GEGL-0.4.42 2023-02-20
Operations:
-
raw-load: add conditional support for 0.21.0 API
-
rgb-clib: improved ui-ranges
-
perlin, mosiac, c2g, long-shadow: small cleanups
-
gif-load: update vendored dependency, including code updates.
Build:
-
keeping up with babl
Contributors to this release:
Alan Mortensen, Chris Mayo, Jehan, Jordi Mas, Michael Drakei, Øyvind Kolås and Sabri Ünal
GEGL-0.4.40 2022-11-13
Operations:
-
newsprint: do not drop alpha channel in RGB modes.
Core
-
fixes for legacy build, still supporting building latest GEGL for stable GIMP with old babl (not recommended if distros can have recent enough meson.).
-
avoiding possible multi thread race in babl garbage collection
Contributors to this release
Asier Sarasua Garmendia, Enrico Nicoletto, Fran Dieguez, Jehan, Jordi Mas, Luming Zh, Marco Ciampa, Matheus Barbosa, nikita, Øyvind Kolås, Sabri Ünal and Sam James.
GEGL-0.4.38 2022-07-13
Operations:
-
denoise-dct: a NEW denoising operation which decomposes the input buffer to sliding overlapping patches, calculates the DCT denoising in each patch, and then aggregates the denoised patches to the output buffer averaging the overlapped pixels.
-
ff-load and ff-save: big cleanup API cleanup, now ffmpeg-5.0 compatible
-
gif-load: updated to latest upstream libnsgif version
-
slic: progress reporting and improved parameter handling.
-
vector-fill: updated to latest upstream ctx version
-
oilify: clamp inputs to avoid nan in output
-
gegl:load fix possible double free
-
rgbe-write: plug leaks in error paths
Core:
Fixes of some potential NULL dereferences on custom dataformats for malformed inputs.
Build
gegl can now be used as a subproject.
The CI rule used on gitlab is now a much simpler one based on fedora, the arch based build started being fragile in working around kernel version dependancy.
Contributors to this release:
Alan Mortensen, Anders Jonsson, Axel Viala, Behnam Momeni, darnuria, dimspingos, Felix Schwarz, Hugo Carvalho, Jehan, Jordi Mas, Luming Zh, Marcia van den Hout, Marco Ciampa, Martin, Michael Drake, Øyvind Kolås, Piotr Drąg, Rodrigo Lledó, Sabri Ünal, Thomas Manni, Tim Sabsch, Xavier Claessens, Yuri Chornoivan and Zurab Kargareteli.
GEGL-0.4.36 2022-02-21
The primary addition in this release is SIMD dispatch, buffer access functions get built for x86-64 v2 and v3 microarchitecture levels as well as ARM neon. the same applies to core operations where we have .so bundles for the variants as well as a generic build.
Operations:
The npd operation fixed with multiple threads.
Contributors to this release
Alan Mortensen, Anders Jonsson, Caleb Xu, Jehan, Øyvind Kolås and zamfofex.
GEGL-0.4.34 2021-12-19
Operations:
magick-load: use g_spawn_async instead of system to run the image magick convert fallback - preventing shell expansion on non-escaped/filtered paths in constructed commandline.
ripple: avoid a possible division by zero.
Build:
Explicit dependency specification in meson for generated CL files.
ctx has been moved to be part of gegl-common.so nearer to the operations using it avoiding polluting the GEGL library symbol table with the ctx symbols.
Contributors to this release:
Anders Jonsson, Asier Sarasua Garmendia, Boyuan Yang, dimspingos, Gavin Cui, Hugo Carvalho, Jehan Jordi Mas, krzygorz, Lukas Oberhuber, Marco Ciampa, Matej Urbančič, Øyvind Kolås, Rodrigo Lledó, Rūdolfs Mazurs and Simon McVittie.
GEGL-0.4.32 2021-09-05
Docs/testing:
Fixes and improvements to the test system ands it integration for comparing/keeping track of hashes of operation output by John Marshall including and overhaul of reference compositions and hashes, improving development as well as documentation.
Operations:
-
distance-transform:
-
openexr: fix build with newer OpenEXR 3
-
negative-darkroom: contrast boost and illuminant adjustment parameter, reworked emulsion dye model, UI improvements, more black and white paper presets.
-
fill-path: 32bit float RGB and CMYK color processing, using ctx as renderer.
Core:
Fixing of leaks and dead code removal in json op handling, branch hints in buffer tile handling.
Contributors to this release:
Alexandre Prokoudine, Anders Jonsson, Andrzej Hunt, Antonio Rojas, Asier Sarasua Garmendia, Boyuan Yang, Daniel Mustieles, dimspingos, John Marshall, JonnyRobbie, Jordi Mas, Marco Ciampa, Matej Urbančič, Øyvind Kolås, Piotr Drąg, Richard Kreckel, Rodrigo Lledó, woob and Yuri Chornoivan,
GEGL-0.4.30 2021-03-27
Build:
libjpeg and libpng are now required.
Reduce babl requirement to 0.1.78, it is highly recommended to build with a newer babl though despite it being what wants to increase the meson version requirement. Improved gir build options, fixes for cross compilation.
Large improvements of integration of tests with meson. XML tests are now outputting in TAP format. As well as many related cleanups of tests that have been in a state of slight bitrot since the migration to meson.
Upload meson logs as artifcats, use release and debugoptimized build types.
Pin glibc package in arch CI build to a version before it hard depends on linux 4.4.
New Operations:
negative-darkroom: This operation is for artists who use hybrid workflow
technique of analog photography. After scanning a developed negative, this
operation is used to invert the scan to create a positive image by simulating
the light behaviour of darkroom enlarger and common photographic papers.
Operations:
jpg-load,png-load,tiff-load: show ICC relatd problems as warnings.
rgbe-load : fix handling of exponent=0 in RLE encoded files.
color-reduction : Added blue-noise modes, the patents are expired for some
years already, also added fix to levels parameters; enabling bi-level.
fattal02 : ensure gaussian pyramid has at least one level.
paint-select (in workshop): multi-level banded graphcut, sub-region rather
than full buffer rcomputations and other improvements.
Contributors to this release:
dimspingos, Jan Vesely, John Marshall, JonnyRobbie, Jordi Mas, Милош Поповић, Olivier Tilloy, Øyvind Kolås, Philipp Kiemle, Piotr Drąg, Sabri Ünal, scootergrisen, Thomas Manni, Yuri Chornoivan.
GEGL-0.4.28 2020-12-20
Core:
Infrastructure to opt out of listing ops with unmet runtime dependencies.
Build:
Set vapi metadata better under meson. Make graphviz' dot library is a run-time not build time dependency.
Operations:
gegl:convert-space and gegl:cast-space operations fixed, and added to build. icc:load new operation that treats .icc files as images, permitting loading a space into the graph from file.
Workshop:
New operation paint-select for an foreground select tool in development.
Contributors to this release:
Asier Sarasua Garmendia, Christian Kirbach, Corentin Noël, Cristian Secară, Dimitris Spingos (Δημήτρης Σπίγγος), Jehan, Jordi Mas, Milo Ivir, Niels De Graef, Øyvind Kolås, Piotr Drąg, Rico Tzschichholz, Sabri Ünal, Stephan Woidowski, Thomas Manni, Yacine Bouklif and Yuri Chornoivan.
GEGL-0.4.26 2020-08-02
Core:
Allow unsetting of property keys on paramspecs.
Operations:
perlin-noise: use a local random generator, avoiding resetting the global one. transform: decrease epsilon for affine-detection, and round some affine matrices to be exact.
Contributors to this release:
Alan Mortensen, Jan Vesely, Øyvind Kolås, Jordi Mas, Michael Natterer, Asier Sarasua Garmendia, Ell, Tim Sabsh and Tatsuki Makino.
GEGL-0.4.24 2020-06-07
Build:
CI creation and caching of container images for build. Force support for 64bit file offsets on win32, re-enabling >2GB tile swap, fallout from migration to meson. Operations in workshop are now built as a bundle.
Core:
cppcheck liniting fixes. Report pixel-formats as part of graphviz debug. New meta-data API, that permit handling non-exif metadata in different file loaders and savers in a generic manner. Use g_ascii_strtod to avoid being tripped up by , expected through locale.
Buffer:
Fixed integer overflow in u32 bilinear blit scaler.
