Øyvind Kolås [Tue, 16 Jan 2018 17:38:49 +0000 (18:38 +0100)]
release 0.1.40
Øyvind Kolås [Wed, 17 Jan 2018 02:57:23 +0000 (03:57 +0100)]
babl: add a test iteration multiplier
On my system many of the conversions got measured to costs of 3 4 and 5, these
makes a big room for measurement errors, to increase reliability the tests are
now run 4 times - making it a ranking of contenders with fewer collisions.
This will slightly increase incurred delayed on search for conversions.
Øyvind Kolås [Wed, 17 Jan 2018 02:46:59 +0000 (03:46 +0100)]
babl: make pixel-count instrumentation opt-in
Checking if an integer is 0/1 is lower overhead than incrementing a long stored
kt a memory location. As a side effect, the entries in
~/.cache/babl/babl-fishes will not be sorted by most used to least used unless
the BABL_INSTRUMENT environment variable is set to a value.
Øyvind Kolås [Tue, 16 Jan 2018 17:34:46 +0000 (18:34 +0100)]
babl: set user_data ptr for quick dispatch of 1 step path
Øyvind Kolås [Tue, 16 Jan 2018 17:12:47 +0000 (18:12 +0100)]
babl: indentation and code cleanup
Øyvind Kolås [Mon, 15 Jan 2018 21:41:03 +0000 (22:41 +0100)]
babl: use memcpy dpsatch path for all identity transforms
Øyvind Kolås [Mon, 15 Jan 2018 13:21:59 +0000 (14:21 +0100)]
babl: reduce amount of instrumentation
The integer counter for number of processings overflows in some scenarios, thus
cannot be relied on for more than being informative.
We might ultimately want to also drop the instrumentation.
Øyvind Kolås [Sun, 14 Jan 2018 22:55:53 +0000 (23:55 +0100)]
babl: store pointer to where user data resides instead of copy
This will make the scenario of changing user data work, we do it
unconditionally to avoid introducing a branch.
Øyvind Kolås [Sun, 14 Jan 2018 21:49:40 +0000 (22:49 +0100)]
babl: refactor code to get rid of more branches in babl_process
In the best case scenario now, babl_process increments instrumentation
counters and directly calls the relevant registered fast path function,
with no additional call stack frames in-between.
Øyvind Kolås [Sun, 14 Jan 2018 21:15:12 +0000 (22:15 +0100)]
babl: use babl_conversion_process where appropriate
Instead of overloading babl_process() with two tasks, both fishes and
conversions, use the conversion specific API for conversions directly.
Øyvind Kolås [Sun, 14 Jan 2018 20:44:18 +0000 (21:44 +0100)]
babl: simplify branching in babl_process dispatch
Øyvind Kolås [Sun, 14 Jan 2018 18:19:30 +0000 (19:19 +0100)]
babl: factor out some branches from path fish dispatch
Øyvind Kolås [Sun, 14 Jan 2018 17:36:26 +0000 (18:36 +0100)]
babl: refactor conversion dispatch to do less branches at runtime
Øyvind Kolås [Sat, 13 Jan 2018 14:43:01 +0000 (15:43 +0100)]
babl.map: update symbols for meson build
Øyvind Kolås [Sun, 7 Jan 2018 00:10:54 +0000 (01:10 +0100)]
babl: add babl_process_rows api
Useful when fetching buffers smaller than a tile or partial tiles in APIs like
gegl_buffer_get.
Ell [Sat, 13 Jan 2018 09:02:24 +0000 (04:02 -0500)]
sse2-float: use same TRC for "leftover" samples
The SSE2 versions of the gamma <-> linear TRC functions can produce
slightly different results than the scalar ones. Consequently,
last commit didn't resolve all discrepancies between the main and
leftover samples (and didn't address the gamma -> linear
direction).
Use the same SSE2 TRC functions for converting the leftover
samples, rather than the scalar functions, to fix that.
