s_ubShadowFS, sizeof(s_ubShadowFS), 1000, {}, CGI::ShaderStageFlags::fs);
-	_shader[S_LIGHT]->CreateDescriptorSet(4, &s_ubPerObject, sizeof(s_ubPerObject), 0);
-	_shader[S_LIGHT]->CreatePipelineLayout();
+	_shader[SHADER_LIGHT]->CreateDescriptorSet(2, &s_ubPerMeshVS, sizeof(s_ubPerMeshVS), 1000, {}, CGI::ShaderStageFlags::vs);
+	_shader[SHADER_LIGHT]->CreateDescriptorSet(3, &s_ubShadowFS, sizeof(s_ubShadowFS), 1000, {}, CGI::ShaderStageFlags::fs);
+	_shader[SHADER_LIGHT]->CreateDescriptorSet(4, &s_ubPerObject, sizeof(s_ubPerObject), 0);
+	_shader[SHADER_LIGHT]->CreatePipelineLayout();
 
-	_shader[S_COMPOSE].Init("[Shaders]:DS_Compose.hlsl");
-	_shader[S_COMPOSE]->CreateDescriptorSet(0, &s_ubComposeVS, sizeof(s_ubComposeVS), 1, {}, CGI::ShaderStageFlags::vs);
-	_shader[S_COMPOSE]->CreateDescriptorSet(1, &s_ubComposeFS, sizeof(s_ubComposeFS), 1,
+	_shader[SHADER_COMPOSE].Init("[Shaders]:DS_Compose.hlsl");
+	_shader[SHADER_COMPOSE]->CreateDescriptorSet(0, &s_ubComposeVS, sizeof(s_ubComposeVS), 2, {}, CGI::ShaderStageFlags::vs);
+	_shader[SHADER_COMPOSE]->CreateDescriptorSet(1, &s_ubComposeFS, sizeof(s_ubComposeFS), 2,
 		{
-			Sampler::nearest2D,
-			Sampler::nearest2D,
-			Sampler::nearest2D,
-			Sampler::nearest2D,
-			Sampler::nearest2D,
-			Sampler::nearest2D
+			Sampler::nearestMipN,
+			Sampler::nearestMipN,
+			Sampler::nearestMipN,
+			Sampler::nearestMipN,
+			Sampler::nearestMipN,
+			Sampler::nearestMipN
 		}, CGI::ShaderStageFlags::fs);
-	_shader[S_COMPOSE]->CreatePipelineLayout();
+	_shader[SHADER_COMPOSE]->CreatePipelineLayout();
 
 	{
-		PipelineDesc pipeDesc(renderer.GetGeoQuad(), _shader[S_COMPOSE], "#", renderer.GetRenderPass_SwapChainDepth());
+		PipelineDesc pipeDesc(renderer.GetGeoQuad(), _shader[SHADER_COMPOSE], "#Compose", _rphCompose);
 		pipeDesc._topology = PrimitiveTopology::triangleStrip;
 		pipeDesc._depthTestEnable = false;
 		_pipe[PIPE_COMPOSE].Init(pipeDesc);
 	}
 	{
-		PipelineDesc pipeDesc(renderer.GetGeoQuad(), renderer.GetShaderQuad(), "#", renderer.GetRenderPass_SwapChainDepth());
+		PipelineDesc pipeDesc(renderer.GetGeoQuad(), _shader[SHADER_COMPOSE], "#ToneMapping", renderer.GetRenderPassHandle_SwapChainDepth());
+		pipeDesc._topology = PrimitiveTopology::triangleStrip;
+		pipeDesc._depthTestEnable = false;
+		_pipe[PIPE_TONE_MAPPING].Init(pipeDesc);
+	}
+	{
+		PipelineDesc pipeDesc(renderer.GetGeoQuad(), renderer.GetShaderQuad(), "#", renderer.GetRenderPassHandle_SwapChainDepth());
 		pipeDesc._topology = PrimitiveTopology::triangleStrip;
 		pipeDesc._depthTestEnable = false;
 		_pipe[PIPE_QUAD].Init(pipeDesc);
@@ -121,39 +143,33 @@ void DeferredShading::InitGBuffers(int w, int h)
 	_tex[TEX_GBUFFER_0].Init(texDesc);
 
 	// GB1:
-	texDesc._clearValue = Vector4(1);
-	texDesc._format = Format::floatR32;
+	texDesc._clearValue = Vector4(0.5f, 0.5f, 0.25f, 0.5f);
+	texDesc._format = Format::unormR10G10B10A2;
 	texDesc._width = w;
 	texDesc._height = h;
 	texDesc._flags = TextureDesc::Flags::colorAttachment | TextureDesc::Flags::inputAttachment;
 	_tex[TEX_GBUFFER_1].Init(texDesc);
 
 	// GB2:
-	texDesc._clearValue = Vector4(0.5f, 0.5f, 0.25f, 0.5f);
-	texDesc._format = Format::unormR10G10B10A2;
-	texDesc._width = w;
-	texDesc._height = h;
-	texDesc._flags = TextureDesc::Flags::colorAttachment | TextureDesc::Flags::inputAttachment;
-	_tex[TEX_GBUFFER_2].Init(texDesc);
-
-	// GB3:
 	texDesc._clearValue = Vector4(1, 1, 1, 1);
 	texDesc._format = Format::unormR8G8B8A8;
 	texDesc._width = w;
 	texDesc._height = h;
 	texDesc._flags = TextureDesc::Flags::colorAttachment | TextureDesc::Flags::inputAttachment;
-	_tex[TEX_GBUFFER_3].Init(texDesc);
+	_tex[TEX_GBUFFER_2].Init(texDesc);
 }
 
 void DeferredShading::InitByAtmosphere(TexturePtr texShadow)
 {
+	VERUS_QREF_RENDERER;
+
 	_texShadowAtmo = texShadow;
-	_cshLight = _shader[S_LIGHT]->BindDescriptorSetTextures(1,
+	_cshLight = _shader[SHADER_LIGHT]->BindDescriptorSetTextures(1,
 		{
 			_tex[TEX_GBUFFER_0],
 			_tex[TEX_GBUFFER_1],
 			_tex[TEX_GBUFFER_2],
-			_tex[TEX_GBUFFER_3],
+			renderer.GetTexDepthStencil(),
 			_texShadowAtmo,
 			_texShadowAtmo
 		});
@@ -170,16 +186,18 @@ void DeferredShading::OnSwapChainResized(bool init, bool done)
 
 	if (done)
 	{
-		_shader[S_LIGHT]->FreeDescriptorSet(_cshLight);
+		_shader[SHADER_LIGHT]->FreeDescriptorSet(_cshLight);
 		VERUS_FOR(i, VERUS_COUNT_OF(_cshQuad))
 			renderer.GetShaderQuad()->FreeDescriptorSet(_cshQuad[i]);
-		_shader[S_COMPOSE]->FreeDescriptorSet(_cshCompose);
+		_shader[SHADER_COMPOSE]->FreeDescriptorSet(_cshToneMapping);
+		_shader[SHADER_COMPOSE]->FreeDescriptorSet(_cshCompose);
 
-		renderer->DeleteFramebuffer(_fb);
+		renderer->DeleteFramebuffer(_fbhCompose);
+		renderer->DeleteFramebuffer(_fbh);
 
+		_tex[TEX_COMPOSED].Done();
 		_tex[TEX_LIGHT_ACC_SPEC].Done();
 		_tex[TEX_LIGHT_ACC_DIFF].Done();
-		_tex[TEX_GBUFFER_3].Done();
 		_tex[TEX_GBUFFER_2].Done();
 		_tex[TEX_GBUFFER_1].Done();
 		_tex[TEX_GBUFFER_0].Done();
@@ -198,34 +216,47 @@ void DeferredShading::OnSwapChainResized(bool init, bool done)
 		texDesc._height = renderer.GetSwapChainHeight();
 		texDesc._flags = TextureDesc::Flags::colorAttachment;
 		_tex[TEX_LIGHT_ACC_DIFF].Init(texDesc);
-		texDesc._format = Format::floatR11G11B10;
-		texDesc._width = renderer.GetSwapChainWidth();
-		texDesc._height = renderer.GetSwapChainHeight();
-		texDesc._flags = TextureDesc::Flags::colorAttachment;
 		_tex[TEX_LIGHT_ACC_SPEC].Init(texDesc);
+		_tex[TEX_COMPOSED].Init(texDesc);
 
-		_fb = renderer->CreateFramebuffer(_rp,
+		_fbh = renderer->CreateFramebuffer(_rph,
 			{
 				_tex[TEX_GBUFFER_0],
 				_tex[TEX_GBUFFER_1],
 				_tex[TEX_GBUFFER_2],
-				_tex[TEX_GBUFFER_3],
 				_tex[TEX_LIGHT_ACC_DIFF],
 				_tex[TEX_LIGHT_ACC_SPEC],
 				renderer.GetTexDepthStencil()
 			},
 			renderer.GetSwapChainWidth(),
 			renderer.GetSwapChainHeight());
+		_fbhCompose = renderer->CreateFramebuffer(_rphCompose,
+			{
+				_tex[TEX_COMPOSED],
+				renderer.GetTexDepthStencil()
+			},
+			renderer.GetSwapChainWidth(),
+			renderer.GetSwapChainHeight());
 
-		_cshCompose = _shader[S_COMPOSE]->BindDescriptorSetTextures(1,
+		_cshCompose = _shader[SHADER_COMPOSE]->BindDescriptorSetTextures(1,
 			{
 				_tex[TEX_GBUFFER_0],
 				_tex[TEX_GBUFFER_1],
 				_tex[TEX_GBUFFER_2],
-				_tex[TEX_GBUFFER_3],
+				renderer.GetTexDepthStencil(),
 				_tex[TEX_LIGHT_ACC_DIFF],
 				_tex[TEX_LIGHT_ACC_SPEC]
 			});
+		_cshToneMapping = _shader[SHADER_COMPOSE]->BindDescriptorSetTextures(1,
+			{
+				_tex[TEX_COMPOSED],
+				_tex[TEX_GBUFFER_1],
+				_tex[TEX_GBUFFER_2],
+				_tex[TEX_GBUFFER_0],
+				_tex[TEX_LIGHT_ACC_DIFF],
+				_tex[TEX_LIGHT_ACC_SPEC]
+			});
+
 		VERUS_FOR(i, VERUS_COUNT_OF(_cshQuad))
 			_cshQuad[i] = renderer.GetShaderQuad()->BindDescriptorSetTextures(1, { _tex[TEX_GBUFFER_0 + i] });
 
@@ -313,12 +344,11 @@ void DeferredShading::BeginGeometryPass(bool onlySetRT, bool spriteBaking)
 		_activeGeometryPass = true;
 		_frame = renderer.GetFrameCount();
 
-		renderer.GetCommandBuffer()->BeginRenderPass(_rp, _fb,
+		renderer.GetCommandBuffer()->BeginRenderPass(_rph, _fbh,
 			{
 				_tex[TEX_GBUFFER_0]->GetClearValue(),
 				_tex[TEX_GBUFFER_1]->GetClearValue(),
 				_tex[TEX_GBUFFER_2]->GetClearValue(),
-				_tex[TEX_GBUFFER_3]->GetClearValue(),
 				_tex[TEX_LIGHT_ACC_DIFF]->GetClearValue(),
 				_tex[TEX_LIGHT_ACC_SPEC]->GetClearValue(),
 				renderer.GetTexDepthStencil()->GetClearValue()
@@ -342,7 +372,7 @@ bool DeferredShading::BeginLightingPass()
 	_activeLightingPass = true;
 
 	renderer.GetCommandBuffer()->NextSubpass();
-	_shader[S_LIGHT]->BeginBindDescriptors();
+	_shader[SHADER_LIGHT]->BeginBindDescriptors();
 
 	if (!IsLoaded())
 		return false;
@@ -355,7 +385,7 @@ bool DeferredShading::BeginLightingPass()
 		Scene::RDeferredLights dl = helpers.GetDeferredLights();
 
 		{
-			PipelineDesc pipeDesc(dl.Get(CGI::LightType::dir).GetGeometry(), _shader[S_LIGHT], "#InstancedDir", _rp, 1);
+			PipelineDesc pipeDesc(dl.Get(CGI::LightType::dir).GetGeometry(), _shader[SHADER_LIGHT], "#InstancedDir", _rph, 1);
 			pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ADD;
 			pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_ADD;
 			pipeDesc._vertexInputBindingsFilter = (1 << 0) | (1 << 4);
@@ -363,7 +393,7 @@ bool DeferredShading::BeginLightingPass()
 			_pipe[PIPE_INSTANCED_DIR].Init(pipeDesc);
 		}
 		{
-			PipelineDesc pipeDesc(dl.Get(CGI::LightType::omni).GetGeometry(), _shader[S_LIGHT], "#InstancedOmni", _rp, 1);
+			PipelineDesc pipeDesc(dl.Get(CGI::LightType::omni).GetGeometry(), _shader[SHADER_LIGHT], "#InstancedOmni", _rph, 1);
 			pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ADD;
 			pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_ADD;
 			pipeDesc._rasterizationState._cullMode = CullMode::front;
@@ -373,7 +403,7 @@ bool DeferredShading::BeginLightingPass()
 			_pipe[PIPE_INSTANCED_OMNI].Init(pipeDesc);
 		}
 		{
-			PipelineDesc pipeDesc(dl.Get(CGI::LightType::spot).GetGeometry(), _shader[S_LIGHT], "#InstancedSpot", _rp, 1);
+			PipelineDesc pipeDesc(dl.Get(CGI::LightType::spot).GetGeometry(), _shader[SHADER_LIGHT], "#InstancedSpot", _rph, 1);
 			pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ADD;
 			pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_ADD;
 			pipeDesc._rasterizationState._cullMode = CullMode::front;
@@ -394,28 +424,71 @@ void DeferredShading::EndLightingPass()
 	VERUS_RT_ASSERT(!_activeGeometryPass && _activeLightingPass);
 	_activeLightingPass = false;
 
-	_shader[S_LIGHT]->EndBindDescriptors();
+	_shader[SHADER_LIGHT]->EndBindDescriptors();
 	renderer.GetCommandBuffer()->EndRenderPass();
 }
 
-void DeferredShading::Compose(RcVector4 bgColor)
+void DeferredShading::BeginCompose(RcVector4 bgColor)
 {
 	VERUS_QREF_RENDERER;
 	VERUS_QREF_SM;
+	VERUS_QREF_ATMO;
+
+	renderer.GetCommandBuffer()->BeginRenderPass(_rphCompose, _fbhCompose,
+		{
+			_tex[TEX_COMPOSED]->GetClearValue(),
+			renderer.GetTexDepthStencil()->GetClearValue()
+		});
 
 	renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_COMPOSE]);
 
-	_shader[S_COMPOSE]->BeginBindDescriptors();
+	const Matrix4 matInvVP = VMath::inverse(sm.GetCamera()->GetMatrixVP());
+
+	_shader[SHADER_COMPOSE]->BeginBindDescriptors();
 	s_ubComposeVS._matW = Math::QuadMatrix().UniformBufferFormat();
 	s_ubComposeVS._matV = Math::ToUVMatrix().UniformBufferFormat();
 	s_ubComposeFS._matInvV = sm.GetCamera()->GetMatrixVi().UniformBufferFormat();
-	s_ubComposeFS._colorAmbient = Vector4(0.1f, 0.1f, 0.2f, 1).GLM();
+	s_ubComposeFS._matInvVP = matInvVP.UniformBufferFormat();
+	s_ubComposeFS._colorAmbient = float4(atmo.GetAmbientColor().GLM(), 0);
 	s_ubComposeFS._colorBackground = bgColor.GLM();
+	s_ubComposeFS._fogColor = Vector4(atmo.GetFogColor(), atmo.GetFogDensity()).GLM();
+	s_ubComposeFS._zNearFarEx = sm.GetCamera()->GetZNearFarEx().GLM();
+	memcpy(&s_ubComposeFS._toneMappingConfig.x, &_toneMappingConfig, sizeof(_toneMappingConfig));
+	renderer.GetCommandBuffer()->BindDescriptors(_shader[SHADER_COMPOSE], 0);
+	renderer.GetCommandBuffer()->BindDescriptors(_shader[SHADER_COMPOSE], 1, _cshCompose);
+	_shader[SHADER_COMPOSE]->EndBindDescriptors();
+
+	renderer.DrawQuad();
+
+	renderer.GetCommandBuffer()->NextSubpass();
+}
+
+void DeferredShading::EndCompose()
+{
+	VERUS_QREF_RENDERER;
+
+	renderer.GetCommandBuffer()->EndRenderPass();
+}
+
+void DeferredShading::ToneMapping()
+{
+	VERUS_QREF_RENDERER;
+	VERUS_QREF_SM;
+	VERUS_QREF_ATMO;
+
+	renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_TONE_MAPPING]);
+
+	_shader[SHADER_COMPOSE]->BeginBindDescriptors();
+	s_ubComposeVS._matW = Math::QuadMatrix().UniformBufferFormat();
+	s_ubComposeVS._matV = Math::ToUVMatrix().UniformBufferFormat();
+	s_ubComposeFS._matInvV = sm.GetCamera()->GetMatrixVi().UniformBufferFormat();
+	s_ubComposeFS._colorAmbient = float4(atmo.GetAmbientColor().GLM(), 0);
 	s_ubComposeFS._fogColor = Vector4(0.5f, 0.5f, 0.5f, 0.002f).GLM();
 	s_ubComposeFS._zNearFarEx = sm.GetCamera()->GetZNearFarEx().GLM();
-	renderer.GetCommandBuffer()->BindDescriptors(_shader[S_COMPOSE], 0);
-	renderer.GetCommandBuffer()->BindDescriptors(_shader[S_COMPOSE], 1, _cshCompose);
-	_shader[S_COMPOSE]->EndBindDescriptors();
+	memcpy(&s_ubComposeFS._toneMappingConfig.x, &_toneMappingConfig, sizeof(_toneMappingConfig));
+	renderer.GetCommandBuffer()->BindDescriptors(_shader[SHADER_COMPOSE], 0);
+	renderer.GetCommandBuffer()->BindDescriptors(_shader[SHADER_COMPOSE], 1, _cshToneMapping);
+	_shader[SHADER_COMPOSE]->EndBindDescriptors();
 
 	renderer.DrawQuad();
 }
@@ -437,36 +510,38 @@ void DeferredShading::OnNewLightType(LightType type, bool wireframe)
 	VERUS_QREF_RENDERER;
 	VERUS_QREF_ATMO;
 
+	auto cb = renderer.GetCommandBuffer();
+
 	switch (type)
 	{
 	case CGI::LightType::dir:
 	{
-		renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_INSTANCED_DIR]);
+		cb->BindPipeline(_pipe[PIPE_INSTANCED_DIR]);
 	}
 	break;
 	case CGI::LightType::omni:
 	{
-		renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_INSTANCED_OMNI]);
+		cb->BindPipeline(_pipe[PIPE_INSTANCED_OMNI]);
 	}
 	break;
 	case CGI::LightType::spot:
 	{
-		renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_INSTANCED_SPOT]);
+		cb->BindPipeline(_pipe[PIPE_INSTANCED_SPOT]);
 	}
 	break;
 	}
 
-	renderer.GetCommandBuffer()->BindDescriptors(_shader[S_LIGHT], 0);
-	renderer.GetCommandBuffer()->BindDescriptors(_shader[S_LIGHT], 1, _cshLight);
+	cb->BindDescriptors(_shader[SHADER_LIGHT], 0);
+	cb->BindDescriptors(_shader[SHADER_LIGHT], 1, _cshLight);
 
 	// Shadow:
 	s_ubShadowFS._matSunShadow = atmo.GetShadowMap().GetShadowMatrixDS(0).UniformBufferFormat();
 	s_ubShadowFS._matSunShadowCSM1 = atmo.GetShadowMap().GetShadowMatrixDS(1).UniformBufferFormat();
 	s_ubShadowFS._matSunShadowCSM2 = atmo.GetShadowMap().GetShadowMatrixDS(2).UniformBufferFormat();
 	s_ubShadowFS._matSunShadowCSM3 = atmo.GetShadowMap().GetShadowMatrixDS(3).UniformBufferFormat();
-	s_ubShadowFS._shadowConfig = atmo.GetShadowMap().GetConfig().GLM();
+	memcpy(&s_ubShadowFS._shadowConfig, &atmo.GetShadowMap().GetConfig(), sizeof(s_ubShadowFS._shadowConfig));
 	s_ubShadowFS._splitRanges = atmo.GetShadowMap().GetSplitRanges().GLM();
-	renderer.GetCommandBuffer()->BindDescriptors(_shader[S_LIGHT], 3);
+	cb->BindDescriptors(_shader[SHADER_LIGHT], 3);
 }
 
 void DeferredShading::UpdateUniformBufferPerFrame()
@@ -486,7 +561,7 @@ void DeferredShading::UpdateUniformBufferPerFrame()
 void DeferredShading::BindDescriptorsPerMeshVS()
 {
 	VERUS_QREF_RENDERER;
-	renderer.GetCommandBuffer()->BindDescriptors(_shader[S_LIGHT], 2);
+	renderer.GetCommandBuffer()->BindDescriptors(_shader[SHADER_LIGHT], 2);
 }
 
 void DeferredShading::Load()

Verus/src/CGI/DeferredShading.h

@@ -17,11 +17,11 @@ namespace verus
 #include "../Shaders/DS.inc.hlsl"
 #include "../Shaders/DS_Compose.inc.hlsl"
 
-			enum S
+			enum SHADER
 			{
-				S_LIGHT,
-				S_COMPOSE,
-				S_MAX
+				SHADER_LIGHT,
+				SHADER_COMPOSE,
+				SHADER_COUNT
 			};
 
 			enum PIPE
@@ -30,19 +30,20 @@ namespace verus
 				PIPE_INSTANCED_OMNI,
 				PIPE_INSTANCED_SPOT,
 				PIPE_COMPOSE,
+				PIPE_TONE_MAPPING,
 				PIPE_QUAD,
-				PIPE_MAX
+				PIPE_COUNT
 			};
 
 			enum TEX
 			{
-				TEX_GBUFFER_0, // Albedo RGB, A = motion blur.
-				TEX_GBUFFER_1, // Depth.
-				TEX_GBUFFER_2, // RG - Normals in screen space, Emission, Metallicity.
-				TEX_GBUFFER_3, // LamScale, LamBias, Specular, Gloss.
+				TEX_GBUFFER_0, // Albedo, Spec.
+				TEX_GBUFFER_1, // RG - Normals, Emission, Motion.
+				TEX_GBUFFER_2, // LamScale, LamBias, Metallicity, Gloss.
 				TEX_LIGHT_ACC_DIFF,
 				TEX_LIGHT_ACC_SPEC,
-				TEX_MAX
+				TEX_COMPOSED,
+				TEX_COUNT
 			};
 
 			static UB_PerFrame   s_ubPerFrame;
@@ -53,21 +54,31 @@ namespace verus
 			static UB_ComposeVS  s_ubComposeVS;
 			static UB_ComposeFS  s_ubComposeFS;
 
-			ShaderPwns<S_MAX>      _shader;
-			PipelinePwns<PIPE_MAX> _pipe;
-			TexturePwns<TEX_MAX>   _tex;
+			ShaderPwns<SHADER_COUNT> _shader;
+			PipelinePwns<PIPE_COUNT> _pipe;
+			TexturePwns<TEX_COUNT>   _tex;
 			TexturePtr               _texShadowAtmo;
 			UINT64                   _frame = 0;
-			int                    _rp = -1;
-			int                    _fb = -1;
-			int                    _cshLight = -1;
-			int                    _cshCompose = -1;
-			int                    _cshQuad[6];
+			RPHandle                 _rph;
+			RPHandle                 _rphCompose;
+			FBHandle                 _fbh;
+			FBHandle                 _fbhCompose;
+			CSHandle                 _cshLight;
+			CSHandle                 _cshCompose;
+			CSHandle                 _cshToneMapping;
+			CSHandle                 _cshQuad[6];
 			bool                     _activeGeometryPass = false;
 			bool                     _activeLightingPass = false;
 			bool                     _async_initPipe = false;
 
 		public:
+			struct ToneMappingConfig
+			{
+				float _exposure = 0.005f;
+				float _filmicLook = 0.5f;
+				float _albedoScale = 0.5f;
+			} _toneMappingConfig;
+
 			DeferredShading();
 			~DeferredShading();
 
@@ -85,13 +96,17 @@ namespace verus
 			bool IsActiveGeometryPass() const { return _activeGeometryPass; }
 			bool IsActiveLightingPass() const { return _activeLightingPass; }
 
-			int GetRenderPassHandle() const { return _rp; }
+			RPHandle GetRenderPassHandle() const { return _rph; }
+			RPHandle GetRenderPassHandle_Compose() const { return _rphCompose; }
+			int GetSubpass_Compose() const { return 1; }
 
 			void BeginGeometryPass(bool onlySetRT = false, bool spriteBaking = false);
 			void EndGeometryPass(bool resetRT = false);
 			bool BeginLightingPass();
 			void EndLightingPass();
-			void Compose(RcVector4 bgColor = Vector4(0));
+			void BeginCompose(RcVector4 bgColor = Vector4(0));
+			void EndCompose();
+			void ToneMapping();
 			void AntiAliasing();
 
 			static bool IsLightUrl(CSZ url);

Verus/src/CGI/Renderer.cpp

@@ -52,35 +52,15 @@ void Renderer::Init(PRendererDelegate pDelegate)
 
 	_commandBuffer.Init();
 
-	GeometryDesc geoDesc;
-	const InputElementDesc ied[] =
+	_rphSwapChain = _pBaseRenderer->CreateRenderPass(
 		{
-		{0, offsetof(Vertex, _pos), IeType::floats, 2, IeUsage::position, 0},
-		InputElementDesc::End()
-	};
-	geoDesc._pInputElementDesc = ied;
-	const int strides[] = { sizeof(Vertex), 0 };
-	geoDesc._pStrides = strides;
-	_geoQuad.Init(geoDesc);
-
-	_shader[S_GENERATE_MIPS].Init("[Shaders]:GenerateMips.hlsl");
-	_shader[S_GENERATE_MIPS]->CreateDescriptorSet(0, &_ubGenerateMips, sizeof(_ubGenerateMips), 100,
-		{ Sampler::linearClamp2D, Sampler::storage, Sampler::storage, Sampler::storage, Sampler::storage }, ShaderStageFlags::cs);
-	_shader[S_GENERATE_MIPS]->CreatePipelineLayout();
-
-	_shader[S_QUAD].Init("[Shaders]:Quad.hlsl");
-	_shader[S_QUAD]->CreateDescriptorSet(0, &_ubQuadVS, sizeof(_ubQuadVS), 100, {}, ShaderStageFlags::vs);
-	_shader[S_QUAD]->CreateDescriptorSet(1, &_ubQuadFS, sizeof(_ubQuadFS), 100, { Sampler::linearClamp2D }, ShaderStageFlags::fs);
-	_shader[S_QUAD]->CreatePipelineLayout();
-
-	PipelineDesc pipeDesc(_shader[S_GENERATE_MIPS], "#");
-	_pipeGenerateMips.Init(pipeDesc);
-
-	_rpSwapChain = _pBaseRenderer->CreateRenderPass(
-		{ RP::Attachment("Color", Format::srgbB8G8R8A8).LoadOpClear().Layout(ImageLayout::undefined, ImageLayout::presentSrc) },
-		{ RP::Subpass("Sp0").Color({RP::Ref("Color", ImageLayout::colorAttachment)}) },
+			RP::Attachment("Color", Format::srgbB8G8R8A8).LoadOpClear().Layout(ImageLayout::undefined, ImageLayout::presentSrc)
+		},
+		{
+			RP::Subpass("Sp0").Color({RP::Ref("Color", ImageLayout::colorAttachment)})
+		},
 		{});
-	_rpSwapChainDepth = _pBaseRenderer->CreateRenderPass(
+	_rphSwapChainDepth = _pBaseRenderer->CreateRenderPass(
 		{
 			RP::Attachment("Color", Format::srgbB8G8R8A8).LoadOpClear().Layout(ImageLayout::undefined, ImageLayout::presentSrc),
 			RP::Attachment("Depth", Format::unormD24uintS8).Layout(ImageLayout::depthStencilAttachment),
@@ -93,9 +73,40 @@ void Renderer::Init(PRendererDelegate pDelegate)
 		},
 		{});
 
+	GeometryDesc geoDesc;
+	const VertexInputAttrDesc viaDesc[] =
+	{
+		{0, offsetof(Vertex, _pos), IeType::floats, 2, IeUsage::position, 0},
+		VertexInputAttrDesc::End()
+	};
+	geoDesc._pVertexInputAttrDesc = viaDesc;
+	const int strides[] = { sizeof(Vertex), 0 };
+	geoDesc._pStrides = strides;
+	_geoQuad.Init(geoDesc);
+
+	_shader[S_GENERATE_MIPS].Init("[Shaders]:GenerateMips.hlsl");
+	_shader[S_GENERATE_MIPS]->CreateDescriptorSet(0, &_ubGenerateMips, sizeof(_ubGenerateMips), 100,
+		{
+			Sampler::linearClampMipN,
+			Sampler::storage,
+			Sampler::storage,
+			Sampler::storage,
+			Sampler::storage
+		},
+		ShaderStageFlags::cs);
+	_shader[S_GENERATE_MIPS]->CreatePipelineLayout();
+
+	_shader[S_QUAD].Init("[Shaders]:Quad.hlsl");
+	_shader[S_QUAD]->CreateDescriptorSet(0, &_ubQuadVS, sizeof(_ubQuadVS), 100, {}, ShaderStageFlags::vs);
+	_shader[S_QUAD]->CreateDescriptorSet(1, &_ubQuadFS, sizeof(_ubQuadFS), 100, { Sampler::linearClampMipN }, ShaderStageFlags::fs);
+	_shader[S_QUAD]->CreatePipelineLayout();
+
+	PipelineDesc pipeDesc(_shader[S_GENERATE_MIPS], "#");
+	_pipeGenerateMips.Init(pipeDesc);
+
 	OnSwapChainResized(true, false);
 
-	_pBaseRenderer->ImGuiInit(_rpSwapChainDepth);
+	_pBaseRenderer->ImGuiInit(_rphSwapChainDepth);
 	ImGuiUpdateStyle();
 
 	_ds.Init();
@@ -126,6 +137,7 @@ void Renderer::Done()
 		_geoQuad.Done();
 		_commandBuffer.Done();
 
+		Scene::Grass::DoneStatic();
 		Scene::Terrain::DoneStatic();
 		Scene::Mesh::DoneStatic();
 
@@ -151,7 +163,7 @@ void Renderer::Present()
 	_frameCount++;
 
 	VERUS_QREF_TIMER;
-	_fps = _fps * 0.75f + timer.GetDeltaTimeInv() * 0.25f;
+	_fps = Math::Lerp(_fps, timer.GetDeltaTimeInv(), 0.25f);
 }
 
 void Renderer::OnWindowResized(int w, int h)
@@ -170,10 +182,10 @@ void Renderer::OnSwapChainResized(bool init, bool done)
 {
 	if (done)
 	{
-		VERUS_FOR(i, _fbSwapChainDepth.size())
-			_pBaseRenderer->DeleteFramebuffer(_fbSwapChainDepth[i]);
-		VERUS_FOR(i, _fbSwapChain.size())
-			_pBaseRenderer->DeleteFramebuffer(_fbSwapChain[i]);
+		VERUS_FOR(i, _fbhSwapChainDepth.size())
+			_pBaseRenderer->DeleteFramebuffer(_fbhSwapChainDepth[i]);
+		VERUS_FOR(i, _fbhSwapChain.size())
+			_pBaseRenderer->DeleteFramebuffer(_fbhSwapChain[i]);
 
 		_texDepthStencil.Done();
 	}
@@ -185,14 +197,15 @@ void Renderer::OnSwapChainResized(bool init, bool done)
 		texDesc._format = Format::unormD24uintS8;
 		texDesc._width = _swapChainWidth;
 		texDesc._height = _swapChainHeight;
+		texDesc._flags = TextureDesc::Flags::inputAttachment | TextureDesc::Flags::depthSampledW;
 		_texDepthStencil.Init(texDesc);
 
-		_fbSwapChain.resize(_pBaseRenderer->GetSwapChainBufferCount());
-		VERUS_FOR(i, _fbSwapChain.size())
-			_fbSwapChain[i] = _pBaseRenderer->CreateFramebuffer(_rpSwapChain, {}, _swapChainWidth, _swapChainHeight, i);
-		_fbSwapChainDepth.resize(_pBaseRenderer->GetSwapChainBufferCount());
-		VERUS_FOR(i, _fbSwapChainDepth.size())
-			_fbSwapChainDepth[i] = _pBaseRenderer->CreateFramebuffer(_rpSwapChainDepth, { _texDepthStencil }, _swapChainWidth, _swapChainHeight, i);
+		_fbhSwapChain.resize(_pBaseRenderer->GetSwapChainBufferCount());
+		VERUS_FOR(i, _fbhSwapChain.size())
+			_fbhSwapChain[i] = _pBaseRenderer->CreateFramebuffer(_rphSwapChain, {}, _swapChainWidth, _swapChainHeight, i);
+		_fbhSwapChainDepth.resize(_pBaseRenderer->GetSwapChainBufferCount());
+		VERUS_FOR(i, _fbhSwapChainDepth.size())
+			_fbhSwapChainDepth[i] = _pBaseRenderer->CreateFramebuffer(_rphSwapChainDepth, { _texDepthStencil }, _swapChainWidth, _swapChainHeight, i);
 	}
 }
 

Verus/src/CGI/Renderer.h

@@ -44,10 +44,10 @@ namespace verus
 			int               _swapChainWidth = 0;
 			int               _swapChainHeight = 0;
 			float             _fps = 30;
-			int               _rpSwapChain = 0;
-			int               _rpSwapChainDepth = 0;
-			Vector<int>       _fbSwapChain;
-			Vector<int>       _fbSwapChainDepth;
+			RPHandle          _rphSwapChain;
+			RPHandle          _rphSwapChainDepth;
+			Vector<FBHandle>  _fbhSwapChain;
+			Vector<FBHandle>  _fbhSwapChainDepth;
 			UB_GenerateMips   _ubGenerateMips;
 			UB_QuadVS         _ubQuadVS;
 			UB_QuadFS         _ubQuadFS;
@@ -78,16 +78,17 @@ namespace verus
 
 			RDeferredShading GetDS() { return _ds; }
 			CommandBufferPtr GetCommandBuffer() const { return _commandBuffer; }
-			GeometryPtr GetGeoQuad() const { return _geoQuad; }
 			TexturePtr GetTexDepthStencil() const { return _texDepthStencil; }
 
 			void OnShaderError(CSZ s);
 			void OnShaderWarning(CSZ s);
 
+			// Window:
 			App::PWindow GetMainWindow() const { return _pMainWindow; }
 			App::PWindow SetMainWindow(App::PWindow p) { return Utils::Swap(_pMainWindow, p); }
 			float GetSwapChainAspectRatio() const;
 
+			// ImGui:
 			virtual void ImGuiSetCurrentContext(ImGuiContext* pContext);
 			void ImGuiUpdateStyle();
 