Changed cubic interpolation to be a bit softer, reducing induced moire seen in issue #167. Speeded up cubic sampler with custom inlinable fabsf.
Gracefully fail when running out of swap space. issue #229 Avoid possible deadlock in file monitoring, working around upstream issue. Reduced contention on tile allocation. Allow creating buffers with uninitialized data, and tracking of empty tiles in tile backends. Sleep on tile lock contention, reducing contention when hyperthreading. Cppcheck liniting fixes.
Operations:
Meta-operations now have an additional vfunc, update(), which gets invoked on property changes, all meta ops have been refactored to use the new cleaner API. gegl_operation_meta_watch_node(s) were already effectively nops, they are marked as deprecated and left as stubs for thid party ops still calling them.
We now provide our own inlined versions of some basic single precision floats ops, this enables inlining of performance critical code also with lower optimization levels.
crop: with no parameters set, infer rectangle to crop from graph.
crop: cropping rectangle can be implied from graph. hue-chroma: avoid modifying hue/chroma of neutrals dropshadow: added option for shadow growing median-blur: accept negative radius values rgb-clip: add parameters for specifying clipping ranges vignette: new shapes, horizontal and vertical. imgcmp: make max-diff property conform to gobject standard. save-pixbuf: use a GObject for consumer read instead of pointer property. gegl-graph: add a gegl-chain tutorial as default script.
New Ops:
border-align: place a buffer within the borders of another one. pack: joins two buffers into one, with optional gap. bloom: adds a glow around highlights. piecewise-blend: uses a grayscale map as index into array of buffers used as LUT. variable-blur: uses piecewise-blend piecewise-blend and gaussian blur to approximate blurs with per-pixel blur radius. focus-blur: a vignetting blur, using variable-blur. reset-origin: moves upper left of extent to 0,0 band-tune: parametric band equalizer for tuning frequency bands of image.
Contributors to this release
Alexandre Prokoudine, Andre Klapper, Asier Sarasua Garmendia, Brian Stafford, Daniel Mustieles, Debarshi Ray, Dimitris Spingos (Δημήτρης Σπίγγος), Ell, Elle Stone, Jan Vesely, Jehan, Jordi Mas, lillolollo, Marco Ciampa, Michael Schumacher, Nathan Follens, Niels De Graef, Øyvind Kolås, Piotr Drąg, Rodrigo Lledó, Senya, Simon McVittie, suzu urek, woob, Yuri Chornoivan,
GEGL-0.4.22 2020-02-18
Build
Updates to python gobject introspection tests, and made them able to look up babl typelib. Build pdf:load again; missing since meson migration. Fix OpenCL include file generation to work in non-utf8 locales.
Operations
matting-{global,levin}: fix crash when bounding boxes of input and aux differ.
Contributors to this release
Anders Jonsson, band-a-prend, Ell, Félix Piédallu and Øyvind Kolås.
GEGL-0.4.20 2020-01-17
Build
Improvements and refinements of meson builds and CI. Fixing major issues for OSX and win32.
Buffer
Pass O_BINARY where relevant on windows, fixing crashes occuring when swap is used. Avoid 32bit overflows in computations for gegl_buffer_set and gegl_buffer_get.
Core
Avoid running more thread jobs than there are pixels to process. Faster teardown of buffer caches when bounding box shrinks. only do in-place processing if ROI fits in input abyss.
Operations
Improved edge handling for gegl:distance-transform gegl:crop removed workaround for incorrect in-place handling.
Contributors to this release
Alan Mortensen, Alex Samorukov, Christoph Reiter, Dimitris Spingos, Ell, Jan Vesely, Jordi Mas, Julien Hardelin, Koop Mast, Marco Ciampa, Michael Schumacher, Øyvind Kolås and woob.
GEGL-0.4.18 2019-10-27
Build
Drop autotools, GEGL is now using meson, most things are working again now - documentation/website build is a work in progress, many issues have been fixed, more a sure to surface with this first meson only release following the babl one.
Core
Support ui labels and descriptions for aux pads of ops. Guard against uint32_t overflows in memory computations, improved memory usage stat on FreeBSD. Added support for hygon dhyana architecture. Added gegl_try_malloc and use it to allocate storage for tiles. gegl-stats gained active- and assigned-threads members.
Buffer
New API for aligning to regular tile grid, gegl_rectangle_align[_to_buffer] added gegl_rectangle_xor gegl_buffer_create_sub_bufer now always returns a new object. Improved tile compression, more accurate accounting and avoidiance when it is determinetal. gegl_buffer_save : store BablSpace take into consideration concurrent use of different tile sizes in gegl_tile_alloc Fast path for gegl_buffer_copy is entirely in abyss. Improved api misuse handling in gegl_buffer_set|get Improved consistency of results when doing COW of abyss. Improved Gobject Intospection annotations. Use a binary tree for tile swap, improving tile removal speed. Avoid gegl_buffer_set avoid set_pixel fast path if level>0
Operations
Dynamic computation of per-operation thread cost, making GEGL fully use available cores on more operations.
new operations: bayer-matrix, linear-sinusoid, color-assimilation-grid, normal-map. new operations in workshop: boxblur, alpha-inpaint
pass-through for infinite planes for: apply-lens, polar-coordinates, whirl-pinch, convolution-matrix, maze, cartoon, waterpixels, tile, distance-transform, spherize, tile-paper, dither, mantiuk06, reinhard05, fattal02, fractal-trace, motion-blur-zoom, motion-blur-circular, wind, little-planet, mirrors, cubism and long-shdow. improved property ranges/meta-data or copyright for: motion-blur-zoom, edge-neon, cartoon, photocopy, softglow, oilify, bump-map, remap,
convolution: fix source-buffer offsets of filtered-out components, avoiding excessive shifting. jpg-save: force grayscale if incoming buffer is grayscale sdl-display: port to SDL2 gamma: handle negative input values apply-lens; enable auto threading matting-global: disable auto threading, allocate less temporay memory. maze: reimplemented without recursion newsprint: added black pullout and per colorant pattern and period parameters, enable threading. saturation: add a colormodel property which permits operating in RGB instead of CIE Lab, enabling hue preserving saturation changes. exr-load/exr-save: only use associated alpha with EXR files ff-load: fix audio track decoding for pure audio files c2g: fixed initialization data race and out of bounds array access when threaded. color-enhance: fixed a buffer leak perlin-noise: fix initialization race domain-transform: faster nop for edge-preservation=1.0, support BablSpace oilify: support specifying mask-radius with aux buffer linear-gradient, readial-gradient: use colors in correct order. pixelize: avoid potential glitches weighted-blend: harmonize opencl implementation with C edge-neon: performane improvements
GEGL UI
gcut video editor ui has been removed, all its functionality is latent in the new media browser/manipulator/viewer UI.
Permit start/end times for playlist items, enabling use of folder as EDL. Add ability loop currently shown media element. Nearest neighbor decoding; for realtime playback of HD video content. Frame-cache for rendered video frames (with associated PCM data) Drag and drop from file-managers to collection view.
Contributors to this release:
Abderrahim Kitouni, Alexandre Prokoudine, Asier Sarasua Garmendia, Bruce Cowan, Debarshi Ray, Dimitris Spingos, Ell, fanjinke, Félix Piédallu, Jan Vesely, Jordi Mas, Marco Ciampa, Martin Srebotnjak, Michael Natterer, Michael Schumacher, Niels De Graef, Øyvind Kolås, Pawan Chitrakar, Piotr Drąg, Rodrigo Lledó, Sabri Ünal, Schwan Carl, Shlomi Fish, Stefan Brüns, Thomas Manni and Ting-Wei Lan.
GEGL-0.4.16 2019-05-07
Buffer
Swapped behavior of cubic interpolation from cubic to mithcell coefficients, leading to crisper results in all uses of the cubic sampler.
Use a custom allocator for tile data which aligns data and groups allocations in blocks - improving performance on windows by keeping more slack allocation memory in-process. On linux the opposite was needed and is achieved by using the GNU extension malloc_trim which permits forcing invokation of the glibc malloc/free allocators garbage collection function.
Permit tiles to be unreffed after GEGL shut-down without crash, which is likely when using bindings to dynamic languages.
Improved deadlock prevention in gegl_buffer_copy() when using copy on write.
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.