Ell [Thu, 11 Jan 2018 14:22:46 +0000 (09:22 -0500)]
sse2-float: use babl_linear_to_gamma_2_2f() (instead of double)
Use the single-precision babl_linear_to_gamma_2_2f() function,
instead of the double-precision version, when converting "leftover"
samples, so that the result is consistent with the vectorized
samples.
Ell [Thu, 11 Jan 2018 14:19:50 +0000 (09:19 -0500)]
util, sse2-float: make sure 1 maps to 1 in linear -> gamma conversions
Add a small offset to the result of the linear -> gamma
conversions, such that an input value of 1.0 maps to an output
value of 1.0, exactly.
Øyvind Kolås [Sun, 7 Jan 2018 04:37:39 +0000 (05:37 +0100)]
babl: shuffle some dispatch functions to be inlineable
Øyvind Kolås [Sat, 6 Jan 2018 23:42:33 +0000 (00:42 +0100)]
babl: strip out some internal length returning code
Øyvind Kolås [Sat, 6 Jan 2018 23:35:22 +0000 (00:35 +0100)]
babl: prepare babl_fish_process for direct calling
Øyvind Kolås [Wed, 3 Jan 2018 16:12:41 +0000 (17:12 +0100)]
palette: expect palettes/formats to be gamma corrected in 8bit
A correctness correction related to bug #763581 the formats wanted
by GIMP and expected are gamma corrected not linear, I expect this
fix either does not much for GIMP or fixes an unknown 8bit linear
vs gamma quantization problem that has been lurking in INDEXED mode.
Øyvind Kolås [Tue, 2 Jan 2018 20:05:02 +0000 (21:05 +0100)]
babl: improve babl_trc_formula_srgb sRGB detection constants
For bug #791816. Use constants as passed from GIMP when its built in sRGB ICC
profile is passed.
Debarshi Ray [Thu, 21 Dec 2017 09:14:53 +0000 (10:14 +0100)]
CIE: Use a faster cbrtf implementation
This is the approximate cube root of an IEEE float implementation from
Hacker's Delight. The elimination of all conditional branches probably
makes it a better candidate for future SIMD accelerated code paths.
On an Intel i7 Haswell, it now takes 0.27s to convert a 15 megapixel
buffer from "RGBA float" to "CIE Lab alpha float" instead of the
earlier 0.35s. A "Y float" to "CIE L float" conversion takes 0.085s
instead of 0.102s.
Original code: http://www.hackersdelight.org/hdcodetxt/acbrt.c.txt
Permissions: http://www.hackersdelight.org/permissions.htm
https://bugzilla.gnome.org/show_bug.cgi?id=791837
Øyvind Kolås [Wed, 20 Dec 2017 02:22:22 +0000 (03:22 +0100)]
extensions-gggl: clamp when converting float to 16bit
Improve the quality of a conversion that was picked in error for bug #790658
Øyvind Kolås [Wed, 20 Dec 2017 02:22:00 +0000 (03:22 +0100)]
babl: re-enable earlier bailing for big path errors
This is one way of dealing with 790658 - that in general is a good idea as
well.
Øyvind Kolås [Tue, 19 Dec 2017 01:13:26 +0000 (02:13 +0100)]
meson.build: babl's license is LGPL3+
Félix Piédallu [Fri, 15 Dec 2017 09:54:49 +0000 (10:54 +0100)]
Simpler library handling. PKGBUILD+release, lto.
Félix Piédallu [Thu, 14 Dec 2017 14:52:57 +0000 (15:52 +0100)]
fix check version
Félix Piédallu [Thu, 14 Dec 2017 14:52:47 +0000 (15:52 +0100)]
Fix library name
Félix Piédallu [Thu, 14 Dec 2017 14:12:25 +0000 (15:12 +0100)]
Added PKGBUILD for archlinux, based on AUR.
Félix Piédallu [Wed, 6 Dec 2017 14:22:41 +0000 (15:22 +0100)]
Remove useless join_paths arg
Félix Piédallu [Sat, 2 Dec 2017 00:03:56 +0000 (01:03 +0100)]
use find_program. Use env for multiple commands as it allows to split into string list.