@@ -95,15 +96,19 @@ namespace verus
 			float GetFps() const { return _fps; }
 			UINT64 GetFrameCount() const { return _frameCount; }
 
-			int GetRenderPass_SwapChain() const { return _rpSwapChain; }
-			int GetRenderPass_SwapChainDepth() const { return _rpSwapChainDepth; }
-			int GetFramebuffer_SwapChain(int index) const { return _fbSwapChain[index]; }
-			int GetFramebuffer_SwapChainDepth(int index) const { return _fbSwapChainDepth[index]; }
+			// RenderPass & Framebuffer:
+			RPHandle GetRenderPassHandle_SwapChain() const { return _rphSwapChain; }
+			RPHandle GetRenderPassHandle_SwapChainDepth() const { return _rphSwapChainDepth; }
+			FBHandle GetFramebufferHandle_SwapChain(int index) const { return _fbhSwapChain[index]; }
+			FBHandle GetFramebufferHandle_SwapChainDepth(int index) const { return _fbhSwapChainDepth[index]; }
 
+			// Generate mips:
 			PipelinePtr GetPipelineGenerateMips() { return _pipeGenerateMips; }
 			ShaderPtr GetShaderGenerateMips() { return _shader[S_GENERATE_MIPS]; }
 			UB_GenerateMips& GetUbGenerateMips() { return _ubGenerateMips; }
 
+			// Quad:
+			GeometryPtr GetGeoQuad() const { return _geoQuad; }
 			ShaderPtr GetShaderQuad() { return _shader[S_QUAD]; }
 			UB_QuadVS& GetUbQuadVS() { return _ubQuadVS; }
 			UB_QuadFS& GetUbQuadFS() { return _ubQuadFS; }

Verus/src/CGI/TextureRAM.cpp

@@ -0,0 +1,48 @@
+#include "verus.h"
+
+using namespace verus;
+using namespace verus::CGI;
+
+TextureRAM::TextureRAM()
+{
+}
+
+TextureRAM::~TextureRAM()
+{
+	Done();
+}
+
+void TextureRAM::Init(RcTextureDesc desc)
+{
+	VERUS_INIT();
+	VERUS_RT_ASSERT(desc._width > 0 && desc._height > 0);
+	VERUS_QREF_RENDERER;
+
+	_size = Vector4(
+		float(desc._width),
+		float(desc._height),
+		1.f / desc._width,
+		1.f / desc._height);
+	_desc = desc;
+	_desc._mipLevels = desc._mipLevels ? desc._mipLevels : Math::ComputeMipLevels(desc._width, desc._height, desc._depth);
+	_bytesPerPixel = FormatToBytesPerPixel(desc._format);
+
+	_vBuffer.resize(_desc._width * _desc._height * _bytesPerPixel);
+}
+
+void TextureRAM::Done()
+{
+	VERUS_DONE(TextureRAM);
+}
+
+void TextureRAM::UpdateSubresource(const void* p, int mipLevel, int arrayLayer, PBaseCommandBuffer pCB)
+{
+	if (mipLevel || arrayLayer)
+		return;
+	memcpy(_vBuffer.data(), p, _vBuffer.size());
+}
+
+void TextureRAM::GenerateMips(PBaseCommandBuffer pCB)
+{
+	VERUS_RT_ASSERT(__FUNCTION__);
+}

Verus/src/CGI/TextureRAM.h

@@ -0,0 +1,26 @@
+#pragma once
+
+namespace verus
+{
+	namespace CGI
+	{
+		class TextureRAM : public BaseTexture
+		{
+			Vector<BYTE> _vBuffer;
+
+		public:
+			TextureRAM();
+			virtual ~TextureRAM() override;
+
+			virtual void Init(RcTextureDesc desc) override;
+			virtual void Done() override;
+
+			virtual void UpdateSubresource(const void* p, int mipLevel, int arrayLayer, PBaseCommandBuffer pCB) override;
+
+			virtual void GenerateMips(PBaseCommandBuffer pCB) override;
+
+			const BYTE* GetData() const { VERUS_RT_ASSERT(IsLoaded()); return _vBuffer.data(); }
+		};
+		VERUS_TYPEDEFS(TextureRAM);
+	}
+}

Verus/src/CGI/Types.h

@@ -37,7 +37,9 @@ namespace verus
 			lineList,
 			lineStrip,
 			triangleList,
-			triangleStrip
+			triangleStrip,
+			patchList3,
+			patchList4
 		};
 
 		struct PipelineRasterizationState
@@ -72,7 +74,9 @@ namespace verus
 			gs = (1 << 3),
 			fs = (1 << 4),
 			cs = (1 << 5),
-			vs_fs = vs | fs
+			vs_fs = vs | fs,
+			vs_hs_ds = vs | hs | ds,
+			vs_hs_ds_fs = vs | hs | ds | fs
 		};
 
 		enum class IeType : int
@@ -82,22 +86,24 @@ namespace verus
 			shorts
 		};
 
-		// See: https://docs.microsoft.com/en-us/windows/desktop/direct3d9/d3ddeclusage
+		// See: https://docs.microsoft.com/en-us/windows/win32/direct3d9/d3ddeclusage
 		enum class IeUsage : int
 		{
 			position,
-			blendWeight,
+			blendWeights,
 			blendIndices,
 			normal,
-			psize,
-			texCoord,
 			tangent,
 			binormal,
-			color
+			color,
+			psize,
+			texCoord,
+			instData,
+			attr
 		};
 
-		// See: https://docs.microsoft.com/en-us/windows/desktop/direct3d9/d3dvertexelement9
-		struct InputElementDesc
+		// See: https://docs.microsoft.com/en-us/windows/win32/direct3d9/d3dvertexelement9
+		struct VertexInputAttrDesc
 		{
 			int     _binding;
 			int     _offset;
@@ -106,9 +112,9 @@ namespace verus
 			IeUsage _usage;
 			int     _usageIndex;
 
-			static constexpr InputElementDesc End() { return { -1, -1, IeType::floats, 0, IeUsage::position, 0 }; }
+			static constexpr VertexInputAttrDesc End() { return { -1, -1, IeType::floats, 0, IeUsage::position, 0 }; }
 		};
-		VERUS_TYPEDEFS(InputElementDesc);
+		VERUS_TYPEDEFS(VertexInputAttrDesc);
 
 		enum class Sampler : int
 		{
@@ -117,15 +123,51 @@ namespace verus
 			input,
 			shadow,
 			aniso, // Most common sampler for 3D.
-			linear3D,
-			nearest3D,
-			linear2D,
-			nearest2D,
-			linearClamp3D,
-			nearestClamp3D,
-			linearClamp2D,
-			nearestClamp2D,
+			linearMipL,
+			nearestMipL,
+			linearMipN,
+			nearestMipN,
+			linearClampMipL,
+			nearestClampMipL,
+			linearClampMipN,
+			nearestClampMipN,
 			count
 		};
+
+		template<typename T>
+		class BaseHandle
+		{
+			int _h = -1;
+
+		protected:
+			static T Make(int value)
+			{
+				T ret;
+				ret._h = value;
+				return ret;
+			}
+
+		public:
+			int Get() const { return _h; }
+			bool IsSet() const { return _h >= 0; }
+		};
+
+		class RPHandle : public BaseHandle<RPHandle>
+		{
+		public:
+			static RPHandle Make(int value) { return BaseHandle::Make(value); }
+		};
+
+		class FBHandle : public BaseHandle<FBHandle>
+		{
+		public:
+			static FBHandle Make(int value) { return BaseHandle::Make(value); }
+		};
+
+		class CSHandle : public BaseHandle<CSHandle>
+		{
+		public:
+			static CSHandle Make(int value) { return BaseHandle::Make(value); }
+		};
 	}
 }

Verus/src/Global/Convert.cpp

@@ -71,16 +71,31 @@ UINT16 Convert::Uint8x4ToUint4x4(UINT32 in)
 	return (x[0] << 12) | (x[1] << 8) | (x[2] << 4) | (x[3]);
 }
 
-UINT32 Convert::Uint4x4ToUint8x4(UINT16 in)
+UINT32 Convert::Uint4x4ToUint8x4(UINT16 in, bool scaleTo255)
 {
 	int x[4] =
 	{
-		(in >> 12) & 0xF,
-		(in >> 8) & 0xF,
-		(in >> 4) & 0xF,
-		(in >> 0) & 0xF
+		((in >> 12) & 0xF) << 4,
+		((in >> 8) & 0xF) << 4,
+		((in >> 4) & 0xF) << 4,
+		((in >> 0) & 0xF) << 4
 	};
-	return (x[0] << 28) | (x[1] << 20) | (x[2] << 12) | (x[3] << 4);
+	if (scaleTo255)
+	{
+		VERUS_FOR(i, 4)
+			x[i] = x[i] * 255 / 0xF0;
+	}
+	return (x[0] << 24) | (x[1] << 16) | (x[2] << 8) | (x[3] << 0);
+}
+
+float Convert::SRGBToLinear(float color)
+{
+	return color < 0.04045f ? color * (1 / 12.92f) : pow((color + 0.055f) * (1 / 1.055f), 2.4f);
+}
+
+float Convert::LinearToSRGB(float color)
+{
+	return color < 0.0031308f ? 12.92f * color : 1.055f * pow(color, 1 / 2.4f) - 0.055f;
 }
 
 UINT32 Convert::Color16To32(UINT16 in)
@@ -93,20 +108,35 @@ UINT32 Convert::Color16To32(UINT16 in)
 
 void Convert::ColorInt32ToFloat(UINT32 in, float* out, bool sRGB)
 {
-	const float gamma = sRGB ? 2.2f : 1.f;
-	out[0] = pow(float(int((in >> 0) & 0xFF))* (1 / 255.f), gamma);
-	out[1] = pow(float(int((in >> 8) & 0xFF))* (1 / 255.f), gamma);
-	out[2] = pow(float(int((in >> 16) & 0xFF))* (1 / 255.f), gamma);
-	out[3] = float(int((in >> 24) & 0xFF))* (1 / 255.f);
+	out[0] = ((in >> 0) & 0xFF) * (1 / 255.f);
+	out[1] = ((in >> 8) & 0xFF) * (1 / 255.f);
+	out[2] = ((in >> 16) & 0xFF) * (1 / 255.f);
+	out[3] = ((in >> 24) & 0xFF) * (1 / 255.f);
+	if (sRGB)
+	{
+		out[0] = SRGBToLinear(out[0]);
+		out[1] = SRGBToLinear(out[1]);
+		out[2] = SRGBToLinear(out[2]);
+	}
 }
 
 UINT32 Convert::ColorFloatToInt32(const float* in, bool sRGB)
 {
-	const float gamma = sRGB ? 1 / 2.2f : 1.f;
-	const int r = Math::Clamp(int(pow(in[0], gamma) * 255.5f), 0, 255);
-	const int g = Math::Clamp(int(pow(in[1], gamma) * 255.5f), 0, 255);
-	const int b = Math::Clamp(int(pow(in[2], gamma) * 255.5f), 0, 255);
-	const int a = Math::Clamp(int(in[3] * 255.5f), 0, 255);
+	float out[4];
+	out[0] = in[0];
+	out[1] = in[1];
+	out[2] = in[2];
+	out[3] = in[3];
+	if (sRGB)
+	{
+		out[0] = LinearToSRGB(out[0]);
+		out[1] = LinearToSRGB(out[1]);
+		out[2] = LinearToSRGB(out[2]);
+	}
+	const int r = Math::Clamp(static_cast<int>(out[0] * 255 + 0.5f), 0, 255);
+	const int g = Math::Clamp(static_cast<int>(out[1] * 255 + 0.5f), 0, 255);
+	const int b = Math::Clamp(static_cast<int>(out[2] * 255 + 0.5f), 0, 255);
+	const int a = Math::Clamp(static_cast<int>(out[3] * 255 + 0.5f), 0, 255);
 	return VERUS_COLOR_RGBA(r, g, b, a);
 }
 
@@ -140,11 +170,16 @@ void Convert::ColorTextToFloat4(CSZ sz, float* out, bool sRGB)
 		if (3 == count)
 			color[3] = 255;
 	}
-	const float gamma = sRGB ? 2.2f : 1.f;
-	out[0] = pow(float(color[0]) * (1 / 255.f), gamma);
-	out[1] = pow(float(color[1]) * (1 / 255.f), gamma);
-	out[2] = pow(float(color[2]) * (1 / 255.f), gamma);
-	out[3] = float(color[3]) * (1 / 255.f);
+	out[0] = color[0] * (1 / 255.f);
+	out[1] = color[1] * (1 / 255.f);
+	out[2] = color[2] * (1 / 255.f);
+	out[3] = color[3] * (1 / 255.f);
+	if (sRGB)
+	{
+		out[0] = SRGBToLinear(out[0]);
+		out[1] = SRGBToLinear(out[1]);
+		out[2] = SRGBToLinear(out[2]);
+	}
 }
 
 UINT32 Convert::ColorTextToInt32(CSZ sz)
@@ -321,6 +356,8 @@ String Convert::ToMd5String(const Vector<BYTE>& vBin)
 
 void Convert::Test()
 {
+	const float e = 1e-6f;
+
 	BYTE dataUint8[] = { 0, 127, 255 };
 	UINT16 dataUint16[] = { 0, 127, 65535, 65534 };
 	char dataSint8[] = { 64, -127, -128 };
@@ -336,6 +373,32 @@ void Convert::Test()
 	Sint16ToSnorm(dataSint16, dataFloat, 4);
 	SnormToSint16(dataFloat, dataSint16, 4);
 
+	BYTE colorIn[20];
+	VERUS_FOR(i, 16)
+		colorIn[i] = i;
+	colorIn[16] = 64;
+	colorIn[17] = 128;
+	colorIn[18] = 192;
+	colorIn[19] = 255;
+	float colorOut[20];
+	VERUS_FOR(i, 20)
+		colorOut[i] = SRGBToLinear(colorIn[i] * (1 / 255.f));
+	BYTE colorTest[20];
+	VERUS_FOR(i, 20)
+		colorTest[i] = static_cast<BYTE>(LinearToSRGB(colorOut[i]) * 255 + 0.5f);
+	VERUS_FOR(i, 20)
+	{
+		if (i < 19)
+		{
+			const float d = colorOut[i + 1] - colorOut[i];
+			if (i < 10)
+				VERUS_RT_ASSERT(glm::epsilonEqual(d, colorOut[1], e));
+			else
+				VERUS_RT_ASSERT(glm::epsilonNotEqual(d, colorOut[1], e));
+		}
+		VERUS_RT_ASSERT(colorIn[i] == colorTest[i]);
+	}
+
 	Vector<BYTE> vBin;
 	vBin.resize(16);
 

Verus/src/Global/Convert.h

@@ -43,9 +43,11 @@ namespace verus
 
 		// 4 bits per channel:
 		static UINT16 Uint8x4ToUint4x4(UINT32 in);
-		static UINT32 Uint4x4ToUint8x4(UINT16 in);
+		static UINT32 Uint4x4ToUint8x4(UINT16 in, bool scaleTo255 = false);
 
 		// Colors:
+		static float SRGBToLinear(float color);
+		static float LinearToSRGB(float color);
 		static UINT32 Color16To32(UINT16 in);
 		static   void ColorInt32ToFloat(UINT32 in, float* out, bool sRGB = true);
 		static UINT32 ColorFloatToInt32(const float* in, bool sRGB = true);

Verus/src/Global/EngineInit.cpp

@@ -86,6 +86,7 @@ void EngineInit::Init(Input::PKeyMapperDelegate pKeyMapperDelegate, CGI::Rendere
 	{
 		Scene::Mesh::InitStatic();
 		Scene::Terrain::InitStatic();
+		Scene::Grass::InitStatic();
 		//Scene::CForest::InitStatic();
 	}
 

Verus/src/Global/QuickRefs.h

@@ -7,8 +7,8 @@
 #define VERUS_QREF_BLUR           Effects::RBlur          blur     = Effects::Blur::I()
 #define VERUS_QREF_BULLET         Physics::RBullet        bullet   = Physics::Bullet::I()
 #define VERUS_QREF_CONST_SETTINGS App::RcSettings         settings = App::Settings::IConst()
-#define VERUS_QREF_DEPTH          Effects::RDepth         depth    = Effects::Depth::I()
 #define VERUS_QREF_DD             CGI::RDebugDraw         dd       = CGI::DebugDraw::I()
+#define VERUS_QREF_DEPTH          Effects::RDepth         depth    = Effects::Depth::I()
 #define VERUS_QREF_FSYS           IO::RFileSystem         fsys     = IO::FileSystem::I()
 #define VERUS_QREF_GAME           Game::RGame             game     = Game::Game::I()
 #define VERUS_QREF_GRASS          Scene::RGrass           grass    = Scene::Grass::I()

Verus/src/Math/Math.cpp

@@ -9,6 +9,18 @@ bool Math::IsPowerOfTwo(int x)
 	return !(x & (x - 1));
 }
 
+UINT32 Math::NextPowerOfTwo(UINT32 x)
+{
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x++;
+	return x;
+}
+
 int Math::HighestBit(int x)
 {
 	int bit = -1;
@@ -59,6 +71,11 @@ float Math::LinearToCos(float t)
 	return cos(t * VERUS_PI) * -0.5f + 0.5f; // [1 to -1] -> [0 to 1].
 }
 
+float Math::EaseInOutSine(float x)
+{
+	return -(cos(VERUS_PI * x) - 1) * 0.5f;
+}
+
 Vector3 Math::TriangleNormal(RcPoint3 a, RcPoint3 b, RcPoint3 c)
 {
 	return VMath::normalize(VMath::cross(a - b, a - c));
@@ -77,15 +94,15 @@ float Math::TriangleArea(
 	return sqrt(s * (s - a) * (s - b) * (s - c));
 }
 
-int Math::GetStripGridIndexCount(int polyCountWidth, int polyCountHeight)
+int Math::StripGridIndexCount(int polyCountWidth, int polyCountHeight)
 {
 	const int vertCountWidth = polyCountWidth + 1;
 	return (vertCountWidth * 2) * polyCountHeight + (polyCountHeight - 1) * 2; // Strip and primitive restart values.
 }
 
-void Math::BuildStripGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices)
+void Math::CreateStripGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices)
 {
-	vIndices.resize(GetStripGridIndexCount(polyCountWidth, polyCountHeight));
+	vIndices.resize(StripGridIndexCount(polyCountWidth, polyCountHeight));
 	int offset = 0;
 	const int vertCountWidth = polyCountWidth + 1;
 	VERUS_FOR(h, polyCountHeight)
@@ -114,7 +131,7 @@ void Math::BuildStripGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16
 	}
 }
 
-void Math::BuildListGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices)
+void Math::CreateListGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices)
 {
 	vIndices.resize(polyCountWidth * polyCountHeight * 6);
 	int offset = 0;
@@ -148,6 +165,23 @@ void Math::BuildListGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>
 	}
 }
 
+bool Math::CheckIndexBuffer(Vector<UINT16>& vIndices, int maxIndex)
+{
+	Vector<UINT16> vTemp(vIndices.begin(), vIndices.end());
+	std::sort(vTemp.begin(), vTemp.end());
+	auto itRestart = std::find(vTemp.begin(), vTemp.end(), 0xFFFF);
+	if (itRestart != vTemp.end())
+		vTemp.erase(itRestart, vTemp.end());
+	if (vTemp.empty())
+		return false;
+	if (vTemp.front() != 0)
+		return false;
+	if (vTemp.back() != maxIndex)
+		return false;
+	auto it = std::adjacent_find(vTemp.begin(), vTemp.end(), [](UINT16 a, UINT16 b) {return (b - a) > 1; });
+	return vTemp.end() == it;
+}
+
 Transform3 Math::BoundsDrawMatrix(RcPoint3 mn, RcPoint3 mx)
 {
 	const Vector3 d = mx - mn;
@@ -161,6 +195,13 @@ float Math::ComputeOnePixelDistance(float size, float screenHeight, float fovY)
 	return viewSliceSize / (tan(fovY * 0.5f) * 2);
 }
 
+float Math::ComputeDistToMipScale(float texSize, float screenSize, float realSize, float fovY)
+{
+	const float fillScreenScale = screenSize / texSize;
+	const float a = realSize * fillScreenScale * 0.5f;
+	return tan(fovY * 0.5f) / a;
+}
+
 void Math::Quadrant(const int** ppSrcMinMax, int** ppDestMinMax, int half, int id)
 {
 	VERUS_RT_ASSERT(id >= 0 && id < 4);
@@ -395,7 +436,7 @@ void Math::Test()
 
 	{
 		Vector<UINT16> v;
-		BuildStripGrid(2, 2, v);
+		CreateStripGrid(2, 2, v);
 		VERUS_RT_ASSERT(v.size() == 14);
 		VERUS_RT_ASSERT(0 == v[0] && 3 == v[1] && 1 == v[2]); // First triangle is CCW.
 	}

Verus/src/Math/Math.h

@@ -48,6 +48,7 @@ namespace verus
 	{
 		// Bits:
 		bool IsPowerOfTwo(int x);
+		UINT32 NextPowerOfTwo(UINT32 x);
 		int HighestBit(int x);
 		bool IsNaN(float x);
 
@@ -61,19 +62,22 @@ namespace verus
 		float SmoothStep(float a, float b, float t);
 		float LinearToSin(float t);
 		float LinearToCos(float t);
+		float EaseInOutSine(float x);
 
 		// Shapes:
 		Vector3 TriangleNormal(RcPoint3 a, RcPoint3 b, RcPoint3 c);
 		float TriangleArea(const glm::vec3& p0, const glm::vec3& p1, const glm::vec3& p2);
 
 		// Geometry:
-		int GetStripGridIndexCount(int polyCountWidth, int polyCountHeight);
-		void BuildStripGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices);
-		void BuildListGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices);
+		int StripGridIndexCount(int polyCountWidth, int polyCountHeight);
+		void CreateStripGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices);
+		void CreateListGrid(int polyCountWidth, int polyCountHeight, Vector<UINT16>& vIndices);
+		bool CheckIndexBuffer(Vector<UINT16>& vIndices, int maxIndex);
 
 		// Scene:
 		Transform3 BoundsDrawMatrix(RcPoint3 mn, RcPoint3 mx);
 		float ComputeOnePixelDistance(float size, float screenHeight = 120, float fovY = VERUS_PI / 4);
+		float ComputeDistToMipScale(float texSize, float screenSize, float realSize, float fovY);
 		void Quadrant(const int** ppSrcMinMax, int** ppDestMinMax, int half, int id);
 
 		int ComputeMipLevels(int w, int h, int d = 1);

Verus/src/Scene/Atmosphere.cpp

@@ -3,6 +3,11 @@
 using namespace verus;
 using namespace verus::Scene;
 
+Atmosphere::UB_PerFrame      Atmosphere::s_ubPerFrame;
+Atmosphere::UB_PerMaterialFS Atmosphere::s_ubPerMaterialFS;
+Atmosphere::UB_PerMeshVS     Atmosphere::s_ubPerMeshVS;
+Atmosphere::UB_PerObject     Atmosphere::s_ubPerObject;
+
 Atmosphere::Atmosphere()
 {
 }
@@ -17,20 +22,214 @@ void Atmosphere::Init()
 	VERUS_INIT();
 	VERUS_QREF_RENDERER;
 	VERUS_QREF_CONST_SETTINGS;
+	VERUS_QREF_UTILS;
+
+	_skyDome.Init("[Models]:SkyDome.x3d");
+
+	_shader.Init("[Shaders]:Sky.hlsl");
+	_shader->CreateDescriptorSet(0, &s_ubPerFrame, sizeof(s_ubPerFrame), 100);
+	_shader->CreateDescriptorSet(1, &s_ubPerMaterialFS, sizeof(s_ubPerMaterialFS), 100,
+		{
+			CGI::Sampler::aniso,
+			CGI::Sampler::aniso,
+			CGI::Sampler::aniso,
+			CGI::Sampler::aniso
+		}, CGI::ShaderStageFlags::fs);
+	_shader->CreateDescriptorSet(2, &s_ubPerMeshVS, sizeof(s_ubPerMeshVS), 100, {}, CGI::ShaderStageFlags::vs);
+	_shader->CreateDescriptorSet(3, &s_ubPerObject, sizeof(s_ubPerObject), 100);
+	_shader->CreatePipelineLayout();
 
 	if (settings._sceneShadowQuality > App::Settings::ShadowQuality::none)
 		_shadowMap.Init(4096);
 
 	renderer.GetDS().InitByAtmosphere(_shadowMap.GetTexture());
 
-	_sun._dirTo = VMath::normalize(Vector3(1.3f, 1, 1.3f));
+	_texSky.LoadDDS("[Textures]:Sky/Sky.dds");
+
+	_tex[TEX_SKY].Init("[Textures]:Sky/Sky.dds");
+	_tex[TEX_STARS].Init("[Textures]:Sky/Stars.dds");
+	_tex[TEX_SUN].Init("[Textures]:Sky/Sun.dds");
+	_tex[TEX_MOON].Init("[Textures]:Sky/Moon.dds");
+	_tex[TEX_CLOUDS].Init("[Textures]:Sky/Clouds.FX.dds");
+	_tex[TEX_CLOUDS_NM].Init("[Textures]:Sky/Clouds.NM.dds");
+
+	_clouds._phaseA.setX(utils.GetRandom().NextFloat());
+	_clouds._phaseA.setY(utils.GetRandom().NextFloat());
+	_clouds._phaseB.setX(utils.GetRandom().NextFloat());
+	_clouds._phaseB.setY(utils.GetRandom().NextFloat());
+
+	_sun._matTilt = Matrix3::rotationX(-VERUS_PI / 10);
+	_sun._dirToMidnight = _sun._matTilt * Vector3(0, -1, 0);
+	_sun._dirTo = VMath::mulPerElem(_sun._dirToMidnight, Vector3(1, -1, 1));
 }
 
 void Atmosphere::Done()
 {
+	_shader->FreeDescriptorSet(_cshSkyFS);
 	VERUS_DONE(Atmosphere);
 }
 
+void Atmosphere::UpdateSun(float time)
+{
+	// Update time and sun's color before calling this method.
+
+	_sun._dirTo = Matrix3::rotationZ(time * VERUS_2PI) * _sun._dirToMidnight;
+
+	_night = false;
+	if (_sun._dirTo.getY() < 0) // Moon light:
+	{
+		_night = true;
+		_sun._dirTo = Matrix3::rotationZ(time * VERUS_2PI + VERUS_PI) * _sun._dirToMidnight;
+		_sun._color = VMath::mulPerElem(_sun._color, Vector3(0.19f, 0.21f, 0.45f));
+	}
+
+	// Reduce light's intensity when near horizon:
+	const float intensity = Math::Clamp<float>(abs(_sun._dirTo.getY()) * 4, 0, 1);
+	_sun._color *= intensity;
+}
+
+void Atmosphere::Update()
+{
+	VERUS_QREF_RENDERER;
+	VERUS_QREF_TIMER;
+
+	if (!_async_loaded)
+	{
+		bool allTexLoaded = true;
+		VERUS_FOR(i, TEX_COUNT)
+		{
+			if (!_tex[i]->IsLoaded())
+				allTexLoaded = false;
+		}
+		if (_skyDome.IsLoaded() && allTexLoaded)
+		{
+			_async_loaded = true;
+
+			{
+				CGI::PipelineDesc pipeDesc(_skyDome.GetGeometry(), _shader, "#Sky", renderer.GetDS().GetRenderPassHandle_Compose(), renderer.GetDS().GetSubpass_Compose());
+				pipeDesc._vertexInputBindingsFilter = (1 << 0);
+				pipeDesc._depthWriteEnable = false;
+				pipeDesc._depthCompareOp = CGI::CompareOp::lessOrEqual;
+				_pipe[PIPE_SKY].Init(pipeDesc);
+			}
+			{
+				CGI::PipelineDesc pipeDesc(_skyDome.GetGeometry(), _shader, "#Clouds", renderer.GetDS().GetRenderPassHandle_Compose(), renderer.GetDS().GetSubpass_Compose());
+				pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ALPHA;
+				pipeDesc._vertexInputBindingsFilter = (1 << 0);
+				pipeDesc._depthWriteEnable = false;
+				pipeDesc._depthCompareOp = CGI::CompareOp::lessOrEqual;
+				_pipe[PIPE_CLOUDS].Init(pipeDesc);
+			}
+
+			_shader->FreeDescriptorSet(_cshSkyFS);
+			_cshSkyFS = _shader->BindDescriptorSetTextures(1, { _tex[TEX_SKY], _tex[TEX_STARS], _tex[TEX_CLOUDS], _tex[TEX_CLOUDS_NM] });
+		}
+	}
+	if (!_async_loaded)
+		return;
+
+	_time = glm::fract(_time + dt * _timeSpeed);
+
+	_clouds._cloudiness.Update();
+
+	float color[3];
+	GetColor(64, color, 1.5f);  _fog._color = Vector3::MakeFromPointer(color) * _hdrScale;
+	GetColor(208, color, 0.3f); _ambientColor = Vector3::MakeFromPointer(color) * _hdrScale;
+	GetColor(240, color, 1);    _sun._color = Vector3::MakeFromPointer(color) * _hdrScale;
+
+	glm::vec3 ambientColor = _ambientColor.GLM();
+	glm::vec3 fogColor = _fog._color.GLM();
+	_ambientColor = glm::saturation(0.75f, ambientColor);
+	const float clearSky = Math::Min(1.f, 1.5f - _clouds._cloudiness);
+	_fog._color = glm::saturation(clearSky * 0.75f, fogColor);
+	_fog._colorInit = _fog._color;
+
+	UpdateSun(_time);
+}
+
+void Atmosphere::DrawSky()
+{
+	if (!_async_loaded)
+		return;
+
+	VERUS_QREF_RENDERER;
+	VERUS_QREF_SM;
+
+	auto cb = renderer.GetCommandBuffer();
+
+	RcCamera cam = *sm.GetCamera();
+	Matrix3 matS = Matrix3::scale(Vector3(4, 1, 4)); // Stretch sky dome.
+	const Matrix4 matSkyDome = cam.GetMatrixVP() * Transform3(matS, Vector3(cam.GetEyePosition()));
+
+	s_ubPerFrame._time_cloudiness_expo.x = _time;
+	s_ubPerFrame._time_cloudiness_expo.y = _clouds._cloudiness;
+	s_ubPerFrame._time_cloudiness_expo.z = 0;
+	s_ubPerFrame._dirToSun = float4(_sun._dirTo.GLM(), 0);
+	s_ubPerFrame._phaseAB = float4(
+		_clouds._phaseA.getX(),
+		_clouds._phaseA.getY(),
+		_clouds._phaseB.getX(),
+		_clouds._phaseB.getY());
+	s_ubPerFrame._fogColor = float4(_fog._color.GLM(), 0);
+	_skyDome.CopyPosDeqScale(&s_ubPerMeshVS._posDeqScale.x);
+	_skyDome.CopyPosDeqBias(&s_ubPerMeshVS._posDeqBias.x);
+	s_ubPerObject._matWVP = matSkyDome.UniformBufferFormat();
+
+	_skyDome.BindGeo(cb);
+
+	// <Sky>
+	cb->BindPipeline(_pipe[PIPE_SKY]);
+
+	_shader->BeginBindDescriptors();
+	cb->BindDescriptors(_shader, 0);
+	cb->BindDescriptors(_shader, 1, _cshSkyFS);
+	cb->BindDescriptors(_shader, 2);
+	cb->BindDescriptors(_shader, 3);
+	_shader->EndBindDescriptors();
+
+	cb->DrawIndexed(_skyDome.GetIndexCount());
+	// </Sky>
+
+	// <Clouds>
+	cb->BindPipeline(_pipe[PIPE_CLOUDS]);
+
+	_shader->BeginBindDescriptors();
+	cb->BindDescriptors(_shader, 0);
+	cb->BindDescriptors(_shader, 1, _cshSkyFS);
+	cb->BindDescriptors(_shader, 2);
+	cb->BindDescriptors(_shader, 3);
+	_shader->EndBindDescriptors();
+
+	cb->DrawIndexed(_skyDome.GetIndexCount());
+	// </Clouds>
+}
+
+void Atmosphere::GetColor(int level, float* pOut, float scale)
+{
+	if (!_texSky.IsLoaded())
+		return;
+	const float texCoord = _time * 256;
+	const int intPartA = static_cast<int>(texCoord) & 0xFF;
+	const int intPartB = (intPartA + 1) & 0xFF;
+	const float fractPart = texCoord - intPartA;
+
+	const int offsetA = (level << 10) + (intPartA << 2);
+	const int offsetB = (level << 10) + (intPartB << 2);
+	auto LoadFromSky = [this](float rgb[3], int offset)
+	{
+		rgb[0] = Convert::SRGBToLinear(Convert::Uint8ToUnorm(_texSky.GetData()[offset + 0]));
+		rgb[1] = Convert::SRGBToLinear(Convert::Uint8ToUnorm(_texSky.GetData()[offset + 1]));
+		rgb[2] = Convert::SRGBToLinear(Convert::Uint8ToUnorm(_texSky.GetData()[offset + 2]));
+	};
+	float rgbA[3];
+	float rgbB[3];
+	LoadFromSky(rgbA, offsetA);
+	LoadFromSky(rgbB, offsetB);
+
+	VERUS_FOR(i, 3)
+		pOut[i] = Math::Clamp<float>(Math::Lerp(rgbA[i], rgbB[i], fractPart) * 0.5f * scale, 0, 1);
+}
+
 RcVector3 Atmosphere::GetDirToSun() const
 {
 	return _sun._dirTo;

Verus/src/Scene/Atmosphere.h

@@ -6,6 +6,37 @@ namespace verus
 	{
 		class Atmosphere : public Singleton<Atmosphere>, public Object
 		{
+		public:
+#include "../Shaders/Sky.inc.hlsl"
+
+			enum PIPE
+			{
+				PIPE_SKY,
+				PIPE_CLOUDS,
+				PIPE_COUNT
+			};
+
+			enum TEX
+			{
+				TEX_SKY,
+				TEX_STARS,
+				TEX_SUN,
+				TEX_MOON,
+				TEX_CLOUDS,
+				TEX_CLOUDS_NM,
+				TEX_COUNT
+			};
+
+		private:
+			struct Clouds
+			{
+				Vector4              _phaseA = Vector4(0);
+				Vector4              _phaseB = Vector4(0);
+				float                _speedPhaseA = 0.017f;
+				float                _speedPhaseB = 0.003f;
+				Anim::Damping<float> _cloudiness = 0.25f;
+			};
+
 			struct Fog
 			{
 				Vector3 _color = Vector3(0);
@@ -14,7 +45,7 @@ namespace verus
 				float   _end = 100000;
 				float   _startEx = 0;
 				float   _endEx = 0;
-				float   _density = 1 / 1500.f;
+				float   _density = 0.0002f;
 			};
 
 			struct Sun
@@ -25,12 +56,27 @@ namespace verus
 				Vector3 _color = Vector3(0);
 			};
 
