2019-07-02 Alex Tutubalin <lexa@lexa.ru>
* Several fixes inspired by OSS-Fuzz
* LibRaw 0.19.3
* Note: ABI slightly changed, new variable added at end of class LibRaw
Generally, all old code should run w/o recompile, but you're warned...
2018-12-24 Alex Tutubalin <lexa@lexa.ru>
* Fixed possible buffer overrun at Fuji makernotes parser
* Fixed possible write to NULL pointer at raw2image/raw2image_ex calls.
Details:
a) Three different CVE numbers was assigned for single problem:
CVE-2018-20363, CVE-2018-20364, CVE-2018-20365
b) The POCs exploits inconsistency in Sinar-4Shot files handling.
LibRaw 0.19 does not support this files format, so it is not
subject of exactly same problem
c) However, additional checks for bayer raw data presence are
backported from LibRaw-master (development) branch.
* LibRaw 0.19.2
2018-11-22 Alex Tutubalin <lexa@lexa.ru>
* Finally: got Sinar 4shot sample, works fine now
* OpenMP critical sections for malloc/free; extra #ifdefs removed; bin/dcraw_dist could be built again using Makefile.devel
* additional checks in parse_phase_one()
* more checks on file offsets/tag len in parse_minolta
* more checks in parse_ciff
* Mempool check reworked
* Old Leaf (16bit/3color/TIFF) support
* Fix cameraWB->autoWB fallback
* Polaroid x530 channel swap; get metadata pointer for Foveon files
* Fixed Secunia Advisory SA86384
- possible infinite loop in unpacked_load_raw()
- possible infinite loop in parse_rollei()
- possible infinite loop in parse_sinar_ia()
Credits: Laurent Delosieres, Secunia Research at Flexera
* LibRaw 0.19.1-Release
2018-06-28 Alex Tutubalin <lexa@lexa.ru>
* changed wrong fix for Canon D30 white balance
* fixed possible stack overrun while reading zero-sized strings
* fixed possible integer overflow
* LibRaw 0.19.0-Release
2018-06-11 Alex Tutubalin <lexa@lexa.ru>
* Sony uncompressed/untiled DNGs: do not set bits-per-sample to 14 bit
* Do not change tiff_bps for DNG files
* Another possible stack overflow in kodak radc reader
* Secunia Advisory SA83507, credits Kasper Leigh Haabb,
Secunia Research at Flexera"
- parse_qt: possible integer overflow
- reject broken/crafted NOKIARAW files
* LibRaw 0.19-Beta6
2018-05-10 Alex Tutubalin <lexa@lexa.ru>
* Put rogue printf's behind #ifdef DCRAW_VERBOSE
* Exceptions was not caught in x3f_new_from_file resulting in x3f handle leak
* packed_load_raw(): EOF check on each row
* define LIBRAW_USE_CALLOC_INSTEAD_OF_MALLOC to use ::calloc instead of
::malloc in LibRaw_mem_mgr malloc calls;
Note: realloc is not changed, so only partial fix
* Fixed possible div by zero in EOS D30 WB data parse
* U-suffix for filter-var manipulation consts
* restored static specifier for utf2char() lost in previous bugfix
* Fixed stack overrun in kodak_radc_load_raw
* Secunia Advisory SA83050: possible infinite loop in parse_minolta()
* LibRaw 0.19-Beta5
2018-05-03 Alex Tutubalin <lexa@lexa.ru>
* CVE-2018-10529 fixed: out of bounds read in X3F parser
* CVE-2018-10528 fixed: possible stack overrun in X3F parser
* LibRaw 0.19-Beta4
2018-04-24 Alex Tutubalin <lexa@lexa.ru>
* LibRaw 0.19-Beta3
* fixed lot of bugs reported by ImageMagic/oss-fuzz
* fixed several bugs reported by Secunia team (adv 81800,
Credit: Laurent Delosieres, Secunia Research at Flexera)
2018-03-22 Alex Tutubalin <lexa@lexa.ru>
* LibRaw 0.19-Beta2
* Better handling of broken JPEG thumbnails
* Panasonic GH5S/G9-hires decoder, thanks to Alexey Danilchenko
Note: ABI has changed due to this patch, so shlib version increased
* Fujifilm X-A5/A20 metadata parsing fix
* New error code LIBRAW_TOO_BIG: image data size excess LIBRAW_MAX_ALLOC_MB