Félix Piédallu [Fri, 1 Dec 2017 23:50:41 +0000 (00:50 +0100)]
use env instead of bash
Félix Piédallu [Fri, 1 Dec 2017 14:57:41 +0000 (15:57 +0100)]
Use formatting instead of weird path reconstruction
Félix Piédallu [Fri, 1 Dec 2017 14:31:43 +0000 (15:31 +0100)]
Dispatch the custom_target in two custom_targets for cleaner dependency declaration.
Félix Piédallu [Fri, 1 Dec 2017 13:47:10 +0000 (14:47 +0100)]
Fix push_web, add index_html generation. TODO: simplify this someday
Félix Piédallu [Tue, 28 Nov 2017 14:53:04 +0000 (15:53 +0100)]
Add linker script for symbols export
Félix Piédallu [Sun, 26 Nov 2017 12:51:31 +0000 (13:51 +0100)]
some small cleanup, export script for scp
Félix Piédallu [Thu, 23 Nov 2017 20:43:16 +0000 (21:43 +0100)]
add docs. fix install libs. Seems finished.~
Félix Piédallu [Thu, 23 Nov 2017 17:18:56 +0000 (18:18 +0100)]
add extensions, tools
Félix Piédallu [Thu, 23 Nov 2017 17:12:09 +0000 (18:12 +0100)]
Add tests
Félix Piédallu [Thu, 23 Nov 2017 17:01:10 +0000 (18:01 +0100)]
Fix build, links.
Félix Piédallu [Thu, 23 Nov 2017 16:58:37 +0000 (17:58 +0100)]
Added basic meson files
Øyvind Kolås [Sat, 18 Nov 2017 00:35:09 +0000 (01:35 +0100)]
babl: make conversion to model double more generic
Iterate over the components of the model, and fetch the relevant
component from the source format or use a filler value if not found.
Some changes by Debarshi Ray.
https://bugzilla.gnome.org/show_bug.cgi?id=790860
Øyvind Kolås [Sat, 9 Dec 2017 22:38:38 +0000 (23:38 +0100)]
CIE: refactor away some intermediate variables/assignments
Øyvind Kolås [Sat, 9 Dec 2017 19:51:38 +0000 (20:51 +0100)]
CIE: add CIE XYZ and CIE XYZ alpha models
Even more useful now with custom RGB primaries - and being able to use babl for
conversions when needed is convenient. This is with reference paths - which
means fast paths for double precision float - single precision would have to be
added for it to be used and higher throughput in most common scenarios.
Jehan [Mon, 4 Dec 2017 02:33:47 +0000 (03:33 +0100)]
tests: not all UNIX-like OSes have a libpthread.
In particular Android systems don't need to link with -lpthread
(actually the link would fail with "ld: cannot find -lpthread").
Use the $(THREAD_LIB) variable which is set correctly during configure
since commit
c02af82.
Debarshi Ray [Sat, 18 Nov 2017 01:09:03 +0000 (02:09 +0100)]
CIE: Simplify code
Exploit the fact that fy is not needed to calculate the a and b
components. This is consistent with the other monochrome conversions.
Øyvind Kolås [Wed, 15 Nov 2017 22:18:20 +0000 (23:18 +0100)]
configure.ac: post release version bump
Øyvind Kolås [Wed, 15 Nov 2017 22:16:36 +0000 (23:16 +0100)]
configure.ac: release 0.1.38
Øyvind Kolås [Wed, 15 Nov 2017 22:15:25 +0000 (23:15 +0100)]
NEWS: update
Edward E [Tue, 14 Nov 2017 04:43:29 +0000 (22:43 -0600)]
babl_pow_24f(): add missing float suffix
Debarshi Ray [Thu, 9 Nov 2017 06:48:26 +0000 (07:48 +0100)]
CIE: Add "CIE L float"
Some of these conversions will be leveraged by gegl:shadows-highlights
which needs to go from "Y float" or "YaA float" to "CIE L float".