+			static UB_PerFrame      s_ubPerFrame;
+			static UB_PerMaterialFS s_ubPerMaterialFS;
+			static UB_PerMeshVS     s_ubPerMeshVS;
+			static UB_PerObject     s_ubPerObject;
+
+			CGI::ShaderPwn                _shader;
+			CGI::PipelinePwns<PIPE_COUNT> _pipe;
+			CGI::TexturePwns<TEX_COUNT>   _tex;
+			Clouds                        _clouds;
 			Fog                           _fog;
 			Sun                           _sun;
-			Vector3           _colorAmbient = Vector3(0);
+			Vector3                       _ambientColor = Vector3(0);
 			CascadedShadowMap             _shadowMap;
+			Mesh                          _skyDome;
+			CGI::TextureRAM               _texSky;
 			float                         _time = 0.5f;
 			float                         _timeSpeed = 1 / 300.f;
+			float                         _hdrScale = 1000;
+			CGI::CSHandle                 _cshSkyFS;
+			bool                          _night = false;
+			bool                          _async_loaded = false;
 
 		public:
 			Atmosphere();
@@ -39,13 +85,28 @@ namespace verus
 			void Init();
 			void Done();
 
+			void UpdateSun(float time);
 			void Update();
+			void DrawSky();
+
+			void GetColor(int level, float* pOut, float scale);
 
 			// Time:
 			float GetTime() const { return _time; }
 			void SetTime(float x) { _time = x; }
 			void SetTimeSpeed(float x) { _timeSpeed = x; }
 
+			RcVector3 GetAmbientColor() const { return _ambientColor; }
+
+			// Clouds:
+			Anim::Damping<float>& GetCloudiness() { return _clouds._cloudiness; }
+			void SetCloudiness(float x) { _clouds._cloudiness = x; _clouds._cloudiness.ForceTarget(); }
+
+			// Fog:
+			RcVector3 GetFogColor() const { return _fog._color; }
+			float GetFogDensity() const { return _fog._density; }
+			void SetFogDensity(float d) { _fog._density = d; }
+
 			// Sun:
 			RcVector3 GetDirToSun() const;
 			RcVector3 GetSunColor() const;

Verus/src/Scene/BaseMesh.cpp

@@ -328,7 +328,7 @@ void BaseMesh::LoadPrimaryBones()
 			VERUS_FOR(i, 4)
 				it->bi[i] = _skeleton.RemapBoneIndex(it->bi[i]);
 		}
-		BufferDataVB(_vBinding1.data(), 1);
+		UpdateVertexBuffer(_vBinding1.data(), 1);
 	}
 #endif
 }
@@ -367,8 +367,8 @@ void BaseMesh::LoadWarp()
 			_vBinding2[i]._tan[3] = g * SHRT_MAX / 9;
 			_vBinding2[i]._bin[3] = b * SHRT_MAX / 9;
 		}
-		BufferDataVB(_vBinding0.data(), 0);
-		BufferDataVB(_vBinding2.data(), 2);
+		UpdateVertexBuffer(_vBinding0.data(), 0);
+		UpdateVertexBuffer(_vBinding2.data(), 2);
 	}
 
 	vData.clear();

Verus/src/Scene/BaseMesh.h

@@ -96,7 +96,7 @@ namespace verus
 
 			// GPU:
 			virtual void CreateDeviceBuffers() {}
-			virtual void BufferDataVB(const void* p, int binding) {}
+			virtual void UpdateVertexBuffer(const void* p, int binding) {}
 
 			// Bounds:
 			void GetBounds(RPoint3 mn, RPoint3 mx) const;

Verus/src/Scene/Grass.cpp

@@ -0,0 +1,488 @@
+#include "verus.h"
+
+using namespace verus;
+using namespace verus::Scene;
+
+CGI::ShaderPwn    Grass::s_shader;
+Grass::UB_GrassVS Grass::s_ubGrassVS;
+Grass::UB_GrassFS Grass::s_ubGrassFS;
+
+Grass::Grass()
+{
+	VERUS_CT_ASSERT(16 == sizeof(Vertex));
+	VERUS_FOR(i, 4)
+	{
+		VERUS_FOR(j, 4)
+		{
+			const char u0 = j * 25 + 1;
+			const char v0 = i * 25 + 1;
+			const char u1 = u0 + 25 - 2;
+			const char v1 = v0 + 25 - 2;
+			const int index = (i << 2) + j;
+			_bushTexCoords[index][0][0] = u0; _bushTexCoords[index][0][1] = v1;
+			_bushTexCoords[index][1][0] = u0; _bushTexCoords[index][1][1] = v0;
+			_bushTexCoords[index][2][0] = u1; _bushTexCoords[index][2][1] = v0;
+			_bushTexCoords[index][3][0] = u1; _bushTexCoords[index][3][1] = v1;
+			_bushTexCoords[index][4][0] = u0; _bushTexCoords[index][4][1] = v1;
+			_bushTexCoords[index][5][0] = u0; _bushTexCoords[index][5][1] = v0;
+			_bushTexCoords[index][6][0] = u1; _bushTexCoords[index][6][1] = v0;
+			_bushTexCoords[index][7][0] = u1; _bushTexCoords[index][7][1] = v1;
+		}
+	}
+}
+
+Grass::~Grass()
+{
+	Done();
+}
+
+void Grass::InitStatic()
+{
+	s_shader.Init("[Shaders]:DS_Grass.hlsl");
+	s_shader->CreateDescriptorSet(0, &s_ubGrassVS, sizeof(s_ubGrassVS), 100,
+		{
+			CGI::Sampler::linearMipL,
+			CGI::Sampler::linearMipN,
+			CGI::Sampler::nearestMipN,
+		}, CGI::ShaderStageFlags::vs);
+	s_shader->CreateDescriptorSet(1, &s_ubGrassFS, sizeof(s_ubGrassFS), 100,
+		{
+			CGI::Sampler::aniso
+		}, CGI::ShaderStageFlags::fs);
+	s_shader->CreatePipelineLayout();
+}
+
+void Grass::DoneStatic()
+{
+	s_shader.Done();
+}
+
+void Grass::Init(RTerrain terrain)
+{
+	VERUS_INIT();
+	VERUS_QREF_RENDERER;
+	VERUS_QREF_CONST_SETTINGS;
+
+	_pTerrain = &terrain;
+
+	_bushMask = 0;
+
+	if (!Math::IsPowerOfTwo(_pTerrain->GetMapSide()))
+		throw VERUS_RECOVERABLE << "Init(), mapSide must be power of two";
+
+	_mapSide = _pTerrain->GetMapSide();
+	_mapShift = Math::HighestBit(_mapSide);
+
+	const int patchSide = _mapSide >> 4;
+	const int patchShift = _mapShift - 4;
+	const int patchCount = patchSide * patchSide;
+	const int maxInstances = 128 * 8;
+
+	_vPatches.resize(patchCount);
+
+	CreateBuffers();
+	_vInstanceBuffer.resize(maxInstances);
+
+	CGI::GeometryDesc geoDesc;
+	const CGI::VertexInputAttrDesc viaDesc[] =
+	{
+		{ 0, offsetof(Vertex, _pos),               CGI::IeType::shorts, 4, CGI::IeUsage::position, 0},
+		{ 0, offsetof(Vertex, _tc),                CGI::IeType::shorts, 4, CGI::IeUsage::texCoord, 0},
+		{-1, offsetof(PerInstanceData, _patchPos), CGI::IeType::shorts, 4, CGI::IeUsage::instData, 0},
+		CGI::VertexInputAttrDesc::End()
+	};
+	geoDesc._pVertexInputAttrDesc = viaDesc;
+	const int strides[] = { sizeof(Vertex), sizeof(PerInstanceData), 0 };
+	geoDesc._pStrides = strides;
+	_geo.Init(geoDesc);
+	_geo->CreateVertexBuffer(_vPatchMeshVB.size(), 0);
+	_geo->CreateVertexBuffer(maxInstances, 1);
+	_geo->CreateIndexBuffer(_vPatchMeshIB.size());
+	_geo->UpdateVertexBuffer(_vPatchMeshVB.data(), 0);
+	_geo->UpdateIndexBuffer(_vPatchMeshIB.data());
+
+	{
+		CGI::PipelineDesc pipeDesc(_geo, s_shader, "#", renderer.GetDS().GetRenderPassHandle());
+		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[2] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._rasterizationState._cullMode = CGI::CullMode::none;
+		_pipe[PIPE_MAIN].Init(pipeDesc);
+	}
+	{
+		CGI::PipelineDesc pipeDesc(_geo, s_shader, "#Billboards", renderer.GetDS().GetRenderPassHandle());
+		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[2] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._topology = CGI::PrimitiveTopology::pointList;
+		_pipe[PIPE_BILLBOARDS].Init(pipeDesc);
+	}
+
+	const int texW = 1024;
+	const int texH = 1024;
+	_vTextureSubresData.clear();
+	_vTextureSubresData.resize(texW * texH);
+
+	CGI::TextureDesc texDesc;
+	texDesc._format = CGI::Format::srgbR8G8B8A8;
+	texDesc._width = texW;
+	texDesc._height = texH;
+	texDesc._mipLevels = 0;
+	texDesc._flags = CGI::TextureDesc::Flags::generateMips;
+	_tex.Init(texDesc);
+	_tex->UpdateSubresource(_vTextureSubresData.data());
+	_tex->GenerateMips();
+
+	s_shader->FreeDescriptorSet(_cshVS);
+	_cshVS = s_shader->BindDescriptorSetTextures(0,
+		{
+			_pTerrain->GetHeightmapTexture(),
+			_pTerrain->GetNormalsTexture(),
+			_pTerrain->GetMainLayerTexture(),
+		});
+	s_shader->FreeDescriptorSet(_cshFS);
+	_cshFS = s_shader->BindDescriptorSetTextures(1,
+		{
+			_tex
+		});
+
+	_vMagnets.resize(16);
+	std::fill(_vMagnets.begin(), _vMagnets.end(), Magnet());
+}
+
+void Grass::Done()
+{
+	s_shader->FreeDescriptorSet(_cshVS);
+	s_shader->FreeDescriptorSet(_cshFS);
+	VERUS_DONE(Grass);
+}
+
+void Grass::Update()
+{
+	VERUS_UPDATE_ONCE_CHECK;
+
+	// Async texture loading:
+	VERUS_FOR(i, s_maxBushTypes)
+	{
+		if (_texLoaded[i].IsInitialized() && _texLoaded[i].IsLoaded())
+			OnTextureLoaded(i);
+	}
+}
+
+void Grass::Layout()
+{
+	VERUS_QREF_SM;
+
+	const float zFar = sm.GetCamera()->GetZFar();
+	sm.GetCamera()->SetZFar(128);
+	sm.GetCamera()->Update();
+
+	Math::RQuadtreeIntegral quadtree = _pTerrain->GetQuadtree();
+	_visiblePatchCount = 0;
+	quadtree.SetDelegate(this);
+	quadtree.TraverseVisible();
+	quadtree.SetDelegate(_pTerrain);
+
+	sm.GetCamera()->SetZFar(zFar);
+	sm.GetCamera()->Update();
+}
+
+void Grass::Draw()
+{
+	if (!_visiblePatchCount)
+		return;
+
+	VERUS_QREF_ATMO;
+	VERUS_QREF_RENDERER;
+	VERUS_QREF_SM;
+	VERUS_RT_ASSERT(_pTerrain->GetMapSide() == _mapSide);
+
+	auto cb = renderer.GetCommandBuffer();
+
+	const bool drawingDepth = Scene::SceneManager::IsDrawingDepth(Scene::DrawDepth::automatic);
+
+	const UINT32 bushMask = _bushMask & 0xFF; // Use only first 8 bushes, next 8 are reserved.
+	const int half = _mapSide >> 1;
+
+	s_ubGrassVS._matW = Transform3::UniformBufferFormatIdentity();
+	s_ubGrassVS._matWV = sm.GetCamera()->GetMatrixV().UniformBufferFormat();
+	s_ubGrassVS._matVP = sm.GetCamera()->GetMatrixVP().UniformBufferFormat();
+	s_ubGrassVS._matP = sm.GetCamera()->GetMatrixP().UniformBufferFormat();
+	s_ubGrassVS._phase_mapSideInv_bushMask.x = _phase;
+	s_ubGrassVS._phase_mapSideInv_bushMask.y = 1.f / _mapSide;
+	s_ubGrassVS._phase_mapSideInv_bushMask.z = *(float*)&bushMask;
+	s_ubGrassVS._posEye = float4(atmo.GetEyePosition().GLM(), 0);
+	s_ubGrassVS._viewportSize = cb->GetViewportSize().GLM();
+	//s_ubGrassVS._wind = Vector3(VMath::normalize(atmo.GetWindVelocity()) * bendByWind);
+	//s_ubGrassVS._wind.w = m_shaker.Get();
+
+	cb->BindVertexBuffers(_geo);
+	cb->BindIndexBuffer(_geo);
+
+	s_shader->BeginBindDescriptors();
+
+	cb->BindPipeline(_pipe[PIPE_BILLBOARDS]);
+	cb->BindDescriptors(s_shader, 0, _cshVS);
+	cb->BindDescriptors(s_shader, 1, _cshFS);
+	int pointSpriteInstCount = 0;
+	VERUS_FOR(i, _visiblePatchCount)
+	{
+		RcPatch patch = _vPatches[i];
+		if (patch.type & 0x2)
+		{
+			const int at = _instanceCount + pointSpriteInstCount;
+			_vInstanceBuffer[at]._patchPos[0] = patch.j - half;
+			_vInstanceBuffer[at]._patchPos[1] = patch.h;
+			_vInstanceBuffer[at]._patchPos[2] = patch.i - half;
+			_vInstanceBuffer[at]._patchPos[3] = 0;
+			pointSpriteInstCount++;
+		}
+	}
+	cb->Draw(_bbVertCount, pointSpriteInstCount, _vertCount, _instanceCount);
+	_instanceCount += pointSpriteInstCount;
+
+	cb->BindPipeline(_pipe[PIPE_MAIN]);
+	cb->BindDescriptors(s_shader, 0, _cshVS);
+	cb->BindDescriptors(s_shader, 1, _cshFS);
+	int meshInstCount = 0;
+	VERUS_FOR(i, _visiblePatchCount)
+	{
+		RcPatch patch = _vPatches[i];
+		if (patch.type & 0x1)
+		{
+			const int at = _instanceCount + meshInstCount;
+			_vInstanceBuffer[at]._patchPos[0] = patch.j - half;
+			_vInstanceBuffer[at]._patchPos[1] = patch.h;
+			_vInstanceBuffer[at]._patchPos[2] = patch.i - half;
+			_vInstanceBuffer[at]._patchPos[3] = 0;
+			meshInstCount++;
+		}
+	}
+	cb->DrawIndexed(_vPatchMeshIB.size(), meshInstCount, 0, 0, _instanceCount);
+	_instanceCount += meshInstCount;
+
+	s_shader->EndBindDescriptors();
+
+	_geo->UpdateVertexBuffer(_vInstanceBuffer.data(), 1);
+}
+
+void Grass::ResetInstanceCount()
+{
+	_instanceCount = 0;
+}
+
+void Grass::QuadtreeIntegral_ProcessVisibleNode(const short* ij, RcPoint3 center)
+{
+	VERUS_QREF_ATMO;
+
+	const RcPoint3 eyePos = atmo.GetEyePosition();
+	const float distSq = VMath::distSqr(eyePos, center);
+	if (distSq >= 128 * 128.f)
+		return;
+
+	Patch patch;
+	patch.i = ij[0];
+	patch.j = ij[1];
+	patch.h = center.getY() * 100;
+	patch.type = 0;
+	if (distSq <= 64 * 64.f)
+		patch.type |= 0x1;
+	if (distSq >= 16 * 16.f)
+		patch.type |= 0x2;
+	_vPatches[_visiblePatchCount++] = patch;
+}
+
+void Grass::QuadtreeIntegral_GetHeights(const short* ij, float height[2])
+{
+	VERUS_RT_FAIL(__FUNCTION__);
+}
+
+void Grass::CreateBuffers()
+{
+	VERUS_QREF_CONST_SETTINGS;
+
+	const int density = settings._sceneGrassDensity;
+	const int side = static_cast<int>(sqrt(float(density)));
+	const int count = side * side;
+	const float step = 16.f / side;
+
+	Random random(9052);
+	auto RandomStep = [&random, step]()
+	{
+		return (0.1f + 0.8f * random.NextFloat()) * step;
+	};
+	auto RandomScale = [&random]()
+	{
+		return random.NextFloat() * 0.4f;
+	};
+
+	static const Point3 bushVerts[8] =
+	{
+		Point3(-0.5f, 0.f, 0.f), // A (0)
+		Point3(-0.5f, 1.f, 0.f), // B (1)
+		Point3(+0.5f, 1.f, 0.f), // C (2)
+		Point3(+0.5f, 0.f, 0.f), // D (3)
+		Point3(0.f, 0.f, -0.5f), // A2 (4)
+		Point3(0.f, 1.f, -0.5f), // B2 (5)
+		Point3(0.f, 1.f, +0.5f), // C2 (6)
+		Point3(0.f, 0.f, +0.5f)  // D2 (7)
+	};
+	static const UINT16 bushIndices[12] =
+	{
+		1, 0, 2,
+		2, 0, 3,
+		5, 6, 4,
+		4, 6, 7
+	};
+
+	struct Bush
+	{
+		float _x, _z;
+		Point3 _vertPos[8];
+		float _phaseShift;
+	};
+	Vector<Bush> vBushes;
+	vBushes.reserve(count);
+
+	VERUS_FOR(i, side)
+	{
+		VERUS_FOR(j, side)
+		{
+			const Point3 bushPos(
+				j * step + RandomStep(),
+				0,
+				i * step + RandomStep());
+			const Vector3 scale(
+				0.8f + RandomScale(),
+				0.8f + RandomScale(),
+				0.8f + RandomScale());
+			const float phaseSin =
+				sin(bushPos.getX() * (VERUS_2PI / 16)) *
+				sin(bushPos.getZ() * (VERUS_2PI / 16)) * 0.5f + 0.5f;
+			const float phase = phaseSin * 0.75f + random.NextFloat() * 0.25f;
+			const float angle = random.NextFloat() * VERUS_2PI;
+
+			const Matrix3 matR = Matrix3::rotationY(angle);
+			const Transform3 tr = VMath::appendScale(Transform3(matR, Vector3(bushPos)), scale);
+
+			Bush bush;
+			bush._x = bushPos.getX();
+			bush._z = bushPos.getZ();
+			VERUS_FOR(k, 8)
+				bush._vertPos[k] = tr * bushVerts[k];
+			bush._phaseShift = phase;
+			vBushes.push_back(bush);
+		}
+	}
+
+	std::shuffle(vBushes.begin(), vBushes.end(), random.GetGenerator());
+
+	const int indexedMeshVertCount = count * 8;
+	_vertCount = 0;
+	_bbVertCount = 0;
+	_vPatchMeshVB.clear();
+	_vPatchMeshVB.resize(indexedMeshVertCount + count);
+	_vPatchMeshIB.clear();
+	_vPatchMeshIB.reserve(count * 12);
+	for (const auto& bush : vBushes)
+	{
+		const int indexOffset = _vertCount;
+		VERUS_FOR(i, 8)
+		{
+			Vertex vertex;
+			vertex._pos[0] = static_cast<short>(bush._vertPos[i].getX() * 1000);
+			vertex._pos[1] = static_cast<short>(bush._vertPos[i].getY() * 1000);
+			vertex._pos[2] = static_cast<short>(bush._vertPos[i].getZ() * 1000);
+			vertex._pos[3] = bush._phaseShift * 99;
+			vertex._tc[0] = _bushTexCoords[0][i][0];
+			vertex._tc[1] = _bushTexCoords[0][i][1];
+			vertex._tc[2] = static_cast<short>(bush._x * 1000);
+			vertex._tc[3] = static_cast<short>(bush._z * 1000);
+			_vPatchMeshVB[_vertCount++] = vertex;
+
+			if (1 == i) // Index with top-left tex coords:
+			{
+				const int offset = indexedMeshVertCount + _bbVertCount;
+				_vPatchMeshVB[offset] = vertex;
+				_bbVertCount++;
+			}
+		}
+		VERUS_FOR(i, 12)
+			_vPatchMeshIB.push_back(bushIndices[i] + indexOffset);
+	}
+}
+
+void Grass::ResetAllTextures()
+{
+	_bushMask = 0;
+}
+
+void Grass::SetTexture(int layer, CSZ url, CSZ url2)
+{
+	VERUS_RT_ASSERT(url);
+	_texLoaded[layer].LoadDDS(url);
+	if (layer < 8)
+		SetTexture(layer + 8, url2 ? url2 : url);
+}
+
+void Grass::OnTextureLoaded(int layer)
+{
+	VERUS_QREF_RENDERER;
+
+	const int srcW = 256;
+	const int srcH = 256;
+	const int dstW = 1024;
+	const int xFrom = (layer & 0x3) << 8;
+	const int yFrom = (layer >> 2) << 8;
+
+	const UINT32* pSrc = reinterpret_cast<const UINT32*>(_texLoaded[layer].GetData());
+
+	VERUS_FOR(y, srcH)
+	{
+		const int ySrcOffset = y * srcW;
+		const int yDstOffset = (yFrom + y) * dstW;
+		memcpy(_vTextureSubresData.data() + yDstOffset + xFrom, pSrc + ySrcOffset, srcW * sizeof(UINT32));
+	}
+	_texLoaded[layer].Done();
+
+	_tex->UpdateSubresource(_vTextureSubresData.data());
+	_tex->GenerateMips();
+
+	_bushMask |= (1 << layer);
+}
+
+int Grass::BeginMagnet(RcPoint3 pos, float radius)
+{
+	const int count = _vMagnets.size();
+	VERUS_FOR(i, count)
+	{
+		if (!_vMagnets[i]._active)
+		{
+			_vMagnets[i]._active = true;
+			_vMagnets[i]._pos = pos;
+			_vMagnets[i]._radius = radius;
+			_vMagnets[i]._radiusSq = radius * radius;
+			_vMagnets[i]._radiusSqInv = 1 / _vMagnets[i]._radiusSq;
+			_vMagnets[i]._strength = 0.5f;
+			return i;
+		}
+	}
+	return -1;
+}
+
+void Grass::EndMagnet(int index)
+{
+	_vMagnets[index]._active = false;
+}
+
+void Grass::UpdateMagnet(int index, RcPoint3 pos, float radius)
+{
+	_vMagnets[index]._active = true;
+	_vMagnets[index]._pos = pos;
+	if (radius)
+	{
+		_vMagnets[index]._radius = radius;
+		_vMagnets[index]._radiusSq = radius * radius;
+		_vMagnets[index]._radiusSqInv = 1 / _vMagnets[index]._radiusSq;
+	}
+}

Verus/src/Scene/Grass.h

@@ -0,0 +1,114 @@
+#pragma once
+
+namespace verus
+{
+	namespace Scene
+	{
+		class Grass : public Object, public Math::QuadtreeIntegralDelegate
+		{
+		public:
+#include "../Shaders/DS_Grass.inc.hlsl"
+
+			enum PIPE
+			{
+				PIPE_MAIN,
+				PIPE_BILLBOARDS,
+				PIPE_COUNT
+			};
+
+			struct Vertex
+			{
+				short _pos[4]; // {xyz, phaseShift}.
+				short _tc[4]; // {tc0, center.xz}.
+			};
+			VERUS_TYPEDEFS(Vertex);
+
+			struct PerInstanceData
+			{
+				short _patchPos[4];
+			};
+			VERUS_TYPEDEFS(PerInstanceData);
+
+			struct Patch
+			{
+				short i;
+				short j;
+				short h;
+				short type;
+			};
+			VERUS_TYPEDEFS(Patch);
+
+			class Magnet
+			{
+			public:
+				Point3 _pos;
+				float  _radius = 1;
+				float  _radiusSq = 1;
+				float  _radiusSqInv = 1;
+				float  _strength = 1;
+				bool   _active = false;
+			};
+			VERUS_TYPEDEFS(Magnet);
+
+			static const int s_maxBushTypes = 16;
+
+		private:
+			static CGI::ShaderPwn       s_shader;
+			static UB_GrassVS           s_ubGrassVS;
+			static UB_GrassFS           s_ubGrassFS;
+
+			CGI::GeometryPwn              _geo;
+			CGI::PipelinePwns<PIPE_COUNT> _pipe;
+			CGI::TexturePwn               _tex;
+			CGI::TextureRAM               _texLoaded[s_maxBushTypes];
+			PTerrain                      _pTerrain = nullptr;
+			Vector<Vertex>                _vPatchMeshVB;
+			Vector<UINT16>                _vPatchMeshIB;
+			Vector<PerInstanceData>       _vInstanceBuffer;
+			Vector<UINT32>                _vTextureSubresData;
+			Vector<Patch>                 _vPatches;
+			Vector<Magnet>                _vMagnets;
+			int                           _mapSide = 0;
+			int                           _mapShift = 0;
+			int                           _vertCount = 0;
+			int                           _bbVertCount = 0;
+			int                           _instanceCount = 0;
+			int                           _visiblePatchCount = 0;
+			CGI::CSHandle                 _cshVS;
+			CGI::CSHandle                 _cshFS;
+			char                          _bushTexCoords[s_maxBushTypes][8][2];
+			UINT32                        _bushMask = 0;
+			float                         _phase = 0;
+
+		public:
+			Grass();
+			~Grass();
+
+			static void InitStatic();
+			static void DoneStatic();
+
+			void Init(RTerrain terrain);
+			void Done();
+
+			void Update();
+			void Layout();
+			void Draw();
+
+			void ResetInstanceCount();
+
+			virtual void QuadtreeIntegral_ProcessVisibleNode(const short* ij, RcPoint3 center) override;
+			virtual void QuadtreeIntegral_GetHeights(const short* ij, float height[2]) override;
+
+			VERUS_P(void CreateBuffers());
+
+			void ResetAllTextures();
+			void SetTexture(int layer, CSZ url, CSZ url2 = nullptr);
+			VERUS_P(void OnTextureLoaded(int layer));
+
+			int BeginMagnet(RcPoint3 pos, float radius);
+			void EndMagnet(int index);
+			void UpdateMagnet(int index, RcPoint3 pos, float radius = 0);
+		};
+		VERUS_TYPEDEFS(Grass);
+	}
+}

Verus/src/Scene/MaterialManager.cpp

@@ -14,7 +14,7 @@ Texture::~Texture()
 	Done();
 }
 
-void Texture::Init(CSZ url, bool streamParts, bool sync)
+void Texture::Init(CSZ url, bool streamParts, bool sync, CGI::PcSamplerDesc pSamplerDesc)
 {
 	if (_refCount)
 		return;
@@ -26,6 +26,7 @@ void Texture::Init(CSZ url, bool streamParts, bool sync)
 	CGI::TextureDesc texDesc;
 	texDesc._url = url;
 	texDesc._texturePart = _streamParts ? 512 : 0;
+	texDesc._pSamplerDesc = pSamplerDesc;
 	if (sync)
 		texDesc._flags = CGI::TextureDesc::Flags::sync;
 	_texTiny.Init(texDesc);
@@ -81,12 +82,12 @@ void Texture::Update()
 
 // TexturePtr:
 
-void TexturePtr::Init(CSZ url, bool streamParts, bool sync)
+void TexturePtr::Init(CSZ url, bool streamParts, bool sync, CGI::PcSamplerDesc pSamplerDesc)
 {
 	VERUS_QREF_MM;
 	VERUS_RT_ASSERT(!_p);
 	_p = mm.InsertTexture(url);
-	_p->Init(url, streamParts, sync);
+	_p->Init(url, streamParts, sync, pSamplerDesc);
 }
 
 void TexturePwn::Done()
@@ -368,9 +369,10 @@ void Material::FromString(CSZ txt)
 
 void Material::BindDescriptorSetTextures()
 {
-	VERUS_RT_ASSERT(-1 == _csh);
+	VERUS_RT_ASSERT(!_csh.IsSet());
 	VERUS_RT_ASSERT(IsLoaded());
-	_csh = Scene::Mesh::GetShader()->BindDescriptorSetTextures(1, { _texAlbedo->GetTex(), _texNormal->GetTex() });
+	VERUS_QREF_MM;
+	_csh = Scene::Mesh::GetShader()->BindDescriptorSetTextures(1, { _texAlbedo->GetTex(), _texNormal->GetTex(), mm.GetDetailTexture(), mm.GetStrassTexture() });
 }
 
 bool Material::UpdateMeshUniformBuffer(float motionBlur)
@@ -449,12 +451,23 @@ void MaterialManager::Init()
 
 void MaterialManager::InitCmd()
 {
+	CGI::SamplerDesc strassSamplerDesc;
+	strassSamplerDesc.SetFilter("a");
+	strassSamplerDesc.SetAddressMode("rr");
+	strassSamplerDesc._mipLodBias = -2;
+
 	_texDefaultAlbedo.Init("[Textures]:Default.dds", false, true);
 	_texDefaultNormal.Init("[Textures]:Default.NM.dds", false, true);
-	//_texDetail.Init("[Textures]:Detail.FX.dds");
-	//_texStrass.Init("[Textures]:Strass.dds");
+	_texDetail.Init("[Textures]:Detail.FX.dds", false, true);
+	_texStrass.Init("[Textures]:Strass.dds", false, true, &strassSamplerDesc);
 
-	_cshDefault = Scene::Mesh::GetShader()->BindDescriptorSetTextures(1, { _texDefaultAlbedo->GetTex(), _texDefaultNormal->GetTex() });
+	_cshDefault = Scene::Mesh::GetShader()->BindDescriptorSetTextures(1,
+		{
+			_texDefaultAlbedo->GetTex(),
+			_texDefaultNormal->GetTex(),
+			_texDetail->GetTex(),
+			_texStrass->GetTex()
+		});
 }
 
 void MaterialManager::Done()

Verus/src/Scene/MaterialManager.h

@@ -17,7 +17,7 @@ namespace verus
 			Texture();
 			~Texture();
 
-			void Init(CSZ url, bool streamParts = false, bool sync = false);
+			void Init(CSZ url, bool streamParts = false, bool sync = false, CGI::PcSamplerDesc pSamplerDes = nullptr);
 			bool Done();
 			void AddRef() { _refCount++; }
 
@@ -33,7 +33,7 @@ namespace verus
 		class TexturePtr : public Ptr<Texture>
 		{
 		public:
-			void Init(CSZ url, bool streamParts = false, bool sync = false);
+			void Init(CSZ url, bool streamParts = false, bool sync = false, CGI::PcSamplerDesc pSamplerDes = nullptr);
 		};
 		VERUS_TYPEDEFS(TexturePtr);
 
@@ -127,7 +127,7 @@ namespace verus
 			glm::vec4      _userColor = glm::vec4(0, 0, 0, 0);
 			Pick           _userPick;
 			Blending       _blending = Blending::opaque;
-			int            _csh = -1;
+			CGI::CSHandle  _csh;
 			int            _refCount = 0;
 
 			Material();
@@ -152,7 +152,7 @@ namespace verus
 			void FromString(CSZ txt);
 
 			// Mesh interaction:
-			int GetComplexSetHandle() const { return _csh; }
+			CGI::CSHandle GetComplexSetHandle() const { return _csh; }
 			void BindDescriptorSetTextures();
 			bool UpdateMeshUniformBuffer(float motionBlur = 1);
 
@@ -183,7 +183,7 @@ namespace verus
 			TexturePwn    _texDefaultNormal;
 			TexturePwn    _texDetail;
 			TexturePwn    _texStrass;
-			int        _cshDefault = -1; // For missing, non-mandatory materials.
+			CGI::CSHandle _cshDefault; // For missing, non-mandatory materials.
 
 		public:
 			MaterialManager();
@@ -200,7 +200,7 @@ namespace verus
 			CGI::TexturePtr GetDetailTexture() { return _texDetail->GetTex(); }
 			CGI::TexturePtr GetStrassTexture() { return _texStrass->GetTex(); }
 
-			int GetDefaultComplexSetHandle() const { return _cshDefault; }
+			CGI::CSHandle GetDefaultComplexSetHandle() const { return _cshDefault; }
 
 			// Textures:
 			PTexture InsertTexture(CSZ url);

Verus/src/Scene/Mesh.cpp

@@ -4,7 +4,7 @@ using namespace verus;
 using namespace verus::Scene;
 
 CGI::ShaderPwn                      Mesh::s_shader;
-CGI::PipelinePwns<Mesh::PIPE_MAX> Mesh::s_pipe;
+CGI::PipelinePwns<Mesh::PIPE_COUNT> Mesh::s_pipe;
 