Build
Added gitlab continuous integration.
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.
Custom collection order and per-file key-value metadata.
Contributors to this release:
Alex Samorukov, Brian Stafford, Carles Ferrando Garcia, Dimitris Spingos, Edward E, Ell, Jan Vesely, Jehan, Jordi Mas, lumingzh, Marco Ciampa, Martin Srebotnjak, Massimo Valentini, Nuno Ferreira, Øyvind Kolås, Piotr Drąg, sabri ünal and Sveinn í Felli.
GEGL-0.4.14 2019-03-01
Core
New Scratch allocator for short lived buffers.
Removed unneeded buffer copies in some processing code paths in GeglNode and GeglProcessor.
Add gegl_babl_variant API for getting variations on an existin pixel-format/encoding.
Expose gegl_node_is_graph()
GeglBuffer
Compression of tile data.
Handle dynamically changing swap dir.
Improvements to mipmap damage propagation.
Fixes to threading race conditions.
Low-level tile handling more adapted to buffers creating from existing linear data.
New iterator API is now the default, GEGL_BUFFER_ITERATOR2_API is no longer needed.
CMYK/grayscale handling for scaled blits, code paths in samplers, serialization.
New default tile-cache size, 50% of RAM.
Initialize the first OpenCL device with IMAGE support.
Added gegl_parallell_ from gimp_parallell_ API to distribute processing for multiple cores.
Operations
crop, added aux-pad - if connected the incoming bounding box determines the crop size.
Input-format specific processing for performance in watershed-transform and invert ops. CMYK handling in jpg-load/save, tiff-load/save, gaussian-blur, opacity, most composing and blending operations, text, vector-stroke, path, and transform operations.
Improved or fixed mipmap preview rendering of emboss, linear-gradient and radial-gradient.
Better ui strings/range/defaults for: distort/waves and dropshadow, value-propagate.
gaussian-blur: fix mistake in implementation of IIR Young blur 1d, which caused non circular halos.
New operation: pdf-load with build dependency on poppler-glib
Added offset properties to pixelize filter.
new ops in workshop: voroni diagram, Grey Color Removal, spyrograph.
Handle mipmap rendering for linear-gradient and radial-gradient.
UI
The existing microraptor-gui (immediate mode touch UI+CSS with cairo) image viewer that can be built as part of the gegl binary has been revisted, a graph/property editor has been added, and migration/extension to lua has begun for parts of the code.
Warning
|
many parts of the UI is of exploratory proof of concept/alpha quality but are released since the code is part of the GEGL repository and important development tools both for creating and testing new GEGL operations as well as GEGL itself. |
Some screenshots showing some the process towards the state and capabilties of the UI in this release can be seen at https://www.patreon.com/posts/24123574
To build with this UI the following dependencies also need to be satisifed, for now this is not intended to be packaged by distributions yet.
libluajit-dev
lua-lgi (debian package name) upstream:
mmm https://github.com/hodefoting/mmm
mrg https://github.com/hodefoting/mrg
Features: folder view resolution independent UI touch oriented navigation GIF / video playback PDF pagination goats that expose performance and capabilty issues in GEGL efficient (though not sandboxed) thumbnailing embedded simple commandline graph editor zoomable automatic, consistent graph layout autogenerated property-inspector ui visualization of color model/bitdepth through styling of edges live-extendable with lua code for per-operation canvas UI
Contributors to this release:
Alan Mortensen, Alexandre Prokoudine, Anders Jonsson, Andre Klapper, Debarshi Ray, elad, Ell, Jehan, Marco Ciampa, Michael Natterer, Øyvind Kolås, Piotr Drąg, Richard B. Kreckel, Rodrigo Lledó, sabri ünal, thetoastcaper, Thomas Manni and woob.
This release depends on the new features of babl-0.1.62 which has the following change summary for the last release:
Continuous integration with gitlab. Initial CMYK spaces with lcms2 based ICC support, much room for optimization. Improved custom space support for palette formats. scRGB space, works like sRGB but always with linear TRCs. Model introspection API permitting low overhead checks whether a format/model is RGB, CMYK type of alpha and similar.
GEGL-0.4.12 2018-10-23
GeglBuffer
Fix fix of scaled blitting, the fix from last release introduced scaling artifacts for all formats that are non-8bit.
Contributions to this relase
Ell, Øyvind Kolås and Rodrigo Lledó
GEGL-0.4.10 2018-10-16
GeglBuffer
Fix integer overflow issues for u32 blit scaling code paths.
Reduced memory clearing overhead for temporary buffers during blit.
Round alpha component during u8 bilinear/box filtering (this caused artifacts on OSX through interaction without unknown platform differences in libc/compilter/graphics stack)
gegl_buffer_dup now matches source buffer abyss, shift and tile-size more closely.
Added GEGL_TILE_COPY command for backends, for more efficient buffer duplication/copies.
New GeglBufferIterator API.
Move type-generic scaling algorithms to GeglBuffer folder, move opencl integration bits out of buffer.
Operations
gif-load: new operation for decoding GIF images/animations, with frame delay exposed as an out-property.
exr-save: do not hard-code a version name space long-shadow: add Fading (fixed length) style with a new user-defined Midpoint parameter
maniuk06: use double precision, permitting to work on larger images.
watershed-transform: improved description
unsharp-mask: permit std-dev radius of up to 1500px (from 300).
Updated all operations to use new iterator API
Contributors to this release
Alexandre Prokoudine, Anders Jonsson, Debarshi Ray, Ell, Hussam Al-Tayeb, Marco Ciampa, Massimo Valentini, Øyvind Kolås, Piotr Drąg, Rodrigo Lledó and Tim Sabsch.
GEGL-0.4.8 2018-08-17
Core/GeglBuffer
Fixes to races during buffer/tile storage destruction, improve swap usage for stored empty tiles.
Operations
motion-blur-circular - improve/clarify property ui
median-blur - added abyss-policy property
long-shadow - new operation
little-planet - adapt reference composition
Contributors to this release
Alexandre Prokoudine, Anders Jonsson, Daniel Mustieles, Ell, Marco Ciampa, Martin Srebotnjak, Matej Urbančič, Øyvind Kolås, Piotr Drąg, and Rodrigo Lledó
GEGL-0.4.6 2018-07-23
Up until now GEGL has been using a color space corresponding to scRGB as an unbounded device independent/possibly scene-referred HDR color space - with a similar approach to to how ACEScg works but with a worse set of RGB primaries. babl formats, represented by a pointer and a corresponding encoding/format string have been used to specify the specific encoding of pixel values. The encoding including component order, data type and TRC encoding. Where "RGBA float" means 32bit float data and "R’G'B' u8" the ' indicates non-linear, and thus this is sRGB. "RaGaBaA half" gives premultiplied linear half data. Other encodings and conversions are also provided through these formats including "CIE Lab float" and "HSV float".
As a color management workflow for scene-referred imaging the above could be sufficient, but GIMP needs data in the 0.0-1.0 range for some display referred blending modes to work properly. As a consequence of this recognized short-coming GIMP has been passing the pixels of for instance ProPhoto "R’G'B’A float" off as "R’G'B’A float" and linear ProPhoto "RGBA float" as "RGBA float" this works for single operations, but falls apart when the colors are converted to CIE Lab. This is the good enough state where the other benefits of having a stable release powered by GEGL outweighed not being entirely correct.
Since babl 0.1.32 of october 2017, all babl formats have an associated unchangable space associated with them, and since then GeglBuffer has worked correctly with it - since GeglBuffers use of babl API did not change. GIMP is already using these parts of babl for ICC matrix based conversions since using babl for ICC profile transforms is an order of magnitude faster than using the lcms2 library. It took time to come up with the above scheme of integrating arbitrary primaries and curves for spaces with babl in a maintainable manner, and it has taken until the last month to come up with a full plan for the rest of GEGL to be aware of and handling arbirary parametric ICC v2/v4 based color spaces for operations; without limiting the ability to extend and use the code for a wide range of scenarios.
A space can be constructed from a preferenced name/specification, loaded/saved from ICC matrix profiles or constructed and serialized to whitepoint + rgb chromaticities / xyz matrix. More recently an additional trc mark has been added ~, giving this vocabulary for RGB formats, in addition to variants with alpha and pre-multiplied alpha variants of the same:
"RGB" linear primaries from space, linear data
"R'G'B'" non-linear primaries from space, TRCs from space
"R~G~B~" perceptual primaries from space, sRGB TRC
When creating device independent CIE based spaces they also get passed a space, this means that we can convert CIE Lab to RGB float, keeping track of which space / ICC profile the data correspond to.