The conversion from "RGBA float" was added to aid "YaA float" to
"CIE L float" fishes. They go via:
"YaA float" to "RaGaBaA float"
"RaGaBaA float" to "RGBA float"
"RGBA float" to "Y float"
"Y float" to "CIE L float"
A direct conversion from "YaA float" to "Y float" in simple C is
hindered by the need to check every pixel's alpha value to avoid
dividing by zero. The pipeline stalls make it lose out to the look-up
table and SIMD based RGB conversions to unassociated alpha.
However, we can trivially cut out the third step and still reduce some
memory traffic.
https://bugzilla.gnome.org/show_bug.cgi?id=790111
Tobias Stoeckmann [Fri, 13 Oct 2017 16:32:52 +0000 (18:32 +0200)]
babl: fix memory leaks on error paths
The database-focused functions of babl-cache are prone to
memory leaks when errors occur.
Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org>
Øyvind Kolås [Fri, 10 Nov 2017 09:44:21 +0000 (10:44 +0100)]
extensions: protect another conversion against /0.0
Øyvind Kolås [Fri, 10 Nov 2017 09:34:57 +0000 (10:34 +0100)]
extensions: protect gray unpremuls from gggl against division by 0.0
Øyvind Kolås [Fri, 10 Nov 2017 00:08:10 +0000 (01:08 +0100)]
configure.ac: post release version bump
Øyvind Kolås [Fri, 10 Nov 2017 00:05:32 +0000 (01:05 +0100)]
release: 0.1.36
Øyvind Kolås [Fri, 10 Nov 2017 00:03:37 +0000 (01:03 +0100)]
NEWS: update for 0.1.36
Debarshi Ray [Tue, 7 Nov 2017 09:52:44 +0000 (10:52 +0100)]
CIE: Make the Lab to RGB fast paths work with non-sRGB primaries
This is similar to commit
72a074df6545fcb3.
https://bugzilla.gnome.org/show_bug.cgi?id=790011
Debarshi Ray [Tue, 7 Nov 2017 09:38:12 +0000 (10:38 +0100)]
CIE: Make the Lab to RGB fast paths as accurate as the reference
This is similar to commit
e17807f235557d5d.
https://bugzilla.gnome.org/show_bug.cgi?id=790011
Ell [Tue, 7 Nov 2017 20:44:39 +0000 (15:44 -0500)]
babl, sse2-float: fall back to slow, accurate path for large pow-2.4 inputs
The approximations we use for pow_24() and pow_1_24() diverge from
the actual function for large-enough input values. This can lead
not just to inaccurate results, but also to infinities and NaNs,
especially when multiple conversions are strung in a row.
When the input value is large enough to produce notable divergence
(the difference between the approximate and actual values is ~1% at
the chosen limits,) fall back to a slower, but more accurate
version.
For the SSE2 float conversions, this results in an increase of ~5%
in conversion time, when all values are below the limit. When most
values are above the limit, performance can be 10x slower or worse.
Øyvind Kolås [Tue, 31 Oct 2017 19:48:23 +0000 (20:48 +0100)]
export-symbols: add babl_format_exists
Øyvind Kolås [Tue, 31 Oct 2017 19:48:23 +0000 (20:48 +0100)]
babl: add babl_format_exists
A function to check if a specific format - by name - is already known by babl.
Should be used in conjunction with babl_format() for user supplied format
strings.