 Mesh::UB_PerFrame      Mesh::s_ubPerFrame;
 Mesh::UB_PerMaterialFS Mesh::s_ubPerMaterialFS;
@@ -28,11 +28,13 @@ void Mesh::InitStatic()
 	VERUS_QREF_CONST_SETTINGS;
 
 	s_shader.Init("[Shaders]:DS_Mesh.hlsl");
-	s_shader->CreateDescriptorSet(0, &s_ubPerFrame, sizeof(s_ubPerFrame), settings.GetLimits()._mesh_ubPerFrameCapacity);
+	s_shader->CreateDescriptorSet(0, &s_ubPerFrame, sizeof(s_ubPerFrame), settings.GetLimits()._mesh_ubPerFrameCapacity, {}, CGI::ShaderStageFlags::vs_hs_ds_fs);
 	s_shader->CreateDescriptorSet(1, &s_ubPerMaterialFS, sizeof(s_ubPerMaterialFS), settings.GetLimits()._mesh_ubPerMaterialFSCapacity,
 		{
 			CGI::Sampler::aniso,
-			CGI::Sampler::aniso
+			CGI::Sampler::aniso,
+			CGI::Sampler::aniso,
+			CGI::Sampler::custom
 		}, CGI::ShaderStageFlags::fs);
 	s_shader->CreateDescriptorSet(2, &s_ubPerMeshVS, sizeof(s_ubPerMeshVS), settings.GetLimits()._mesh_ubPerMeshVSCapacity, {}, CGI::ShaderStageFlags::vs);
 	s_shader->CreateDescriptorSet(3, &s_ubSkeletonVS, sizeof(s_ubSkeletonVS), settings.GetLimits()._mesh_ubSkinningVSCapacity, {}, CGI::ShaderStageFlags::vs);
@@ -69,29 +71,92 @@ void Mesh::Done()
 void Mesh::BindPipeline(PIPE pipe, CGI::CommandBufferPtr cb)
 {
 	VERUS_QREF_RENDERER;
+	VERUS_QREF_CONST_SETTINGS;
+	VERUS_QREF_ATMO;
+
+	if (!settings._gpuTessellation) // Fallback:
+	{
+		switch (pipe)
+		{
+		case PIPE_TESS:         pipe = PIPE_MAIN; break;
+		case PIPE_TESS_ROBOTIC: pipe = PIPE_ROBOTIC; break;
+		case PIPE_TESS_SKINNED: pipe = PIPE_SKINNED; break;
+		};
+	}
+
 	if (!s_pipe[pipe])
 	{
 		static CSZ branches[] =
 		{
 			"#",
+			"#Robotic",
+			"#Skinned",
+
+			"#Instanced",
+
+			"#Depth",
 			"#DepthRobotic",
 			"#DepthSkinned",
-			"#Instanced",
-			"#Robotic",
-			"#Skinned"
+
+			"#Tess",
+			"#TessRobotic",
+			"#TessSkinned",
+
+			"#SolidColor",
+			"#SolidColorRobotic",
+			"#SolidColorSkinned",
 		};
+
 		CGI::PipelineDesc pipeDesc(_geo, s_shader, branches[pipe], renderer.GetDS().GetRenderPassHandle());
-		switch (pipe)
+
+		auto SetBlendEqsForDS = [&pipeDesc]()
 		{
-		case PIPE_DEPTH_ROBOTIC:
-		case PIPE_DEPTH_SKINNED:
-			pipeDesc._colorAttachBlendEqs[0] = "";
-			pipeDesc.EnableDepthBias();
-		default:
 			pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_OFF;
 			pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_OFF;
 			pipeDesc._colorAttachBlendEqs[2] = VERUS_COLOR_BLEND_OFF;
-			pipeDesc._colorAttachBlendEqs[3] = VERUS_COLOR_BLEND_OFF;
+		};
+
+		switch (pipe)
+		{
+		case PIPE_MAIN:
+		case PIPE_ROBOTIC:
+		case PIPE_SKINNED:
+		{
+			SetBlendEqsForDS();
+		}
+		break;
+		case PIPE_DEPTH:
+		case PIPE_DEPTH_ROBOTIC:
+		case PIPE_DEPTH_SKINNED:
+		{
+			pipeDesc._colorAttachBlendEqs[0] = "";
+			pipeDesc._renderPassHandle = atmo.GetShadowMap().GetRenderPassHandle();
+		}
+		break;
+		case PIPE_TESS:
+		case PIPE_TESS_ROBOTIC:
+		case PIPE_TESS_SKINNED:
+		{
+			SetBlendEqsForDS();
+			if (settings._gpuTessellation)
+				pipeDesc._topology = CGI::PrimitiveTopology::patchList3;
+		}
+		break;
+		case PIPE_WIREFRAME:
+		case PIPE_WIREFRAME_ROBOTIC:
+		case PIPE_WIREFRAME_SKINNED:
+		{
+			SetBlendEqsForDS();
+			pipeDesc._rasterizationState._polygonMode = CGI::PolygonMode::line;
+			pipeDesc._rasterizationState._depthBiasEnable = true;
+			pipeDesc._rasterizationState._depthBiasConstantFactor = -1000;
+			pipeDesc._depthCompareOp = CGI::CompareOp::lessOrEqual;
+		}
+		break;
+		default:
+		{
+			SetBlendEqsForDS();
+		}
 		}
 		pipeDesc._vertexInputBindingsFilter = _bindingsMask;
 		s_pipe[pipe].Init(pipeDesc);
@@ -99,6 +164,29 @@ void Mesh::BindPipeline(PIPE pipe, CGI::CommandBufferPtr cb)
 	cb->BindPipeline(s_pipe[pipe]);
 }
 
+void Mesh::BindPipeline(CGI::CommandBufferPtr cb, bool allowTess)
+{
+	VERUS_QREF_ATMO;
+	if (_skeleton.IsInitialized())
+	{
+		if (atmo.GetShadowMap().IsRendering())
+			BindPipeline(Scene::Mesh::PIPE_DEPTH_SKINNED, cb);
+		else if (allowTess)
+			BindPipeline(Scene::Mesh::PIPE_TESS_SKINNED, cb);
+		else
+			BindPipeline(Scene::Mesh::PIPE_SKINNED, cb);
+	}
+	else
+	{
+		if (atmo.GetShadowMap().IsRendering())
+			BindPipeline(Scene::Mesh::PIPE_DEPTH, cb);
+		else if (allowTess)
+			BindPipeline(Scene::Mesh::PIPE_TESS, cb);
+		else
+			BindPipeline(Scene::Mesh::PIPE_MAIN, cb);
+	}
+}
+
 void Mesh::BindGeo(CGI::CommandBufferPtr cb)
 {
 	BindGeo(cb, _bindingsMask);
@@ -112,10 +200,12 @@ void Mesh::BindGeo(CGI::CommandBufferPtr cb, UINT32 bindingsFilter)
 
 void Mesh::UpdateUniformBufferPerFrame()
 {
+	VERUS_QREF_RENDERER;
 	VERUS_QREF_SM;
 	s_ubPerFrame._matV = sm.GetCamera()->GetMatrixV().UniformBufferFormat();
-	s_ubPerFrame._matP = sm.GetCamera()->GetMatrixP().UniformBufferFormat();
 	s_ubPerFrame._matVP = sm.GetCamera()->GetMatrixVP().UniformBufferFormat();
+	s_ubPerFrame._matP = sm.GetCamera()->GetMatrixP().UniformBufferFormat();
+	s_ubPerFrame._viewportSize = renderer.GetCommandBuffer()->GetViewportSize().GLM();
 }
 
 void Mesh::UpdateUniformBufferPerMaterialFS()
@@ -144,17 +234,17 @@ void Mesh::UpdateUniformBufferSkeletonVS()
 	_skeleton.UpdateUniformBufferArray(s_ubSkeletonVS._vMatBones);
 }
 
-void Mesh::UpdateUniformBufferPerObject(RcTransform3 tr)
+void Mesh::UpdateUniformBufferPerObject(RcTransform3 tr, RcVector4 color)
 {
 	s_ubPerObject._matW = tr.UniformBufferFormat();
-	s_ubPerObject._userColor = Vector4(1, 0, 0, 1).GLM();
+	s_ubPerObject._userColor = color.GLM();
 }
 
-void Mesh::UpdateUniformBufferPerObject(Point3 pos)
+void Mesh::UpdateUniformBufferPerObject(Point3 pos, RcVector4 color)
 {
 	const Transform3 matT = Transform3::translation(Vector3(pos));
 	s_ubPerObject._matW = matT.UniformBufferFormat();
-	s_ubPerObject._userColor = Vector4(1, 0, 0, 1).GLM();
+	s_ubPerObject._userColor = color.GLM();
 }
 
 void Mesh::CreateDeviceBuffers()
@@ -162,25 +252,25 @@ void Mesh::CreateDeviceBuffers()
 	_bindingsMask = 0;
 
 	CGI::GeometryDesc geoDesc;
-	const CGI::InputElementDesc ied[] =
+	const CGI::VertexInputAttrDesc viaDesc[] =
 	{
 		{0, offsetof(VertexInputBinding0, _pos), CGI::IeType::shorts, 4, CGI::IeUsage::position, 0},
 		{0, offsetof(VertexInputBinding0, _tc0), CGI::IeType::shorts, 2, CGI::IeUsage::texCoord, 0},
 		{0, offsetof(VertexInputBinding0, _nrm), CGI::IeType::ubytes, 4, CGI::IeUsage::normal, 0},
-		{1, offsetof(VertexInputBinding1, _bw),  CGI::IeType::shorts, 4, CGI::IeUsage::texCoord, 4},
-		{1, offsetof(VertexInputBinding1, _bi),  CGI::IeType::shorts, 4, CGI::IeUsage::texCoord, 5},
-		{2, offsetof(VertexInputBinding2, _tan), CGI::IeType::shorts, 4, CGI::IeUsage::tangent,  6},
-		{2, offsetof(VertexInputBinding2, _bin), CGI::IeType::shorts, 4, CGI::IeUsage::binormal, 7},
+		{1, offsetof(VertexInputBinding1, _bw),  CGI::IeType::shorts, 4, CGI::IeUsage::blendWeights, 0},
+		{1, offsetof(VertexInputBinding1, _bi),  CGI::IeType::shorts, 4, CGI::IeUsage::blendIndices, 0},
+		{2, offsetof(VertexInputBinding2, _tan), CGI::IeType::shorts, 4, CGI::IeUsage::tangent, 0},
+		{2, offsetof(VertexInputBinding2, _bin), CGI::IeType::shorts, 4, CGI::IeUsage::binormal, 0},
 		{3, offsetof(VertexInputBinding3, _tc1), CGI::IeType::shorts, 2, CGI::IeUsage::texCoord, 1},
 		{3, offsetof(VertexInputBinding3, _clr), CGI::IeType::ubytes, 4, CGI::IeUsage::color, 0},
 
-		{-4, offsetof(PerInstanceData, _matPart0), CGI::IeType::floats, 4, CGI::IeUsage::texCoord, 8},
-		{-4, offsetof(PerInstanceData, _matPart1), CGI::IeType::floats, 4, CGI::IeUsage::texCoord, 9},
-		{-4, offsetof(PerInstanceData, _matPart2), CGI::IeType::floats, 4, CGI::IeUsage::texCoord, 10},
-		{-4, offsetof(PerInstanceData, _instData), CGI::IeType::floats, 4, CGI::IeUsage::texCoord, 11},
-		CGI::InputElementDesc::End()
+		{-4, offsetof(PerInstanceData, _matPart0), CGI::IeType::floats, 4, CGI::IeUsage::instData, 0},
+		{-4, offsetof(PerInstanceData, _matPart1), CGI::IeType::floats, 4, CGI::IeUsage::instData, 1},
+		{-4, offsetof(PerInstanceData, _matPart2), CGI::IeType::floats, 4, CGI::IeUsage::instData, 2},
+		{-4, offsetof(PerInstanceData, _instData), CGI::IeType::floats, 4, CGI::IeUsage::instData, 3},
+		CGI::VertexInputAttrDesc::End()
 	};
-	geoDesc._pInputElementDesc = ied;
+	geoDesc._pVertexInputAttrDesc = viaDesc;
 	const int strides[] = { sizeof(VertexInputBinding0), sizeof(VertexInputBinding1), sizeof(VertexInputBinding2), sizeof(VertexInputBinding3), sizeof(PerInstanceData), 0 };
 	geoDesc._pStrides = strides;
 	geoDesc._32BitIndices = _vIndices.empty();
@@ -236,7 +326,7 @@ void Mesh::CreateDeviceBuffers()
 	}
 }
 
-void Mesh::BufferDataVB(const void* p, int binding)
+void Mesh::UpdateVertexBuffer(const void* p, int binding)
 {
 	_geo->UpdateVertexBuffer(p, binding);
 }

Verus/src/Scene/Mesh.h

@@ -12,12 +12,24 @@ namespace verus
 			enum PIPE
 			{
 				PIPE_MAIN,
-				PIPE_DEPTH_ROBOTIC,
-				PIPE_DEPTH_SKINNED,
-				PIPE_INSTANCED,
 				PIPE_ROBOTIC,
 				PIPE_SKINNED,
-				PIPE_MAX
+
+				PIPE_INSTANCED,
+
+				PIPE_DEPTH,
+				PIPE_DEPTH_ROBOTIC,
+				PIPE_DEPTH_SKINNED,
+
+				PIPE_TESS,
+				PIPE_TESS_ROBOTIC,
+				PIPE_TESS_SKINNED,
+
+				PIPE_WIREFRAME,
+				PIPE_WIREFRAME_ROBOTIC,
+				PIPE_WIREFRAME_SKINNED,
+
+				PIPE_COUNT
 			};
 
 			struct PerInstanceData
@@ -30,7 +42,7 @@ namespace verus
 
 		private:
 			static CGI::ShaderPwn                s_shader;
-			static CGI::PipelinePwns<PIPE_MAX> s_pipe;
+			static CGI::PipelinePwns<PIPE_COUNT> s_pipe;
 			static UB_PerFrame                   s_ubPerFrame;
 			static UB_PerMaterialFS              s_ubPerMaterialFS;
 			static UB_PerMeshVS                  s_ubPerMeshVS;
@@ -64,6 +76,7 @@ namespace verus
 			void Done();
 
 			void BindPipeline(PIPE pipe, CGI::CommandBufferPtr cb);
+			void BindPipeline(CGI::CommandBufferPtr cb, bool allowTess = true);
 			void BindGeo(CGI::CommandBufferPtr cb);
 			void BindGeo(CGI::CommandBufferPtr cb, UINT32 bindingsFilter);
 
@@ -73,12 +86,12 @@ namespace verus
 			void UpdateUniformBufferPerMaterialFS();
 			void UpdateUniformBufferPerMeshVS();
 			void UpdateUniformBufferSkeletonVS();
-			void UpdateUniformBufferPerObject(RcTransform3 tr);
-			void UpdateUniformBufferPerObject(Point3 pos);
+			void UpdateUniformBufferPerObject(RcTransform3 tr, RcVector4 color = Vector4(0.5f, 0.5f, 0.5f, 1));
+			void UpdateUniformBufferPerObject(Point3 pos, RcVector4 color = Vector4(0.5f, 0.5f, 0.5f, 1));
 
 			CGI::GeometryPtr GetGeometry() const { return _geo; }
 			virtual void CreateDeviceBuffers() override;
-			virtual void BufferDataVB(const void* p, int binding) override;
+			virtual void UpdateVertexBuffer(const void* p, int binding) override;
 
 			// Instancing:
 			void PushInstance(RcTransform3 matW, RcVector4 instData);

Verus/src/Scene/Scene.cpp

@@ -8,9 +8,11 @@ namespace verus
 		Scene::MaterialManager::Make();
 		Scene::SceneManager::Make();
 		Scene::Atmosphere::Make();
+		Scene::Water::Make();
 	}
 	void Free_Scene()
 	{
+		Scene::Water::Free();
 		Scene::Atmosphere::Free();
 		Scene::SceneManager::Free();
 		Scene::MaterialManager::Free();

Verus/src/Scene/Scene.h

@@ -6,8 +6,10 @@
 #include "Mesh.h"
 #include "Terrain.h"
 #include "EditorTerrain.h"
+#include "Grass.h"
 #include "ShadowMap.h"
 #include "Atmosphere.h"
+#include "Water.h"
 #include "MaterialManager.h"
 #include "SceneManager.h"
 #include "Helpers.h"

Verus/src/Scene/ShadowMap.cpp

@@ -24,24 +24,17 @@ void ShadowMap::Init(int side)
 
 	_side = side;
 
-	_rp = renderer->CreateRenderPass(
-		{
-			CGI::RP::Attachment("Depth", CGI::Format::unormD16).LoadOpClear().Layout(CGI::ImageLayout::depthStencilReadOnly),
-		},
-		{
-			CGI::RP::Subpass("Sp0").Color({}).DepthStencil(CGI::RP::Ref("Depth", CGI::ImageLayout::depthStencilAttachment)),
-		},
-		{});
+	_rph = renderer->CreateShadowRenderPass(CGI::Format::unormD24uintS8);
 
 	CGI::TextureDesc texDesc;
 	texDesc._clearValue = Vector4(1);
-	texDesc._format = CGI::Format::unormD16;
+	texDesc._format = CGI::Format::unormD24uintS8;
 	texDesc._width = _side;
 	texDesc._height = _side;
-	texDesc._flags = CGI::TextureDesc::Flags::depthSampled;
+	texDesc._flags = CGI::TextureDesc::Flags::depthSampledR;
 	_tex.Init(texDesc);
 
-	_fb = renderer->CreateFramebuffer(_rp,
+	_fbh = renderer->CreateFramebuffer(_rph,
 		{
 			_tex
 		},
@@ -61,9 +54,9 @@ void ShadowMap::Begin(RcVector3 dirToSun, float zNear, float zFar)
 
 	const Vector3 up(0, 1, 0);
 	Point3 eye, at;
-	const float size = 4096 * 0.01f;
+	const float size = 4096 * 0.005f;
 	if (0 == zFar)
-		zFar = 0x10000 * 0.005f;
+		zFar = 1001;
 
 	at = sm.GetCamera()->GetEyePosition() + sm.GetCamera()->GetFrontDirection() * (size * (1 / 3.f));
 	if (_snapToTexels)
@@ -77,7 +70,7 @@ void ShadowMap::Begin(RcVector3 dirToSun, float zNear, float zFar)
 		atPosShadowSpaceSnapped.setY(atPosShadowSpaceSnapped.getY() - fmod(static_cast<float>(atPosShadowSpaceSnapped.getY()), texelSize));
 		at = (matFromShadowSpace * atPosShadowSpaceSnapped).getXYZ();
 	}
-	eye = at + dirToSun * ((zNear + zFar) * 0.5f);
+	eye = at + dirToSun * ((zFar - zNear) * 0.5f);
 
 	// Setup light space camera and use it (used for terrain draw, etc.):
 	_camera.SetUpDirection(up);
@@ -91,9 +84,9 @@ void ShadowMap::Begin(RcVector3 dirToSun, float zNear, float zFar)
 	_camera.Update();
 	_pSceneCamera = sm.SetCamera(&_camera);
 
-	_config.setX((zFar - zNear) * 2);
+	_config._penumbraScale = (zFar - zNear) * 2;
 
-	renderer.GetCommandBuffer()->BeginRenderPass(_rp, _fb,
+	renderer.GetCommandBuffer()->BeginRenderPass(_rph, _fbh,
 		{
 			_tex->GetClearValue()
 		});
@@ -199,25 +192,20 @@ void CascadedShadowMap::Init(int side)
 	VERUS_QREF_RENDERER;
 
 	_side = side;
+	if (settings._sceneShadowQuality >= App::Settings::ShadowQuality::ultra)
+		_side *= 2;
 
-	_rp = renderer->CreateRenderPass(
-		{
-			CGI::RP::Attachment("Depth", CGI::Format::floatD32).LoadOpClear().Layout(CGI::ImageLayout::depthStencilReadOnly),
-		},
-		{
-			CGI::RP::Subpass("Sp0").Color({}).DepthStencil(CGI::RP::Ref("Depth", CGI::ImageLayout::depthStencilAttachment)),
-		},
-		{});
+	_rph = renderer->CreateShadowRenderPass(CGI::Format::unormD24uintS8);
 
 	CGI::TextureDesc texDesc;
 	texDesc._clearValue = Vector4(1);
-	texDesc._format = CGI::Format::floatD32;
+	texDesc._format = CGI::Format::unormD24uintS8;
 	texDesc._width = _side;
 	texDesc._height = _side;
-	texDesc._flags = CGI::TextureDesc::Flags::depthSampled;
+	texDesc._flags = CGI::TextureDesc::Flags::depthSampledR;
 	_tex.Init(texDesc);
 
-	_fb = renderer->CreateFramebuffer(_rp,
+	_fbh = renderer->CreateFramebuffer(_rph,
 		{
 			_tex
 		},
@@ -251,12 +239,14 @@ void CascadedShadowMap::Begin(RcVector3 dirToSun, float zNear, float zFar, int s
 
 	const Vector3 up(0, 1, 0);
 	Point3 eye, at;
+	int sideW, sideH, side;
 	float sizeW, sizeH;
 
 	float zNearFrustum, zFarFrustum;
-	const float closerToLight = 10000;
+	const float closerToLight = 1500;
 	const Matrix4 matToLightSpace = Matrix4::lookAt(Point3(0), Point3(-dirToSun), up);
-	const float range = Math::Min(1000.f, cam.GetZFar() - cam.GetZNear()); // Clip terrain, etc.
+	const float maxRange = (settings._sceneShadowQuality >= App::Settings::ShadowQuality::ultra) ? 850 : 350;
+	const float range = Math::Min<float>(maxRange, cam.GetZFar() - cam.GetZNear()); // Clip terrain, etc.
 	_camera = cam;
 
 	if (0 == _currentSplit)
@@ -277,8 +267,10 @@ void CascadedShadowMap::Begin(RcVector3 dirToSun, float zNear, float zFar, int s
 		if (0 == zFar)
 			zFar = abs(zFarFrustum - zNearFrustum) + closerToLight;
 
-		sizeW = static_cast<float>(int(frustumBounds.getZ() - frustumBounds.getX() + 0.5f) + 2);
-		sizeH = static_cast<float>(int(frustumBounds.getW() - frustumBounds.getY() + 0.5f) + 2);
+		sideW = static_cast<int>(frustumBounds.getZ() - frustumBounds.getX() + 2.5f);
+		sideH = static_cast<int>(frustumBounds.getW() - frustumBounds.getY() + 2.5f);
+		sizeW = sideW;
+		sizeH = sideH;
 
 		// Setup CSM light space camera for full range (used for terrain layout, etc.):
 		_cameraCSM.SetUpDirection(up);
@@ -327,14 +319,26 @@ void CascadedShadowMap::Begin(RcVector3 dirToSun, float zNear, float zFar, int s
 		(frustumBounds.getY() + frustumBounds.getW()) * 0.5f,
 		zNearFrustum);
 
-	sizeW = static_cast<float>(int(frustumBounds.getZ() - frustumBounds.getX() + 0.5f) + 2);
-	sizeH = static_cast<float>(int(frustumBounds.getW() - frustumBounds.getY() + 0.5f) + 2);
+	sideW = static_cast<int>(frustumBounds.getZ() - frustumBounds.getX() + 2.5f);
+	sideH = static_cast<int>(frustumBounds.getW() - frustumBounds.getY() + 2.5f);
+	if (_currentSplit < 3)
+	{
+		side = (sideW < sideH) ? sideH : sideW;
+		side = Math::NextPowerOfTwo(side);
+		sizeW = side;
+		sizeH = side;
+	}
+	else
+	{
+		sizeW = sideW;
+		sizeH = sideH;
+	}
 
 	if (_snapToTexels)
 	{
 		const float sideInv = 2.f / _side;
-		pos.setX(pos.getX() - fmod(static_cast<float>(pos.getX()), sizeW* sideInv));
-		pos.setY(pos.getY() - fmod(static_cast<float>(pos.getY()), sizeH* sideInv));
+		pos.setX(pos.getX() - fmod(static_cast<float>(pos.getX()), sizeW * sideInv));
+		pos.setY(pos.getY() - fmod(static_cast<float>(pos.getY()), sizeH * sideInv));
 	}
 	pos = (VMath::orthoInverse(matToLightSpace) * pos).getXYZ(); // To world space.
 
@@ -357,11 +361,11 @@ void CascadedShadowMap::Begin(RcVector3 dirToSun, float zNear, float zFar, int s
 	_camera.Update();
 	_pSceneCamera = sm.SetCamera(&_camera);
 
-	_config.setX((zFar - zNear) * 2);
+	_config._penumbraScale = (zFar - zNear) * 2;
 
 	if (0 == _currentSplit)
 	{
-		renderer.GetCommandBuffer()->BeginRenderPass(_rp, _fb,
+		renderer.GetCommandBuffer()->BeginRenderPass(_rph, _fbh,
 			{
 				_tex->GetClearValue()
 			});

Verus/src/Scene/ShadowMap.h

@@ -6,16 +6,26 @@ namespace verus
 	{
 		class ShadowMap : public Object
 		{
+		public:
+			struct Config
+			{
+				float _penumbraScale = 0;
+				float _penumbraLamBiasScale = 2.5f;
+				float _normalDepthBias = 0.012f;
+				float _csmContrastScale = 2.8f;
+			};
+			VERUS_TYPEDEFS(Config);
+
 		protected:
 			Matrix4         _matShadow;
 			Matrix4         _matShadowDS;
-			Vector4         _config = Vector4(0);
+			Config          _config;
 			CGI::TexturePwn _tex;
 			Camera          _camera;
 			PCamera         _pSceneCamera = nullptr;
 			int             _side = 0;
-			int             _rp = -1;
-			int             _fb = -1;
+			CGI::RPHandle   _rph;
+			CGI::FBHandle   _fbh;
 			bool            _snapToTexels = true;
 			bool            _rendering = false;
 
@@ -29,7 +39,7 @@ namespace verus
 			void SetSnapToTexels(bool b) { _snapToTexels = b; }
 			bool IsRendering() const { return _rendering; }
 
-			int GetRenderPassHandle() const { return _rp; }
+			CGI::RPHandle GetRenderPassHandle() const { return _rph; }
 
 			void Begin(RcVector3 dirToSun, float zNear = 1, float zFar = 0);
 			void BeginLight(RcPoint3 pos, RcPoint3 target, float side = 10);
@@ -38,7 +48,7 @@ namespace verus
 			RcMatrix4 GetShadowMatrix() const;
 			RcMatrix4 GetShadowMatrixDS() const;
 
-			RcVector4 GetConfig() { return _config; }
+			RConfig GetConfig() { return _config; }
 
 			CGI::TexturePtr GetTexture() const;
 

Verus/src/Scene/Terrain.cpp

@@ -3,9 +3,11 @@
 using namespace verus;
 using namespace verus::Scene;
 
-CGI::ShaderPwn            Terrain::s_shader;
-Terrain::UB_DrawDepth     Terrain::s_ubDrawDepth;
-Terrain::UB_PerMaterialFS Terrain::s_ubPerMaterialFS;
+CGI::ShaderPwns<Terrain::SHADER_COUNT> Terrain::s_shader;
+Terrain::UB_TerrainVS                  Terrain::s_ubTerrainVS;
+Terrain::UB_TerrainFS                  Terrain::s_ubTerrainFS;
+Terrain::UB_SimpleTerrainVS            Terrain::s_ubSimpleTerrainVS;
+Terrain::UB_SimpleTerrainFS            Terrain::s_ubSimpleTerrainFS;
 
 // TerrainPhysics:
 
@@ -76,19 +78,67 @@ void TerrainPhysics::EnableDebugDraw(bool b)
 
 // TerrainLOD:
 
-void TerrainLOD::Init(int sidePoly, int step)
+void TerrainLOD::Init(int sidePoly, int step, bool addEdgeTess)
 {
 	_side = sidePoly + 1;
 	_vertCount = _side * _side;
 	if (1 == step)
-		Math::BuildListGrid(sidePoly, sidePoly, _vIB);
+		Math::CreateListGrid(sidePoly, sidePoly, _vIB);
 	else
-		Math::BuildStripGrid(sidePoly, sidePoly, _vIB);
-	_indexCount = Utils::Cast32(_vIB.size());
+		Math::CreateStripGrid(sidePoly, sidePoly, _vIB);
 	_firstIndex = 0;
+
+	if (addEdgeTess)
+	{
+		const int extraVerts = sidePoly * 4;
+		_vIB.reserve(_vIB.capacity() + extraVerts * 4);
+		_vIB.push_back(0xFFFF);
+		auto ComputeIndex = [this](int i, int j)
+		{
+			return i * _side + j;
+		};
+
+		// Bottom, forwards:
+		VERUS_FOR(i, sidePoly)
+		{
+			_vIB.push_back(ComputeIndex(sidePoly, i));
+			_vIB.push_back(_vertCount + i);
+			_vIB.push_back(ComputeIndex(sidePoly, i + 1));
+			_vIB.push_back(0xFFFF);
+		}
+		// Right, backwards:
+		VERUS_FOR(i, sidePoly)
+		{
+			_vIB.push_back(ComputeIndex(sidePoly - i, sidePoly));
+			_vIB.push_back(_vertCount + sidePoly + i);
+			_vIB.push_back(ComputeIndex(sidePoly - i - 1, sidePoly));
+			_vIB.push_back(0xFFFF);
+		}
+		// Top, backwards:
+		VERUS_FOR(i, sidePoly)
+		{
+			_vIB.push_back(ComputeIndex(0, sidePoly - i));
+			_vIB.push_back(_vertCount + sidePoly * 2 + i);
+			_vIB.push_back(ComputeIndex(0, sidePoly - i - 1));
+			_vIB.push_back(0xFFFF);
+		}
+		// Left, forwards:
+		VERUS_FOR(i, sidePoly)
+		{
+			_vIB.push_back(ComputeIndex(i, 0));
+			_vIB.push_back(_vertCount + sidePoly * 3 + i);
+			_vIB.push_back(ComputeIndex(i + 1, 0));
+			if (_vIB.size() < _vIB.capacity())
+				_vIB.push_back(0xFFFF);
+		}
+
+		_vertCount += extraVerts;
+		VERUS_RT_ASSERT(Math::CheckIndexBuffer(_vIB, _vertCount - 1));
+	}
+	_indexCount = Utils::Cast32(_vIB.size());
 }
 
-void TerrainLOD::InitGeo(short* pV, UINT16* pI, int vertexOffset)
+void TerrainLOD::InitGeo(short* pV, UINT16* pI, int vertexOffset, bool addEdgeTess)
 {
 	const int edge = _side - 1;
 	const int step = 16 / edge;
@@ -121,6 +171,39 @@ void TerrainLOD::InitGeo(short* pV, UINT16* pI, int vertexOffset)
 		if (_vIB[i] != 0xFFFF)
 			_vIB[i] -= vertexOffset;
 	}
+
+	if (addEdgeTess)
+	{
+		const int halfStep = step / 2;
+		const int vertCount = _side * _side;
+		const int sidePoly = _side - 1;
+		const short minPos = 0;
+		const short maxPos = 16;
+		VERUS_FOR(i, sidePoly)
+		{
+			pV[((vertCount + i) << 2) + 0] = halfStep + i * step;
+			pV[((vertCount + i) << 2) + 2] = maxPos;
+			pV[((vertCount + i) << 2) + 3] = -1;
+		}
+		VERUS_FOR(i, sidePoly)
+		{
+			pV[((vertCount + sidePoly + i) << 2) + 0] = maxPos;
+			pV[((vertCount + sidePoly + i) << 2) + 2] = halfStep + (sidePoly - i - 1) * step;
+			pV[((vertCount + sidePoly + i) << 2) + 1] = -1;
+		}
+		VERUS_FOR(i, sidePoly)
+		{
+			pV[((vertCount + sidePoly * 2 + i) << 2) + 0] = halfStep + (sidePoly - i - 1) * step;
+			pV[((vertCount + sidePoly * 2 + i) << 2) + 2] = minPos;
+			pV[((vertCount + sidePoly * 2 + i) << 2) + 3] = 1;
+		}
+		VERUS_FOR(i, sidePoly)
+		{
+			pV[((vertCount + sidePoly * 3 + i) << 2) + 0] = minPos;
+			pV[((vertCount + sidePoly * 3 + i) << 2) + 2] = halfStep + i * step;
+			pV[((vertCount + sidePoly * 3 + i) << 2) + 1] = 1;
+		}
+	}
 }
 
 // TerrainPatch:
@@ -297,17 +380,35 @@ void Terrain::InitStatic()
 {
 	VERUS_QREF_CONST_SETTINGS;
 
-	s_shader.Init("[Shaders]:DS_Terrain.hlsl");
-	s_shader->CreateDescriptorSet(0, &s_ubDrawDepth, sizeof(s_ubDrawDepth), settings.GetLimits()._terrain_ubDrawDepthCapacity,
+	s_shader[SHADER_MAIN].Init("[Shaders]:DS_Terrain.hlsl");
+	s_shader[SHADER_MAIN]->CreateDescriptorSet(0, &s_ubTerrainVS, sizeof(s_ubTerrainVS), settings.GetLimits()._terrain_ubDrawDepthCapacity,
 		{
-			CGI::Sampler::linear2D
-		});
-	s_shader->CreateDescriptorSet(1, &s_ubPerMaterialFS, sizeof(s_ubPerMaterialFS), 1000,
+			CGI::Sampler::linearMipN, // Height
+			CGI::Sampler::linearMipN // Normal
+		}, CGI::ShaderStageFlags::vs_hs_ds);
+	s_shader[SHADER_MAIN]->CreateDescriptorSet(1, &s_ubTerrainFS, sizeof(s_ubTerrainFS), 1000,
 		{
-			CGI::Sampler::aniso, CGI::Sampler::aniso,
-			CGI::Sampler::aniso, CGI::Sampler::aniso
+			CGI::Sampler::aniso, // Normal
+			CGI::Sampler::aniso, // Blend
+			CGI::Sampler::aniso, // Layers
+			CGI::Sampler::aniso, // LayersNM
+			CGI::Sampler::aniso // Detail
 		}, CGI::ShaderStageFlags::fs);
-	s_shader->CreatePipelineLayout();
+	s_shader[SHADER_MAIN]->CreatePipelineLayout();
+
+	s_shader[SHADER_SIMPLE].Init("[Shaders]:SimpleTerrain.hlsl");
+	s_shader[SHADER_SIMPLE]->CreateDescriptorSet(0, &s_ubSimpleTerrainVS, sizeof(s_ubSimpleTerrainVS), settings.GetLimits()._terrain_ubDrawDepthCapacity,
+		{
+			CGI::Sampler::linearMipN, // Height
+			CGI::Sampler::linearMipN // Normal
+		}, CGI::ShaderStageFlags::vs);
+	s_shader[SHADER_SIMPLE]->CreateDescriptorSet(1, &s_ubSimpleTerrainFS, sizeof(s_ubSimpleTerrainFS), 1000,
+		{
+			CGI::Sampler::linearMipN, // Normal
+			CGI::Sampler::linearMipN, // Blend
+			CGI::Sampler::linearMipN, // Layers
+		}, CGI::ShaderStageFlags::fs);
+	s_shader[SHADER_SIMPLE]->CreatePipelineLayout();
 }
 
 void Terrain::DoneStatic()
@@ -319,6 +420,7 @@ void Terrain::Init(RcDesc desc)
 {
 	VERUS_INIT();
 	VERUS_QREF_RENDERER;
+	VERUS_QREF_CONST_SETTINGS;
 
 	if (!Math::IsPowerOfTwo(desc._mapSide))
 		throw VERUS_RECOVERABLE << "Init(), mapSide must be power of two";
@@ -329,6 +431,8 @@ void Terrain::Init(RcDesc desc)
 	const int patchSide = _mapSide >> 4;
 	const int patchShift = _mapShift - 4;
 	const int patchCount = patchSide * patchSide;
+	const int maxInstances = patchCount * 8;
+
 	_vPatches.resize(patchCount);
 	_vPatchTBNs.resize(patchCount * 3);
 	VERUS_P_FOR(i, patchSide)
@@ -345,6 +449,22 @@ void Terrain::Init(RcDesc desc)
 		}
 	});
 	if (desc._debugHills)
+	{
+		if (desc._debugHills < 0)
+		{
+			VERUS_P_FOR(i, _mapSide)
+			{
+				VERUS_FOR(j, _mapSide)
+				{
+					const int ij[] = { i, j };
+					if (((i >> 4) + (j >> 4)) & 0x1)
+						SetHeightAt(ij, (i & 0xF) * 100);
+					else
+						SetHeightAt(ij, (j & 0xF) * 100);
+				}
+			});
+		}
+		else
 		{
 			const float scale = 1.f / desc._debugHills;
 			VERUS_P_FOR(i, _mapSide)
@@ -358,6 +478,7 @@ void Terrain::Init(RcDesc desc)
 					SetHeightAt(ij, short(res * 1000.f));
 				}
 			});
+		}
 
 		VERUS_FOR(lod, 5)
 		{
@@ -368,11 +489,10 @@ void Terrain::Init(RcDesc desc)
 		}
 	}
 	_vSortedPatchIndices.resize(patchCount);
-	const int maxInstances = patchCount * 8;
 
 	// Init LODs:
 	VERUS_FOR(i, VERUS_COUNT_OF(_lods))
-		_lods[i].Init(16 >> i, 1 << i);
+		_lods[i].Init(16 >> i, 1 << i, i > 0);
 