GEGL operations now construct their desired encodings of formats by taking the space of buffers on input pads into account. By default, for composers "input" wins over "aux" to determine ops space. If an operation is not ported, data will be converted to sRGB on input and sRGB will come out of the node.
Buffer loaders PNG, JPG, TIFF and EXR generate custom spaces based on ICC profiles/primaries. The corresponding savers saves color space information. A new save handler for the .icc extension, acts like an image storer but only saves the ICC profile of the buffer it gets on input.
With no additional operations inserted, this now means that GEGL graphs operate on linear / non-linear variants of the color space used in the input images without conversion. The new operations gegl:cast-space and gegl:convert-space provide means of overriding this behavior, see the new section about color management in the gegl-chain syntax documentation at http://gegl.org/gegl-chain.html
Initial work has started on making GIMP also use of and propagate color space information along with encoding in babl formats, changes which also will be integrated in the 2.10 branch.
Other changes to operations:
vignette: fixes to gamma property
New operations:
cast-space: assign/override color space
convert-space: convert to a different color space
litte-planet: stereographic-mapping split out of the panorama-projection
New operation in workshop:
acrs-rrt: ACES RRT based HDR to SDR proof/preview point-filter tonemapping op.
Contributors to this release:
Alan Mortensen, Daniel Korostil, Daniel Mustieles, Ell, Francisco Mariano-Neto, Marco Ciampa, Matej Urbančič, Michael Natterer, Michael Schumacher, Øyvind Kolås, Piotr Drąg, Rodrigo and Snehalata B Shirude.
GEGL-0.4.4 2018-07-04
Core
Added swap-reading, swap-writing, swap-read-total and swap-write-total properties to GeglStats.
Buffer
Trim tile-cache when shrinking. Added reading locks to tiles further improving concurrent access tolerance of GeglBuffer.
Operations
Point filters with need for format conversion of input are now parallel.
threshold: operating format changed from linear to perceptual unsharp-mask: add rename standard-deviation and scale properties to Radius and Amount, add Threshold property. recursive-transform now supports applying multiple transformations simultanously. Fixed integer overflow due to logic error in pixelize. New operation: gegl:abs which negates negative values. New operation in workshop: selective-hue-saturation.
Contributors
Alan Mortensen, Alexandre Prokoudine, Daniel Korostil, Debarshi Ray, Dimitris Spingos (Δημήτρης Σπίγγος), Ell, Marco Ciampa, Michael Schumacher, Miroslav Talasek, orbea, Øyvind Kolås, Øyvind "pippin" Kolås, Piotr Drąg, Rūdolfs Mazurs, and Tobias.
GEGL-0.4.2 2018-05-15
Build
Abort early if autoreconf fails, remove unused bits, default to -Ofast as CFLAGS.
GeglBuffer
Improve concurrency for trimming and destruction of tile caches. Improve cache invalidation during partial mipmap regeneration.
Do new cheap clones of buffers with new internal gegl-buffer backed tile-backend.
Do not keep cached sampler in buffer it makes cache invalidation hard, and for performance/threading it is better to create ones own samplers anyways. The old API still exists, though parts of it is now deprecated. The single special case where gegl_buffer_sample remains somewhat performant is with the NEAREST sampler, for all other samplers creating a caching sampler is better.
Operations
operation: add GeglOperationAreaFilter::get_abyss_policy() vfunc Copyright notice improvements to spherize, color-overlay. ff-save: implement defines handling compilation with ffmpeg 2.3-2.7, 4.0 compat.
Improved multi-threaded performance of panorama-projection and other transformation operations through optimizations in buffer and base-classes.
Contributors
code improvements and translations from: Daniel Korostil, Debarshi Ray, Dimitris Spingos (Δημήτρης Σπίγγος), Ell, Franz Brausse, Jehan, Jordi Mas, Marco Ciampa, Øyvind Kolås, and Sebastian Pipping.
GEGL-0.4.0 2018-04-26
GEGL 0.4.x is a stable release series of GEGL to have a stable API and ABI to go with GIMP-2.10. When the need arises to break API or ABI compatibility the 0.4.x series will get its own branch and unstable API experiments and development, 0.5.x will happen in master. For now though master will track the 0.4.x series.
Core
Change GeglParamSpecSeed from int to uint to match the value range of GeglRandom’s seed.
Operations
ripple fixed remaining issues needed to remove legacy GIMP plug-in in GIMP. spherize, requrise transform and color-overlay moved out of workshop. ppm-load ported to GIO.
Contributors to this release: Ell, Piotr Drąg, Anders Mortensen, Anders Jonsson, Marco Ciampa, Daniel Korostil, Øyvind Kolås, Jon Nordby and Simon Budig.
GEGL-0.3.32 2018-04-16
Core
Change GeglParamSpecSeed from int to uint to match the value range of GeglRandom’s seed.
Operations
Limit allocations in ppm-load to 2GB CVE-2018-10114
Contributors to this release: Piotr Drąg, Øyvind Kolås, Sebastian Rasmussen and Michael Natterer.
GEGL-0.3.32 2018-04-13
Operations
panorama-projection: added reverse transform, which permits using GIMP for retouching zenith, nadir or other arbitrary gaze directions in equirectangular, also known as 360x180 panoramas.
Added abyss-policy to base class for scale ops, making it possible to achieve hard edges on rescaled buffers.
GeglBuffer
Improved performance and correctness, avoid incorrectly gamma/ungamma correcting alpha in u8 formats, for a tiny 2-3% performance boost.
Keep track of valid/invalid areas on smaller granularity than tiles in mipmap.
Various micro-optimizations in display paths, with minuscle performance impact.
Contributors to this release: Alexandre Prokoudine, Ell, Jehan, Marco Ciampa, Michael Natterer, Øyvind Kolås and Piotr Drąg.
GEGL-0.3.30 2018-03-19
GeglBuffer:
gegl_buffer_get now treats the final argument, which used to control only abyss behavior as a bit-field that also can have interpolation method to use when scaling down from the next larger mipmap level, valid values are GEGL_BUFFER_FILTER_NEAREST, GEGL_BUFFER_FILTER_BILINEAR, GEGL_BUFFER_FILTER_BOXFILTER and GEGL_BUFFER_FILTER_AUTO.
All scaling is now done on linear data, making future results of mipmap computations more valid as well as providing correct results for scaled down thumbnails in user interfaces. Optimized code paths have been added for 8bit buffers (using 12bit linear as processing space) and other formats like 16bit float are now correctly handled by going through generic code paths.
The scaling code paths has also been instrumented and optimized for performance. Improvements to pixel data fetching and setting functions leads to performance boosts across many GEGL operations, as well as display code paths in GEGL using applications. For some performance critical display cases performance should have improved 2-3x over last release.
Improvements to logic switching to boxfiltering behavior and improve cubic samplers box filtering by using a linear sampler and a nearest sampler for the boxfiltering for the linear sampler. This slows down scaling down with the cubic sampler by half, but is less prone to aliasing, while keeping the faster code paths for the fast but not as bad as nearest - bilinear.
gegl_buffer_copy gegl_buffer_clear and gegl_buffer_set color now only emit a single changed signal on the buffer.
Operations:
c2g and stress have gained the ability to toggle the influence of the shadow neighborhood or not, the new default is to not improve shadow detail; yielding a bit more natural renditions.
color-to-alpha now contains improvements from experiment in workshop which is gone.
Fixed crahses in median-blur.
Transform ops:
Only rasterize inside the transformed polygon, saving cpu. The optimizations mentioned for GeglBuffer giving boosts and improved quality.
New ops:
Mean-curvature blur.
News ops in workshop:
color-overlay, enlarge, inpaint, domain-transform, recursive=transform.
Contributors for this release were: Alan Mortensen, Alexandre Prokoudine, Debarshi Ray, Dimitris Spingos, Ell, Felipeek, Inaki Larranaga Murgoitio, Jehan, Jordi Mas, Mario Blättermann, Michael Natterer, Øyvind Kolås, Piotr Drąg, Rūdolfs Mazurs, and Thomas Manni.