* winsock2 included before windows.h to make MinGW happy
2018-02-23 Alex Tutubalin <lexa@lexa.ru>
* LibRaw 0.19-Beta1
* 84 cameras added compared to 0.18 (1014 total):
Apple
iPhone 8(*), iPhone 8 plus, iPhone X
BlackMagic
URSA Mini 4k, URSA Mini 4.6k, URSA Mini Pro 4.6k
Canon CHDK hack
PowerShot A410, A540, D10, ELPH 130 IS, ELPH 160 IS, SD750,
SX100 IS,SX130 IS, SX160 IS, SX510 HS, SX10 IS, IXUS 900Ti
Canon
PowerShot G1 X Mark III, G9 X Mark II, EOS 6D Mark II, EOS 77D,
EOS 200D, EOS 800D, EOS M6, EOS M100
Casio EX-ZR4100/5100
DJI
Phantom4 Pro/Pro+, Zenmuse X5, Zenmuse X5R
FujiFilm
S6500fd, GFX 50S, X100f, X-A3, X-A5, X-A10, X-A20, X-E3, X-H1, X-T20
GITUP GIT2P
Hasselblad
H6D-100c, A6D-100c
Huawei
P9 (EVA-L09/AL00), Honor6a, Honor9, Mate10 (BLA-L29)
Leica
CL, M10, TL2
LG
V20 (F800K), VS995,
Nikon
D850, D5600, D7500, Coolpix B700
Olympus
E-PL9, E-M10 Mark III, TG-5
OnePlus
One, A3303, A5000
Panasonic
DMC-FZ45, DMC-FZ72, DC-FZ80/82, DC-G9 (std. res mode only), DC-GF10/GF90,
DC-GH5, DC-GX9, DC-GX800/850/GF9, DMC-LX1, DC-ZS70 (DC-TZ90/91/92, DC-T93),
DC-TZ100/101/ZS100, DC-TZ200/ZS200
PARROT
Bebop 2, Bebop Drone
Pentax KP
PhaseOne IQ3 100MP Trichromatic
Samsung
Galaxy Nexus, Galaxy S3, S6 (SM-G920F), S7, S7 Edge, S8 (SM-G950U),
Sony
A7R III, A9, DSC-RX0, DSC-RX10IV
Yi M1
YUNEEC
CGO3, CGO3P
Xiaoyi YIAC3 (YI 4k)
Note(*): for mobile phones with DNG format recording, only really tested cameras
are added to supported camera list. Really LibRaw should support any correct DNG.
* No more built-in support for LibRaw demosaic packs (GPL2/GPL3).
We're unable to support this (very old code), so we'll be happy to transfer this
code to some maintainer who wish to work with it.
In LibRaw 0.19 we provide extension API: user-settable callbacks to be called in
code points where demosaic pack code was called.
- int callbacks.pre_identify_cb(void *) => to be called in LibRaw::open_datastream
before call to (standard) identify() function. If this call returns 1, this means
that RAW file is identified and all metadata fields are set, so no need to run
standard identify code.
- void callbacks.post_identify_cb(void*) => called just after identify(), but before
any cleanup code;
- dcraw_process() callbacks are called before dcraw_process phases (name speaks for itself):
pre_subtractblack_cb, pre_scalecolors_cb, pre_preinterpolate_cb, pre_interpolate_cb,
interpolate_bayer_cb, interpolate_xtrans_cb, post_interpolate_cb, pre_converttorgb_cb,
post_converttorgb_cb
All these new callbacks are called with (this) as the only arg.
To continue LibRaw-demosaic-pack-GPLx support one need to subclass LibRaw, set needed
callbacks in (e.g.) constructor code, than these callbacks to be called
* Better DNG parser:
- support for DefaultCrop Origin/Size tags (add LIBRAW_PROCESSING_USE_DNG_DEFAULT_CROP to raw_processing_options to enable)
- better parsing for nested DNG tags (use tag from RAW IFD, fallback to IFD0 if no tag in current IFD)
- DNG PreviewColorspace extracted into dng_levels.preview_colorspace
* Metadata extraction:
- Better extraction of camera measured balance (LIBRAW_WBI_Auto and WBI_Measured),
this not the same as 'as shot' if some preset/manual tune is used.
- Extraction of camera custom balances (LIBRAW_WBI_CustomN)
- Nikon data compression tag extracted into makernotes.nikon.NEFCompression
- Hasselblad BaseISO and Gain extracted into makernotes.hasselblad
- Canon multishot params extracted into makernotes.canon.multishot
- lot of other vendor-specific makernotes data (see data structures definitions for details).