 	Vector<short> vVB;
 	Vector<UINT16> vIB;
@@ -387,21 +507,22 @@ void Terrain::Init(RcDesc desc)
 	vbSize = 0, ibSize = 0;
 	VERUS_FOR(i, VERUS_COUNT_OF(_lods))
 	{
-		_lods[i].InitGeo(&vVB[vbSize * 4], &vIB[ibSize], vbSize);
+		_lods[i].InitGeo(&vVB[vbSize * 4], &vIB[ibSize], vbSize, i > 0);
 		_lods[i]._firstIndex = ibSize;
 		vbSize += _lods[i]._vertCount;
 		ibSize += _lods[i]._indexCount;
 	}
+	_vInstanceBuffer.resize(maxInstances);
 
 	CGI::GeometryDesc geoDesc;
-	const CGI::InputElementDesc ied[] =
+	const CGI::VertexInputAttrDesc viaDesc[] =
 	{
 		{ 0, 0,                                    CGI::IeType::shorts, 4, CGI::IeUsage::position, 0},
-		{-1, offsetof(PerInstanceData, _posPatch), CGI::IeType::shorts, 4, CGI::IeUsage::texCoord, 8},
-		{-1, offsetof(PerInstanceData, _layers),   CGI::IeType::shorts, 4, CGI::IeUsage::texCoord, 9},
-		CGI::InputElementDesc::End()
+		{-1, offsetof(PerInstanceData, _posPatch), CGI::IeType::shorts, 4, CGI::IeUsage::instData, 0},
+		{-1, offsetof(PerInstanceData, _layers),   CGI::IeType::shorts, 4, CGI::IeUsage::instData, 1},
+		CGI::VertexInputAttrDesc::End()
 	};
-	geoDesc._pInputElementDesc = ied;
+	geoDesc._pVertexInputAttrDesc = viaDesc;
 	const int strides[] = { sizeof(short) * 4, sizeof(PerInstanceData), 0 };
 	geoDesc._pStrides = strides;
 	_geo.Init(geoDesc);
@@ -411,14 +532,11 @@ void Terrain::Init(RcDesc desc)
 	_geo->UpdateVertexBuffer(vVB.data(), 0);
 	_geo->UpdateIndexBuffer(vIB.data());
 
-	_vInstanceBuffer.resize(maxInstances);
-
 	{
-		CGI::PipelineDesc pipeDesc(_geo, s_shader, "#", renderer.GetDS().GetRenderPassHandle());
+		CGI::PipelineDesc pipeDesc(_geo, s_shader[SHADER_MAIN], "#", renderer.GetDS().GetRenderPassHandle());
 		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_OFF;
 		pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_OFF;
 		pipeDesc._colorAttachBlendEqs[2] = VERUS_COLOR_BLEND_OFF;
-		pipeDesc._colorAttachBlendEqs[3] = VERUS_COLOR_BLEND_OFF;
 		_pipe[PIPE_LIST].Init(pipeDesc);
 		pipeDesc._topology = CGI::PrimitiveTopology::triangleStrip;
 		pipeDesc._primitiveRestartEnable = true;
@@ -426,14 +544,36 @@ void Terrain::Init(RcDesc desc)
 	}
 	{
 		VERUS_QREF_ATMO;
-		CGI::PipelineDesc pipeDesc(_geo, s_shader, "#Depth", atmo.GetShadowMap().GetRenderPassHandle());
+		CGI::PipelineDesc pipeDesc(_geo, s_shader[SHADER_MAIN], "#Depth", atmo.GetShadowMap().GetRenderPassHandle());
 		pipeDesc._colorAttachBlendEqs[0] = "";
-		pipeDesc.EnableDepthBias();
 		_pipe[PIPE_DEPTH_LIST].Init(pipeDesc);
 		pipeDesc._topology = CGI::PrimitiveTopology::triangleStrip;
 		pipeDesc._primitiveRestartEnable = true;
 		_pipe[PIPE_DEPTH_STRIP].Init(pipeDesc);
 	}
+	{
+		CGI::PipelineDesc pipeDesc(_geo, s_shader[SHADER_MAIN], "#SolidColor", renderer.GetDS().GetRenderPassHandle());
+		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[2] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._rasterizationState._polygonMode = CGI::PolygonMode::line;
+		pipeDesc._rasterizationState._depthBiasEnable = true;
+		pipeDesc._rasterizationState._depthBiasConstantFactor = -50;
+		pipeDesc._depthCompareOp = CGI::CompareOp::lessOrEqual;
+		_pipe[PIPE_WIREFRAME_LIST].Init(pipeDesc);
+		pipeDesc._topology = CGI::PrimitiveTopology::triangleStrip;
+		pipeDesc._primitiveRestartEnable = true;
+		_pipe[PIPE_WIREFRAME_STRIP].Init(pipeDesc);
+	}
+	if (settings._gpuTessellation)
+	{
+		CGI::PipelineDesc pipeDesc(_geo, s_shader[SHADER_MAIN], "#Tess", renderer.GetDS().GetRenderPassHandle());
+		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[1] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._colorAttachBlendEqs[2] = VERUS_COLOR_BLEND_OFF;
+		pipeDesc._topology = CGI::PrimitiveTopology::patchList3;
+		_pipe[PIPE_TESS].Init(pipeDesc);
+	}
 
 	CGI::TextureDesc texDesc;
 	texDesc._format = CGI::Format::floatR16;
@@ -447,12 +587,20 @@ void Terrain::Init(RcDesc desc)
 	texDesc._width = _mapSide;
 	texDesc._height = _mapSide;
 	texDesc._mipLevels = 0;
-	texDesc._flags = CGI::TextureDesc::Flags::generateMips;
+	texDesc._flags = CGI::TextureDesc::Flags::anyShaderResource | CGI::TextureDesc::Flags::generateMips;
 	_tex[TEX_NORMALS].Init(texDesc);
+	texDesc._flags = CGI::TextureDesc::Flags::generateMips;
 	_tex[TEX_BLEND].Init(texDesc);
 
-	s_shader->FreeDescriptorSet(_cshVS);
-	_cshVS = s_shader->BindDescriptorSetTextures(0, { _tex[TEX_HEIGHTMAP] });
+	texDesc.Reset();
+	texDesc._format = CGI::Format::unormR8;
+	texDesc._width = _mapSide;
+	texDesc._height = _mapSide;
+	texDesc._flags = CGI::TextureDesc::Flags::anyShaderResource;
+	_tex[TEX_MAIN_LAYER].Init(texDesc);
+
+	s_shader[SHADER_MAIN]->FreeDescriptorSet(_cshVS);
+	_cshVS = s_shader[SHADER_MAIN]->BindDescriptorSetTextures(0, { _tex[TEX_HEIGHTMAP], _tex[TEX_NORMALS] });
 
 	_vBlendBuffer.resize(_mapSide * _mapSide);
 	VERUS_P_FOR(i, _vBlendBuffer.size())
@@ -460,6 +608,7 @@ void Terrain::Init(RcDesc desc)
 		_vBlendBuffer[i] = VERUS_COLOR_RGBA(255, 0, 0, 255);
 	});
 	UpdateBlendTexture();
+	UpdateMainLayerTexture();
 
 	OnHeightModified();
 	AddNewRigidBody();
@@ -469,8 +618,8 @@ void Terrain::Init(RcDesc desc)
 
 void Terrain::Done()
 {
-	s_shader->FreeDescriptorSet(_cshVS);
-	s_shader->FreeDescriptorSet(_cshFS);
+	s_shader[SHADER_MAIN]->FreeDescriptorSet(_cshVS);
+	s_shader[SHADER_MAIN]->FreeDescriptorSet(_cshFS);
 
 	_tex.Done();
 	_pipe.Done();
@@ -481,17 +630,6 @@ void Terrain::Done()
 	VERUS_DONE(Terrain);
 }
 
-int Terrain::UserPtr_GetType()
-{
-	return 0;
-	//return +NodeType::terrain;
-}
-
-void Terrain::ResetInstanceCount()
-{
-	_instanceCount = 0;
-}
-
 void Terrain::Layout()
 {
 	VERUS_QREF_CONST_SETTINGS;
@@ -520,39 +658,48 @@ void Terrain::Layout()
 		sm.SetCamera(pPrevCamera);
 }
 
-void Terrain::Draw()
+void Terrain::Draw(RcDrawDesc dd)
 {
 	VERUS_QREF_RENDERER;
 	VERUS_QREF_SM;
 	VERUS_QREF_ATMO;
+	VERUS_QREF_CONST_SETTINGS;
 
 	if (!_visiblePatchCount)
 		return;
 
+	auto cb = renderer.GetCommandBuffer();
+
 	const bool drawingDepth = Scene::SceneManager::IsDrawingDepth(Scene::DrawDepth::automatic);
 
-	s_ubDrawDepth._matW = mataff(1);
-	s_ubDrawDepth._matWV = sm.GetCamera()->GetMatrixV().UniformBufferFormat();
-	s_ubDrawDepth._matVP = sm.GetCamera()->GetMatrixVP().UniformBufferFormat();
-	s_ubDrawDepth._posEye_mapSideInv = float4(atmo.GetEyePosition().GLM(), 0);
-	s_ubDrawDepth._posEye_mapSideInv.w = 1.f / _mapSide;
+	const Transform3 matW = Transform3::identity();
+	s_ubTerrainVS._matW = matW.UniformBufferFormat();
+	s_ubTerrainVS._matWV = Transform3(sm.GetCamera()->GetMatrixV() * matW).UniformBufferFormat();
+	s_ubTerrainVS._matV = sm.GetCamera()->GetMatrixV().UniformBufferFormat();
+	s_ubTerrainVS._matVP = sm.GetCamera()->GetMatrixVP().UniformBufferFormat();
+	s_ubTerrainVS._matP = sm.GetCamera()->GetMatrixP().UniformBufferFormat();
+	s_ubTerrainVS._eyePos_mapSideInv = float4(atmo.GetEyePosition().GLM(), 0);
+	s_ubTerrainVS._eyePos_mapSideInv.w = 1.f / _mapSide;
+	s_ubTerrainVS._viewportSize = cb->GetViewportSize().GLM();
 
+	s_ubTerrainFS._matWV = s_ubTerrainVS._matWV;
 	VERUS_FOR(i, VERUS_COUNT_OF(_layerData))
 	{
-		s_ubPerMaterialFS._vSpecStrength[i >> 2][i & 0x3] = _layerData[i]._specStrength;
-		s_ubPerMaterialFS._vDetailStrength[i >> 2][i & 0x3] = _layerData[i]._detailStrength;
+		s_ubTerrainFS._vSpecStrength[i >> 2][i & 0x3] = _layerData[i]._specStrength;
+		s_ubTerrainFS._vDetailStrength[i >> 2][i & 0x3] = _layerData[i]._detailStrength;
 	}
+	s_ubTerrainFS._lamScaleBias.x = _lamScale;
+	s_ubTerrainFS._lamScaleBias.y = _lamBias;
 
-	renderer.GetCommandBuffer()->BindVertexBuffers(_geo);
-	renderer.GetCommandBuffer()->BindIndexBuffer(_geo);
+	cb->BindVertexBuffers(_geo);
+	cb->BindIndexBuffer(_geo);
 
-	s_shader->BeginBindDescriptors();
-
-	bool strip = false;
+	bool bindStrip = true;
 	const int half = _mapSide >> 1;
 	int firstInstance = 0;
 	int edge = _visiblePatchCount - 1;
 	int lod = _vPatches[_vSortedPatchIndices.front()]._quadtreeLOD;
+	s_shader[SHADER_MAIN]->BeginBindDescriptors();
 	VERUS_FOR(i, _visiblePatchCount)
 	{
 		RcTerrainPatch patch = _vPatches[_vSortedPatchIndices[i]];
@@ -563,35 +710,28 @@ void Terrain::Draw()
 
 			if (!lod)
 			{
-				if (drawingDepth)
-				{
-					renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_DEPTH_LIST]);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 0, _cshVS);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 1, _cshFS);
-				}
+				if (dd._wireframe)
+					cb->BindPipeline(_pipe[PIPE_WIREFRAME_LIST]);
+				else if (drawingDepth)
+					cb->BindPipeline(_pipe[PIPE_DEPTH_LIST]);
+				else if (dd._tess && settings._gpuTessellation)
+					cb->BindPipeline(_pipe[PIPE_TESS]);
 				else
-				{
-					renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_LIST]);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 0, _cshVS);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 1, _cshFS);
+					cb->BindPipeline(_pipe[PIPE_LIST]);
+				cb->BindDescriptors(s_shader[SHADER_MAIN], 0, _cshVS);
+				cb->BindDescriptors(s_shader[SHADER_MAIN], 1, _cshFS);
 			}
-			}
-			else if (!strip)
+			else if (bindStrip)
 			{
-				if (drawingDepth)
-				{
-					strip = true;
-					renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_DEPTH_STRIP]);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 0, _cshVS);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 1, _cshFS);
-				}
+				bindStrip = false;
+				if (dd._wireframe)
+					cb->BindPipeline(_pipe[PIPE_WIREFRAME_STRIP]);
+				else if (drawingDepth)
+					cb->BindPipeline(_pipe[PIPE_DEPTH_STRIP]);
 				else
-				{
-					strip = true;
-					renderer.GetCommandBuffer()->BindPipeline(_pipe[PIPE_STRIP]);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 0, _cshVS);
-					renderer.GetCommandBuffer()->BindDescriptors(s_shader, 1, _cshFS);
-				}
+					cb->BindPipeline(_pipe[PIPE_STRIP]);
+				cb->BindDescriptors(s_shader[SHADER_MAIN], 0, _cshVS);
+				cb->BindDescriptors(s_shader[SHADER_MAIN], 1, _cshFS);
 			}
 
 			const int instanceCount = i - firstInstance; // Drawing patches [firstInstance, i).
@@ -603,22 +743,43 @@ void Terrain::Draw()
 				_vInstanceBuffer[at]._posPatch[1] = patchToDraw._patchHeight;
 				_vInstanceBuffer[at]._posPatch[2] = patchToDraw._ijCoord[0] - half;
 				_vInstanceBuffer[at]._posPatch[3] = 0;
+				VERUS_ZERO_MEM(_vInstanceBuffer[at]._layers);
+				if (dd._wireframe)
+				{
+					switch (patchToDraw._usedChannelCount)
+					{
+					case 1: _vInstanceBuffer[at]._layers[0] = 0; _vInstanceBuffer[at]._layers[1] = 0; _vInstanceBuffer[at]._layers[2] = 1; break;
+					case 2: _vInstanceBuffer[at]._layers[0] = 0; _vInstanceBuffer[at]._layers[1] = 1; _vInstanceBuffer[at]._layers[2] = 0; break;
+					case 3: _vInstanceBuffer[at]._layers[0] = 1; _vInstanceBuffer[at]._layers[1] = 1; _vInstanceBuffer[at]._layers[2] = 0; break;
+					case 4: _vInstanceBuffer[at]._layers[0] = 1; _vInstanceBuffer[at]._layers[1] = 0; _vInstanceBuffer[at]._layers[2] = 0; break;
+					}
+				}
+				else
+				{
 					VERUS_FOR(ch, 4)
 						_vInstanceBuffer[at]._layers[ch] = patchToDraw._layerForChannel[ch];
 				}
+			}
 
-			renderer.GetCommandBuffer()->DrawIndexed(_lods[lod]._indexCount, instanceCount, _lods[lod]._firstIndex, 0, _instanceCount + firstInstance);
+			cb->DrawIndexed(_lods[lod]._indexCount, instanceCount, _lods[lod]._firstIndex, 0, _instanceCount + firstInstance);
 
 			lod = patch._quadtreeLOD;
 			firstInstance = i;
 		}
 	}
+	s_shader[SHADER_MAIN]->EndBindDescriptors();
 
 	_instanceCount += _visiblePatchCount;
-
 	_geo->UpdateVertexBuffer(_vInstanceBuffer.data(), 1);
+}
 
-	s_shader->EndBindDescriptors();
+void Terrain::DrawReflection()
+{
+}
+
+void Terrain::ResetInstanceCount()
+{
+	_instanceCount = 0;
 }
 
 void Terrain::SortVisiblePatches()
@@ -635,6 +796,12 @@ void Terrain::SortVisiblePatches()
 		});
 }
 
+int Terrain::UserPtr_GetType()
+{
+	return 0;
+	//return +NodeType::terrain;
+}
+
 void Terrain::QuadtreeIntegral_ProcessVisibleNode(const short ij[2], RcPoint3 center)
 {
 	VERUS_QREF_ATMO;
@@ -647,6 +814,9 @@ void Terrain::QuadtreeIntegral_ProcessVisibleNode(const short ij[2], RcPoint3 ce
 	const float lodF = log2(Math::Clamp((dist - 12) * (2 / 100.f), 1.f, 18.f));
 	const int lod = Math::Clamp<int>(int(lodF), 0, 4);
 
+	if (4 == lod && center.getY() <= -20) // Cull underwater patches:
+		return;
+
 	// Locate this patch:
 	const int shiftPatch = _mapShift - 4;
 	const int iPatch = ij[0] >> 4;
@@ -850,7 +1020,9 @@ void Terrain::LoadLayerTextures()
 	if (_vLayerUrls.empty())
 		return;
 
+	VERUS_QREF_MM;
 	VERUS_QREF_RENDERER;
+
 	renderer->WaitIdle();
 
 	_tex[TEX_LAYERS].Done();
@@ -884,8 +1056,8 @@ void Terrain::LoadLayerTextures()
 	texDesc._urls = vLayerUrlsPtrNM.data();
 	_tex[TEX_LAYERS_NM].Init(texDesc);
 
-	s_shader->FreeDescriptorSet(_cshFS);
-	_cshFS = s_shader->BindDescriptorSetTextures(1, { _tex[TEX_NORMALS], _tex[TEX_BLEND], _tex[TEX_LAYERS], _tex[TEX_LAYERS_NM] });
+	s_shader[SHADER_MAIN]->FreeDescriptorSet(_cshFS);
+	_cshFS = s_shader[SHADER_MAIN]->BindDescriptorSetTextures(1, { _tex[TEX_NORMALS], _tex[TEX_BLEND], _tex[TEX_LAYERS], _tex[TEX_LAYERS_NM], mm.GetDetailTexture() });
 }
 
 int Terrain::GetMainLayerAt(const int ij[2]) const
@@ -1019,7 +1191,7 @@ void Terrain::UpdateMainLayerTexture()
 			_vMainLayerSubresData[offset + j] = GetMainLayerAt(ij) * 16 + 4;
 		}
 	});
-	//_tex[TEX_MAIN_LAYER]->UpdateSubresource(_vMainLayerSubresData.data());
+	_tex[TEX_MAIN_LAYER]->UpdateSubresource(_vMainLayerSubresData.data());
 }
 
 CGI::TexturePtr Terrain::GetMainLayerTexture() const
@@ -1039,6 +1211,7 @@ void Terrain::OnHeightModified()
 
 void Terrain::AddNewRigidBody()
 {
+	_physics.Done();
 	_physics.Init(this, _mapSide, _mapSide, _vHeightBuffer.data(), ConvertHeight(static_cast<short>(1)));
 }
 
@@ -1175,7 +1348,7 @@ void Terrain::Deserialize(IO::RStream stream)
 			}
 			else
 			{
-				_vPatches[i]._layerForChannel[j] = -1;
+				_vPatches[i]._layerForChannel[j] = 0;
 			}
 		}
 	}
@@ -1194,12 +1367,13 @@ void Terrain::Deserialize(IO::RStream stream)
 		{
 			UINT16 data;
 			stream >> data;
-			_vBlendBuffer[offset + j] = Convert::Uint4x4ToUint8x4(data);
+			_vBlendBuffer[offset + j] = Convert::Uint4x4ToUint8x4(data, true);
 			const int ij[] = { i, j };
 			UpdateMainLayerAt(ij);
 		}
 	}
 	UpdateBlendTexture();
+	UpdateMainLayerTexture();
 	// </Blend>
 
 	OnHeightModified();

Verus/src/Scene/Terrain.h

@@ -40,8 +40,8 @@ namespace verus
 			int            _indexCount = 0;
 			int            _firstIndex = 0;
 
-			void Init(int sidePoly, int step);
-			void InitGeo(short* pV, UINT16* pI, int vertexOffset);
+			void Init(int sidePoly, int step, bool addEdgeTess);
+			void InitGeo(short* pV, UINT16* pI, int vertexOffset, bool addEdgeTess);
 		};
 		VERUS_TYPEDEFS(TerrainLOD);
 
@@ -80,6 +80,14 @@ namespace verus
 		{
 		public:
 #include "../Shaders/DS_Terrain.inc.hlsl"
+#include "../Shaders/SimpleTerrain.inc.hlsl"
+
+			enum SHADER
+			{
+				SHADER_MAIN,
+				SHADER_SIMPLE,
+				SHADER_COUNT
+			};
 
 			enum PIPE
 			{
@@ -87,7 +95,10 @@ namespace verus
 				PIPE_STRIP,
 				PIPE_DEPTH_LIST,
 				PIPE_DEPTH_STRIP,
-				PIPE_MAX
+				PIPE_WIREFRAME_LIST,
+				PIPE_WIREFRAME_STRIP,
+				PIPE_TESS,
+				PIPE_COUNT
 			};
 
 			enum TEX
@@ -98,7 +109,7 @@ namespace verus
 				TEX_LAYERS,
 				TEX_LAYERS_NM,
 				TEX_MAIN_LAYER,
-				TEX_MAX
+				TEX_COUNT
 			};
 
 			static const int s_maxLayers = 32;
@@ -116,13 +127,15 @@ namespace verus
 			};
 
 		protected:
-			static CGI::ShaderPwn   s_shader;
-			static UB_DrawDepth     s_ubDrawDepth;
-			static UB_PerMaterialFS s_ubPerMaterialFS;
+			static CGI::ShaderPwns<SHADER_COUNT> s_shader;
+			static UB_TerrainVS                  s_ubTerrainVS;
+			static UB_TerrainFS                  s_ubTerrainFS;
+			static UB_SimpleTerrainVS            s_ubSimpleTerrainVS;
+			static UB_SimpleTerrainFS            s_ubSimpleTerrainFS;
 
 			CGI::GeometryPwn              _geo;
-			CGI::PipelinePwns<PIPE_MAX> _pipe;
-			CGI::TexturePwns<TEX_MAX>   _tex;
+			CGI::PipelinePwns<PIPE_COUNT> _pipe;
+			CGI::TexturePwns<TEX_COUNT>   _tex;
 			Vector<TerrainPatch>          _vPatches;
 			Vector<TerrainPatch::TBN>     _vPatchTBNs;
 			Vector<UINT16>                _vSortedPatchIndices;
@@ -136,18 +149,21 @@ namespace verus
 			float                         _quadtreeFatten = 0.5f;
 			int                           _mapSide = 0;
 			int                           _mapShift = 0;
-			int                         _visiblePatchCount = 0;
 			int                           _vertCount = 0;
 			int                           _indexCount = 0;
 			int                           _instanceCount = 0;
-			int                         _cshVS = -1;
-			int                         _cshFS = -1;
+			int                           _visiblePatchCount = 0;
+			CGI::CSHandle                 _cshVS;
+			CGI::CSHandle                 _cshFS;
 			TerrainLOD                    _lods[5]; // Level of detail data for (16x16, 8x8, 4x4, 2x2, 1x1).
 			LayerData                     _layerData[s_maxLayers];
 			TerrainPhysics                _physics;
 			Math::QuadtreeIntegral        _quadtree;
 
 		public:
+			float _lamScale = 1.9f;
+			float _lamBias = -0.9f;
+
 			struct Desc
 			{
 				int   _mapSide = 256;
@@ -159,6 +175,18 @@ namespace verus
 			};
 			VERUS_TYPEDEFS(Desc);
 
+			struct DrawDesc
+			{
+				bool _tess = false;
+				bool _wireframe = false;
+
+				void Reset()
+				{
+					*this = DrawDesc();
+				}
+			};
+			VERUS_TYPEDEFS(DrawDesc);
+
 			Terrain();
 			virtual ~Terrain();
 
@@ -168,14 +196,16 @@ namespace verus
 			void Init(RcDesc desc = Desc());
 			void Done();
 
-			virtual int UserPtr_GetType() override;
+			void Layout();
+			void Draw(RcDrawDesc dd = DrawDesc());
+			void DrawReflection();
 
 			void ResetInstanceCount();
-			void Layout();
-			void Draw();
 
 			void SortVisiblePatches();
 
+			virtual int UserPtr_GetType() override;
+
 			int GetMapSide() const { return _mapSide; }
 