Debarshi Ray [Tue, 31 Oct 2017 11:20:45 +0000 (12:20 +0100)]
CIE: Make the RGB to Lab fast paths work with non-sRGB primaries
The older constants were tied to sRGB primaries. Using the matrices
from the Babl space removes this restriction.
https://bugzilla.gnome.org/show_bug.cgi?id=789695
Debarshi Ray [Tue, 31 Oct 2017 08:43:53 +0000 (09:43 +0100)]
CIE: Make the RGB to Lab fast paths as accurate as the reference
The matrices used by the fast paths to convert from RGB to XYZ didn't
exactly match the reference matrices in the Babl space. This caused a
measurable error in their output.
https://bugzilla.gnome.org/show_bug.cgi?id=789695
Debarshi Ray [Tue, 31 Oct 2017 08:36:39 +0000 (09:36 +0100)]
CIE: Use cbrt instead of pow for the reference XYZ to LAB conversion
The fast-paths use an inlining-friendly version of cbrt(3). Using
something similar removes superficial differences between the two
conversion paths. It's not like the C library's cbrt(3) will perform
any worse than its own pow(3).
https://bugzilla.gnome.org/show_bug.cgi?id=789695
Debarshi Ray [Sun, 29 Oct 2017 12:52:26 +0000 (13:52 +0100)]
CIE: Add conversion from "RGBA float" to "CIE Lab float"
Conversions from "RaGaBaA float" to "CIE Lab float", as seen when
using gegl:shadows-highlights" go via:
"RaGaBaA float" to "RGBA float"
"RGBA float" to "RGB float"
"RGB float" to "CIE Lab float"
A direct conversion from "RaGaBaA float" to "CIE Lab float" in simple
C is hindered by the need to check every pixel's alpha value to avoid
dividing by zero. The pipeline stalls make it lose out to the look-up
table and SIMD based conversions to unassociated alpha.
However, we can trivially cut out the second step and still reduce
some memory traffic.
https://bugzilla.gnome.org/show_bug.cgi?id=789695
Tobias Stoeckmann [Tue, 24 Oct 2017 18:02:15 +0000 (20:02 +0200)]
babl: use snprintf instead of sprintf
Using sprintf with environment variables is dangerous, because it can
easily lead to out of boundary writes on heap space.
While at it, replace sprintf calls with snprintf where proper
boundary checks are possible and required.
Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org>
Tobias Stoeckmann [Fri, 13 Oct 2017 16:39:02 +0000 (18:39 +0200)]
babl: properly handle large files on 32 bit systems
If large file support is enabled on 32 bit systems, it is possible
to trigger an out of boundary write with files larger than 2 GB.
Always check if fseek and ftell are successful and if the file is
small enough to fit into memory.
Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org>
Tobias Stoeckmann [Fri, 13 Oct 2017 17:25:01 +0000 (19:25 +0200)]
babl: always end strncpy strings with NUL
The function strncpy(3) does not guarantee to end the destination string
with NUL character if not enough space was available. This could happen
on systems which allow paths which are longer than 4096 characters.
Signed-off-by: Tobias Stoeckmann <tobias@stoeckmann.org>
Øyvind Kolås [Fri, 6 Oct 2017 13:16:25 +0000 (15:16 +0200)]
post release version bump
Øyvind Kolås [Fri, 6 Oct 2017 13:15:57 +0000 (15:15 +0200)]
release 0.1.34, update NEWS
Øyvind Kolås [Thu, 5 Oct 2017 21:12:57 +0000 (23:12 +0200)]
export-symbols: add babl_backtrack for gimp-2.8
Øyvind Kolås [Wed, 4 Oct 2017 14:47:39 +0000 (16:47 +0200)]
babl: rearrange BablModel struct
Babl is doing casting that relies on BablFormat / BablModel having the same
layout for shared members.
Øyvind Kolås [Wed, 4 Oct 2017 14:16:45 +0000 (16:16 +0200)]
babl: do not use model.data for storing original model backpointer
Fixing bug #788471
Øyvind Kolås [Tue, 3 Oct 2017 15:31:11 +0000 (17:31 +0200)]
update NEWS
Øyvind Kolås [Tue, 3 Oct 2017 15:17:00 +0000 (17:17 +0200)]
post-release version bump
Øyvind Kolås [Tue, 3 Oct 2017 15:11:44 +0000 (17:11 +0200)]
release 0.1.32
Øyvind Kolås [Tue, 3 Oct 2017 15:14:02 +0000 (17:14 +0200)]
docs: update with api changes
Øyvind Kolås [Tue, 3 Oct 2017 15:06:06 +0000 (17:06 +0200)]
babl: acknowledge some used trademarks
Øyvind Kolås [Tue, 3 Oct 2017 14:41:26 +0000 (16:41 +0200)]
babl: move to/from xyz and rgbtoxyz matrix functions out of babl.h
Both of these are useful, but we might want a wider range, permitting
both linear and non-linear variants to be converted to both XYZ and LAB.
