Commit Graph

2664 Commits

Author SHA1 Message Date
Pascal Massimino
f66955de5f WEBP_REDUCE_CSP: restrict colorspace support
only supported ones are: RGBA/BGRA/rgbA/bgrA (decoder)
as well as: WebPPictureImportRGB/RGBX/RGBA (encoder).

(note: extras/get_disto is affected too)

Change-Id: If6c4f95054ca15759c4e289fb3b4c352b3521c2c
(cherry picked from commit 6de20df02c)
2017-11-28 00:15:15 -08:00
Pascal Massimino
1b27bf8b76 WEBP_REDUCE_SIZE: disable all rescaler code
BUG=webp:355

Change-Id: Id87cb11902e3fb8544a214308526ea9665ce8440
(cherry picked from commit 0df22b9eed)
2017-11-24 22:40:15 -08:00
James Zern
9add62b581 bump version to 0.6.1
libwebp{,decoder} - 0.6.1
libwebp libtool - 7.1.0
libwebpdecoder libtool - 3.1.0

mux - 0.4.1
libtool - 3.1.0

demux - 0.3.3
libtool - 2.3.0

Change-Id: If4a95c6e9829d4a608028ee9258b5c2b7af60c37
2017-11-24 21:03:05 +00:00
James Zern
9ca568ef82 Merge "right-size some tables" 2017-11-24 06:28:02 +00:00
Pascal Massimino
a80c46bd87 SSE2 implementation of HasAlphaXXX
Change-Id: I2548d9a0c252e20ee3cf5f4be736a3703671ecb4
HasAlpha32b: ~3-4x faster
HasAlpha8b: ~7-8x faster
2017-11-23 15:02:21 +01:00
James Zern
083507f244 right-size some tables
Change-Id: I5e894fd3f0f4b666512108495d8cecb34f65e119
2017-11-23 08:40:58 +00:00
James Zern
b299c47eac add WEBP_REDUCE_SIZE
remove auto-filter (-af) support and make WebPPictureCopy,
WebPPictureIsView, WebPPictureView, WebPPictureCrop, and
WebPPictureRescale noops.

Change-Id: If39d512cc268a0015298a1138dbc94feb86575e5
2017-11-22 17:35:39 -08:00
James Zern
f593d71a64 enc: disable pic->stats/extra_info w/WEBP_DISABLE_STATS
Change-Id: I4ca3fa45710bd6bbe131b2ae047b1e268241657c
2017-11-22 17:04:01 -08:00
James Zern
541179a9a5 Merge "predictor_enc: fix build w/--disable-near-lossless" 2017-11-23 01:02:59 +00:00
James Zern
5755a7ec53 predictor_enc: fix build w/--disable-near-lossless
Change-Id: I0d01cdc77e72663f1cf778e3cf3066cd54aa5439
2017-11-22 15:02:21 -08:00
James Zern
eab5bab74f add WEBP_DISABLE_STATS
use to to make WebPPictureDistortion & WebPPlaneDistortion noops and
clear some ssim code.

Change-Id: I9b50b2318b7a114632e5a237a4002f64e95afbbc
2017-11-22 12:41:17 -08:00
Pascal Massimino
c245343dcb move LOAD8x4 and STORE8x2 closer to their use location
Change-Id: I674821732d3e607123070e4bbba87d9359c9a4ec
2017-11-21 23:44:39 -08:00
James Zern
b9e734fd5c dec,cosmetics: normalize function naming style
Change-Id: I33a2d1b4133db7a6d56d506f5c19670f0268cecd
2017-11-21 14:31:34 -08:00
James Zern
c188d546b3 dec: harmonize function suffixes
BUG=webp:355

Change-Id: Iabdfd3fbde906c2e35a7d7c080a8512425eb8ccb
2017-11-21 13:00:25 -08:00
James Zern
28c5ac8104 dec_sse41: harmonize function suffixes
BUG=webp:355

Change-Id: Id55f7b2e6288d1d0885d8451fbc59771222073d6
2017-11-21 12:47:06 -08:00
Pascal Massimino
e65b72a368 Merge "introduce WebPHasAlpha8b and WebPHasAlpha32b" 2017-11-21 06:21:44 +00:00
James Zern
b94cee98fb dec_sse2: remove HE8uv_SSE2
with gcc-4.8, clang-4.0.1/5 this is no faster (actually up to 2x slower)
than the code generated for memset (0x01010... * dst[-1]). shuffles in
sse4 recover a bit, but performance is still down.

Change-Id: Ie85e8353f8ede559d0b05a1d388787fd18ecc80f
2017-11-20 20:34:05 -08:00
Pascal Massimino
44a0ee3fa7 introduce WebPHasAlpha8b and WebPHasAlpha32b
Rewrote WebPPictureHasTransparency() to use them (even for argb).
This is 10% faster, for some reasons.

SSE2 version should be straightforward.
Removes a TODO.

Change-Id: I7ad5848fc5e355e2df505dbcd5a0f42fb6cbab41
2017-11-20 15:20:29 +01:00
Pascal Massimino
aebf59ac50 Merge "WebPPictureAllocARGB: align argb allocation" 2017-11-17 07:27:41 +00:00
James Zern
c184665ecd WebPPictureAllocARGB: align argb allocation
Change-Id: Ib390e8bbb97b38316a38af6a33e8a26bd050ee16
2017-11-16 20:19:09 -08:00
James Zern
3daf7509c2 WebPParseHeaders: remove obsolete animation TODO
The WebPDemux and WebPAnimDecoder APIs are provided for the purpose of
animated webp parsing and decoding. No major changes are currently
planned for the libwebp API.