 			RTerrainPatch GetPatch(int index) { return _vPatches[index]; }

Verus/src/Scene/Water.cpp

@@ -0,0 +1,513 @@
+#include "verus.h"
+
+using namespace verus;
+using namespace verus::Scene;
+
+Water::UB_WaterVS        Water::s_ubWaterVS;
+Water::UB_WaterFS        Water::s_ubWaterFS;
+Water::UB_Gen            Water::s_ubGen;
+Water::UB_GenHeightmapFS Water::s_ubGenHeightmapFS;
+Water::UB_GenNormalsFS   Water::s_ubGenNormalsFS;
+
+
+Water::Water()
+{
+}
+
+Water::~Water()
+{
+	Done();
+}
+
+void Water::Init(RTerrain terrain)
+{
+	VERUS_INIT();
+
+	VERUS_QREF_RENDERER;
+	VERUS_QREF_CONST_SETTINGS;
+
+	_pTerrain = &terrain;
+
+	_rphGenHeightmap = renderer->CreateSimpleRenderPass(CGI::Format::floatR16, CGI::ImageLayout::xsReadOnly);
+	_rphGenNormals = renderer->CreateSimpleRenderPass(CGI::Format::unormR10G10B10A2);
+	_rphReflection = renderer->CreateRenderPass(
+		{
+			CGI::RP::Attachment("Color", CGI::Format::floatR11G11B10).LoadOpDontCare().Layout(CGI::ImageLayout::fsReadOnly),
+			CGI::RP::Attachment("Depth", CGI::Format::unormD24uintS8).LoadOpClear().Layout(CGI::ImageLayout::depthStencilAttachment),
+		},
+		{
+			CGI::RP::Subpass("Sp0").Color(
+				{
+					CGI::RP::Ref("Color", CGI::ImageLayout::colorAttachment)
+				}).DepthStencil(CGI::RP::Ref("Depth", CGI::ImageLayout::depthStencilAttachment)),
+		},
+		{});
+
+	_shader[SHADER_MAIN].Init("[Shaders]:Water.hlsl");
+	_shader[SHADER_MAIN]->CreateDescriptorSet(0, &s_ubWaterVS, sizeof(s_ubWaterVS), 100,
+		{
+			CGI::Sampler::linearClampMipL,
+			CGI::Sampler::linearMipL
+		}, CGI::ShaderStageFlags::vs);
+	_shader[SHADER_MAIN]->CreateDescriptorSet(1, &s_ubWaterFS, sizeof(s_ubWaterFS), 100,
+		{
+			CGI::Sampler::linearClampMipL,
+			CGI::Sampler::linearMipL,
+			CGI::Sampler::custom,
+			CGI::Sampler::aniso
+		}, CGI::ShaderStageFlags::fs);
+	_shader[SHADER_MAIN]->CreatePipelineLayout();
+
+	_shader[SHADER_GEN].Init("[Shaders]:WaterGen.hlsl");
+	_shader[SHADER_GEN]->CreateDescriptorSet(0, &s_ubGen, sizeof(s_ubGen), 100, {}, CGI::ShaderStageFlags::vs);
+	_shader[SHADER_GEN]->CreateDescriptorSet(1, &s_ubGenHeightmapFS, sizeof(s_ubGenHeightmapFS), 100, { CGI::Sampler::linearMipN }, CGI::ShaderStageFlags::fs);
+	_shader[SHADER_GEN]->CreateDescriptorSet(2, &s_ubGenNormalsFS, sizeof(s_ubGenNormalsFS), 100, { CGI::Sampler::linearMipN }, CGI::ShaderStageFlags::fs);
+	_shader[SHADER_GEN]->CreatePipelineLayout();
+
+	Vector<UINT16> vIndices;
+	Math::CreateStripGrid(_gridWidth - 1, _gridHeight - 1, vIndices);
+	_indexCount = vIndices.size();
+
+	CGI::GeometryDesc geoDesc;
+	const CGI::VertexInputAttrDesc viaDesc[] =
+	{
+		{0, offsetof(Vertex, _pos), CGI::IeType::floats, 4, CGI::IeUsage::position, 0},
+		CGI::VertexInputAttrDesc::End()
+	};
+	geoDesc._pVertexInputAttrDesc = viaDesc;
+	const int strides[] = { sizeof(Vertex), 0 };
+	geoDesc._pStrides = strides;
+	_geo.Init(geoDesc);
+	_geo->CreateVertexBuffer(_gridWidth * _gridHeight, 0);
+	_geo->CreateIndexBuffer(vIndices.size());
+	_geo->UpdateIndexBuffer(vIndices.data());
+
+	CreateWaterPlane();
+
+	{
+		CGI::PipelineDesc pipeDesc(_geo, _shader[SHADER_MAIN], "#0", renderer.GetDS().GetRenderPassHandle_Compose(), renderer.GetDS().GetSubpass_Compose());
+		pipeDesc._colorAttachBlendEqs[0] = VERUS_COLOR_BLEND_ALPHA;
+		//pipeDesc._rasterizationState._polygonMode = CGI::PolygonMode::line;
+		pipeDesc._topology = CGI::PrimitiveTopology::triangleStrip;
+		_pipe[PIPE_MAIN].Init(pipeDesc);
+	}
+	{
+		CGI::PipelineDesc pipeDesc(renderer.GetGeoQuad(), _shader[SHADER_GEN], "#GenHeightmap", _rphGenHeightmap);
+		pipeDesc._topology = CGI::PrimitiveTopology::triangleStrip;
+		pipeDesc._depthTestEnable = false;
+		_pipe[PIPE_GEN_HEIGHTMAP].Init(pipeDesc);
+	}
+	{
+		CGI::PipelineDesc pipeDesc(renderer.GetGeoQuad(), _shader[SHADER_GEN], "#GenNormals", _rphGenNormals);
+		pipeDesc._topology = CGI::PrimitiveTopology::triangleStrip;
+		pipeDesc._depthTestEnable = false;
+		_pipe[PIPE_GEN_NORMALS].Init(pipeDesc);
+	}
+
+	CGI::TextureDesc texDesc;
+	texDesc._url = "[Textures]:Water/Foam.dds";
+	_tex[TEX_FOAM].Init(texDesc);
+
+	texDesc._url = "[Textures]:Water/Heightmap.FX.dds";
+	_tex[TEX_SOURCE_HEIGHTMAP].Init(texDesc);
+
+	CGI::SamplerDesc normalsSamplerDesc;
+	normalsSamplerDesc.SetFilter("a");
+	normalsSamplerDesc.SetAddressMode("rr");
+	normalsSamplerDesc._mipLodBias = 0.5f;
+	normalsSamplerDesc._minLod = 1;
+	texDesc.Reset();
+	texDesc._format = CGI::Format::floatR16;
+	texDesc._width = _genSide;
+	texDesc._height = _genSide;
+	texDesc._mipLevels = 0;
+	texDesc._flags = CGI::TextureDesc::Flags::colorAttachment | CGI::TextureDesc::Flags::anyShaderResource | CGI::TextureDesc::Flags::generateMips;
+	_tex[TEX_GEN_HEIGHTMAP].Init(texDesc);
+	texDesc._format = CGI::Format::unormR10G10B10A2;
+	texDesc._flags = CGI::TextureDesc::Flags::colorAttachment | CGI::TextureDesc::Flags::generateMips;
+	texDesc._pSamplerDesc = &normalsSamplerDesc;
+	_tex[TEX_GEN_NORMALS].Init(texDesc);
+
+	_fbhGenHeightmap = renderer->CreateFramebuffer(_rphGenHeightmap, { _tex[TEX_GEN_HEIGHTMAP] }, _genSide, _genSide);
+	_fbhGenNormals = renderer->CreateFramebuffer(_rphGenNormals, { _tex[TEX_GEN_NORMALS] }, _genSide, _genSide);
+
+	_cshWaterVS = _shader[SHADER_MAIN]->BindDescriptorSetTextures(0, { _pTerrain->GetHeightmapTexture(), _tex[TEX_GEN_HEIGHTMAP] });
+
+	_cshGenNormals = _shader[SHADER_GEN]->BindDescriptorSetTextures(2, { _tex[TEX_GEN_HEIGHTMAP] });
+
+	texDesc.Reset();
+	texDesc._format = CGI::Format::floatR11G11B10;
+	texDesc._width = settings._screenSizeWidth;
+	texDesc._height = settings._screenSizeHeight;
+	texDesc._flags = CGI::TextureDesc::Flags::colorAttachment;
+	_tex[TEX_REFLECTION].Init(texDesc);
+	texDesc._format = CGI::Format::unormD24uintS8;
+	texDesc._flags = CGI::TextureDesc::Flags::none;
+	_tex[TEX_REFLECTION_DEPTH].Init(texDesc);
+
+	_fbhReflection = renderer->CreateFramebuffer(_rphReflection,
+		{
+			_tex[TEX_REFLECTION],
+			_tex[TEX_REFLECTION_DEPTH]
+		},
+		texDesc._width,
+		texDesc._height);
+
+	// Texture for rendering reflection onto:
+	/*
+	if (settings._sceneWaterQuality >= App::Settings::WaterQuality::simpleReflection)
+	{
+		texDesc.Reset();
+		texDesc._format = CGI::Format::unormR8G8B8A8;
+		texDesc._width = 512;
+		texDesc._height = 512;
+		texDesc._mipLevels = 0;
+		texDesc._flags = CGI::TextureDesc::Flags::generateMips;
+		_tex[TEX_REFLECT].Init(texDesc);
+
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::trueWavesRefraction)
+			_tex[TEX_REFRACT].Init(texDesc);
+
+		texDesc._format = CGI::Format::unormD16;
+		_tex[TEX_REFLECT_DEPTH].Init(texDesc);
+	}
+	*/
+
+	VERUS_FOR(i, s_maxHarmonics)
+		_amplitudes[i] = PhillipsSpectrum(0.5f + i / 2.f);
+}
+
+void Water::Done()
+{
+	VERUS_DONE(Water);
+}
+
+void Water::Update()
+{
+	VERUS_UPDATE_ONCE_CHECK;
+
+	VERUS_QREF_TIMER;
+	_phase = fmod(_phase + dt * (1 / 29.f), 1.f);
+	_phaseWave = fmod(_phaseWave + dt * (1 / 61.f), 1.f);
+}
+
+void Water::Draw()
+{
+	VERUS_UPDATE_ONCE_CHECK_DRAW;
+
+	if (!_cshWaterFS.IsSet())
+	{
+		if (_tex[TEX_FOAM]->IsLoaded())
+		{
+			_cshWaterFS = _shader[SHADER_MAIN]->BindDescriptorSetTextures(1,
+				{
+					_pTerrain->GetHeightmapTexture(),
+					_tex[TEX_GEN_HEIGHTMAP],
+					_tex[TEX_GEN_NORMALS],
+					_tex[TEX_FOAM]
+				});
+		}
+		else
+			return;
+	}
+
+	VERUS_QREF_RENDERER;
+	VERUS_QREF_SM;
+	VERUS_QREF_CONST_SETTINGS;
+	VERUS_QREF_TIMER;
+	VERUS_QREF_ATMO;
+
+	auto cb = renderer.GetCommandBuffer();
+
+	RCamera cam = *sm.GetCamera();
+
+	Transform3 matW;
+
+	Point3 eyePosAtGroundLevel = cam.GetEyePosition();
+	eyePosAtGroundLevel.setY(0);
+
+	Vector3 flatDir = cam.GetFrontDirection();
+	flatDir.setY(0);
+	flatDir = VMath::normalizeApprox(flatDir);
+	const float angle = atan2(
+		static_cast<float>(flatDir.getX()),
+		static_cast<float>(flatDir.getZ())); // Align with high density sector.
+	const Matrix3 matR = Matrix3::rotationY(angle);
+	matW = Transform3(matR, Vector3(eyePosAtGroundLevel - flatDir * 5 * abs(cam.GetFrontDirection().getY())));
+
+	const Matrix4 matWVP = cam.GetMatrixVP() * matW;
+
+	s_ubWaterVS._matW = matW.UniformBufferFormat();
+	s_ubWaterVS._matVP = cam.GetMatrixVP().UniformBufferFormat();
+	s_ubWaterVS._eyePos = float4(cam.GetEyePosition().GLM(), 0);
+	s_ubWaterVS._waterScale_distToMipScale_mapSideInv_landDistToMipScale.x = 1 / _patchSide;
+	s_ubWaterVS._waterScale_distToMipScale_mapSideInv_landDistToMipScale.y = Math::ComputeDistToMipScale(_genSide << 6, renderer.GetSwapChainHeight(), _patchSide, cam.GetFOV());
+	s_ubWaterVS._waterScale_distToMipScale_mapSideInv_landDistToMipScale.z = 1.f / _pTerrain->GetMapSide();
+	s_ubWaterVS._waterScale_distToMipScale_mapSideInv_landDistToMipScale.w = Math::ComputeDistToMipScale(_pTerrain->GetMapSide(), renderer.GetSwapChainHeight(), _pTerrain->GetMapSide(), cam.GetFOV());
+
+	s_ubWaterFS._phase.x = _phase;
+	s_ubWaterFS._dirToSun = float4(atmo.GetDirToSun().GLM(), 0);
+	s_ubWaterFS._sunColor = float4(atmo.GetSunColor().GLM(), 0);
+
+	cb->BindVertexBuffers(_geo);
+	cb->BindIndexBuffer(_geo);
+	cb->BindPipeline(_pipe[PIPE_MAIN]);
+
+	_shader[SHADER_MAIN]->BeginBindDescriptors();
+	cb->BindDescriptors(_shader[SHADER_MAIN], 0, _cshWaterVS);
+	cb->BindDescriptors(_shader[SHADER_MAIN], 1, _cshWaterFS);
+	cb->DrawIndexed(_indexCount);
+	_shader[SHADER_MAIN]->EndBindDescriptors();
+
+	//ms_cbPerObject.matWVP = matWVP.ConstBufferFormat();
+	//ms_cbPerObject.matW = matW.ConstBufferFormat();
+	//ms_cbPerObject.matV = cam.GetMatrixV().ConstBufferFormat();
+
+	if (false)
+	{
+		//_sb[SB_DEPTH]->Apply(sbad);
+
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::trueWavesReflection)
+		{
+			//renderer->SetTextures({ _tex[TEX_HEIGHTMAP] }, 0 + VERUS_ST_VS_OFFSET);
+			//renderer->SetTextures({ _texLand }, 5 + VERUS_ST_VS_OFFSET);
+		}
+
+		//_s->Bind("TDepth");
+		//_s->UpdateBuffer(0);
+
+		//_geo->BeginDraw(0x1);
+		//renderer->DrawIndexedPrimitive(CGI::PT_TRIANGLESTRIP, 0, _indexCount);
+		//_geo->EndDraw(0x1);
+	}
+	else
+	{
+		VERUS_QREF_ATMO;
+
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::trueWavesRefraction) // True refraction:
+		{
+			//renderer->StretchRect(m_tex[TEX_REFRACT], renderer.GetOffscreenColorTexture());
+			//_tex[TEX_REFRACT]->GenerateMipmaps();
+		}
+
+		//renderer->SetRenderTargets({ renderer.GetOffscreenColorTexture(), renderer.GetDS().GetGBuffer(1) }, renderer.GetOffscreenDepthTexture());
+
+		//_sb[SB_MASTER]->Apply(sbad);
+
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::trueWavesReflection)
+		{
+			//renderer->SetTextures({ _tex[TEX_HEIGHTMAP] }, 0 + VERUS_ST_VS_OFFSET);
+			//renderer->SetTextures({ _texLand }, 5 + VERUS_ST_VS_OFFSET);
+		}
+
+		Matrix4 matTexSpace = Matrix4::identity();
+
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::solidColor)
+		{
+			//renderer->SetTextures({ _tex[TEX_HEIGHTMAP_BASE] });
+			//renderer->SetTextures({ _tex[TEX_FOAM] }, 2);
+			//renderer->SetTextures({ _texLand }, 5);
+			//renderer->SetTextures({ atmo.GetSunShadowTexture() }, 6);
+		}
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::simpleReflection)
+		{
+			//renderer->SetTextures({ _tex[TEX_REFLECT] }, 3);
+			const Transform3 matShift = Transform3::translation(Vector3(0, -0.01f, 0)); // Sky-waterline bleeding fix.
+			//matTexSpace = Matrix4(Math::ToUVMatrix(0, Math::TOUV_RENDER_TARGET_VFLIP | Math::TOUV_MAP_TEXELS_TO_PIXELS, _tex[TEX_REFLECT]->GetSize()) * matShift) * matWVP;
+		}
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::trueWavesReflection)
+		{
+			//renderer->SetTextures({ _tex[TEX_HEIGHTMAP], _tex[TEX_NORMALMAP] });
+		}
+		if (settings._sceneWaterQuality >= App::Settings::WaterQuality::trueWavesRefraction)
+		{
+			//renderer->SetTextures({ _tex[TEX_REFRACT] }, 4);
+		}
+
+		//ms_cbPerFrame.matReflect = matTexSpace.ConstBufferFormat();
+		//ms_cbPerFrame.shadowTexSize = atmo.GetSunShadowTexture() ? atmo.GetSunShadowTexture()->GetSize() : CVector4(0);
+		//ms_cbPerFrame.matSunShadow = atmo.GetSunShadowMatrix(0).ConstBufferFormat();
+		//ms_cbPerFrame.matSunShadowCSM1 = atmo.GetSunShadowMatrix(1).ConstBufferFormat();
+		//ms_cbPerFrame.matSunShadowCSM2 = atmo.GetSunShadowMatrix(2).ConstBufferFormat();
+		//ms_cbPerFrame.matSunShadowCSM3 = atmo.GetSunShadowMatrix(3).ConstBufferFormat();
+		//ms_cbPerFrame.csmParams = atmo.GetParamsCSM();
+		//ms_cbPerFrame.phases_mapSideInv.x = _phase;
+		//ms_cbPerFrame.phases_mapSideInv.y = _phaseWave;
+		//ms_cbPerFrame.phases_mapSideInv.z = 1.f / _mapSide;
+		//ms_cbPerFrame.eyePos = cam.GetPositionEye();
+		//ms_cbPerFrame.colorAmbient = atmo.GetAmbientColor();
+		//ms_cbPerFrame.dirToSun = atmo.GetDirToSun();
+		//ms_cbPerFrame.colorSun = atmo.GetSunColor();
+		//ms_cbPerFrame.colorSunSpec = atmo.GetSunColor();
+		//ms_cbPerFrame.fogColor = atmo.GetFogColor();
+
+		//char tech[8];
+		//sprintf_s(tech, "T_%d", settings._sceneWaterQuality);
+		//_s->Bind(tech);
+		//_s->UpdateBuffer(0);
+		//_s->UpdateBuffer(1);
+
+		//_geo->BeginDraw(0x1);
+		//renderer->DrawIndexedPrimitive(CGI::PT_TRIANGLESTRIP, 0, _indexCount);
+		//_geo->EndDraw(0x1);
+	}
+
+	///renderer->SetTextures({ nullptr }, 0 + VERUS_ST_VS_OFFSET);
+
+	//renderer.DrawQuad(m_tex[HTEX_HEIGHTMAP]);
+}
+
+void Water::BeginReflection(CGI::PBaseCommandBuffer pCB)
+{
+	VERUS_QREF_SM;
+	VERUS_QREF_RENDERER;
+
+	if (!pCB)
+		pCB = &(*renderer.GetCommandBuffer());
+
+	_reflectionMode = sm.GetCamera()->GetEyePosition().getY() >= 0;
+	_renderToTexture = true;
+
+	pCB->BeginRenderPass(_rphReflection, _fbhReflection,
+		{
+			_tex[TEX_REFLECTION]->GetClearValue(),
+			_tex[TEX_REFLECTION_DEPTH]->GetClearValue()
+		});
+}
+
+void Water::EndReflection(CGI::PBaseCommandBuffer pCB)
+{
+	VERUS_QREF_RENDERER;
+
+	if (!pCB)
+		pCB = &(*renderer.GetCommandBuffer());
+
+	pCB->EndRenderPass();
+
+	_reflectionMode = false;
+	_renderToTexture = false;
+}
+
+void Water::GenerateTextures()
+{
+	if (!_cshGenHeightmap.IsSet())
+	{
+		if (_tex[TEX_SOURCE_HEIGHTMAP]->IsLoaded())
+			_cshGenHeightmap = _shader[SHADER_GEN]->BindDescriptorSetTextures(1, { _tex[TEX_SOURCE_HEIGHTMAP] });
+		else
+			return;
+	}
+
+	GenerateHeightmapTexture();
+	GenerateNormalsTexture();
+}
+
+void Water::GenerateHeightmapTexture()
+{
+	VERUS_QREF_RENDERER;
+
+	s_ubGen._matW = Math::QuadMatrix().UniformBufferFormat();
+	s_ubGen._matV = Math::ToUVMatrix().UniformBufferFormat();
+	s_ubGenHeightmapFS._phase.x = _phase;
+	memcpy(&s_ubGenHeightmapFS._amplitudes, _amplitudes, sizeof(_amplitudes));
+
+	auto cb = renderer.GetCommandBuffer();
+
+	cb->BeginRenderPass(_rphGenHeightmap, _fbhGenHeightmap, { _tex[TEX_GEN_HEIGHTMAP]->GetClearValue(), });
+
+	cb->BindPipeline(_pipe[PIPE_GEN_HEIGHTMAP]);
+
+	_shader[SHADER_GEN]->BeginBindDescriptors();
+	cb->BindDescriptors(_shader[SHADER_GEN], 0);
+	cb->BindDescriptors(_shader[SHADER_GEN], 1, _cshGenHeightmap);
+	_shader[SHADER_GEN]->EndBindDescriptors();
+
+	renderer.DrawQuad(&(*cb));
+
+	cb->EndRenderPass();
+
+	_tex[TEX_GEN_HEIGHTMAP]->GenerateMips();
+}
+
+void Water::GenerateNormalsTexture()
+{
+	VERUS_QREF_RENDERER;
+
+	s_ubGen._matW = Math::QuadMatrix().UniformBufferFormat();
+	s_ubGen._matV = Math::ToUVMatrix().UniformBufferFormat();
+	s_ubGenNormalsFS._textureSize = _tex[TEX_GEN_HEIGHTMAP]->GetSize().GLM();
+	s_ubGenNormalsFS._waterScale.x = 1 / _patchSide;
+
+	auto cb = renderer.GetCommandBuffer();
+
+	cb->BeginRenderPass(_rphGenNormals, _fbhGenNormals, { _tex[TEX_GEN_NORMALS]->GetClearValue(), });
+
+	cb->BindPipeline(_pipe[PIPE_GEN_NORMALS]);
+
+	_shader[SHADER_GEN]->BeginBindDescriptors();
+	cb->BindDescriptors(_shader[SHADER_GEN], 0);
+	cb->BindDescriptors(_shader[SHADER_GEN], 2, _cshGenNormals);
+	_shader[SHADER_GEN]->EndBindDescriptors();
+
+	renderer.DrawQuad(&(*cb));
+
+	cb->EndRenderPass();
+
+	_tex[TEX_GEN_NORMALS]->GenerateMips();
+}
+
+CGI::TexturePtr Water::GetCausticsTexture() const
+{
+	VERUS_QREF_CONST_SETTINGS;
+	if (settings._sceneWaterQuality >= App::Settings::WaterQuality::trueWavesReflection)
+		return _tex[TEX_GEN_NORMALS];
+	else
+		return _tex[TEX_SOURCE_HEIGHTMAP];
+}
+
+void Water::CreateWaterPlane()
+{
+	VERUS_QREF_SM;
+
+	const float maxRadius = Math::Min(sm.GetCamera()->GetZFar(), 5000.f);
+	const float stepW = 1.f / (_gridWidth - 1);
+	const float stepH = 1.f / (_gridHeight - 1);
+
+	float xPolar = 0;
+	float zPolar = 0;
+
+	Vector<glm::vec4> vVB;
+	vVB.resize(_gridWidth * _gridHeight);
+
+	const float maxAngle = atan(maxRadius / 10); // Assume best height is 10 meters.
+	VERUS_FOR(i, _gridHeight)
+	{
+		const float radius = tan(zPolar * maxAngle) * 10;
+		VERUS_FOR(j, _gridWidth)
+		{
+			float xSqueeze = xPolar;
+			VERUS_FOR(k, 6)
+				xSqueeze = Math::EaseInOutSine(xSqueeze);
+			xSqueeze = Math::Lerp(xSqueeze, xPolar, 0.2f);
+
+			const float xCartesian = radius * sin(xSqueeze * VERUS_2PI);
+			const float zCartesian = radius * cos(xSqueeze * VERUS_2PI);
+			vVB[i * _gridWidth + j] = glm::vec4(xCartesian, 0, zCartesian, 1);
+			xPolar += stepW;
+		}
+		xPolar = 0;
+		zPolar += stepH;
+	}
+	// Watertight mesh, pun intended:
+	VERUS_FOR(i, _gridHeight)
+		vVB[i * _gridWidth + (_gridWidth - 1)] = vVB[i * _gridWidth];
+
+	_geo->UpdateVertexBuffer(vVB.data(), 0);
+}
+
+float Water::PhillipsSpectrum(float k)
+{
+	const float kl = k;
+	const float k2 = k * k;
+	return exp(-1 / (kl * kl)) / (k2 * k2);
+}

Verus/src/Scene/Water.h

@@ -0,0 +1,119 @@
+#pragma once
+
+namespace verus
+{
+	namespace Scene
+	{
+		class Water : public Singleton<Water>, public Object
+		{
+		public:
+#include "../Shaders/Water.inc.hlsl"
+#include "../Shaders/WaterGen.inc.hlsl"
+
+			enum SHADER
+			{
+				SHADER_MAIN,
+				SHADER_GEN,
+				SHADER_COUNT
+			};
+
+			enum PIPE
+			{
+				PIPE_MAIN,
+				PIPE_GEN_HEIGHTMAP,
+				PIPE_GEN_NORMALS,
+				PIPE_COUNT
+			};
+
+			enum TEX
+			{
+				TEX_FOAM,
+				TEX_SOURCE_HEIGHTMAP,
+				TEX_GEN_HEIGHTMAP,
+				TEX_GEN_NORMALS,
+				TEX_REFLECTION,
+				TEX_REFLECTION_DEPTH,
+
+				TEX_REFRACT,
+				TEX_REFLECT,
+				TEX_REFLECT_DEPTH,
+				TEX_COUNT
+			};
+
+		private:
+			static const int s_maxHarmonics = 4;
+
+			struct Vertex
+			{
+				glm::vec4 _pos;
+			};
+
+			static UB_WaterVS        s_ubWaterVS;
+			static UB_WaterFS        s_ubWaterFS;
+			static UB_Gen            s_ubGen;
+			static UB_GenHeightmapFS s_ubGenHeightmapFS;
+			static UB_GenNormalsFS   s_ubGenNormalsFS;
+
+			PTerrain                      _pTerrain = nullptr;
+			Vector<Vertex>                _vSwapBuffer;
+			CGI::GeometryPwn              _geo;
+			CGI::ShaderPwns<SHADER_COUNT> _shader;
+			CGI::PipelinePwns<PIPE_COUNT> _pipe;
+			CGI::TexturePwns<TEX_COUNT>   _tex;
+			CGI::TexturePtr               _texLand;
+			CGI::CSHandle                 _cshWaterVS;
+			CGI::CSHandle                 _cshWaterFS;
+			CGI::RPHandle                 _rphGenHeightmap;
+			CGI::FBHandle                 _fbhGenHeightmap;
+			CGI::CSHandle                 _cshGenHeightmap;
+			CGI::RPHandle                 _rphGenNormals;
+			CGI::FBHandle                 _fbhGenNormals;
+			CGI::CSHandle                 _cshGenNormals;
+			CGI::RPHandle                 _rphReflection;
+			CGI::FBHandle                 _fbhReflection;
+			const int                     _genSide = 1024;
+			const int                     _sideReflect = 256;
+			int                           _gridWidth = 128;
+			int                           _gridHeight = 512;
+			int                           _indexCount = 0;
+			const float                   _patchSide = 64;
+			float                         _phase = 0;
+			float                         _phaseWave = 0;
+			float                         _amplitudes[s_maxHarmonics];
+			bool                          _reflectionMode = false;
+			bool                          _renderToTexture = false;
+			bool                          _geoReady = false;
+
+		public:
+			Water();
+			~Water();
+
+			void Init(RTerrain terrain);
+			void Done();
+
+			void Update();
+			void Draw();
+
+			void BeginReflection(CGI::PBaseCommandBuffer pCB = nullptr);
+			void EndReflection(CGI::PBaseCommandBuffer pCB = nullptr);
+
+			void GenerateTextures();
+			VERUS_P(void GenerateHeightmapTexture());
+			VERUS_P(void GenerateNormalsTexture());
+
+			// Caustics are highlights on ocean floor.
+			CGI::TexturePtr GetCausticsTexture() const;
+
+			// Render reflection when above or normal view when below water level.
+			bool IsReflectionMode() const { return _reflectionMode; }
+
+			// Is rendering done to water's renderer target?
+			bool IsRenderToTexture() const { return _renderToTexture; }
+
+			VERUS_P(void CreateWaterPlane());
+
+			static float PhillipsSpectrum(float k);
+		};
+		VERUS_TYPEDEFS(Water);
+	}
+}

Verus/src/Shaders/DS.hlsl

@@ -18,7 +18,7 @@ ConstantBuffer<UB_PerObject>  g_ubPerObject  : register(b0, space4);
 VK_SUBPASS_INPUT(0, g_texGBuffer0, g_samGBuffer0, t1, s1, space1);
 VK_SUBPASS_INPUT(1, g_texGBuffer1, g_samGBuffer1, t2, s2, space1);
 VK_SUBPASS_INPUT(2, g_texGBuffer2, g_samGBuffer2, t3, s3, space1);
-VK_SUBPASS_INPUT(3, g_texGBuffer3, g_samGBuffer3, t4, s4, space1);
+VK_SUBPASS_INPUT(3, g_texDepth, g_samDepth, t4, s4, space1);
 Texture2D              g_texShadowCmp : register(t5, space1);
 SamplerComparisonState g_samShadowCmp : register(s5, space1);
 Texture2D              g_texShadow    : register(t6, space1);
@@ -26,18 +26,16 @@ SamplerState           g_samShadow    : register(s6, space1);
 
 struct VSI
 {
-	// Binding 0:
 	VK_LOCATION_POSITION int4 pos   : POSITION;
 	VK_LOCATION_NORMAL   float3 nrm : NORMAL;
 	VK_LOCATION(8)       int4 tc0   : TEXCOORD0;
-
 	_PER_INSTANCE_DATA
 };
 
 struct VSO
 {
 	float4 pos							/**/ : SV_Position;
-	float4 posFS						/**/ : TEXCOORD0;
+	float4 clipSpacePos					/**/ : TEXCOORD0;
 #if defined(DEF_OMNI) || defined(DEF_SPOT)
 	float3 radius_radiusSq_invRadiusSq	/**/ : TEXCOORD1;
 	float3 lightPosWV					/**/ : TEXCOORD2;
@@ -56,8 +54,6 @@ VSO mainVS(VSI si)
 {
 	VSO so;
 
-	const float3 intactPos = DequantizeUsingDeq3D(si.pos.xyz, g_ubPerMeshVS._posDeqScale.xyz, g_ubPerMeshVS._posDeqBias.xyz);
-
 	// World matrix, instance data:
 #ifdef DEF_INSTANCED
 	const mataff matW = GetInstMatrix(
@@ -84,13 +80,15 @@ VSO mainVS(VSI si)
 	const float3x3 matV33 = (float3x3)g_ubPerFrame._matV;
 	const float3x3 matWV33 = (float3x3)matWV;
 
+	const float3 intactPos = DequantizeUsingDeq3D(si.pos.xyz, g_ubPerMeshVS._posDeqScale.xyz, g_ubPerMeshVS._posDeqBias.xyz);
+
 #ifdef DEF_DIR
 	so.pos = float4(mul(float4(intactPos, 1), g_ubPerFrame._matQuad), 1);
-	so.posFS = float4(intactPos, 1);
+	so.clipSpacePos = float4(intactPos, 1);
 #else
 	const float3 posW = mul(float4(intactPos, 1), matW);
 	so.pos = mul(float4(posW, 1), g_ubPerFrame._matVP);
-	so.posFS = so.pos;
+	so.clipSpacePos = so.pos;
 #endif
 
 	// <MoreLightParams>
@@ -128,15 +126,13 @@ DS_ACC_FSO mainFS(VSO si)
 {
 	DS_ACC_FSO so;
 
-	// Clip-space position:
 #ifdef DEF_DIR
-	const float3 posCS = si.posFS.xyz;
+	const float3 ndcPos = si.clipSpacePos.xyz;
 #else
-	const float3 posCS = si.posFS.xyz / si.posFS.w;
+	const float3 ndcPos = si.clipSpacePos.xyz / si.clipSpacePos.w;
 #endif
 
-	// GBuffer tex. coord. from posCS:
-	const float2 tc0 = mul(float4(posCS.xy, 0, 1), g_ubPerFrame._matToUV).xy;
+	const float2 tc0 = mul(float4(ndcPos.xy, 0, 1), g_ubPerFrame._matToUV).xy;
 
 	// For Omni & Spot: light's radius and position:
 #if defined(DEF_OMNI) || defined(DEF_SPOT)
@@ -163,8 +159,8 @@ DS_ACC_FSO mainFS(VSO si)
 #endif
 
 	// GBuffer1:
-	const float rawGBuffer1 = VK_SUBPASS_LOAD(g_texGBuffer1, g_samGBuffer1, tc0).r;
-	const float3 posWV = DS_GetPosition(rawGBuffer1, g_ubPerFrame._matInvP, posCS.xy);
+	const float depth = VK_SUBPASS_LOAD(g_texDepth, g_samDepth, tc0).r;
+	const float3 posWV = DS_GetPosition(depth, g_ubPerFrame._matInvP, ndcPos.xy);
 
 	so.target0 = 0.0;
 	so.target1 = 0.0;
@@ -174,35 +170,34 @@ DS_ACC_FSO mainFS(VSO si)
 #endif
 	{
 		// Light's diffuse & specular color:
-		const float3 colorDiff = si.color_coneOut.rgb;
-		const float3 colorSpec = saturate(colorDiff + dot(colorDiff, 0.1));
+		const float3 hdrLightColor = si.color_coneOut.rgb;
 
-		// GBuffer0 (color, spec):
+		// GBuffer0 (albedo, spec):
 		const float4 rawGBuffer0 = VK_SUBPASS_LOAD(g_texGBuffer0, g_samGBuffer0, tc0);
 		const float spec = rawGBuffer0.a;
 		const float3 dirToEyeWV = normalize(-posWV);
 
 		// GBuffer2 (normal, emission, motion):
-		const float4 rawGBuffer2 = VK_SUBPASS_LOAD(g_texGBuffer2, g_samGBuffer2, tc0);
-		const float3 normalWV = DS_GetNormal(rawGBuffer2);
-		const float2 emission = DS_GetEmission(rawGBuffer2);
+		const float4 rawGBuffer1 = VK_SUBPASS_LOAD(g_texGBuffer1, g_samGBuffer1, tc0);
+		const float3 normalWV = DS_GetNormal(rawGBuffer1);
+		const float2 emission = DS_GetEmission(rawGBuffer1);
 
 		// GBuffer3 (lam, metal, gloss):
-		const float4 rawGBuffer3 = VK_SUBPASS_LOAD(g_texGBuffer3, g_samGBuffer3, tc0);
-		const float3 anisoWV = DS_GetAnisoSpec(rawGBuffer3);
-		const float2 lamScaleBias = DS_GetLamScaleBias(rawGBuffer3);
-		const float2 metal = DS_GetMetallicity(rawGBuffer3);
-		const float gloss64 = rawGBuffer3.a * 64.0;
+		const float4 rawGBuffer2 = VK_SUBPASS_LOAD(g_texGBuffer2, g_samGBuffer2, tc0);
+		const float3 anisoWV = DS_GetAnisoSpec(rawGBuffer2);
+		const float2 lamScaleBias = DS_GetLamScaleBias(rawGBuffer2);
+		const float2 metal = DS_GetMetallicity(rawGBuffer2);
+		const float gloss64 = rawGBuffer2.a * 64.0;
 		const float gloss64Scaled = gloss64 * glossScale;
 
 		// Special:
-		const float skinAlpha = emission.y;
-		const float hairAlpha = metal.y;
-		const float eyeAlpha = saturate(1.0 - gloss64);
-		const float rim = pow(1.0 - normalWV.z, 2.0) * (0.05 + 0.2 * skinAlpha);
-		const float glossEye = 25.0 * glossScale;
+		const float skinMask = emission.y;
+		const float hairMask = metal.y;
+		const float eyeMask = saturate(1.0 - gloss64);
+		const float eyeGloss = 25.0 * glossScale;
+		const float2 lamScaleBiasWithHair = lerp(lamScaleBias, float2(1, 0.4), hairMask);
 
-		const float gloss = lerp(gloss64Scaled * gloss64Scaled * gloss64Scaled, glossEye * glossEye * glossEye, eyeAlpha);
+		const float gloss = lerp(gloss64Scaled * gloss64Scaled * gloss64Scaled, eyeGloss * eyeGloss * eyeGloss, eyeMask);
 
 #ifdef DEF_DIR
 		const float3 dirToLightWV = -lightDirWV;
@@ -225,63 +220,65 @@ DS_ACC_FSO mainFS(VSO si)
 #endif
 
 		const float4 litRet = VerusLit(dirToLightWV, normalWV, dirToEyeWV,
-			lerp(gloss * (2.0 - lightFalloff), 12.0, hairAlpha),
-			lerp(lamScaleBias, float2(1, 0.4), hairAlpha),
-			float4(anisoWV, hairAlpha));
+			lerp(gloss * (2.0 - lightFalloff), 12.0, hairMask),
+			lamScaleBiasWithHair,
+			float4(anisoWV, hairMask));
 
-		// Shadow:
+		// <Shadow>
+		float shadowMask = 1.0;
+		{
 #ifdef DEF_DIR
-		const float4 tcShadow = ShadowCoords(float4(posWV, 1), g_ubShadowFS._matSunShadow, -posWV.z);
-		const float shadowAlpha = ShadowMapCSM(
+			float4 config = g_ubShadowFS._shadowConfig;
+			const float lamBiasMask = saturate(lamScaleBiasWithHair.y * config.y);
+			config.y = 1.0 - lamBiasMask; // Keep penumbra blurry.
+			const float scale = -posWV.z - 5.0;
+			const float3 posForShadow = posWV + normalWV * 0.012 * max(1.0, scale * 0.2) + dirToLightWV * max(0.0, scale * 0.002);
+			const float4 tcShadow = ShadowCoords(float4(posForShadow, 1), g_ubShadowFS._matSunShadow, -posForShadow.z);
+			shadowMask = ShadowMapCSM(
 				g_texShadowCmp,
 				g_samShadowCmp,
 				g_texShadow,
 				g_samShadow,
 				tcShadow,
-			g_ubShadowFS._shadowConfig,
+				config,
 				g_ubShadowFS._splitRanges,
 				g_ubShadowFS._matSunShadow,
 				g_ubShadowFS._matSunShadowCSM1,
 				g_ubShadowFS._matSunShadowCSM2,
 				g_ubShadowFS._matSunShadowCSM3);
-#else
-		const float shadowAlpha = 1.0;
 #endif
+		}
+		// </Shadow>
 
-		const float intensityDiff = lightFalloff * coneIntensity * shadowAlpha;
-		const float intensitySpec = saturate(intensityDiff * 2.0) * shadowAlpha;
-		const float maxSpecAdd = lerp(0.2 * spec, 0.1 + spec * 0.5, metal.x);
+		const float lightFalloffWithCone = lightFalloff * coneIntensity;
+		const float diffLightMask = lightFalloffWithCone * shadowMask;
+		const float specLightMask = saturate(lightFalloffWithCone * 2.0) * shadowMask;
+		const float maxSpecAdd = (1.0 - spec) * lerp(0.01, 0.5, metal.x);
 
 		// Subsurface scattering effect for front & back faces:
-		const float3 colorSkinSSS_Face = float3(1.6, 1.2, 0.3);
-		const float3 colorSkinSSS_Back = float3(1.6, 0.8, 0.5);
-		const float3 colorDiffSSS_Face = lerp(colorDiff, colorDiff * colorSkinSSS_Face, skinAlpha * pow(1.0 - litRet.y, 2.0));
-		const float3 colorDiffSSS_Back = lerp(colorDiffSSS_Face, colorDiff * colorSkinSSS_Back, skinAlpha * litRet.x);
+		const float3 frontSkinSSSColor = float3(1.6, 1.2, 0.3);
+		const float3 backSkinSSSColor = float3(1.6, 0.8, 0.5);
+		const float frontSkinSSSMask = 1.0 - litRet.y;
+		const float3 hdrLightFaceColor = hdrLightColor * lerp(1.0, frontSkinSSSColor, skinMask * frontSkinSSSMask * frontSkinSSSMask);
+		const float3 hdrLightColorSSS = lerp(hdrLightFaceColor, hdrLightColor * backSkinSSSColor, skinMask * litRet.x);
 
 		const float gray = Grayscale(rawGBuffer0.rgb);
 		const float3 metalSpec = saturate(rawGBuffer0.rgb / (gray + _SINGULARITY_FIX));
 
-		const float3 matSpec = lerp(float3(1, 1, 1), metalSpec, saturate(metal.x + (hairAlpha - eyeAlpha) * 0.4)) * FresnelSchlick(spec, maxSpecAdd, litRet.w);
-		const float3 maxDiff = colorDiffSSS_Back * intensityDiff;
-		const float3 maxSpec = matSpec * colorSpec + float3(-0.03, 0.0, 0.05) * skinAlpha;
-#ifdef DEF_DIR
-		const float skyIntensity = 1.0;
-		const float3 skySpec = skyColor * matSpec * skyIntensity;
-#endif
-
-		// Very bright light:
-		const float3 overbright = saturate(dot(maxDiff, 1.0 / 3.0) - 1.0); // Intensity above one.
-		const float3 diffBoost = overbright * maxDiff * litRet.y * (0.25 + rawGBuffer0.rgb);
+		const float3 specHue = saturate(lerp(float3(1, 1, 1), metalSpec, saturate(metal.x + (hairMask - eyeMask) * 0.4)) +
+			float3(-0.03, 0.0, 0.05) * skinMask);
+		const float3 hdrMaxDiff = hdrLightColorSSS * diffLightMask;
+		const float3 hdrMaxSpec = hdrLightColorSSS * specLightMask * specHue * FresnelSchlick(spec, maxSpecAdd, litRet.w);
 
 #ifdef DEF_DIR
-		so.target0.rgb = maxDiff * litRet.y; // Lambert's cosine law.
-		so.target1.rgb = maxSpec * saturate(litRet.z * intensitySpec + rim) + diffBoost;
+		so.target0.rgb = hdrMaxDiff * litRet.y;
+		so.target1.rgb = hdrMaxSpec * litRet.z;
 #else
-		so.target0.rgb = maxDiff * litRet.y; // Lambert's cosine law.
-		so.target1.rgb = maxSpec * saturate(litRet.z * intensitySpec) + diffBoost;
+		so.target0.rgb = hdrMaxDiff * litRet.y;
+		so.target1.rgb = hdrMaxSpec * litRet.z;
 #endif
-		so.target0.rgb = max(so.target0.rgb, emission.x);
-		so.target1.rgb = min(so.target1.rgb, 1.0 - emission.x);
+		//so.target0.rgb = max(so.target0.rgb, emission.x);
+		//so.target1.rgb = min(so.target1.rgb, 1.0 - emission.x);
 		so.target0.a = 1.0;
 		so.target1.a = 1.0;
 	}

Verus/src/Shaders/DS_Compose.hlsl

@@ -15,8 +15,8 @@ Texture2D    g_texGBuffer1 : register(t2, space1);
 SamplerState g_samGBuffer1 : register(s2, space1);
 Texture2D    g_texGBuffer2 : register(t3, space1);
 SamplerState g_samGBuffer2 : register(s3, space1);
-Texture2D    g_texGBuffer3 : register(t4, space1);
-SamplerState g_samGBuffer3 : register(s4, space1);
+Texture2D    g_texDepth    : register(t4, space1);
+SamplerState g_samDepth    : register(s4, space1);
 Texture2D    g_texAccDiff  : register(t5, space1);
 SamplerState g_samAccDiff  : register(s5, space1);
 Texture2D    g_texAccSpec  : register(t6, space1);
@@ -31,6 +31,7 @@ struct VSO
 {
 	float4 pos : SV_Position;
 	float2 tc0 : TEXCOORD0;
+	float2 clipSpacePos : TEXCOORD1;
 };
 
 struct FSO
@@ -46,6 +47,7 @@ VSO mainVS(VSI si)
 	// Standard quad:
 	so.pos = float4(mul(si.pos, g_ubComposeVS._matW), 1);
 	so.tc0 = mul(si.pos, g_ubComposeVS._matV).xy;
+	so.clipSpacePos = so.pos.xy;
 
 	return so;
 }
@@ -56,14 +58,17 @@ FSO mainFS(VSO si)
 {
 	FSO so;
 
+#ifdef DEF_COMPOSE
 	const float3 colorAmbient = g_ubComposeFS._colorAmbient.rgb;
 