* New LibRaw::open_bayer call allows to pass sensor dump w/o metadata directly to LibRaw:
virtual int open_bayer(unsigned char *data, unsigned datalen,
ushort _raw_width, ushort _raw_height, ushort _left_margin, ushort _top_margin,
ushort _right_margin, ushort _bottom_margin,
unsigned char procflags, unsigned char bayer_pattern, unsigned unused_bits, unsigned otherflags,
unsigned black_level);
Parameters:
data, datalen - buffer passed
width/height/margins - speaks for itself
procflags:
for 10-bit format:
1: "4 pixels in 5 bytes" packing is used
0: "6 pixels in 8 bytes" packing is used
for 16-bit format:
1: Big-endian data
bayer_pattern: one of LIBRAW_OPENBAYER_RGGB,LIBRAW_OPENBAYER_BGGR,
LIBRAW_OPENBAYER_GRBG,LIBRAW_OPENBAYER_GBRG
unused_bits: count of upper zero bits
otherflags:
Bit 1 - filter (average neighbors) for pixels with values of zero
Bits 2-4 - the orientation of the image (0=do not rotate, 3=180, 5=90CCW, 6=90CW)
black_level: file black level (it also may be specified via imgdata.params)
see samples/openbayer_sample.cpp for usage sample (note, this sample is 'sample only', suited for
Kodak KAI-0340 sensor, you'll need change open_bayer() params for your data).
* Color data added/updated/fixed for many cameras
* Correct data maximum for Fuji X-* cameras
* Thumbnail processing:
- JPEG thumbnails: if compiled with libjpeg, color count is extracted into imgdata.thumbnail.tcolors
- PPM (bitmap) thumbnails: color count is set according to thumbnail IFD tag
- PPM16 thumbnails: if LIBRAW_PROCESSING_USE_PPM16_THUMBS set in raw_processing_options, than thumbnail will be extracted
as is, not converted to 8 bit. thumbnail.tformat is set to LIBRAW_THUMBNAIL_BITMAP16 in this case.
Untested, because it is hard to find RAWs with 16-bit bitmaps.
== Compatibility fixes
* struct tiff_tag renamed to libraw_tiff_tag
* pow64f renamed to libraw_pow64f
== Bugs fixed:
* COLOR(r,c) works correctly on X-Trans files
== Security fixes:
Secunia #81000:
Credit: Laurent Delosieres, Secunia Research at Flexera
* leaf_hdr_load_raw: check for image pointer for demosaiced raw
* NOKIARAW parser: check image dimensions readed from file
* quicktake_100_load_raw: check width/height limits
Secunia #79000:
Credit: Laurent Delosieres, Secunia Research at Flexera
* All legacy (RGB raw) image loaders checks for imgdata.image is not NULL
* kodak_radc_load_raw: check image size before processing
* legacy memory allocator: allocate max(widh,raw_width)*max(height,raw_height)
Secunia #76000:
* Fixed fuji_width handling if file is neither fuji nor DNG
* Fixed xtrans interpolate for broken xtrans pattern
* Fixed panasonic decoder
* LibRaw 0.18.6
Other fixes:
* Checks for width+left_margin/height+top_margin not larger than 64k
* LIBRAW_MAX_ALLOC_MB define limits maximum image/raw_image allocation
(default is 2048 so 2Gb per array)
* LibRaw::read_shorts item count is now unsigned
* Fixed possible out of bound access in Kodak 65000 loader
* CVE-2017-14348: Fix for possible heap overrun in Canon makernotes parser
Credit: Henri Salo from Nixu Corporation
* Fix for CVE-2017-13735
* CVE-2017-14265: Additional check for X-Trans CFA pattern data
* Fixed several errors (Secunia advisory SA75000)
* ACES colorspace output option included in dcraw_emu help page
* Avoided possible 32-bit overflows in Sony metadata parser
* Phase One flat field code called even for half-size
4.1.0 Release
Added new renders Alto, LSTMBox, WordStrBox.
Added character boxes in hOCR output.
Added python training scripts (experimental) as alternative shell scripts.
Better support AVX / AVX2 / SSE.
Disable OpenMP support by default.
Fix for bounding box problem.
Implemented support for whitelist/blacklist in LSTM engine.
Improved cmake configuration.
Code modernization and improvements.
A lot of bug fixes...
From David Weller-Fahy in PR pkg/54340.
Leather is the Python charting library for those who need charts now
and don't care if they're perfect.
Leather isn't picky. It's rough. It gets dirty. It looks sexy just
hanging on the back of a chair. Leather doesn't need your accessories.
Leather is how Snake Plissken would make charts.
Get it?
* Use 1024 instead of NAME_MAX. NAME_MAX is not defined OmniOS-derived OS
and MAXNAMELEN is not portable too, _POSIX_NAME_MAX is too short.
Discussed with tron@.