GEGL-0.3.28 2018-01-23
Core
A new GeglStats object, which provides realtime statistics/profiling for use in UIs. Automatically copy-on-write clone input buffers for auto-threading of operations.
Buffer
Many significant optimizations as well as a correctness improvement, mipmap scaling and rendering is now done correctly in linear space, thus GIMP will not only produce the right results but also avoid gamma errors in thumbnails and viewport.
Among the signifcant performance boosters are per buffer-instance tile queues, adaptive sampler pre-fetch caching and dedicated code paths for fetching columns of pixels. Along with many multi-threadedness improvements due to more significant testing.
Operations
Mipmap tuning and performance improvements to gaussian blur added dedicated code paths for nearest sampler in transform ops. more generic RGBE file handling ff-load: permit decoding audio files as well as video files. various code cleanups
GEGL-0.3.28 2018-01-23
Core
A new GeglStats object, which provides realtime statistics/profiling for use in UIs. Automatically copy-on-write clone input buffers for auto-threading of operations.
Buffer
Many significant optimizations as well as a correctness improvement, mipmap scaling and rendering is now done correctly in linear space, thus GIMP will not only produce the right results but also avoid gamma errors in thumbnails and viewport.
Among the signifcant performance boosters are per buffer-instance tile queues, adaptive sampler pre-fetch caching and dedicated code paths for fetching columns of pixels. Along with many multi-threadedness improvements due to more significant testing.
Operations
Mipmap tuning and performance improvements to gaussian blur added dedicated code paths for nearest sampler in transform ops. more generic RGBE file handling ff-load: permit decoding audio files as well as video files. various code cleanups
GEGL-0.3.26 2017-12-14
Core
Make gegl-chain parser more robust wrt parsing babl formats, and improved - translatable - error handling.
Add capabilitity to use per thread input buffers for multi-threaded filter/composer/composer3 ops, this reduces contention over the buffer tile locks and improves multi-threaded performance in some cases without reducing it for the majority of cases.
Buffer
Threads no longer blocked while locking tiles, and fixes to gegL_buffer_copy. Better clone accounting in tile cache.
Operations
exp-combine: code cleanup/simplification
libraw: make bounding box handling consistent with rest of code.
transform ops: re-enabled multi-threading, reduced some format
conversions and added wiggle-room in required/invalidated
calculations.
GEGL-0.3.24 2017-11-24
Core
Flush opencl side buffers before multi threaded iteration for point ops, making multi threading + opencl work.
Add meta-op graph case to gegl_operation_get_source_node
Versioning of installed .mo files restored.
Enable multi-threaded processes by default.
Stop enabling opencl by default, performance with beignet is worse than multi-threading, and with nvidia blank results are being observed in GIMP; maybe improved opencl kernels would make it competetive with multi threaded cpu?
Deprecate gegl_operation_context_get_source, replace usage with gegl_operation_context_get_object.
Buffer
More multi-thread proofing.
Operations
median-blur: now op coming out of the GEGL workshop.
exposure: increase old lingering constant enabling larger range of valid inputs.
transform: turn off multi-threading, it was currently degrading overall performance.
ppm-load: fix file-loader registration.
Contributors
Debarshi Ray, Ell, Jordi Mas, Øyvind Kolås, Richard Kreckel, Thomas Manni, Marco Ciampa, Alexandre Prokoudine and Piotr Drąg.
GEGL-0.3.22 2017-11-12
Core
Refactored graph traversals to use topological sorts.
Completely avoid loading plug-ins .so with -gpl3 in filename unless the application using GEGL explicitly has enabled GPL3 operations.
Replace custom property redirection for meta-ops with GBinding.
Reduced temporary buffer allocation overhead in mipmap downscaling.
Do divide and conquer on higher level in point filter/composer op base-classes.
Migrated from intltool to GNU gettext
Buffer
Tweak nohalo and lohalo samplers to be orders of magnitude faster for scale factors > ~3% while also removing possible hangs.
Operations
Improved threading awareness/capabilities of gegl:copy-buffer, gegl:wind, gegl:color-enhance, gegl:photocopy, gegl:matting-levin, npd, contrast-curve, noise-spread, noise-pick, noise- slur, transform ops.
Use of more modern GLib APIs like g_clear_ and make use of the fact that g_free is already NULL safe.
Fixed grayscale behavior of gegl:noise-hurl. OpenCL vs CPU consistency fix for gegl:threshold, and optimization of gegl:wavelet-blur by decomposing to two 1d passes. median-blur, perform binning in a more perceptual space. hue-chroma: added OpenCL support threshold: make opencl match cpu wavelet-blur: decompose for performance illusion: performance improvement permitting partial result computation grey: add fast path for alpha-less input buffers.
new ops in workshop: spherize, color-to-alpha-plus, shadows-highlights
With contributions from:
Debarshi Ray, Ell, Jehan, Jordi Mas, Marco Ciampa, Massimo, Nicolas Robidoux, Niels De Graef, Øyvind Kolås, Piotr Drąg and Thomas Manni.
GEGL-0.3.20 2017-08-15
Core
Increased max thread limit to 64 threads, and improvements to both buffer and progress reporting thread safety, point sized buffers are handled better and COW-tiles are now lock free. Serialization and parsing has been made more robust.
Operations
Improved property meta data for many opertions, two new operations gegl:spiral and gegl:mix.
ff-load/ff-save: make audio frame handling more generic.
c2g, avoid possible infinite loop on transparent areas.
Fixes to gaussian-blur and ops core for multi-threaded blurring.
New binary
gcut, a video editing engine for GEGL has also been added to the repo. It permits reusing image filters written as GEGL operations for video, and basic motion graphics. The engines data model is currently fully functional for text-file based input and offline rendering, an experimental user interface for this video editing system is also currently being worked on.
Contributors
Øyvind Kolås, Michael Natterer, Jehan, Dimitris Spingos, Piotr Drąg, Alexandre Prokoudine, Eugene Hopkinson, Ell, Thomas Manni, Jordi Mas, Edward E, Jan Vesely, Arfrever Frehtes Taifersar Arahesis.
GEGL-0.3.18 2017-05-31
Operations
warp: performance improvements displace, map-absolute, map-relative: abyss handling and more buffer-source: improved invalidation signalling text,path: use endian agnostic cairo pixel formats opencl implementations of: noise-hsv, diffraction-patterns, gaussian-blur-selective and motion-blur-linear.
Core
rerrange connection handling for nodes when changing operation. gegl-path: improvements for corner cases found in warp add validation of roi in gegl_operation_get_bounding_box
Contributors
Ell, Anton Gorenko, Øyvind Kolås, Piotr Drąg, Mario Blättermann
GEGL-0.3.16 2017-05-15
Core
Improvements to gegl-chain serialization and documentation. Remove remnants of deprecated gegl-graph syntactic sugar for C
Buffer
Avoid choking on 0x0 rectangles, handle implicit mipmap generation for 15bpc and half precision float pixelformate.
Operations
Many operations have gotten an additiona meta-ui description layer sprinkled on top. And The C source code of the operations is no longer kept as a meta-data key on the binary, with a shrink of ~10mb it might even impact startup time.
New operations: newsprint/spachrotyzer. Moved out of workshop: unpremultiply New in workshop: spiral drop-shadow: improve edge mode of blur, to work on tiny radiuses. transform-ops: ported to the new chanting system ff-load: readjust frame accurate mp4 seeking for newer ffmpeg. file-loading ops: stop some crashes happening due to invalid/lacking properties. npy-save, jpg-save, png-save: port to GIO. raw-load: rotate bounding box based on image meta-data.
Tests
Improve testing by using hashes of autogenerated reference images for regression testing, thus causing fewer manual steps to add an op with its tests. Some XML based tests removed as they were fully superceded by new automated tests.
Documentation
Replaced dynamic client side js content with static operation gallery, incremental improvements to website content. This includes updates to the meta-data in the .C files of operations.