+	const float2 ndcPos = si.clipSpacePos.xy;
+
 	const float4 rawGBuffer0 = g_texGBuffer0.Sample(g_samGBuffer0, si.tc0);
-	const float1 rawGBuffer1 = g_texGBuffer1.Sample(g_samGBuffer1, si.tc0).r; // Depth for fog.
+	const float4 rawGBuffer1 = g_texGBuffer1.Sample(g_samGBuffer1, si.tc0);
 	const float4 rawGBuffer2 = g_texGBuffer2.Sample(g_samGBuffer2, si.tc0);
-	const float4 rawGBuffer3 = g_texGBuffer3.Sample(g_samGBuffer3, si.tc0);
-	const float3 normalWV = DS_GetNormal(rawGBuffer2);
-	const float2 emission = DS_GetEmission(rawGBuffer2);
+	const float1 rawDepth = g_texDepth.Sample(g_samDepth, si.tc0).r;
+	const float3 normalWV = DS_GetNormal(rawGBuffer1);
+	const float2 emission = DS_GetEmission(rawGBuffer1);
 
 	const float4 rawAccDiff = g_texAccDiff.Sample(g_samAccDiff, si.tc0);
 	const float4 rawAccSpec = g_texAccSpec.Sample(g_samAccSpec, si.tc0);
@@ -74,28 +79,40 @@ FSO mainFS(VSO si)
 	const float ssaoSpec = 1.0;
 	const float ssaoAmb = 1.0;
 
-	const float3 albedo = rawGBuffer0.rgb;
+	const float3 albedo = max(rawGBuffer0.rgb * g_ubComposeFS._toneMappingConfig.z, 0.0001);
 
-	const float depth = ToLinearDepth(rawGBuffer1, g_ubComposeFS._zNearFarEx);
-	const float fog = ComputeFog(depth, g_ubComposeFS._fogColor.a);
+	const float depth = ToLinearDepth(rawDepth, g_ubComposeFS._zNearFarEx);
+	const float3 posW = DS_GetPosition(rawDepth, g_ubComposeFS._matInvVP, ndcPos);
+	const float fog = ComputeFog(depth, g_ubComposeFS._fogColor.a, posW.y);
 
 	const float3 normalW = mul(normalWV, (float3x3)g_ubComposeFS._matInvV);
-	const float gray = Grayscale(colorAmbient);
-	const float3 colorGround = lerp(colorAmbient, float3(1, 0.88, 0.47) * gray, saturate(gray * 14.0));
+	const float grayAmbient = Grayscale(colorAmbient);
+	const float3 colorGround = lerp(colorAmbient, float3(1, 0.88, 0.47) * grayAmbient, saturate(grayAmbient * 14.0));
 	const float3 colorAmbientFinal = lerp(colorGround, colorAmbient, saturate(normalW.y * 2.0 + 0.5));
 
 	const float3 color =
-		albedo * saturate(accDiff.rgb * ssaoDiff + colorAmbientFinal * ssaoAmb) +
+		albedo * (accDiff.rgb * ssaoDiff + colorAmbientFinal * ssaoAmb) +
 		accSpec.rgb * ssaoSpec +
 		albedo * emission.x;
 
 	const float3 colorWithFog = lerp(color, g_ubComposeFS._fogColor.rgb, fog);
 
-	so.color.rgb = lerp(colorWithFog, g_ubComposeFS._colorBackground.rgb, floor(rawGBuffer1) * g_ubComposeFS._colorBackground.a);
+	so.color.rgb = colorWithFog;
+	so.color.rgb = lerp(so.color.rgb, g_ubComposeFS._colorBackground.rgb, floor(rawDepth) * g_ubComposeFS._colorBackground.a);
 	so.color.a = 1.0;
+#endif
+
+#ifdef DEF_TONE_MAPPING
+	const float4 rawComposed = g_texGBuffer0.Sample(g_samGBuffer0, si.tc0);
+
+	so.color.rgb = VerusToneMapping(rawComposed.rgb * g_ubComposeFS._toneMappingConfig.x, g_ubComposeFS._toneMappingConfig.y);
+	so.color.a = 1.0;
+
+#endif
 
 	return so;
 }
 #endif
 
-//@main:#
+//@main:#Compose COMPOSE
+//@main:#ToneMapping TONE_MAPPING

Verus/src/Shaders/DS_Compose.inc.hlsl

@@ -9,8 +9,10 @@ VERUS_UBUFFER UB_ComposeVS
 VERUS_UBUFFER UB_ComposeFS
 {
 	mataff _matInvV;
+	matrix _matInvVP;
 	float4 _colorAmbient;
 	float4 _colorBackground;
 	float4 _fogColor;
 	float4 _zNearFarEx;
+	float4 _toneMappingConfig;
 };

Verus/src/Shaders/DS_Grass.hlsl

@@ -0,0 +1,237 @@
+// Copyright (C) 2020, Dmitry Maluev (dmaluev@gmail.com)
+
+#include "Lib.hlsl"
+#include "LibColor.hlsl"
+#include "LibDeferredShading.hlsl"
+#include "LibSurface.hlsl"
+#include "LibVertex.hlsl"
+#include "DS_Grass.inc.hlsl"
+
+ConstantBuffer<UB_GrassVS> g_ubGrassVS : register(b0, space0);
+ConstantBuffer<UB_GrassFS> g_ubGrassFS : register(b0, space1);
+
+Texture2D    g_texHeightVS : register(t1, space0);
+SamplerState g_samHeightVS : register(s1, space0);
+Texture2D    g_texNormalVS : register(t2, space0);
+SamplerState g_samNormalVS : register(s2, space0);
+Texture2D    g_texMLayerVS : register(t3, space0);
+SamplerState g_samMLayerVS : register(s3, space0);
+
+Texture2D    g_texAlbedo : register(t1, space1);
+SamplerState g_samAlbedo : register(s1, space1);
+
+struct VSI
+{
+	VK_LOCATION_POSITION int4 pos      : POSITION;
+	VK_LOCATION(8)       int4 tc       : TEXCOORD0;
+	VK_LOCATION(16)      int4 patchPos : INSTDATA0;
+	uint vertexID                      : SV_VertexID;
+};
+
+struct VSO
+{
+	float4 pos                 : SV_Position;
+	float2 tc0                 : TEXCOORD0;
+	float3 tcOffset_phaseShift : TEXCOORD1;
+	float4 normal_top          : TEXCOORD2;
+	float2 psize               : PSIZE;
+};
+
+#ifdef _VS
+VSO mainVS(VSI si)
+{
+	VSO so;
+
+	const float phase = g_ubGrassVS._phase_mapSideInv_bushMask.x;
+	const float mapSideInv = g_ubGrassVS._phase_mapSideInv_bushMask.y;
+	const float bushMask = g_ubGrassVS._phase_mapSideInv_bushMask.z;
+	const float3 warp = g_ubGrassVS._warp_turb.xyz;
+	const float turbulence = g_ubGrassVS._warp_turb.w;
+
+#ifdef DEF_BILLBOARDS
+	const uint bushID = si.vertexID;
+#else
+	const uint bushID = si.vertexID >> 3; // 8 vertices per bush.
+#endif
+
+	// <FromTextures>
+	float3 intactPos;
+	float3 pos;
+	float2 center;
+	float2 pointSpriteScale = 1.0;
+	float groundHeight;
+	float3 normal;
+	float2 tc0;
+	bool bushMaskOK = true;
+	{
+		intactPos = si.pos.xyz * (1.0 / 1000.0);
+		pos = intactPos + float3(si.patchPos.x, 0, si.patchPos.z);
+		center = si.tc.zw * (1.0 / 1000.0) + si.patchPos.xz;
+#ifdef DEF_BILLBOARDS
+		pointSpriteScale = intactPos.y;
+		pos = float3(center.x, 0.45 * pointSpriteScale.y, center.y);
+#endif
+
+		const float bestPrecision = 50.0;
+		const float distToEye = distance(pos + float3(0, si.patchPos.y * 0.01, 0), g_ubGrassVS._posEye.xyz);
+		const float geomipsLod = log2(clamp(distToEye * (2.0 / 100.0), 1.0, 18.0));
+		const float texelCenter = 0.5 * mapSideInv;
+		const float mipTexelCenter = texelCenter * exp2(geomipsLod);
+		const float2 tcMap = pos.xz * mapSideInv + 0.5;
+		const float2 tcUniform = center * mapSideInv + 0.5;
+		groundHeight = g_texHeightVS.SampleLevel(g_samHeightVS, tcMap + mipTexelCenter, geomipsLod).r + bestPrecision;
+		pos.y += groundHeight;
+
+		const float4 rawNormal = g_texNormalVS.SampleLevel(g_samNormalVS, tcUniform + texelCenter, 0.0);
+		normal = float3(rawNormal.x, 0, rawNormal.y) * 2.0 - 1.0;
+		normal.y = ComputeNormalZ(normal.xz);
+
+		const float layer = g_texMLayerVS.SampleLevel(g_samMLayerVS, tcUniform, 0.0).r * 4.0;
+		const float uShift = frac(layer) - (16.0 / 256.0);
+		const float vShift = floor(layer) * 0.25;
+		float vShiftAlt = 0.0;
+		if (!(bushID & 0xF))
+			vShiftAlt = 0.5; // Every 16th bush uses alternative texture.
+		tc0 = si.tc.xy * (1.0 / 100.0) + float2(uShift, vShift + vShiftAlt);
+
+		// Cull blank bushes:
+		const int mainLayer = round(layer * 4.0);
+		const int ibushMask = asint(bushMask);
+		if (!((ibushMask >> mainLayer) & 0x1))
+		{
+			bushMaskOK = false;
+		}
+	}
+	// </FromTextures>
+
+	// <Special>
+	float phaseShift = 0.0;
+	float2 windWarp = 0.0;
+	float top = 0.0;
+	{
+#ifdef DEF_BILLBOARDS
+		phaseShift = frac(si.pos.w * (1.0 / 100.0));
+		windWarp = 0.0;
+		top = 0.28;
+#else
+		if (intactPos.y >= 0.1)
+		{
+			phaseShift = frac(si.pos.w * (1.0 / 100.0));
+			windWarp = warp.xz * (1.0 + turbulence * sin((phase + phaseShift) * (_PI * 2.0)));
+			top = 1.0;
+		}
+#endif
+	}
+	// </Special>
+
+	// <Warp>
+	float3 posWarped = pos;
+	{
+		const float distToEye = -mul(float4(posWarped, 1), g_ubGrassVS._matWV).z;
+		const float distant = saturate((distToEye - 50.0) * (1.0 / 50.0)); // [50.0 to 100.0] -> [0.0 to 1.0].
+		const float distExt = saturate((distToEye - 25.0) * (1.0 / 25.0)); // [25.0 to 50.0] -> [0.0 to 1.0].
+
+		const float cliff = dot(normal.xz, normal.xz);
+		float hide = cliff + step(groundHeight, 1.0) + distant;
+		if (!bushMaskOK)
+			hide = 1.0;
+		posWarped.xz -= normal.xz;
+#ifdef DEF_BILLBOARDS
+		const float distExtInv = 1.0 - distExt;
+		posWarped.xz += windWarp * 0.5;
+		hide += distExtInv * distExtInv;
+#else
+		posWarped.xz += windWarp;
+		posWarped.y -= dot(windWarp, windWarp);
+		hide += distExt * distExt;
+#endif
+		hide = saturate(hide);
+
+#ifdef DEF_BILLBOARDS
+		pointSpriteScale = lerp(pointSpriteScale, float2(0.0, pointSpriteScale.y), hide);
+#else
+		posWarped = lerp(posWarped, float3(center.x, posWarped.y, center.y), hide); // Optimize by morphing to center point.
+#endif
+	}
+	// </Warp>
+
+	const float3 posW = mul(float4(posWarped, 1), g_ubGrassVS._matW).xyz;
+	so.pos = mul(float4(posW, 1), g_ubGrassVS._matVP);
+	so.tc0 = tc0;
+	so.tcOffset_phaseShift = float3(0, 0, phaseShift);
+	so.normal_top.xyz = mul(normal, (float3x3)g_ubGrassVS._matWV);
+	so.normal_top.w = top;
+	so.psize = 1.0;
+
+#ifdef DEF_BILLBOARDS
+	so.tcOffset_phaseShift.xy = tc0;
+	const float2 pointSize = g_ubGrassVS._viewportSize.yx * g_ubGrassVS._viewportSize.z * pointSpriteScale;
+	so.psize = pointSize * g_ubGrassVS._matP._m11;
+#else
+	so.normal_top.xyz += float3(0, 0, top * top * 0.25);
+#endif
+
+	return so;
+}
+#endif
+
+#ifdef _GS
+[maxvertexcount(4)]
+void mainGS(point VSO si[1], inout TriangleStream<VSO> stream)
+{
+	VSO so;
+
+	so = si[0];
+	const float2 center = so.pos.xy;
+	for (int i = 0; i < 4; ++i)
+	{
+		so.pos.xy = center + _POINT_SPRITE_POS_OFFSETS[i] * so.psize;
+		so.tc0.xy = _POINT_SPRITE_TEX_COORDS[i];
+		stream.Append(so);
+	}
+}
+#endif
+
+#ifdef _FS
+DS_FSO mainFS(VSO si)
+{
+	DS_FSO so;
+
+	const float3 rand = Rand(si.pos.xy);
+
+	float2 tc = si.tc0;
+#ifdef DEF_BILLBOARDS
+	tc = si.tc0 * 0.23 + si.tcOffset_phaseShift.xy;
+#endif
+
+	const float4 rawAlbedo = g_texAlbedo.Sample(g_samAlbedo, tc);
+	const float3 normal = normalize(si.normal_top.xyz);
+	const float gray = Grayscale(rawAlbedo.rgb);
+	const float mask = saturate((gray - 0.25) * 4.0 + 0.25);
+
+	const float top = si.normal_top.w;
+	const float spec = saturate(top * top * (mask + si.tcOffset_phaseShift.z * 0.1));
+	const float gloss = lerp(4.0, 12.0, spec);
+
+	{
+		DS_Reset(so);
+
+		DS_SetAlbedo(so, rawAlbedo.rgb);
+		DS_SetSpec(so, spec);
+
+		DS_SetNormal(so, normal + NormalDither(rand));
+		DS_SetEmission(so, 0.0, 0.0);
+
+		DS_SetLamScaleBias(so, float2(1.2, -0.2), 0.0);
+		DS_SetMetallicity(so, 0.05, 0.0);
+		DS_SetGloss(so, gloss);
+	}
+
+	clip(rawAlbedo.a - 0.5);
+
+	return so;
+}
+#endif
+
+//@main:#
+//@main:#Billboards BILLBOARDS (VGF)

Verus/src/Shaders/DS_Grass.inc.hlsl

@@ -0,0 +1,18 @@
+// Copyright (C) 2020, Dmitry Maluev (dmaluev@gmail.com)
+
+VERUS_UBUFFER UB_GrassVS
+{
+	mataff _matW;
+	mataff _matWV;
+	matrix _matVP;
+	matrix _matP;
+	float4 _phase_mapSideInv_bushMask;
+	float4 _posEye;
+	float4 _viewportSize;
+	float4 _warp_turb;
+};
+
+VERUS_UBUFFER UB_GrassFS
+{
+	float4 _dummy;
+};

Verus/src/Shaders/DS_Mesh.hlsl

@@ -3,6 +3,8 @@
 #include "Lib.hlsl"
 #include "LibColor.hlsl"
 #include "LibDeferredShading.hlsl"
+#include "LibSurface.hlsl"
+#include "LibTessellation.hlsl"
 #include "LibVertex.hlsl"
 #include "DS_Mesh.inc.hlsl"
 
@@ -17,8 +19,10 @@ Texture2D    g_texAlbedo : register(t1, space1);
 SamplerState g_samAlbedo : register(s1, space1);
 Texture2D    g_texNormal : register(t2, space1);
 SamplerState g_samNormal : register(s2, space1);
-//Texture2D    g_texDetail : register(t3, space1);
-//SamplerState g_samDetail : register(s3, space1);
+Texture2D    g_texDetail : register(t3, space1);
+SamplerState g_samDetail : register(s3, space1);
+Texture2D    g_texStrass : register(t4, space1);
+SamplerState g_samStrass : register(s4, space1);
 
 struct VSI
 {
@@ -28,18 +32,17 @@ struct VSI
 	VK_LOCATION(8)       int4 tc0   : TEXCOORD0;
 	// Binding 1:
 #ifdef DEF_SKINNED
-	VK_LOCATION(12)      int4 bw : TEXCOORD4;
-	VK_LOCATION(13)      int4 bi : TEXCOORD5;
+	VK_LOCATION_BLENDWEIGHTS int4 bw : BLENDWEIGHTS;
+	VK_LOCATION_BLENDINDICES int4 bi : BLENDINDICES;
 #endif
 	// Binding 2:
-	VK_LOCATION(14)      float4 tan : TEXCOORD6;
-	VK_LOCATION(15)      float4 bin : TEXCOORD7;
+	VK_LOCATION_TANGENT  float4 tan : TANGENT;
+	VK_LOCATION_BINORMAL float4 bin : BINORMAL;
 	// Binding 3:
 #if 0
 	VK_LOCATION(9)     int4 tc1     : TEXCOORD1;
 	VK_LOCATION_COLOR0 float4 color : COLOR0;
 #endif
-
 	_PER_INSTANCE_DATA
 };
 
@@ -47,13 +50,14 @@ struct VSO
 {
 	float4 pos     : SV_Position;
 	float2 tc0     : TEXCOORD0;
-#ifndef DEF_DEPTH
-	float4 matTBN0 : TEXCOORD2;
-	float4 matTBN1 : TEXCOORD3;
-	float4 matTBN2 : TEXCOORD4;
+#if !defined(DEF_DEPTH)
 	float4 color0  : COLOR0;
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
+	float4 matTBN0 : TEXCOORD1;
+	float4 matTBN1 : TEXCOORD2;
+	float4 matTBN2 : TEXCOORD3;
+#endif
 #endif
-	float2 depth : TEXCOORD6;
 };
 
 #ifdef _VS
@@ -142,15 +146,76 @@ VSO mainVS(VSI si)
 #endif
 	}
 
-	so.pos = mul(float4(posW, 1), g_ubPerFrame._matVP);
+	so.pos = MulTessPos(float4(posW, 1), g_ubPerFrame._matV, g_ubPerFrame._matVP);
 	so.tc0 = intactTc0;
-#ifndef DEF_DEPTH
+#if !defined(DEF_DEPTH)
+	so.color0 = userColor;
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
 	so.matTBN0 = float4(tanWV, posW.x);
 	so.matTBN1 = float4(binWV, posW.y);
 	so.matTBN2 = float4(nrmWV, posW.z);
-	so.color0 = userColor;
 #endif
-	so.depth = so.pos.zw;
+#endif
+
+	return so;
+}
+#endif
+
+_HSO_STRUCT;
+
+#ifdef _HS
+PCFO PatchConstFunc(const OutputPatch<HSO, 3> outputPatch)
+{
+	PCFO so;
+	_HS_PCF_BODY(g_ubPerFrame._matP);
+	return so;
+}
+
+[domain("tri")]
+//[maxtessfactor(7.0)]
+[outputcontrolpoints(3)]
+[outputtopology("triangle_cw")]
+[partitioning(_PARTITION_METHOD)]
+[patchconstantfunc("PatchConstFunc")]
+HSO mainHS(InputPatch<VSO, 3> inputPatch, uint id : SV_OutputControlPointID)
+{
+	HSO so;
+
+	_HS_PN_BODY(matTBN2, g_ubPerFrame._matP, g_ubPerFrame._viewportSize);
+
+	_HS_COPY(pos);
+	_HS_COPY(tc0);
+#if !defined(DEF_DEPTH)
+	_HS_COPY(color0);
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
+	_HS_COPY(matTBN0);
+	_HS_COPY(matTBN1);
+	_HS_COPY(matTBN2);
+#endif
+#endif
+
+	return so;
+}
+#endif
+
+#ifdef _DS
+[domain("tri")]
+VSO mainDS(_IN_DS)
+{
+	VSO so;
+
+	_DS_INIT_SMOOTH_POS;
+
+	so.pos = ApplyProjection(smoothPosWV, g_ubPerFrame._matP);
+	_DS_COPY(tc0);
+#if !defined(DEF_DEPTH)
+	_DS_COPY(color0);
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
+	_DS_COPY(matTBN0);
+	_DS_COPY(matTBN1);
+	_DS_COPY(matTBN2);
+#endif
+#endif
 
 	return so;
 }
@@ -170,13 +235,8 @@ DS_FSO mainFS(VSO si)
 {
 	DS_FSO so;
 
-#ifdef DEF_SOLID
-	DS_Reset(so);
-	DS_SetAlbedo(so, si.color0.rgb);
-	DS_SetDepth(so, si.depth.x / si.depth.y);
-	DS_SetNormal(so, float3(0, 0, 1));
-	DS_SetEmission(so, 1.0, 0.0);
-	DS_SetMetallicity(so, 1.0, 0.0);
+#ifdef DEF_SOLID_COLOR
+	DS_SolidColor(so, si.color0.rgb);
 #else
 	_TBN_SPACE(
 		si.matTBN0.xyz,
@@ -185,11 +245,16 @@ DS_FSO mainFS(VSO si)
 	const float3 rand = Rand(si.pos.xy);
 
 	// <Albedo>
+	float4 rawAlbedo;
+	{
 		const float texAlbedoEnable = ceil(g_ubPerMaterialFS._texEnableAlbedo.a);
-	float4 rawAlbedo = g_texAlbedo.Sample(g_samAlbedo, si.tc0 * texAlbedoEnable);
+		rawAlbedo = g_texAlbedo.Sample(g_samAlbedo, si.tc0 * texAlbedoEnable);
 		rawAlbedo.rgb = lerp(g_ubPerMaterialFS._texEnableAlbedo.rgb, rawAlbedo.rgb, g_ubPerMaterialFS._texEnableAlbedo.a);
-	const float2 alpha_spec = AlphaSwitch(rawAlbedo, si.tc0, g_ubPerMaterialFS._ssb_as.zw);
+	}
+	const float gray = Grayscale(rawAlbedo.rgb);
+	// </Albedo>
 
+	const float2 alpha_spec = AlphaSwitch(rawAlbedo, si.tc0, g_ubPerMaterialFS._ssb_as.zw);
 	const float emitAlpha = PickAlpha(rawAlbedo.rgb, g_ubPerMaterialFS._emitPick, 16.0);
 	const float emitXAlpha = PickAlpha(rawAlbedo.rgb, g_ubPerMaterialFS._emitXPick, 16.0);
 	const float eyeAlpha = PickAlphaRound(g_ubPerMaterialFS._eyePick, si.tc0);
@@ -199,61 +264,79 @@ DS_FSO mainFS(VSO si)
 	const float skinAlpha = PickAlpha(rawAlbedo.rgb, g_ubPerMaterialFS._skinPick, 16.0);
 	const float userAlpha = PickAlphaHue(rawAlbedo.rgb, g_ubPerMaterialFS._userPick, 32.0);
 
-	// Add details:
-#if 0
-	const float3 colorDetail = (g_texDetail.Sample(g_samDetail, frac(si.tc0 * g_ubPerMaterialFS._ds_scale.zw)).rgb - 0.5) *
-		g_ubPerMaterialFS._ds_scale.x + 0.5;
-	rawAlbedo.rgb = rawAlbedo.rgb * colorDetail * 2.0;
-#endif
-
-	const float gray = Grayscale(rawAlbedo.rgb);
 	rawAlbedo.rgb = lerp(rawAlbedo.rgb, Overlay(gray, si.color0.rgb), userAlpha * si.color0.a);
-	// </Albedo>
+	const float3 hairAlbedo = Overlay(alpha_spec.y, Desaturate(rawAlbedo.rgb, hairAlpha * g_ubPerMaterialFS._hairParams.w));
+
+	// <Gloss>
+	//float gloss = lerp(g_ubPerMaterialFS._lsb_gloss_lp.z, g_ubPerMaterialFS._motionBlur_glossX.y, glossXAlpha);
+	float gloss = lerp(4.4, 16.0, alpha_spec.y);
+	gloss = lerp(gloss, 4.0 + alpha_spec.y, skinAlpha);
+	gloss = lerp(gloss, 0.0, eyeAlpha);
+	// </Gloss>
 
 	// <Normal>
+	float3 normalWV;
+	float toksvigFactor;
+	float lightPassStrength;
+	float3 anisoWV;
+	{
 		const float texNormalEnable = ceil(g_ubPerMaterialFS._texEnableNormal.a);
 		float4 rawNormal = g_texNormal.Sample(g_samNormal, si.tc0 * texNormalEnable);
 		rawNormal = lerp(g_ubPerMaterialFS._texEnableNormal.rgbr, rawNormal, g_ubPerMaterialFS._texEnableNormal.a);
 		const float4 normalAA = NormalMapAA(rawNormal);
 		const float3 normalTBN = normalAA.xyz;
-	const float3 normalWV = normalize(mul(normalTBN, matFromTBN));
-
-	const float3 anisoWV = normalize(mul(cross(normalTBN, cross(g_ubPerMaterialFS._hairParams.xyz, normalTBN)), matFromTBN));
-	// Almost solid hair color is used for picking.
-	// Final albedo is calculated using specualar component.
-	const float3 hairAlbedo = Overlay(alpha_spec.y, Desaturate(rawAlbedo.rgb, hairAlpha * g_ubPerMaterialFS._hairParams.w));
+		normalWV = normalize(mul(normalTBN, matFromTBN));
+		toksvigFactor = ComputeToksvigFactor(normalAA.a, gloss);
+		lightPassStrength = rawNormal.r;
+		anisoWV = normalize(mul(cross(normalTBN, cross(g_ubPerMaterialFS._hairParams.xyz, normalTBN)), matFromTBN));
+	}
 	// </Normal>
 
+	// <Detail>
+	{
+		const float3 rawDetail = g_texDetail.Sample(g_samDetail, si.tc0 * g_ubPerMaterialFS._ds_scale.zw).rgb;
+		rawAlbedo.rgb = rawAlbedo.rgb * lerp(0.5, rawDetail, g_ubPerMaterialFS._ds_scale.x) * 2.0;
+	}
+	// </Detail>
+
+	// <Strass>
+	float strass;
+	{
+		const float rawStrass = g_texStrass.Sample(g_samStrass, si.tc0 * g_ubPerMaterialFS._ds_scale.zw * 2.0).r;
+		strass = saturate(rawStrass * (0.3 + 0.7 * alpha_spec.y) * 4.0) * g_ubPerMaterialFS._ds_scale.y;
+	}
+	// </Strass>
+
 	// <LambertianScaleBias>
-	float2 lamScaleBias = g_ubPerMaterialFS._lsb_gloss_lp.xy + float2(0, rawNormal.r * 8.0 * g_ubPerMaterialFS._lsb_gloss_lp.w);
+	float2 lamScaleBias = g_ubPerMaterialFS._lsb_gloss_lp.xy + float2(0, lightPassStrength * 8.0 * g_ubPerMaterialFS._lsb_gloss_lp.w);
+	lamScaleBias += float2(-0.1, -0.3) * alpha_spec.y + float2(0.1, 0.2); // We bring the noise!
 	lamScaleBias = lerp(lamScaleBias, float2(1, 0.45), skinAlpha);
 	// </LambertianScaleBias>
 
-	// <Gloss>
-	float gloss = lerp(g_ubPerMaterialFS._lsb_gloss_lp.z, g_ubPerMaterialFS._motionBlur_glossX.y, glossXAlpha);
-	gloss = lerp(gloss, 4.0 + alpha_spec.y, skinAlpha);
-	gloss = lerp(gloss, 0.0, eyeAlpha);
-	// </Gloss>
-
-	const float strass = 0.0;
-
 	// <RimAlbedo>
-	//const float rimA = lerp(0.3 + alpha_spec.y*0.3, 0.7, skinAlpha);
-	//const float3 rimAlbedo = saturate(rawAlbedo.rgb*rimA*2.0);
-	//const float rimAlbedoAmount = 1.0 - abs(normalWV.z);
-	//rawAlbedo.rgb = lerp(rawAlbedo.rgb, rimAlbedo, rimAlbedoAmount*rimAlbedoAmount);
+	{
+		const float3 newColor = saturate((rawAlbedo.rgb + gray) * 0.6);
+		const float mask = 1.0 - abs(normalWV.z);
+		rawAlbedo.rgb = lerp(rawAlbedo.rgb, newColor, mask * mask * (1.0 - alpha_spec.y));
+	}
 	// </RimAlbedo>
 
+	{
 		DS_Reset(so);
+
 		DS_SetAlbedo(so, lerp(rawAlbedo.rgb, hairAlbedo, hairAlpha));
-	DS_SetSpec(so, normalAA.w * max(eyeAlpha, max(strass,
+		DS_SetSpec(so, max(eyeAlpha, max(strass,
 			alpha_spec.y * (1.0 + hairAlpha * 0.75) * g_ubPerMaterialFS._ssb_as.x + g_ubPerMaterialFS._ssb_as.y)));
-	DS_SetDepth(so, si.depth.x / si.depth.y);
+
 		DS_SetNormal(so, normalWV + NormalDither(rand));
 		DS_SetEmission(so, max(emitAlpha, emitXAlpha) * alpha_spec.y, skinAlpha);
+
 		DS_SetLamScaleBias(so, lamScaleBias, float4(anisoWV, hairAlpha));
 		DS_SetMetallicity(so, metalAlpha, hairAlpha);
-	DS_SetGloss(so, normalAA.w * gloss);
+		DS_SetGloss(so, gloss * lerp(toksvigFactor, 1.0, eyeAlpha));
+	}
+
+	clip(alpha_spec.x - 0.5);
 #endif
 
 	return so;
@@ -262,8 +345,19 @@ DS_FSO mainFS(VSO si)
 #endif
 
 //@main:#
-//@main:#DepthRobotic DEPTH ROBOTIC (V)
-//@main:#DepthSkinned DEPTH SKINNED (V)
-//@main:#Instanced    INSTANCED
 //@main:#Robotic           ROBOTIC
 //@main:#Skinned           SKINNED
+
+//@main:#Instanced         INSTANCED
+
+//@main:#Depth             DEPTH (V)
+//@main:#DepthRobotic      DEPTH ROBOTIC (V)
+//@main:#DepthSkinned      DEPTH SKINNED (V)
+
+//@main:#SolidColor        SOLID_COLOR
+//@main:#SolidColorRobotic SOLID_COLOR ROBOTIC
+//@main:#SolidColorSkinned SOLID_COLOR SKINNED
+
+//@main:#Tess              TESS (VHDF)
+//@main:#TessRobotic       TESS ROBOTIC (VHDF)
+//@main:#TessSkinned       TESS SKINNED (VHDF)

Verus/src/Shaders/DS_Mesh.inc.hlsl

@@ -3,8 +3,9 @@
 VERUS_UBUFFER UB_PerFrame
 {
 	mataff _matV;
-	matrix _matP;
 	matrix _matVP;
+	matrix _matP;
+	float4 _viewportSize;
 };
 
 VERUS_UBUFFER UB_PerMaterialFS

Verus/src/Shaders/DS_Terrain.hlsl

@@ -3,15 +3,20 @@
 #include "Lib.hlsl"
 #include "LibColor.hlsl"
 #include "LibDeferredShading.hlsl"
+#include "LibSurface.hlsl"
+#include "LibTessellation.hlsl"
 #include "DS_Terrain.inc.hlsl"
 
-#define HEIGHT_SCALE 0.01
+#define DETAIL_TC_SCALE 8.0
 
-ConstantBuffer<UB_DrawDepth>     g_ubDrawDepth     : register(b0, space0);
-ConstantBuffer<UB_PerMaterialFS> g_ubPerMaterialFS : register(b0, space1);
+ConstantBuffer<UB_TerrainVS> g_ubTerrainVS : register(b0, space0);
+ConstantBuffer<UB_TerrainFS> g_ubTerrainFS : register(b0, space1);
+
+Texture2D      g_texHeightVS  : register(t1, space0);
+SamplerState   g_samHeightVS  : register(s1, space0);
+Texture2D      g_texNormalVS  : register(t2, space0);
+SamplerState   g_samNormalVS  : register(s2, space0);
 
-Texture2D      g_texHeight    : register(t1, space0);
-SamplerState   g_samHeight    : register(s1, space0);
 Texture2D      g_texNormal    : register(t1, space1);
 SamplerState   g_samNormal    : register(s1, space1);
 Texture2D      g_texBlend     : register(t2, space1);
@@ -20,60 +25,145 @@ Texture2DArray g_texLayers    : register(t3, space1);
 SamplerState   g_samLayers    : register(s3, space1);
 Texture2DArray g_texLayersNM  : register(t4, space1);
 SamplerState   g_samLayersNM  : register(s4, space1);
+Texture2D      g_texDetail    : register(t5, space1);
+SamplerState   g_samDetail    : register(s5, space1);
 
 struct VSI
 {
 	VK_LOCATION_POSITION int4 pos             : POSITION;
-
-	VK_LOCATION(16)      int4 posPatch        : TEXCOORD8;
-	VK_LOCATION(17)      int4 layerForChannel : TEXCOORD9;
+	VK_LOCATION(16)      int4 posPatch        : INSTDATA0;
+	VK_LOCATION(17)      int4 layerForChannel : INSTDATA1;
 };
 
 struct VSO
 {
 	float4 pos             : SV_Position;
-	float4 layerForChannel : TEXCOORD0;
-#ifndef DEF_DEPTH
-	float3 tcBlend         : TEXCOORD1;
-	float4 tcLayer_tcMap   : TEXCOORD2;
+	float3 nrmWV           : TEXCOORD0;
+#if !defined(DEF_DEPTH)
+	float4 layerForChannel : TEXCOORD1;
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
+	float3 tcBlend         : TEXCOORD2;
+	float4 tcLayer_tcMap   : TEXCOORD3;
+#endif
 #endif
-	float2 depth           : TEXCOORD3;
 };
 
+static const float g_bestPrecision = 50.0;
+static const float g_layerScale = 1.0 / 8.0;
+
 #ifdef _VS
 VSO mainVS(VSI si)
 {
 	VSO so;
 
-	const float3 posEye = g_ubDrawDepth._posEye_mapSideInv.xyz;
-	const float mapSideInv = g_ubDrawDepth._posEye_mapSideInv.w;
+	const float3 eyePos = g_ubTerrainVS._eyePos_mapSideInv.xyz;
+	const float mapSideInv = g_ubTerrainVS._eyePos_mapSideInv.w;
 
 	const float2 edgeCorrection = si.pos.yw;
 	si.pos.yw = 0.0;
-	float3 pos = si.pos.xyz + si.posPatch.xyz * float3(1, HEIGHT_SCALE, 1);
+	float3 pos = si.pos.xyz + si.posPatch.xyz;
+	const float2 tcMap = pos.xz * mapSideInv + 0.5; // Range [0, 1).
 	const float2 posBlend = pos.xz + edgeCorrection * 0.5;
 