* Simplyfy my previous code to fix overrun
6.1.0:
- Deprecate Image.__del__
- Tiff: Add support for JPEG quality
- Respect the PKG_CONFIG environment variable when building
- Use explicit memcpy() to avoid unaligned memory accesses
- Improve encoding of TIFF tags
- Update Py_UNICODE to Py_UCS4
- Consider I;16 pixel size when drawing
- Add TIFFTAG_SAMPLEFORMAT to blocklist
- Create GIF deltas from background colour of GIF frames if disposal mode is 2
- Added ImageSequence all_frames
- Use unsigned int to store TIFF IFD offsets
- Include CPPFLAGS when searching for libraries
- Updated TIFF tile descriptors to match current decoding functionality
- Added an `image.entropy()` method (second revision)
- Pass the correct types to PyArg_ParseTuple
- Fixed crash when loading non-font bytes
- Fix SPARC memory alignment issues in Pack/Unpack functions
- Added CMYK;16B and CMYK;16N unpackers
- Fixed bugs in calculating text size
- Add __main__.py to output basic format and support information
- Added variation font support
- Do not down-convert if image is LA when showing with PNG format
- Improve handling of PSD frames
- Improved ICO and ICNS loading
- Changed Preview application path so that it is no longer static
- Corrected ttb text positioning
- Handle unexpected ICO image sizes
- Fixed bits value for RGB;16N unpackers
- Travis CI: Add Fedora 30, remove Fedora 28
- Added reading of CMYK;16L TIFF images
- Fixed dimensions of 1-bit PDFs
- Fixed opening mmap image through Path on Windows
- Fixed ImageDraw arc gaps
- Expand GIF to include frames with extents outside the image size
- Fixed ImageTk getimage
- Fixed bug in decoding large images
- Fixed reading APP13 marker without Photoshop data
- Added option to include layered windows in ImageGrab.grab on Windows
- Detect libimagequant when installed by pacman on MingW
- Fixed raqm layout bug
- Fixed loading font with non-Unicode path on Windows
- Travis CI: Upgrade PyPy from 6.0.0 to 7.1.1
- Depends: Updated openjpeg to 2.3.1
- Fix numpy bool bug
GCC 9 started implementing the OpenMP 4.0 and later behavior. When not using
default clause or when using default(shared), this makes no difference, but
if using default(none), previously the choice was not specify the const
qualified variables on the construct at all, or specify in firstprivate
clause. In GCC 9 as well as for OpenMP 4.0 compliance, those variables need
to be specified on constructs in which they are used, either in shared or
in firstprivate clause. Specifying them in firstprivate clause is one way to
achieve compatibility with both older GCC versions and GCC 9,
another option is to drop the default(none) clause.
This patch thus drops the default(none) clause.
See https://gcc.gnu.org/gcc-9/porting_to.html#ompdatasharing
patch from fedora.
ChangeLog:
AnsiLove/C 4.0.1 (2019-06-25)
- Fix a few typos in README and ansilove.1
- Simplify display of default font when no font option is specified
- Drop the leading dot when saving input file extension
- Fix the check for readComments() return value
AnsiLove/C 4.0.0 (2019-01-16)
- Switch to using libansilove
- Add a new switch (-d) to toggle DOS aspect ratio
- Add initial support for PabloDraw 24-bit ANSI sequences
- Ansilove now generates only one output file when -r or -R options
are used
- Display scale factor when -r or -R options are used
- Allow to specify the number of columns for ANSI and Tundra files
- Restrict maximum allowed number of columns to 4096
- Check asprintf() return value and error if allocation fails
- Bugfixes and improvements in the SAUCE parser
- Refactor font selection processing
- Refactor rendering mode processing
- Use OpenBSD style(9) for function prototypes and declaration
- Add missing headers and remove unnecessary ones
- Remove versionInfo(), simply print the header with version information
when invoked with -v
- Remove showHelp(), show synopsis instead when invoked with -h
- Add a new example ANSi (Thanks Nail)
- Remove -e switch, examples are now listed in the man page instead
- Make synopsis() only display SYNOPSIS, as it should be.
- Add an EXAMPLES section in ansilove.1
- PC fonts for code pages other than cp437 are now named using the cpNNN
scheme. Albeit not documented anymore, the previously used names will
keep working
OK kamil@
openimageio now uses qt5, not qt4.
This was causing blender to fail to build in bulk builds and on systems
where qt4 was not preinstalled.
XXX needs pullup?
Based on work in pkgsrc-wip.
The Irrlicht Engine is an open source high performance real-time
3D engine written and usable in C++ and also available for .NET
languages.
It is completely cross-platform, using D3D, OpenGL and its own
software renderer, and has all of the state-of-the-art features
which can be found in commercial 3d engines.