Contributors
Daniel Mustieles, Debarshi Ray, Dimitris Spingos (Δημήτρης Σπίγγος), Ell, Jan Vesely, Jean-Baptiste Mayer, Marco Ciampa, Mario Blättermann, Martin Blanchard, Michael Natterer, nick black, Øyvind Kolås, Piotr Drąg, Thomas Manni,
GEGL-0.3.14 2017-02-27
Buffer
-
changed gegl_buffer_set to accept mipmap level scaled rectangles, similar to gegl_buffer_get and gegl_buffer_iterator_new/_add
Operations
-
made mipmap preview rendering capable: gblur-1d/gaussian blur, sinus, transform (rotate, scale, perspective etc), snn-mean
-
noise-perlin: remove unused random seed property
-
exposure: remove gamma property
-
save-pixbuf: allocate less temporary memory
-
load-pixbuf: fix rowstride related crasher
Translations
Slovenian, Russian, Greek, Polish and Catalan translations updated.
Contributors
Alexandre Prokoudine, Debarshi Ray, Dimitris Spingos (Δημήτρης Σπίγγος), Jordi Mas, Martin Srebotnjak, Øyvind Kolås
GEGL-0.3.12 2017-02-13
Operations
-
new ops: edge-neon, image-gradient, slic, wavelet-blur, waterpixels, watershed
-
moved from workshop to common: color-warp, component-extract
-
text: remove now unneeded work-around, ability to control vertical positioning, permit <1.0 font-sizes, handle text-color alpha, other improvements.
-
lens-distortion: default to transparent background
-
crop: bounding box computation simplifications
-
noise-rgb: add gamma and distribution properties
-
dither: renamed from color-reduction and improved ui/property controls.
-
high-pass: do inversion, over and contrast in non-linear RGB
-
noise-rgb: new linear and gaussian properties
-
transform: added a clip-to-input property
-
raw-load: improvements to handling of Sony’s ARW files
-
exposure: replaced offset with black-level
-
moved from common to workshop: bilateral-filter-fast
-
new workshop ops: bayer-matrix, linear-sinusoid, shadows-highlights, integral-image, segment-kmeans,
-
removed ops: gaussian-blur-old
Core
-
less locale dependent serializations/parameters
-
fix local raw file detection of ARW and CR2 files
-
gegl_memset_pattern performance improvement
-
clean up the way we drop references and free memory
-
static caching of some frequently used babl formats/types.
-
mipmap preview render code fixes for the following subset of operations: point operations (filter, and composer subclasses), integer translate, crop.
Translations
Catalan, German, Greek, Italian, Polish, Portuguese, Slovenian, Swedish translations updated
Contributors
Piotr Drąg, Marco Ciampa, Sergey "Shnatsel" Davidoff, Ell, Øyvind Kolås, Anders Jonsson, Thomas Manni, Jordi Mas, Michael Natterer, Peter O’Regan, Jehan Pagès, Sebastian Rasmussen, Debarshi Ray, Dimitris Spingos (Δημήτρης Σπίγγος), Martin Srebotnjak,Miroslav Talasek.
GEGL-0.3.10 2016-11-06
buffer
-
stop caching tiles in a singly-linked list; use a hash table only
-
adjust box filtering threshold of linear/cubic samplers
-
add const qualifier to input rect parameter of gegl_tile_backend_set_extent()
-
don’t acquire tile storage lock in NN sampler when not using multithreading
operations
-
rgbe-load: add image/x-hdr mimetype
-
map-relative: sample from the pixel centers for integer vectors
-
warp: fix and rework the operation, now matches more closely the iWarp behaviour
-
color-reduction: add arithmetic dithers; inline for loop for each different method
binary
speed up some commands by not intializing opencl
json
make bundleable
translations
German, Greek, Italian, Occiatan, Polish, Portuguese, Russian, Serbian, Slovenian, Spanish updated
Contributors
Simon Budig, Marco Ciampa, Piotr Drąg, Ell, Øyvind Kolås, Roman Lebedev, Daniel Mustieles, Michael Natterer, Jehan Pagès, Martin Pelikán, Yolanda Álvarez Pérez, Alexandre Prokoudine, Tiago Santos, Dimitris Spingos, Martin Srebotnjak, Cédric Valmary, Милош Поповић.
GEGL-0.3.8 2016-06-14
Buffer
-
add box filtering downsampling paths to linear and cubic resamplers to improve default results
-
clear hot tile when removing a tile from the cache
operations
-
added: vhsfix (workshop), gegl-gegl (commandline syntax)
-
exr-save: port from deprecated chanting system
-
ff-save: increase max output packet size from 200kb to 1mb
-
gegl-path: add ability to compute y coordinate for a given x
-
transform: treat infinite and empty rectangles correctly
-
webp: port webp-save to GIO for URI handling
-
ops progress:
-
add gegl_operation_progress function to report processing progress
-
use the new function in cartoon and distance-transform ops
-
ops metadata:
-
remove custom multiline paramspec
-
set multiline ui_meta on more ops
-
add reference meta and use it in various ops
gegl-binary
-
add simple video filtering
-
encode audio with video
-
permit dragging and zooming in more ui states
-
deal correctly with integers in ui
-
show prop editor for operation if op chain passed at commmandline
core
-
gegl-chain
-
implement keyframing of strings
-
suggest possible ops in warninge/errors
-
gegl-node: relay property change notifies from operation to node
-
gegl-serialize:
-
improve error handling for parsing of properties
-
start handling relative coordinates
-
use path instead of curve
-
json: misc improvements and fixes
-
opencl: fix RGB luminance constants
-
docs: fixes, update some info in directory overview
-
handlers:
-
allow registration of operations by MIME type
-
register MIME types for loaders
-
guess file types by sniffing content
-
build: fix luajit detection
Contributors
Timm Bäder, Martin Blanchard, Mario Blättermann, Ell, Piotr Drąg, Michael Henning, Øyvind Kolås, Hartmut Kuhse, Michael Natterer, Nils Philippsen, Alexandre Prokoudine, Debarshi Ray, Kristian Rietveld, Dimitris Spingos, Martin Srebotnjak, Pino Toscano, Cédric Valmary, Jan Vesely
GEGL-0.3.6 2016-03-12
buffer
COW (Copy on Write) handling for gegl_buffer_clear.
operations
-
added: saturation, gaussian-blur-selective
-
ff-load/ ff-save: updated to work with ffmpeg-3.0,configuration of codec/format. misc code cleanup, and meta data improvements, gio (url loading) support in more loaders.
-
convolve-generic: optimized paths for common parameters and 3x3 size.
-
removed: dcraw-load
-
jp2, webp, svg:: make use of GIO for URI handling
-
XML: permit serializing a segment/chain of the graph
-
gegl-binary:
-
permit setting properties of commandline passed ops, as well as connecting buffers to aux and other input pads.
-
zoom in/out at cursor position on scroll wheel events.
-
processing: improvements to meta-ops
Contributors
Øyvind Kolås, Thomas Manni, Michael Natterer, Nils Philippsen, Debarshi Ray, Martin Blanchard, Justin Lecher, Massimo Valentini, Jonathan Tait, Mukund Sivaraman, Alexandre Prokoudine, Piotr Drąg, Nanley Chery, Colin Waters
GEGL-0.3.4 2015-11-24
-
build: do not install examples in path
-
ff-save fully store all in-flight codec contexts before closing files
-
ff-load improvements to seeking accuracy
-
transform: make fast paths skip pixel format conversions
Contributors
Øyvind Kolås, Debarshi Ray and Nils Philippsen
GEGL 0.3.2 2015-11-22
Operations
-
new operations: libraw based raw loading op, tiff-save and tif-load, maze, sepia
-
ff-load and ff-save revived, with support from thegrid.io
-
apply-lens uses less memory, higher precision computation.
-
disable automatic threading on many ops where it fails
-
force more operations to prefer operating on linear RGB data for more accurate/physical processing.
Buffer:
-
implement abyss paremeter on gegl_buffer_copy and gegl_buffer_blit
-
Added start of a microraptor gui based image viewer/non destructive editor.
-
Optimizations to scaled blitting (speeds up most GEGL UIs a bit)
Contributiors
Alexandre Prokoudine, André Tupinambá, Claude Paroz, Daniel Mustieles, Debarshi Ray, Dimitris Spingos, Elle Stone, Jehan, Jordi Mas, Marco Ciampa, Martin Blanchard, Martin Srebotnjak, Massimo Valentini, Michael Henning, Michael Natterer, Necdet Yücel, Pedro Albuquerque, Piotr Drąg, Roman Lebedev, Sven Neummann, Thomas Manni, Vilson Vieira, akash akya and Øyvind Kolås.