-	const float bestPrecision = 50.0;
-	const float2 tcMap = pos.xz * mapSideInv + 0.5; // Range [0, 1).
-	const float distToEye = distance(pos, posEye);
+	// <HeightAndNormal>
+	float3 intactNrm;
+	{
+		const float approxHeight = g_texHeightVS.SampleLevel(g_samHeightVS, tcMap + (0.5 * mapSideInv) * 16.0, 4.0).r + g_bestPrecision;
+		pos.y = approxHeight;
+
+		const float distToEye = distance(pos, eyePos);
 		const float geomipsLod = log2(clamp(distToEye * (2.0 / 100.0), 1.0, 18.0));
 		const float geomipsLodFrac = frac(geomipsLod);
 		const float geomipsLodBase = floor(geomipsLod);
 		const float geomipsLodNext = geomipsLodBase + 1.0;
-	const float2 halfTexelAB = (0.5 * mapSideInv) * exp2(float2(geomipsLodBase, geomipsLodNext));
-	const float yA = g_texHeight.SampleLevel(g_samHeight, tcMap + halfTexelAB.xx, geomipsLodBase).r + bestPrecision;
-	const float yB = g_texHeight.SampleLevel(g_samHeight, tcMap + halfTexelAB.yy, geomipsLodNext).r + bestPrecision;
+		const float2 texelCenterAB = (0.5 * mapSideInv) * exp2(float2(geomipsLodBase, geomipsLodNext));
+		const float yA = g_texHeightVS.SampleLevel(g_samHeightVS, tcMap + texelCenterAB.xx, geomipsLodBase).r + g_bestPrecision;
+		const float yB = g_texHeightVS.SampleLevel(g_samHeightVS, tcMap + texelCenterAB.yy, geomipsLodNext).r + g_bestPrecision;
 		pos.y = lerp(yA, yB, geomipsLodFrac);
 
-	so.pos = mul(float4(pos, 1), g_ubDrawDepth._matVP);
+		const float4 rawNormal = g_texNormalVS.SampleLevel(g_samNormalVS, tcMap + texelCenterAB.xx, geomipsLodBase);
+		intactNrm = float3(rawNormal.x, 0, rawNormal.y) * 2.0 - 1.0;
+		intactNrm.y = ComputeNormalZ(intactNrm.xz);
+	}
+	// </HeightAndNormal>
+
+	so.pos = MulTessPos(float4(pos, 1), g_ubTerrainVS._matV, g_ubTerrainVS._matVP);
+	so.nrmWV = mul(intactNrm, (float3x3)g_ubTerrainVS._matWV);
+#if !defined(DEF_DEPTH)
 	so.layerForChannel = si.layerForChannel;
-#ifndef DEF_DEPTH
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
 	so.tcBlend.xy = posBlend * mapSideInv + 0.5;
-	so.tcBlend.z = pos.y + 0.75;
-	so.tcLayer_tcMap.xy = pos.xz * (1.0 / 8.0);
+	so.tcBlend.z = pos.y;
+	so.tcLayer_tcMap.xy = pos.xz * g_layerScale;
 	so.tcLayer_tcMap.zw = (pos.xz + 0.5) * mapSideInv + 0.5; // Texel's center.
 #endif
-	so.depth = so.pos.zw;
+#endif
+
+	return so;
+}
+#endif
+
+_HSO_STRUCT;
+
+#ifdef _HS
+PCFO PatchConstFunc(const OutputPatch<HSO, 3> outputPatch)
+{
+	PCFO so;
+	_HS_PCF_BODY(g_ubTerrainVS._matP);
+	return so;
+}
+
+[domain("tri")]
+//[maxtessfactor(7.0)]
+[outputcontrolpoints(3)]
+[outputtopology("triangle_cw")]
+[partitioning(_PARTITION_METHOD)]
+[patchconstantfunc("PatchConstFunc")]
+HSO mainHS(InputPatch<VSO, 3> inputPatch, uint id : SV_OutputControlPointID)
+{
+	HSO so;
+
+	_HS_PN_BODY(nrmWV, g_ubTerrainVS._matP, g_ubTerrainVS._viewportSize);
+
+	_HS_COPY(pos);
+	_HS_COPY(nrmWV);
+#if !defined(DEF_DEPTH)
+	_HS_COPY(layerForChannel);
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
+	_HS_COPY(tcBlend);
+	_HS_COPY(tcLayer_tcMap);
+#endif
+#endif
+
+	return so;
+}
+#endif
+
+#ifdef _DS
+[domain("tri")]
+VSO mainDS(_IN_DS)
+{
+	VSO so;
+
+	_DS_INIT_FLAT_POS;
+	_DS_INIT_SMOOTH_POS;
+
+	so.pos = ApplyProjection(smoothPosWV, g_ubTerrainVS._matP);
+	_DS_COPY(nrmWV);
+#if !defined(DEF_DEPTH)
+	_DS_COPY(layerForChannel);
+#if !defined(DEF_DEPTH) && !defined(DEF_SOLID_COLOR)
+	_DS_COPY(tcBlend);
+	_DS_COPY(tcLayer_tcMap);
+#endif
+#endif
+
+#ifndef DEF_DEPTH
+	// Fade to non-tess mesh at 80 meters. LOD 1 starts at 100 meters.
+	const float tessStrength = saturate((1.0 - saturate(smoothPosWV.z / -80.0)) * 4.0);
+	const float3 posWV = lerp(flatPosWV, smoothPosWV, tessStrength);
+	so.pos = ApplyProjection(posWV, g_ubTerrainVS._matP);
+#endif
 
 	return so;
 }
@@ -84,15 +174,15 @@ DS_FSO mainFS(VSO si)
 {
 	DS_FSO so;
 
+#ifdef DEF_SOLID_COLOR
+	DS_SolidColor(so, si.layerForChannel.rgb);
+#else
 	const float3 rand = Rand(si.pos.xy);
 
+	const float4 layerForChannel = round(si.layerForChannel);
 	const float2 tcLayer = si.tcLayer_tcMap.xy;
 	const float2 tcMap = si.tcLayer_tcMap.zw;
 
-	const float gloss = 4.0;
-
-	const float4 layerForChannel = round(si.layerForChannel);
-
 	const float4 rawBlend = g_texBlend.Sample(g_samBlend, si.tcBlend.xy);
 	float4 weights = float4(rawBlend.rgb, 1.0 - dot(rawBlend.rgb, float3(1, 1, 1)));
 
@@ -108,9 +198,9 @@ DS_FSO mainFS(VSO si)
 		basisBin = cross(basisTan, basisNrm);
 	}
 	_TBN_SPACE(
-		mul(basisTan, (float3x3)g_ubDrawDepth._matWV),
-		mul(basisBin, (float3x3)g_ubDrawDepth._matWV),
-		mul(basisNrm, (float3x3)g_ubDrawDepth._matWV));
+		mul(basisTan, (float3x3)g_ubTerrainFS._matWV),
+		mul(basisBin, (float3x3)g_ubTerrainFS._matWV),
+		mul(basisNrm, (float3x3)g_ubTerrainFS._matWV));
 	// </Basis>
 
 	// <Albedo>
@@ -140,8 +230,8 @@ DS_FSO mainFS(VSO si)
 		accAlbedo += rawAlbedos[3] * weights.a;
 		albedo = accAlbedo;
 
-		static float vSpecStrength[_MAX_TERRAIN_LAYERS] = (float[_MAX_TERRAIN_LAYERS])g_ubPerMaterialFS._vSpecStrength;
-		static float vDetailStrength[_MAX_TERRAIN_LAYERS] = (float[_MAX_TERRAIN_LAYERS])g_ubPerMaterialFS._vDetailStrength;
+		static float vSpecStrength[_MAX_TERRAIN_LAYERS] = (float[_MAX_TERRAIN_LAYERS])g_ubTerrainFS._vSpecStrength;
+		static float vDetailStrength[_MAX_TERRAIN_LAYERS] = (float[_MAX_TERRAIN_LAYERS])g_ubTerrainFS._vDetailStrength;
 		const float4 specStrengthForChannel = float4(
 			vSpecStrength[layerForChannel.r],
 			vSpecStrength[layerForChannel.g],
@@ -157,26 +247,58 @@ DS_FSO mainFS(VSO si)
 	}
 	// </Albedo>
 
+	// <Water>
+	float waterGlossBoost = 0.0;
+	{
+		const float dryMask = saturate(si.tcBlend.z);
+		const float dryMask3 = dryMask * dryMask * dryMask;
+		const float wetMask = 1.0 - dryMask;
+		const float wetMask3 = wetMask * wetMask * wetMask;
+		albedo.rgb *= dryMask3 * 0.5 + 0.5;
+		specStrength = dryMask * saturate(specStrength + wetMask3 * wetMask3);
+		waterGlossBoost = min(32.0, dryMask * wetMask3 * 128.0);
+	}
+	// </Water>
+
+	const float gloss = lerp(3.3, 15.0, specStrength) + waterGlossBoost;
+
 	// <Normal>
 	float3 normalWV;
-	float4 normalAA;
+	float toksvigFactor;
 	{
 		float4 accNormal = 0.0;
 		accNormal += g_texLayersNM.Sample(g_samLayersNM, float3(tcLayer, layerForChannel.r)) * weights.r;
 		accNormal += g_texLayersNM.Sample(g_samLayersNM, float3(tcLayer, layerForChannel.g)) * weights.g;
 		accNormal += g_texLayersNM.Sample(g_samLayersNM, float3(tcLayer, layerForChannel.b)) * weights.b;
 		accNormal += g_texLayersNM.Sample(g_samLayersNM, float3(tcLayer, layerForChannel.a)) * weights.a;
-		normalAA = NormalMapAA(accNormal);
+		accNormal = lerp(accNormal, float4(0, 0.5, 0.5, 0.5), 0.5);
+		const float4 normalAA = NormalMapAA(accNormal);
 		normalWV = normalize(mul(normalAA.xyz, matFromTBN));
+		toksvigFactor = ComputeToksvigFactor(normalAA.a, gloss);
 	}
 	// </Normal>
 
-	DS_Test(so, 0.0, 0.5, 16.0);
+	// <Detail>
+	{
+		const float3 rawDetail = g_texDetail.Sample(g_samDetail, tcLayer * DETAIL_TC_SCALE).rgb;
+		albedo.rgb = albedo.rgb * lerp(0.5, rawDetail, detailStrength) * 2.0;
+	}
+	// </Detail>
+
+	{
+		DS_Reset(so);
+
 		DS_SetAlbedo(so, albedo.rgb);
-	DS_SetDepth(so, si.depth.x / si.depth.y);
+		DS_SetSpec(so, specStrength);
+
 		DS_SetNormal(so, normalWV + NormalDither(rand));
-	DS_SetSpec(so, normalAA.w * specStrength);
-	DS_SetGloss(so, normalAA.w * gloss);
+		DS_SetEmission(so, 0.0, 0.0);
+
+		DS_SetLamScaleBias(so, g_ubTerrainFS._lamScaleBias.xy, float4(0, 0, 1, 0));
+		DS_SetMetallicity(so, 0.05, 0.0);
+		DS_SetGloss(so, gloss * toksvigFactor);
+	}
+#endif
 
 	return so;
 }
@@ -184,3 +306,5 @@ DS_FSO mainFS(VSO si)
 
 //@main:#
 //@main:#Depth DEPTH (V)
+//@main:#SolidColor SOLID_COLOR
+//@main:#Tess TESS (VHDF)

Verus/src/Shaders/DS_Terrain.inc.hlsl

@@ -1,15 +1,20 @@
 // Copyright (C) 2020, Dmitry Maluev (dmaluev@gmail.com)
 
-VERUS_UBUFFER UB_DrawDepth
+VERUS_UBUFFER UB_TerrainVS
 {
 	mataff _matW;
-	mataff _matWV;
+	mataff _matWV; // For normal.
+	mataff _matV; // For tess.
 	matrix _matVP;
-	float4 _posEye_mapSideInv;
+	matrix _matP; // For tess.
+	float4 _eyePos_mapSideInv;
+	float4 _viewportSize;
 };
 
-VERUS_UBUFFER UB_PerMaterialFS
+VERUS_UBUFFER UB_TerrainFS
 {
+	mataff _matWV; // For basis.
 	float4 _vSpecStrength[8];
 	float4 _vDetailStrength[8];
+	float4 _lamScaleBias;
 };

Verus/src/Shaders/GenerateMips.hlsl

@@ -16,8 +16,8 @@ RWTexture2D<float4> g_uavOutMip4 : register(u5, space0);
 
 struct CSI
 {
-	uint3 _DispatchThreadID : SV_DispatchThreadID;
-	uint  _GroupIndex       : SV_GroupIndex;
+	uint3 dispatchThreadID : SV_DispatchThreadID;
+	uint  groupIndex       : SV_GroupIndex;
 };
 
 #define THREAD_GROUP_SIZE 8
@@ -67,14 +67,14 @@ void mainCS(CSI si)
 	{
 	case DIM_CASE_WE_HE:
 	{
-		const float2 tc = g_ub._texelSize * (si._DispatchThreadID.xy + 0.5);
+		const float2 tc = g_ub._texelSize * (si.dispatchThreadID.xy + 0.5);
 
 		srcColor1 = g_texSrcMip.SampleLevel(g_samSrcMip, tc, g_ub._srcMipLevel);
 	}
 	break;
 	case DIM_CASE_WO_HE:
 	{
-		const float2 tc = g_ub._texelSize * (si._DispatchThreadID.xy + float2(0.25, 0.5));
+		const float2 tc = g_ub._texelSize * (si.dispatchThreadID.xy + float2(0.25, 0.5));
 		const float2 offset = g_ub._texelSize * float2(0.5, 0.0);
 
 		srcColor1 = lerp(
@@ -85,7 +85,7 @@ void mainCS(CSI si)
 	break;
 	case DIM_CASE_WE_HO:
 	{
-		const float2 tc = g_ub._texelSize * (si._DispatchThreadID.xy + float2(0.5, 0.25));
+		const float2 tc = g_ub._texelSize * (si.dispatchThreadID.xy + float2(0.5, 0.25));
 		const float2 offset = g_ub._texelSize * float2(0.0, 0.5);
 
 		srcColor1 = lerp(
@@ -96,7 +96,7 @@ void mainCS(CSI si)
 	break;
 	case DIM_CASE_WO_HO:
 	{
-		const float2 tc = g_ub._texelSize * (si._DispatchThreadID.xy + float2(0.25, 0.25));
+		const float2 tc = g_ub._texelSize * (si.dispatchThreadID.xy + float2(0.25, 0.25));
 		const float2 offset = g_ub._texelSize * 0.5;
 
 		srcColor1 = lerp(
@@ -112,24 +112,24 @@ void mainCS(CSI si)
 	break;
 	}
 
-	g_uavOutMip1[si._DispatchThreadID.xy] = PackColor(srcColor1);
+	g_uavOutMip1[si.dispatchThreadID.xy] = PackColor(srcColor1);
 
 	if (1 == g_ub._mipLevelCount)
 		return;
 
-	StoreColor(si._GroupIndex, srcColor1);
+	StoreColor(si.groupIndex, srcColor1);
 
 	GroupMemoryBarrierWithGroupSync();
 
-	if ((si._GroupIndex & 0x9) == 0) // 16 threads:
+	if ((si.groupIndex & 0x9) == 0) // 16 threads:
 	{
-		const float4 srcColor2 = LoadColor(si._GroupIndex + 0x01); // {+0, +1}
-		const float4 srcColor3 = LoadColor(si._GroupIndex + 0x08); // {+1, +0}
-		const float4 srcColor4 = LoadColor(si._GroupIndex + 0x09); // {+1, +1}
+		const float4 srcColor2 = LoadColor(si.groupIndex + 0x01); // {+0, +1}
+		const float4 srcColor3 = LoadColor(si.groupIndex + 0x08); // {+1, +0}
+		const float4 srcColor4 = LoadColor(si.groupIndex + 0x09); // {+1, +1}
 		srcColor1 = 0.25 * (srcColor1 + srcColor2 + srcColor3 + srcColor4);
 
-		g_uavOutMip2[si._DispatchThreadID.xy >> 1] = PackColor(srcColor1);
-		StoreColor(si._GroupIndex, srcColor1);
+		g_uavOutMip2[si.dispatchThreadID.xy >> 1] = PackColor(srcColor1);
+		StoreColor(si.groupIndex, srcColor1);
 	}
 
 	if (2 == g_ub._mipLevelCount)
@@ -137,15 +137,15 @@ void mainCS(CSI si)
 
 	GroupMemoryBarrierWithGroupSync();
 
-	if ((si._GroupIndex & 0x1B) == 0) // 4 threads:
+	if ((si.groupIndex & 0x1B) == 0) // 4 threads:
 	{
-		const float4 srcColor2 = LoadColor(si._GroupIndex + 0x02); // {+0, +2}
-		const float4 srcColor3 = LoadColor(si._GroupIndex + 0x10); // {+2, +0}
-		const float4 srcColor4 = LoadColor(si._GroupIndex + 0x12); // {+2, +2}
+		const float4 srcColor2 = LoadColor(si.groupIndex + 0x02); // {+0, +2}
+		const float4 srcColor3 = LoadColor(si.groupIndex + 0x10); // {+2, +0}
+		const float4 srcColor4 = LoadColor(si.groupIndex + 0x12); // {+2, +2}
 		srcColor1 = 0.25 * (srcColor1 + srcColor2 + srcColor3 + srcColor4);
 
-		g_uavOutMip3[si._DispatchThreadID.xy >> 2] = PackColor(srcColor1);
-		StoreColor(si._GroupIndex, srcColor1);
+		g_uavOutMip3[si.dispatchThreadID.xy >> 2] = PackColor(srcColor1);
+		StoreColor(si.groupIndex, srcColor1);
 	}
 
 	if (3 == g_ub._mipLevelCount)
@@ -153,14 +153,14 @@ void mainCS(CSI si)
 
 	GroupMemoryBarrierWithGroupSync();
 
-	if (si._GroupIndex == 0) // 1 thread:
+	if (si.groupIndex == 0) // 1 thread:
 	{
-		const float4 srcColor2 = LoadColor(si._GroupIndex + 0x04); // {+0, +4}
-		const float4 srcColor3 = LoadColor(si._GroupIndex + 0x20); // {+4, +0}
-		const float4 srcColor4 = LoadColor(si._GroupIndex + 0x24); // {+4, +4}
+		const float4 srcColor2 = LoadColor(si.groupIndex + 0x04); // {+0, +4}
+		const float4 srcColor3 = LoadColor(si.groupIndex + 0x20); // {+4, +0}
+		const float4 srcColor4 = LoadColor(si.groupIndex + 0x24); // {+4, +4}
 		srcColor1 = 0.25 * (srcColor1 + srcColor2 + srcColor3 + srcColor4);
 
-		g_uavOutMip4[si._DispatchThreadID.xy >> 3] = PackColor(srcColor1);
+		g_uavOutMip4[si.dispatchThreadID.xy >> 3] = PackColor(srcColor1);
 	}
 }
 #endif

Verus/src/Shaders/Lib.hlsl

@@ -6,12 +6,14 @@
 #ifdef _VULKAN
 #	define VK_LOCATION(x)           [[vk::location(x)]]
 #	define VK_LOCATION_POSITION     [[vk::location(0)]]
-#	define VK_LOCATION_BLENDWEIGHT  [[vk::location(1)]]
+#	define VK_LOCATION_BLENDWEIGHTS [[vk::location(1)]]
 #	define VK_LOCATION_BLENDINDICES [[vk::location(6)]]
 #	define VK_LOCATION_NORMAL       [[vk::location(2)]]
-#	define VK_LOCATION_PSIZE        [[vk::location(7)]]
+#	define VK_LOCATION_TANGENT      [[vk::location(14)]]
+#	define VK_LOCATION_BINORMAL     [[vk::location(15)]]
 #	define VK_LOCATION_COLOR0       [[vk::location(3)]]
 #	define VK_LOCATION_COLOR1       [[vk::location(4)]]
+#	define VK_LOCATION_PSIZE        [[vk::location(7)]]
 
 #	define VK_PUSH_CONSTANT [[vk::push_constant]]
 #	define VK_SUBPASS_INPUT(index, tex, sam, t, s, space) layout(input_attachment_index = index) SubpassInput<float4> tex : register(t, space)
@@ -26,12 +28,14 @@
 #else
 #	define VK_LOCATION(x)
 #	define VK_LOCATION_POSITION
-#	define VK_LOCATION_BLENDWEIGHT
+#	define VK_LOCATION_BLENDWEIGHTS
 #	define VK_LOCATION_BLENDINDICES
 #	define VK_LOCATION_NORMAL
-#	define VK_LOCATION_PSIZE
+#	define VK_LOCATION_TANGENT
+#	define VK_LOCATION_BINORMAL
 #	define VK_LOCATION_COLOR0
 #	define VK_LOCATION_COLOR1
+#	define VK_LOCATION_PSIZE
 
 #	define VK_PUSH_CONSTANT
 #	define VK_SUBPASS_INPUT(index, tex, sam, t, s, space)\
@@ -45,10 +49,10 @@
 
 #ifdef DEF_INSTANCED
 #	define _PER_INSTANCE_DATA\
-	VK_LOCATION(16) float4 matPart0 : TEXCOORD8;\
-	VK_LOCATION(17) float4 matPart1 : TEXCOORD9;\
-	VK_LOCATION(18) float4 matPart2 : TEXCOORD10;\
-	VK_LOCATION(19) float4 instData : TEXCOORD11;
+	VK_LOCATION(16) float4 matPart0 : INSTDATA0;\
+	VK_LOCATION(17) float4 matPart1 : INSTDATA1;\
+	VK_LOCATION(18) float4 matPart2 : INSTDATA2;\
+	VK_LOCATION(19) float4 instData : INSTDATA3;
 #else
 #	define _PER_INSTANCE_DATA
 #endif
@@ -57,6 +61,8 @@
 	const float3x3 matFromTBN = float3x3(tan, bin, nrm);\
 	const float3x3 matToTBN   = transpose(matFromTBN);
 
+#define _PI 3.141592654
+
 #define _SINGULARITY_FIX 0.001
 
 #define _MAX_TERRAIN_LAYERS 32
@@ -70,18 +76,6 @@ matrix ToFloat4x4(mataff m)
 		float4(m[3], 1));
 }
 
-float ComputeNormalZ(float2 v)
-{
-	return sqrt(saturate(1.0 - dot(v.rg, v.rg)));
-}
-
-float4 NormalMapAA(float4 rawNormal)
-{
-	float3 normal = rawNormal.agb * -2.0 + 1.0; // Dmitry's reverse!
-	normal.b = ComputeNormalZ(normal.rg);
-	return float4(normal, 0.8 + rawNormal.b * 0.8);
-}
-
 float3 Rand(float2 uv)
 {
 	return frac(sin(dot(uv, float2(12.9898, 78.233)) * float3(1, 2, 3)) * 43758.5453);
@@ -92,3 +86,19 @@ float3 NormalDither(float3 rand)
 	const float2 rr = rand.xy * (1.0 / 333.0) - (0.5 / 333.0);
 	return float3(rr, 0);
 }
+
+static const float2 _POINT_SPRITE_POS_OFFSETS[4] =
+{
+	float2(-0.5,  0.5),
+	float2(-0.5, -0.5),
+	float2(0.5,  0.5),
+	float2(0.5, -0.5)
+};
+
+static const float2 _POINT_SPRITE_TEX_COORDS[4] =
+{
+	float2(0, 0),
+	float2(0, 1),
+	float2(1, 0),
+	float2(1, 1)
+};

Verus/src/Shaders/LibColor.hlsl

@@ -1,37 +1,5 @@
 // Copyright (C) 2020, Dmitry Maluev (dmaluev@gmail.com)
 
-// See: http://chilliant.blogspot.com/2012/08/srgb-approximations-for-hlsl.html
-float4 ColorToLinear(float4 x)
-{
-	const float3 rgb = x.rgb < 0.04045 ? x.rgb * (1.0 / 12.92) : pow(abs(x.rgb + 0.055) / 1.055, 2.4);
-	return float4(rgb, x.a);
-}
-float4 ColorToSRGB(float4 x)
-{
-	const float3 rgb = x.rgb < 0.0031308 ? 12.92 * x.rgb : 1.055 * pow(abs(x.rgb), 1.0 / 2.4) - 0.055;
-	return float4(rgb, x.a);
-}
-
-float4 HDRColorToLinear(float4 x)
-{
-	return float4(x.rgb * x.rgb * 12.0, x.a);
-}
-float4 HDRColorToSRGB(float4 x)
-{
-	return float4(sqrt(saturate(x.rgb * (1.0 / 12.0))), x.a);
-}
-
-// See: https://knarkowicz.wordpress.com/2016/01/06/aces-filmic-tone-mapping-curve/
-float3 ToneMappingACES(float3 x)
-{
-	const float a = 2.51f;
-	const float b = 0.03f;
-	const float c = 2.43f;
-	const float d = 0.59f;
-	const float e = 0.14f;
-	return saturate((x * (a * x + b)) / (x * (c * x + d) + e));
-}
-
 // See: https://en.wikipedia.org/wiki/Grayscale
 float Grayscale(float3 color)
 {
@@ -44,12 +12,6 @@ float3 Desaturate(float3 color, float alpha, float lum = 1.0)
 	return lerp(color, gray, alpha);
 }
 
-float3 DesaturateFilmic(float3 color)
-{
-	const float gray = Grayscale(color);
-	return lerp(color, gray, saturate(gray * gray * gray));
-}
-
 // See: http://www.chilliant.com/rgb2hsv.html
 float3 ConvertRGBtoHCV(float3 color)
 {
@@ -96,3 +58,41 @@ float2 AlphaSwitch(float4 albedo, float2 tc, float2 alphaSwitch)
 	else
 		return saturate(float2(albedo.a * 8.0, (albedo.a - 0.15) / 0.85)); // 5-bit alpha.
 }
+
+// See: http://chilliant.blogspot.com/2012/08/srgb-approximations-for-hlsl.html
+float4 ColorToLinear(float4 x)
+{
+	const float3 rgb = x.rgb < 0.04045 ? x.rgb * (1.0 / 12.92) : pow(abs(x.rgb + 0.055) / 1.055, 2.4);
+	return float4(rgb, x.a);
+}
+float4 ColorToSRGB(float4 x)
+{
+	const float3 rgb = x.rgb < 0.0031308 ? 12.92 * x.rgb : 1.055 * pow(abs(x.rgb), 1.0 / 2.4) - 0.055;
+	return float4(rgb, x.a);
+}
+
+float4 HDRColorToLinear(float4 x)
+{
+	return float4(x.rgb * x.rgb * 12.0, x.a);
+}
+float4 HDRColorToSRGB(float4 x)
+{
+	return float4(sqrt(saturate(x.rgb * (1.0 / 12.0))), x.a);
+}
+
+// See: https://knarkowicz.wordpress.com/2016/01/06/aces-filmic-tone-mapping-curve/
+float3 ToneMappingACES(float3 x)
+{
+	const float a = 2.51f;
+	const float b = 0.03f;
+	const float c = 2.43f;
+	const float d = 0.59f;
+	const float e = 0.14f;
+	return saturate((x * (a * x + b)) / (x * (c * x + d) + e));
+}
+
+float3 VerusToneMapping(float3 hdr, float filmicLook = 1.0)
+{
+	const float3 ldr = lerp(1.0 - exp(-hdr), ToneMappingACES(hdr), filmicLook);
+	return saturate(ldr);
+}

Verus/src/Shaders/LibDeferredShading.hlsl

@@ -2,10 +2,9 @@
 
 struct DS_FSO
 {
-	float4 target0 : SV_Target0; // {color, spec}
-	float4 target1 : SV_Target1; // {depth}
-	float4 target2 : SV_Target2; // {normal, emission, motion}
-	float4 target3 : SV_Target3; // {lam, metal, gloss}
+	float4 target0 : SV_Target0; // {albedo, spec}
+	float4 target1 : SV_Target1; // {normal, emission, motion}
+	float4 target2 : SV_Target2; // {lam, metal, gloss}
 };
 
 struct DS_ACC_FSO
@@ -35,15 +34,20 @@ void DS_Reset(out DS_FSO so)
 	so.target0 = 0.0;
 	so.target1 = 0.0;
 	so.target2 = 0.0;
-	so.target3 = 0.0;
+}
+
+void DS_SolidColor(out DS_FSO so, float3 color)
+{
+	so.target0 = float4(color, 0.0);
+	so.target1 = float4(EncodeNormal(float3(0, 0, 1)), 1.0, 0.0);
+	so.target2 = float4(1.0, 0.0, 0.0, 0.0);
 }
 
 void DS_Test(out DS_FSO so, float3 normal, float spec, float gloss)
 {
 	so.target0 = float4(0.5, 0.5, 0.5, spec);
-	so.target1 = 0.0;
-	so.target2 = float4(EncodeNormal(normal), 0.25, 0.5);
-	so.target3 = float4(1.0 / 8.0, 0.5, 0.5, gloss * (1.0 / 64.0));
+	so.target1 = float4(EncodeNormal(normal), 0.25, 0.5);
+	so.target2 = float4(1.0 / 8.0, 0.5, 0.5, gloss * (1.0 / 64.0));
 }
 
 void DS_SetAlbedo(inout DS_FSO so, float3 albedo)
@@ -56,11 +60,6 @@ void DS_SetSpec(inout DS_FSO so, float spec)
 	so.target0.a = spec;
 }
 
-void DS_SetDepth(inout DS_FSO so, float depth)
-{
-	so.target1 = depth;
-}
-
 float3 DS_GetNormal(float4 gbuffer)
 {
 	return DecodeNormal(gbuffer.rg);
@@ -68,7 +67,7 @@ float3 DS_GetNormal(float4 gbuffer)
 
 void DS_SetNormal(inout DS_FSO so, float3 normal)
 {
-	so.target2.rg = EncodeNormal(normal);
+	so.target1.rg = EncodeNormal(normal);
 }
 
 float2 DS_GetEmission(float4 gbuffer)
@@ -80,7 +79,7 @@ float2 DS_GetEmission(float4 gbuffer)
 void DS_SetEmission(inout DS_FSO so, float emission, float skin)
 {
 	const float em = HDRColorToSRGB(emission).r;
-	so.target2.b = (3.0 * em - skin) * 0.25 + 0.25;
+	so.target1.b = (3.0 * em - skin) * 0.25 + 0.25;
 }
 
 float2 DS_GetLamScaleBias(float4 gbuffer)
@@ -90,30 +89,30 @@ float2 DS_GetLamScaleBias(float4 gbuffer)
 
 void DS_SetLamScaleBias(inout DS_FSO so, float2 lamScaleBias, float4 aniso)
 {
-	so.target3.rg = lerp(lamScaleBias * float2(1.0 / 8.0, 0.25) + float2(0, 0.5), EncodeNormal(aniso.xyz), aniso.w);
+	so.target2.rg = lerp(lamScaleBias * float2(1.0 / 8.0, 0.25) + float2(0, 0.5), EncodeNormal(aniso.xyz), aniso.w);
 }
 
 float2 DS_GetMetallicity(float4 gbuffer)
 {
-	const float x2 = gbuffer.b * 2.0;
+	const float x2 = gbuffer.b * 2.02;
 	return float2(
-		saturate(x2 - 1.0),
+		saturate(x2 - 1.02),
 		saturate(1.0 - x2));
 }
 
 void DS_SetMetallicity(inout DS_FSO so, float metal, float hair)
 {
-	so.target3.b = (metal - hair) * 0.5 + 0.5;
+	so.target2.b = (metal - hair) * 0.5 + 0.5;
 }
 
 void DS_SetGloss(inout DS_FSO so, float gloss)
 {
-	so.target3.a = gloss * (1.0 / 64.0);
+	so.target2.a = gloss * (1.0 / 64.0);
 }
 
-float3 DS_GetPosition(float4 gbuffer, matrix matInvProj, float2 posCS)
+float3 DS_GetPosition(float4 gbuffer, matrix matInvProj, float2 ndcPos)
 {
-	float4 pos = mul(float4(posCS, gbuffer.r, 1), matInvProj);
+	float4 pos = mul(float4(ndcPos, gbuffer.r, 1), matInvProj);
 	pos /= pos.w;
 	return pos.xyz;
 }

Verus/src/Shaders/LibDepth.hlsl

@@ -13,9 +13,10 @@ float4 ToLinearDepth(float4 d, float4 zNearFarEx)
 	return zNearFarEx.w / (d - zNearFarEx.z);
 }
 
-float ComputeFog(float depth, float density)
+float ComputeFog(float depth, float density, float height = 0.0)
 {
-	const float fog = 1.0 / exp(depth * density);
+	const float strength = 1.0 - saturate(height * 0.003);
+	const float fog = 1.0 / exp(depth * lerp(0.0, density, strength * strength));
 	return 1.0 - saturate(fog);
 }
 
@@ -76,9 +77,10 @@ float PCF(
 	if (dzBlockers_blockerCount.y > 0.0)
 	{
 		const float penumbraScale = config.x;
+		const float penumbraContrastMask = config.y;
 		const float dzBlockers = dzBlockers_blockerCount.x / dzBlockers_blockerCount.y;
 		const float penumbra = saturate(dzFrag - dzBlockers);
-		const float contrast = 1.0 + max(0.0, 3.0 - penumbra * penumbraScale);
+		const float contrast = 1.0 + max(0.0, 3.0 - penumbra * penumbraScale) * penumbraContrastMask;
 		return saturate((ret - 0.5) * contrast + 0.5);
 	}
 	else
@@ -122,7 +124,7 @@ float ShadowMapCSM(
 	float ret = 1.0;
 	const float4 p = float4(pos.xyz, 1);
 	float contrast = 1.0;
-	const float contrastScale = 2.8;
+	const float contrastScale = config.w;
 
 	if (pos.w > ranges.x)
 	{

Verus/src/Shaders/LibLighting.hlsl

@@ -11,26 +11,26 @@ float FresnelSchlick(float minRef, float maxAdd, float power)
 	return saturate(minRef + maxAdd * power);
 }
 
-// See: http://www.thetenthplanet.de/archives/255
 float EnergyNorm(float gloss)
 {
-	return (gloss + 2.0) / 8.0;
+	return (gloss + 2.0) * (1.0 / 8.0);
 }
 
 float4 VerusLit(float3 dirToLight, float3 normal, float3 dirToEye, float gloss,
-	float2 lamScaleBias, float4 aniso = 0.0)
+	float2 lamScaleBias, float4 aniso = 0.0, float eyeFresnel = 0.0)
 {
 	float4 ret = float4(1, 0, 0, 1);
 	const float3 hv = normalize(dirToEye + dirToLight);
-	const float nDotL = dot(normal, dirToLight);
+	const float nDotL = dot(normal, dirToLight) * lamScaleBias.x + lamScaleBias.y;
+	const float nDotE = dot(normal, dirToEye);
 	const float nDotH = dot(normal, hv);
 	const float aDotH = SinAcos(dot(aniso.xyz, hv));
-	const float3 spec = saturate(float3(EnergyNorm(gloss).xx, 1) *
-		pow(saturate(float3(nDotH, aDotH, 1.0 - nDotL)), float3(gloss, gloss * 8.0, 5)));
+	const float4 spec = float4(EnergyNorm(gloss).xx, 1, 1) *
+		pow(saturate(float4(nDotH, aDotH, 1.0 - nDotL, 1.0 - nDotE)), float4(gloss, gloss * 8.0, 5, 5));
 	ret.x = saturate(nDotL * -4.0); // For subsurface scattering effect.
-	ret.y = saturate(nDotL * lamScaleBias.x + lamScaleBias.y);
+	ret.y = saturate(nDotL);
 	ret.z = saturate(nDotL * 8.0) * max(spec.x, spec.y * aniso.w);
-	ret.w = spec.z; // For fresnel effect.
+	ret.w = lerp(spec.z, spec.w, eyeFresnel); // For fresnel effect.
 	return ret;
 }