Library for converting ANSI, ASCII, and other formats to PNG, allowing
to render ANSI and artscene-related file formats into PNG images, and
supporting ANSI (.ANS), PCBoard (.PCB), Binary (.BIN), Artworx (.ADF),
iCE Draw (.IDF), Tundra (.TND) and XBin (.XB) formats.
It creates size optimized 4-bit PNG files and supports SAUCE (Standard
Architecture for Universal Comment Extentions), 80x25 and 80x50 PC fonts
(including all the 14 MS-DOS charsets), Amiga fonts, and iCE colors.
OK kamil@
Version 0.11
Add quiet argument to .render() and .pipe() methods of Graph,
Digraph, and Source objects, allowing to suppress stderr of the
layout subprocess (parity with stand-alone render() and pipe()
functions).
The rendering process for render() methods and stand-alone function is now
started from the directory of the renderred dot source file. This allows to
render graph descriptions that use relative paths inline (e.g. for referring to
image files to be included) by using paths relative to the source file
location. Previously, such relative paths would need to be given relative to
the directory from which render() was started, so this change is backwards
incompatible for code that relied on the previous behaviour.
Drop Python 3.4 support.
Changelog:
Overview of Changes from GIMP 2.10.10 to GIMP 2.10.12
=====================================================
Core:
- Add an "Incremental" option to the Dodge/Burn tool, which,
similarly to the Paintbrush, Pencil, and Eraser tools, applies the
effect incrementally as the pointer moves.
- Curves now have a concept of smooth vs corner points: smooth
points produce a smooth curve, while corner points produce a sharp
curve (previously, all points were smooth).
- Search the user font directory path on Windows (since Windows 10,
non-admin users have the ability to install fonts). This is only a
temporary hack until fontconfig adds proper upstream support:
https://gitlab.freedesktop.org/fontconfig/fontconfig/issues/144
- In gimp_brush_core_get_paint_buffer(), when allocating a new paint
buffer, clear the old buffer *before* allocating the new one, to
reduce the amount of simultaneously allocated memory.
- In GimpPaintbrush, avoid refilling the paint buffer at each dab if
the paint color/pixmap hasn't changed, hence allowing faster
painting in some specific cases.
- Add hygon cpu detection and enable MMX/SSE support.
- Add a GimpSymmetry::get_transform() virtual function and a
corresponding gimp_symmetry_get_transform() function which return
the brush transform corresponding to a given symmetry stroke in
terms of the rotation angle and reflection flag (in contrast to
gimp_symmetry_get_operation() which returns the same transforation
in terms of a GeglNode). This allows us to simplify, fix (artifact
bugs, etc.), and improve the painting-code performance for several
symmetry painting cases.
- New gimp-brush-pipe-spacing parasite set by core code to preserve
GIH brush's spacing across reloads.
- Add a new Offset filter tool, as a front-end to gimp:offset. The
tool replaces, and provides the same interface as the drawable-
offset dialog while also providing live preview and on-canvas
interaction.
Plug-ins:
- Exported profile and data format strategy updated in several file
plug-ins.
* When not saving a profile, we always export as sRGB data, as most
viewers would not display the image properly otherwise.
* When saving a profile:
* If a profile was manually assigned, we always export to the
assigned format, hence pixel data is converted accordingly,
whatever the work format.
* If no profile was assigned:
* If the export format support high-bit depth, we export the
work format as-is.
* If the export format is 8-bit max, we convert the work
format to sRGB, except if the work format is 8-bit linear.
The following plug-ins were updated: JPEG, PNG, TIFF and WebP.
- TIFF plug-in now supports layer export.
- Add accelerator on "Save color _profile" item in JPEG export so
that all items in the dialog have keyboard accelerators.
- Add a "Save color profile" checkbox in WebP export and remove the
"Advanced Options" expander (all metadata and profile writing
options are now directly visible).
- Process cases of non-conformant TIFF file with extra channels
while ExtraSamples is not set by assuming the first extra channel
is a non-premultiplied alpha channel and outputting a warning. In
the future, this should be updated further by a dialog asking what
to do with the non-defined dialog (consider as premultiplied or
non-premultiplied alpha channel, or just save as extra channels).
Tools:
- Add input/output spin-buttons to the Curves tool, which allow
setting the selected point's coordinates numerically. When the image
precision is greater than 8-bpc, use a 0.00-100.00 range for the
point-coordinate spin-buttons instead of a 0-255 range.
- In the Curves tool, allow changing the curve's point types (cf. new
curve corner points in Core section). Corner points are displayed
with a diamond shape instead of a circle.
- Free Select tool now creates a selection as soon as the polygon is
closed. This selection is updated when the polygon, or the relevant
tool-options, change, similarly to GimpRectangleSelectTool.