The matrix is for now used in babl for the RGB->Y conversion - this might
warrant a more direct API.
Øyvind Kolås [Tue, 3 Oct 2017 13:54:19 +0000 (15:54 +0200)]
babl: rename babl_space_from_icc to babl_icc_make_space
Thus being more consistent that functions that take icc_data + length have the
prefix babl_icc_, since each icc space in the future potentially can have more
spaces than the relative-colorimetric space defined, _make_ seems better than
_to_ or _new_.
Ell [Tue, 3 Oct 2017 14:01:58 +0000 (10:01 -0400)]
babl: fix clang warnings in babl-polynomial
Cast the input polynomial from BablPolynomial* to
BablBigPolynomial* in the evaluator functions, so that clang
doesn't warn us about out-of-bounds access to its coefficient
array.
Øyvind Kolås [Tue, 3 Oct 2017 13:53:23 +0000 (15:53 +0200)]
HSL: make hue be set in all switch cases for rgb_to_hsl
Edward E [Sat, 30 Sep 2017 19:32:49 +0000 (14:32 -0500)]
babl: revert and clarify libtool flag for win32 builds
Ell [Sat, 30 Sep 2017 11:44:41 +0000 (07:44 -0400)]
babl: improve alpha calc. in pal+alpha -> rgba conversion
... in particular, so that fully opaque pixels remain fully opaque
Ell [Sat, 30 Sep 2017 11:42:56 +0000 (07:42 -0400)]
babl: consistently use unsigned char in u8 palette conversions
... to avoid erroneous sign extensions
Øyvind Kolås [Fri, 29 Sep 2017 23:21:44 +0000 (01:21 +0200)]
sse4-int8: avoid using uninitialized variable
Øyvind Kolås [Thu, 28 Sep 2017 21:44:10 +0000 (23:44 +0200)]
babl: use -no-install to avoid libtool .sh wrappers
Øyvind Kolås [Mon, 25 Sep 2017 22:21:53 +0000 (00:21 +0200)]
babl: fix crash when doing full introspection
also add skeleton functions for introspection of trcs/spaces
Øyvind Kolås [Mon, 25 Sep 2017 22:14:11 +0000 (00:14 +0200)]
babl: do not try to enumerate sampling conversions
Øyvind Kolås [Mon, 25 Sep 2017 20:46:21 +0000 (22:46 +0200)]
babl: set babl_rel_avg_error to a high value on NaN
In relation to bug #787441, if the sum of errors contain a nan the sum
is nan, we return pi as a much higher than expected average error to
indicate that this is not 0.0.
Øyvind Kolås [Mon, 25 Sep 2017 19:41:36 +0000 (21:41 +0200)]
tests/grayscale_to_rgb: increase printed debug precision
Øyvind Kolås [Mon, 25 Sep 2017 19:06:46 +0000 (21:06 +0200)]
tests: move non-test binaries from here to ../tools
Øyvind Kolås [Mon, 25 Sep 2017 19:00:08 +0000 (21:00 +0200)]
babl: check both source/dest bits/component in bad_idea()
Øyvind Kolås [Mon, 25 Sep 2017 18:56:55 +0000 (20:56 +0200)]
babl: add decrease of bits per components to bad_idea()
related to bug #787441
Øyvind Kolås [Mon, 25 Sep 2017 18:33:17 +0000 (20:33 +0200)]
tests: use a epsilon instead of == for float compare
Øyvind Kolås [Mon, 25 Sep 2017 18:04:54 +0000 (20:04 +0200)]
build: link extensions against internal .la again