GEGL-0.3.0 2015-06-03
-
Improvements to thread safety and parallelism.
-
Lower overhead graph travesal due from rewrite of visitors
-
OpenCL support now enabled by default when detected.
-
Experimental multithreading, enable by setting GEGL_THREADS=<number of threads> in the environment.
-
Experimental mipmap rendering, which permits transparent rendering of previews on smaller sized versions, enable by setting GEGL_MIPMAP_RENDERING=true in the environment.
Operations
-
new operations: alien-map, antialias, apply-lens, bilateral-filter, bump.map, cartoon, channel-mixer, color-enhance, color-exchange, color-reduction, color-rotate, convolution-matrix, copy-buffer, cubism, deinterlace, diffraction-patterns, distance-transform, displace, edge, emboss, engrave, exposure, fractal-trace, high-pass, image-compare, illusion, invert-gamma, lens-flare, linear, linear-gradient, mosaic, motion-blur-circular, motion-blur-zoom, noise-cell noise-cie-lch, noise-hsv, noise-hurl, noise-pick, noise-rgb, noise-simplex, noise-spread, n-point deformation ops, oilify, panorama-projection, photocopy, plasma, radial-gradient, red-eye-removal, scale-size-keep-aspect, softglow, stretch-contrast, texturize-canvas, tile-glass, tile-seamless, tile-paper, tile, warp, whirl-pinch, wind, cache, cast-format, lcms-from-profile, npy-save, webp-load, webp-save, scale-ratio, scale-size, seamless-clone, sinus, supernova, value-propagate, video-degradation
-
reimplementation of gaussian-blur faster and more accurate
-
support for using URIs in image loaders
Buffer
-
New default tile backend, doing disk writes in a separate thread.
Contributors
Albert F, Alexandre Prokoudine, Alexia Death, Akash Akya, Anders Jonsson, Andika Triwidada, Andreas Fischer, Angh, Awaw Fumin, Barak Itkin, Bruce Cowan, Carlos Zubieta, Cédric Valmary, Chris Leonard, Christian Kirbach, Clayton Walker, Daniel Mustieles, Daniel Nylander, Daniel Sabo, Debarshi Ray, Denis Knoepfle, Dimitris Papavasiliou, Dimitris Spingos, Djavan Fagundes, Dov Grobgeld, Elle Stone, Enrico Nicoletto, Felix Ulber, Florian Klemme, Francisco Vila, Fran Diéguez, Georges Basile Stavracas Neto, Hans Lo, Harald Korneliussen, Hartmut Kuhse, Inaki Larranaga Murgoitio, Isaac Wagner, Jan Vesely, Jan Vesely, Jehan, Jon Nordby, Jordi Mas, Kalev Lember, Kristjan Schmidt, Marco Ciampa, Marek Dvoroznak, Maria Mavridou, Martijn van Beers, Martin Nordholts, Martin Srebotnjak, Massimo Valentini, Matej Urbančič, Maxime Nicco, Michael Henning, Michael Muré, Michael Natterer, Mikael Magnusson, Miroslav Talasek, Muhammet Kara, Mukund Sivaraman, Nana Chery, Nick Black, Nicolas Robidoux, Nils Philippsen, Norm Murray, Pascal Giessler, Piotr Drąg, Quentin Glidic, Rafael Ferreira, Rasmus, RPG, Rūdolfs Mazurs, Samir Ribic, Samuel Pitoiset, sebul, Simon Budig, Sven Claussner, Téo Mazars, Thomas Manni, Tim Lunn, Tim Mooney, Ting-Wei Lan, Tom Stellard, Ulf-D. Ehlert, Vadim Rutkovsky, Victor Oliveira, Ville Sokk, Vincent Untz, Yongjia Zhang, Yongjia Zhang, Øyvind Kolås and 周 周.
GEGL-0.2.0 2012-04-02
-
OpenCL support
-
build improvements.
-
High level API to apply ops directly to buffers with arguments.
-
Final bits of translation infrastructure.
-
Invalidate regions when disconnecting input pads.
Operations
-
New operation: global-matting
-
Allow transform core to do perspective transforms.
-
Added string based key/value pairs to operations.
-
Added arguments for dealing with scaled down preview rendering.
-
Added human interaction ranges and non-linear mapping to properties.
Buffer
-
Removed broken lanczos sampler.
-
Add gegl_buffer_set_color and gegl_buffer_set_pattern
-
Added ability to drop cached tiles.
-
Added API for handling abyss policy (not implemented yet)
-
Avoid iterating global tile cache when flushing/destroying buffers that have no tiles in the cache.
Contributors
Øyvind Kolås, Martin Nordholts, Hans Petter Jansson, Mikael Magnusson, Victor Oliveira, Nils Philippsen, Kevin Cozens, Michael Muré, Jan Rüegg, Michael Natterer, Michael Henning, Massimo Valentini, Alexandre Prokoudine and Jon Nordby.
GEGL-0.1.8 2011-11-18
-
New operations: spread, vignette, map-relative, noise-reduction, plasma, fractal-trace, exr-save, lens-correct, emboss, cubism, ripple, color-to-alpha, color-rotate, red-eye-removal, convolution-matrix, deinterlace, polar-coordinates, lens-distortion, pixelise.
-
Split GeglView GTK Widget into separate utility library
-
build/test improvements.
Buffer
-
Added lohalo resampler
-
API and infrastructure for doing non-affine resamplings.
Contributors
Øyvind Kolås, Michael Natterer, Barak Itkin, Nils Philippsen, Mukund Sivaraman, Jon Nordby, Emilio Pozuelo Monfort, Étienne Bersac, Martin Nordholts, Victor Oliveira, Michael Muré, Michael Schumacher, Adam Turcotte, Nicolas Robidoux, Robert Sasu, Massimo Valentini, Joao S. O. Bueno, Hans Lo, Zbigniew Chyla, David Evans and Javier Jardón.
GEGL-0.1.6 2011-02-13
-
New operations: max-rgb, pixelise, motion blur.
-
Fixed a bug in matting-levin that made GEGL halt due to errors detected by babl sanity code, this made 0.1.4 be unusable if you had all dependendencies when building.
-
build/test improvements.
Buffer
Added API to use external tile backends, allowing to plug-in alien tilebackends, for GIMP/Krita/OSM or similar.
Contributors
Tobias Ellinghaus, Øyvind Kolås, Barak Itkin and Martin Nordholts
GEGL-0.1.4 2011-01-20
Operations
-
save (chooses delegate save op automatically)
-
rgbe load/save
-
jpeg200 load/save
-
ppm load/save
-
map-absolute (GIMP cage tool gsoc helper op)
-
whirl and pinch
-
mirrors
-
grid render
-
fixed imagemagick fallback load op
-
fixed pipe based dcraw wrapper
-
GSoC 2010 ops: exp-combine, reinhard05, fattal02, mantiuk06, matting-levin
-
Various source and build improvements.
-
improvements towards threaded rendering
Buffer
-
refactored away some constant overhead made GeglBufferIterator API public
Contributors
Michael Natterer, Øyvind Kolås, Vincent Untz, Kaja Liiv, Nils Philippsen, Étienne Bersac, Martin Nordholts, Debarshi Ray, Danny Robson, Stuart Axon, Kao, Mukund Sivaraman, Ruben Vermeersch, Barak Itkin, Michael Muré, Mikael Magnusson, Patrick Horgan and Andy Gill.
GEGL-0.1.2 2010-02-07
-
GeglLookup, configurable floating point lookup tables for lazy computation.
-
Use GFileIOStream in GeglTileBackendFile.
-
Optimizations: in-place processing for point filters/composers, SIMD version of gegl:opacity, avoid making unneccesary sub-buffers, removed some manual instrumentation from critical paths, improved speed of samplers.
-
Added xml composition/reference image based regression tests.
-
Added performance tracking framework.
-
Syntactic sugar using varargs for constructing gegl graphs from C.
-
Build fixes on cygwin,
-
Gegl# fixes.
-
Initial, but unstable code towards multithreading.
-
Improvements to lua op in workshop.
-
Added new resamplers upsize, upsharp, upsmooth, downsize, downsharp and downsmooth.
-
Removed gegl:tonemap and gegl:normal ops.