- Derive GimpEraser from GimpPaintbrush, instead of directly from
GimpBrushCore.
This allows GimpEraser to reuse the paint-buffer content across
dabs, improving performance.
- Allow moving an intersecting pair of guides with the Move tool.
User interface:
- In GimpCurveView, when dragging an existing curve point, don't
immediately move the point to the cursor position upon button
press, but rather move it relative to its current position as the
cursor moves. This allows selecting a point without moving it, and
adjusting its position more easily.
- when the cursor hovers above a point in GimpCurveView, or when
dragging a point, have the coordinate indicator show the point's
position, rather than the cursor's.
- In GimpCurveView, when holding down Ctrl while adding/dragging a
point, snap the y-coordinate to the original curve (at the start
of the drag). This is particularly useful for adding points along
the curve.
- In GimpDashboard, don't show legend for groups with no meter.
- Add a tile-alloc-total variable to the dashboard's memory and misc
groups, showing the total amount of memory used by the tile
allocator.
- Add a "Save Keyboard Shortcuts Now" button to the Configure
Keyboard Shortcuts dialog.
Installers:
- Windows: enable InnoSetup "unofficial" translations. Otherwise we
had a bunch of wasted work from our translators of several
languages whose translations were never used in the installer.
Translations:
- 12 translations were updated: Catalan, Chinese, French, German,
Hungarian, Indonesian, Italian, Polish, Russian, Spanish, Swedish
and Turkish.
Changelog:
GEGL-0.4.16
Not available
GEGL-0.4.15
Build
Added gitlab continuous integration.
Core
Platform specific build fixes for both win32 and OSX.
Avoid in-place processing for cached nodes.
Indentation consitency improvements.
Permit more detailed specifying/overriding of OpenCL device through the
GEGL_USE_OPENCL environment variable.
New call gegl_update_anim_time for applying the animation curves of
properties.
Buffer
Swapped behavior of cubic interpolation from cubic to mithcell
coefficients, leading to crisper results in all uses of the cubic
sampler.
Permit tiles to be unreffed after GEGL shut-down without crash, which
is likely when using bindings to dynamic languages.
Use malloc_trim when compiled with glibc to possibly decrease the
resident size of progress when buffers are destroyed.
Improved deadlock prevention in gegl_buffer_copy() when using copy on
write.
Operations
Use indirect buffer inputs on main thread during multi-processed
processing for point-filters/composers and tranform base classes.
Smaller required/invalided ROI for map-absolute/relative when using
nearest neighbor.
tiff-save and jpg-save fixed dimensions when buffer saved has non-zero
origin
Added missing descriptions of properties for cubism, mosaic,
tile-paper, color, mantiuk06 and vignette operations.
ff-load fix some of the deprecation warnings.
ff-save avoid crashing if audio or video codec fails to initialize.
cache, parent class changed - it is no longer a point operations, also
speed up processing and permit copy on write handling of tiles.
opacity and invert - crash for 8bpc buffers, reimplemented logic for
(non-)associated alpha OpenCL kernel selection.
Tools
Drop direct dependency on exiv2 - we now only transiently depend on it
through the gexiv2 abstraction.
commandline tool/gui
Store per image editing chains in .gegl folder instead of sidecars.
Set window title, permit interaction with opened .lui documents, fixed
playback of audio for opened videos - it now works; if the video is
decoding, scaling and display is fast enough to leave spare cycles.
Support for nearest neighbor scaling of imagry; aiding video playback
without dropping audio frames.
Support for custom collection order and per-file key-value metadata.
Changelog:
2019-06-12 babl-0.1.66
Added API call, babl_space_get_rgb_luminance,
Make most SIMD code-paths optional again on i686.
2019-05-25 babl-0.1.64
Support for Hygon Dhyana, and Y~aA formats for symmetric completeness.
Code consistency, gitlab CI, autotools and meson build improvements.
1.3.32:
Special Issues:
It has been discovered that the 'ICU' library (a perhaps 30MB C++ library) which is now often a libxml2 dependendency causes huge process initialization overhead. This is noticed as unexpected slowness when GraphicsMagick utilities are used to process small to medium sized files. The time to initialize is often longer than the time to read the input file, process the image, and write the output file. If the 'ICU' dependency can not be avoided, then make sure to use the modules build. Please lobby the 'ICU' library developers to change their implementation to avoid long start-up times due to merely linking with the library.
Security Fixes:
GraphicsMagick is now participating in Google's oss-fuzz project due to the contributions and assistance of Alex Gaynor. Since February 4 2018, 343 issues have been opened by oss-fuzz and 331 of those issues have been resolved. The issues list is available at https://bugs.chromium.org/p/oss-fuzz/issues/list under search term "graphicsmagick". Issues are available for anyone to view and duplicate if they have been in "Verified" status for 30 days, or if they have been in "New" status for 90 days. There are too many fixes to list here. Please consult the GraphicsMagick ChangeLog file, Mercurial repository commit log, and the oss-fuzz issues list for details.
BMP reader: Fix heap overflow in 32-bit build due to arithmetic overflow. Only happens if limits are changed from defaults.
BMP reader/writer: Improve buffer-size calculations to guard against buffer overflows.
DIB reader: Reject files which claim more than 8-bits per pixel but also claim to be colormapped.
DIB reader/writer: Improve buffer-size calculations to guard against buffer overflows.
MIFF reader: Detect end of file while reading RLE packets.
MIFF reader: Fix heap overflow (for some files using RLE compression) caused by a typo in the code.
MAT writer: Added missing error handling to avoid heap overflow.
MNG reader: Fixed a small heap buffer overflow.
SVG reader: Fixed a stack buffer overflow.
TGA writer: Fix heap overflow when image rows/columns are larger than 65535.
TIFF reader: Rationalize tile width/height to reject large tile sizes which are much larger than the image dimensions.
TIFF reader: Apply memory resource limits to strip and tile allocations.
WMF reader: Fixed a division by zero problem.
XWD reader: Many heap buffer overflows and uses of uninitialized data were fixed.
Pixel cache: Now apply resource limits to pixel nexus allocations using the same limits (total pixels, width, height, memory) as applied to the whole image since some requests are directly influenced by the input file. More tests are added for arithmetic overflow. Care was taken to minimize performance impact due to the many extra checks.
Bug fixes:
See above note about oss-fuzz fixes.
Fixed include order of magick/api.h vs wand/wand_symbols.h.
WriteImage(): Eliminate use of just-freed memory in clone_info->magick when throwing exception due to no support for format.
Magick++/lib/Magick++/Drawable.h: Fix use of clang diagnostic syntax.
DIB: Preserve PseudoClass opaque representation if ICO mask is opaque.
JPEG reader: Restore ability to access detailed image properties while in 'ping' mode.
JPEG reader: Base test for "Unreasonable dimensions" on original JPEG dimensions and not the scaled dimensions.
JPEG reader: Allow input files to have a compression ratio as high as 2500. Extremely compressed files were being rejected.
FreeType renderer: Fixed a memory leak.
PDF writer: Fixed a memory leak.
PDF writer: Fixed a thread safety problem.
PICT reader: Fix a thread safety problem.
Exception reporting: Throwing an exception was not thread safe. Now it is.
Exception reporting: Handle the case where some passed character strings refer to existing exception character strings.
Command-line parser now does not attempt to read a list of filenames from a file in '@name' syntax if the path '@name' exists. Previously it would attempt to read a list of file names from 'name' even if '@name' did exist.
Rendering: Short-circuit path parsing and return and error immediately if an error occurs.
New Features:
Added support for writing the Braille image format (by Samuel Thibault).
WebP writer: Support WebP 'use_sharp_yuv' option ("if needed, use sharp (and slow) RGB->YUV conversion") via -define webp:use-sharp-yuv=true.
The version command output now reports the OpenMP specification number rather than just the integer version identifier.
API Updates:
ReallocateImageColormap() added to re-allocate an existing colormap.
Some improperly-exposed globals are now static as they should have been.
ChangeLog:
AnsiLove/C 3.0.9 (2018-06-19)
- Add a new example ANSi (Thanks Sylvao)
- Do not use -Werror by default
- Increment maximum ansi sequences length
- Add code to skip PabloDraw 24-bit ANSI sequences
AnsiLove/C 3.0.8 (2017-12-20)
- Adding new example ANSis (Thanks Burps)
- Use asprintf() to create output filenames
- Reflect OpenBSD's pledge() changes
- Add a -R flag allowing to specify a custom retina scale factor
AnsiLove/C 3.0.7 (2017-07-02)
- Print messages to to stderr instead of stdout
- Ensure binary files have an even size
- Better description of supported file formats
- Skip erase in line (EL) sequences (issue reported by Bart Dumon, thanks!)