Contributors
Martin Nordholts, Sven Neumann, Nils Philippsen, Adam Turcotte,
Danny Robson, Michael Schumacher, Ruben Vermeersch, Øyvind Kolås,
Javier Jardón, Yaakov Selkowitz and Michael Natterer, Kaja Liiv
and Eric Daoust.
GEGL-0.1.0 2009-07-01
-
Renamed gegl:load-buffer to gegl:buffer-source and gegl:save-buffer to gegl:buffer-sink (but the old names still work)
-
Represent colors using doubles instead of floats (this change is independent from internal processing)
-
Removed the GTK+ UI parts of the gegl binary and turned gegl into a pure command line tool (which can still visualize stuff with help help the SDL based display operation)
-
Consider G_MININT/2, y=G_MININT/2, width=G_MAXINT, height=G_MAXINT as the only valid region in which processing may occur. Processing outside of this region is undefined behaviour.
-
Added support for storing allocation stack traces for GeglBuffers so that debugging buffer leaks becomes much easier
-
Made small changes and cleanups of the public API, e.g.
-
Removed gegl_node_adapt_child()
-
Made GeglConfig an explicit object
-
Removed most of the ifdeffed stuff to mask away internal structures
-
Added gegl_rectangle_infinite_plane() and gegl_rectangle_is_infinite_plane()
-
Added new sampler GeglSamplerSharp
-
Added format property to gegl:buffer-sink
-
Cleaned up and made gegl:introspect work again
-
Add a bunch of test cases using the automake test system (make check) and also port buffer tests to automake
-
General cleanups, bug fixes, increased robustness and improved documentation
Contributors
Øyvind Kolås, Sven Neumann, Étienne Bersac, Hubert Figuiere, Martin Nordholts, Nicolas Robidoux, Michael Natterer, Ruben Vermeersch, Kevin Cozens, Manish Singh, Gary V. Vaughan, James Legg, Henrik Åkesson, Fryderyk Dziarmagowski, Ozan Caglayan and Tobias Mueller.
GEGL-0.0.22 2008-12-31
GeglOperation
-
operation names are now prefixed, the ops in GEGL use gegl: as prefix.
-
gegl:opacity - combine value and aux mask input when both are available.
-
gegl:src-in - deal correctly with extens.
-
gegl:path - new op covering the stroke/fill needs of SVG.
-
deprecated gegl:shift, the affine family of operations now uses the same fast code paths for integer translations.
GeglBuffer
-
Profiling motivated speed ups in data reading/writing.
-
Remove left-over swapfiles from dead processes at startup. • GeglNode
-
made gegl_node_add_child and gegl_node_remove_child public API. (#507298)
GeglPath
Vector path representation infrastructure, supporting poly lines and beziers by default, the infrastructure allows extensions from applications with other curve types (smooth curves, spiro curves and others.).
Contributors:
Hubert Figuiere, Sven Neumann, Øyvind Kolås, Michael Natterer, Kevin Cozens, Sam Hocevar, Martin Nordholts, Manish Singh, Étienne Bersac and Michael Schumacher.
GEGL-0.0.20 2008-10-05
-
Build and code clean ups and fixes.
-
RAW loader using libopenraw.
GeglBuffer
-
Linear buffer support, amongst other things enabling GeglBuffer API access to external linear buffers in memory.
-
Reworked samplers using a shared caching neighbourhood infrastructure.
-
YAFR - a new resampler contributed by Nicolas Robidoux.
GeglVector
Added SVG parsing capability (no serialisation yet).
GeglOperations
-
Marked user visible strings for translation.
-
Added a fill operation that renders a path.
Contributors
Martin Nordholts, Øyvind Kolås, Sam Hocevar, Manish Singh, Hubert Figuiere, Sven Neumann and yahvuu at gmail.com.
GEGL-0.0.18 2008-06-14
-
Configuration both from commandline arguments and at runtime.
GeglBuffer
-
New raw tiled raster file format, used both as swap and stored buffers.
-
Sharing between processes through synced swap.
-
Babl powered scan iteration infrastructure for efficient access.
-
Cubic and lanczos samplers re-enabled.
Operations
-
Use scan iterator for point-filter, point-composer and point-render base classes internally for minimal amount of copies.
-
Optimized variants of some point and compositing operations reimplemented using a new data type /g4float/ that allows writing CPU agnostic vectorized code for GCC.
-
New temporal filter base class, for operations operating on color values from neighbouring frames in a video stream.
-
Autogenerated operation reference installed for use with devhelp.
-
New operations: write-buffer, v4l, color-temperature.
Contributors
Øyvind Kolås, Kevin Cozens, Sven Neumann, Manish Singh, Martin Nordholts, Étienne Bersac, Hans Petter Jansson, Jan Heller, dmacks@netspace.org, Sven Anders, Hubert Figuiere and Geert Jordaens.
GEGL-0.0.16 2008-02-29
-
Install headers for development of external operation plug-ins.
-
Added rowstride argument to gegl_buffer_set ()
-
Made it possible for sink operations to do chunkwise processing through GeglProcessor. when processed using a GeglProcessor.
-
constification of GeglRectangles and babl formats.
-
Build and dist fixes, plugged annoying buffer leaks, GEGL now also works on OSX
Contributors
Øyvind Kolås, Étienne Bersac, Kevin Cozens, Sven Neumann, Manish Singh, Michael Natterer, Hans Brauer, Deji Akingunola, Bradley Broom and Tor Lillqvist.
GEGL-0.0.14 2007-12-20
-
Relicensed under LGPLv3+.
-
Per node caches (faster recomputation when properties of nodes in composition change)
-
Sampler framework for interpolation.
-
Modified API for retrieving scaled buffers gegl_node_blit / gegl_buffer_get.
-
Renamed XML serialization and parsing API to be descendants of GeglNode.
-
Progress monitoring for GeglProcessor
-
New operation base class GeglOperationAreaFilter
-
API reference installed and browsable using devhelp.
-
New operations: ff-load, mono-mixer, contrast-curve, save-pixbuf, compositing ops from SVG-1.2 draft, value-invert, convert-format, bilateral-filter, snn-mean
Contributors
Étienne Bersac, Kevin Cozens, Manish Singh, Mark Probst, Martin Nordholts, Michael Schumacher, Mukund Sivaraman, Shlomi Fish, Tor Lillqvist and Øyvind Kolås
GEGL-0.0.12 2007-02-18
-
Swapping out of image pyramid.
-
Speedups to text rendering.
-
GEGL compiles on win32
-
GEGL_SWAP now specifies swapdir location.
-
Small refactorings of public API.
-
Changed the XML parser/serializer to not use attributes for node properties.
-
Improvements to documentation.
-
Automagic build environment for operations simplified.
-
Internal API for saving/loading GeglBuffers to/from disk.
-
Ruby, C# and Python bindings added to version control system (not included in distributed tarball)
Contributors
Kevin Cozens, Michael Schumacher. Sven Neumann and Øyvind Kolås,
GEGL-0.0.6 2007-01-07
-
hit detection code
-
handling of relative/absolute path names for XML.
-
merged GeglNode and GeglGraph classes.
-
build fixes for Win32 and FreeBSD
-
temporarily disabled custom cairo in test application.
-
png and EXR improvements.
Contributors
Dominik Ernst, Kevin Cozens, Øyvind Kolås and Michael Schumacher.
GEGL-0.0.4 2006-12-29
This was the first public release of GEGL.
-
8bit, 16bit integer and 32bit floating point, RGB, CIE Lab, YCbCr and naive CMYK output.
-
Extendable through plug-ins.
-
XML, C and Python interfaces.
-
Memory efficient evaluation of subregions.
-
Tiled, sparse, pyramidial and larger than RAM buffers.
Operations
-
PNG, JPEG, SVG, EXR, RAW and other image sources.
-
Arithmetic operations, porter duff compositing operations, SVG blend modes, other blend modes, apply mask.
-
Gaussian blur.
-
Basic color correction tools.
-
Most processing done with High Dynamic Range routines.
-
Text layouting using pango
Contributors
Dominik Ernst, Kevin Cozens, Manish Singh, Martin Nordholts, Michael Natterer. Philip Lafleur, Sven Neumann and Øyvind Kolås.
GEGL-0.0.0
Contributors:
Calvin Williamson, Caroline Dahloff Daniel Rogers, Garry R. Osgood, Jay Cox and Sven Neumann