- Define palettes in config.h
- Use defined palettes instead of hardcoding each color value in the loaders
AnsiLove/C 3.0.6 (2017-04-22)
- Create an output function to unify the way output files are created
- Prevent .png suffix to be appended when using -o flag (*NIX tool
convention compatibility)
AnsiLove/C 3.0.5 (2017-04-16)
- Free memory allocated in the PCBoard loader
- Free font data where appropriate
- Add a trailing NUL character to inputFileBuffer
- Remove the custom substr function, call strndup directly
- Free seqGrab where necessary
- Use the ternary operator when shitfing cursor position
- Use perror when printing errors
- Remove SUBSTITUTE_BREAK and WRAP_COLUMN_80 options
- Use fstat to get the input file size
- Create a retina function instead of duplicating code in all loaders
- Fix OpenBSD's style(9) offenders in compat functions
OK kamil@
Version 2.5.0:
The ffmpeg plugin has been refactored:
The core has been moved to a new library: imageio-ffmpeg.
That library provides platform-specific wheels that includes ffmpeg, so just pip install imageio-ffmpeg instead of the download step.
Note that this new library is py3k only.
Termination of ffmpeg subprocess is now more reliable.
The reader of the ffmpeg plugin now always reports inf as the number of frames. Use reader.count_frames() to get the actual number, or estimate it from the fps and duration in the meta data.
Removed CannotReadFrameError.
Other changes:
The avbin plugin has been depreacted and will be removed in a future version.
Imnproved speed for PIL and FFMPEG plugsins by avoiding memory copies.
Update the included tiffile library.
Support for SimpleITK.
Speed up tiffile plugin when writing to something else than a filename.
Fix that writing to a file object would not work for some plugins.
Can now pass image data to the write functions as anything that resolves to a numpy array with a numeric dtype.
One can now read from a memoryview.
Fix error related to paletted BMP with the Pillow plugin.
Improved logging.
Version 2.4.1:
Fix installation issue on flavors of Ubuntu 14.04 /w Python 2.7.
Use np.frombuffer instead of np.fromstring in some cases.
Version 2.4.0:
Renamed Image class to Array and add documentation for this ndarray subclass.
Reading from HTTP and zipfiles has been improved and better documented.
Improvements to reading and writing of Tiff metadata.
Better dealing of tifffile dependencies on Python 2.7.
Reader for the SPE format.
Better termination of FFMPEG when reading from webcam.
FFMPEG support for reading 16bit videos
v0.9.0:
This is a major release with several substantial and long-desired new features. There are also updates/modifications to the themes and color palettes that give better consistency with matplotlib 2.0 and some notable API changes.
REL: v3.1.0
The first release of the v3.1 series
This version of Matplotlib support Python 3.6+
Highlights of this release include:
- A helper method for scatter legends
- Secondary axis support (ex, degrees and radians)
- A concise date formatter
- No longer requires a framework build of Python to use OSX backend
- Major and minor ticks will no longer collide by default
- A progress callback for animations
- mplot3D is always registered
- deprecated mlab functions have been removed
- Many function calls have become stricter about invalid or ignored input
REL: v3.0.3
The third and last planned bug-fix release of the 3.0 series
* fix several possible memory leaks
* documentation fixes
jquery-ui is no longer bundled in git repo and is downloaded during
the install or sdist process. If you are installing from pypi the
files will be included, but if you are installing from git you will
need internet access the first time you run `pip install .`. The
install process tries to cache the files and will not need internet
access on future installations.
REL: v3.0.2
This is the second bug-fix release for the v3.0 series.
- Un-breaks basemap which was broken by partially restoring
private APIs for cartopy.
- Fixes bug in warning code when used in an embedded context.
- Fixes crash when using Tk and closing the first open window before showing it
- Many documentation improvements.
- Restore a corner case on ColorBar tick usage.
- Change the default behavior of `matplotlib.use` to silently allow
more 'safe' switching after auto-discovery, but before starting an event loop.
- Improvements to bounding box calculations.
- Provide the correct length for RcParams instances.
REL: v3.0.1
This is the first bug fix release for the 3.0 series which fixes
several
- Fix failure to import bug when used with backend-fallback on Python
3.6.7 and 3.7.1
- Fixed a number of failure to import bugs around finding fonts
- Fix Qt4 backend
- Fix bug on OSX that recursively searched current directory for fonts
- Fix bouncing-rocket on OSX when doing backend fallback and not
selecting OSX
- Temporarily restore several private APIs to unbreak cartopy
- Make pyplot more tolerant of varying signatures in 3rd-party
sub-classe
- Improve datetime64 unit handling
- Fixed several poor interactions with tight_layout
REL: v3.0.0
The first release of the Matplotlib 3.0 series
This is the first version of Matplotlib to only support Python 3.
Highlights of this release include:
- GUI backend is selected at run-time based on what toolkits are
installed. A GUI toolkit will not be selected on a headless
server.
- New cyclic color map *twilight*
- Improvements to automatic layout of titles, ticks, and GridSpec
- Many bug fixes!