Change-Id: I2758ecda195b0c4091572d5731a0a85fa3716303
2017-11-16 20:18:21 -08:00
Vincent Rabaud
c462cd0065 Remove useless code.
The casts are to the same type and the #define not used.

Change-Id: I8d69c3b9dde7a1c53c2ba5a026a653d8c2e1d2a7
2017-11-08 10:52:49 +01:00
Pascal Massimino
01a98217ad Merge "remove WebPWorkerImpl declaration from the header" 2017-11-06 20:37:08 +00:00
Pascal Massimino
3c49fc47e7 Merge "thread_utils: fix potentially bad call to Execute" 2017-11-06 20:36:13 +00:00
Pascal Massimino
fde2782ecb thread_utils: fix potentially bad call to Execute
We must use the Interface, and avoid fwd decl.

Change-Id: I18d77a009a29921b6e3694de4df494952b11a83f
2017-11-05 16:39:29 -08:00
Pascal Massimino
2a270c1df5 remove WebPWorkerImpl declaration from the header
BUG=webp:355

Change-Id: Ia4efce4e8f3745e5cdcac495f4a79a8c03062d88
2017-11-05 12:31:48 -08:00
Pascal Massimino
f1f437cc89 remove mention of 'lossy-only parameters' from the doc
it's confusing and mildly accurate.

BUG=webp:355

Change-Id: Ie9667bcdf429f1092b8a523a473391b741164523
2017-11-05 19:38:24 +00:00
Pascal Massimino
3879074d99 Merge "WebPMemToUint32: remove ptr cast to int" 2017-10-31 06:41:05 +00:00
James Zern
04b029d236 WebPMemToUint32: remove ptr cast to int
this can result in an alignment hint on arm causing a SIGBUS. casting
the input ptr to anything aside from its type is unnecessary for memcpy
and is contrary to the intent of this function.

Change-Id: I9a4d3f4be90f80cd8c3e96ccbe557e51e34cf7a5
2017-10-30 17:08:46 -07:00
James Zern
b7971d0e22 dsp: avoid defining _C functions w/NEON builds
when targeting NEON C functions with NEON equivalents won't be used, but
will contribute to binary size. the same goes for sse2, etc., but this
change is primarily concerned with binary sizes for android arm targets.

note '-noasm' or otherwise modifying VP8GetCPUInfo will have no effect
on the use of NEON functions.

this decision can be overridden by defining WEBP_DSP_OMIT_C_CODE to 0.

Change-Id: I47bd453c84a3d341ca39bc986a39eb9c785aface
2017-10-27 10:54:56 -07:00
James Zern
5cfb3b0f6c normalize include guards
some fell out of sync after:
668e1dd4 src/{dec,enc,utils}: give filenames a unique suffix

Change-Id: I280d3b3f44797f3bfb4835784add50a41cdd3793
2017-10-21 00:06:23 -07:00
Pascal Massimino
f433205ee3 Merge changes Ia17c7dfc,I75423abb,Ia2f716b4,I161caa14,I4210081a, ...
* changes:
  {dec,enc}_neon: harmonize function suffixes x2
  upsampling_neon: harmonize function suffixes
  yuv_neon: harmonize function suffixes
  rescaler_neon: harmonize function suffixes
  lossless_neon: harmonize function suffixes
  lossless_enc_neon: harmonize function suffixes
  filters_neon,cosmetics: fix indent
  enc_neon: harmonize function suffixes
  dec_neon: harmonize function suffixes
2017-10-21 06:34:14 +00:00
James Zern
8d033b14d7 {dec,enc}_neon: harmonize function suffixes x2
+ neon.h

BUG=webp:355

Change-Id: Ia17c7dfc7d61742a4758823675a2d556a739c389
2017-10-20 19:00:53 -07:00
James Zern
0295e9815d upsampling_neon: harmonize function suffixes
BUG=webp:355

Change-Id: I75423abbe0bcea3c98a42e412cc2116be81b5d08
2017-10-20 19:00:53 -07:00
James Zern
d572c4e52b yuv_neon: harmonize function suffixes
BUG=webp:355

Change-Id: Ia2f716b459950c18717b062175197d1e6419bf2a
2017-10-20 19:00:53 -07:00
James Zern
ab9c2500db rescaler_neon: harmonize function suffixes
BUG=webp:355

Change-Id: I161caa14f7ebbc3ae978b1722472625a77d0a4a4
2017-10-20 19:00:53 -07:00
James Zern
93e0ce27f4 lossless_neon: harmonize function suffixes
BUG=webp:355

Change-Id: I4210081a39800b5c2589c443da237269908af666
2017-10-20 19:00:53 -07:00
James Zern
22fbc50edd lossless_enc_neon: harmonize function suffixes
BUG=webp:355

Change-Id: I462facaeade4f0f4fc1e96895493306d095a6a9a
2017-10-20 19:00:53 -07:00
James Zern
447875b47b filters_neon,cosmetics: fix indent
BUG=webp:355

Change-Id: I9df1119f1ea94868f75253a92c2e878c9290f744
2017-10-20 19:00:29 -07:00
Pascal Massimino
e51bdd439c remove unused VP8TokenToStats() function
BUG=webp:355

Change-Id: I0ad6f13003ef7201431c54c7db395254191de67c
2017-10-20 04:29:45 -07:00
James Zern
785da7eadd enc_neon: harmonize function suffixes
BUG=webp:355

Change-Id: Ie59efd271d16f12d21f3c800667dfc0980dc2e68
2017-10-20 00:18:32 -07:00
James Zern
bc1a251fcf dec_neon: harmonize function suffixes
BUG=webp:355

Change-Id: I61c9a0c9e24515322955e04afd8c4ea6a44b9319
2017-10-20 00:14:18 -07:00
James Zern
61e535f1ac dsp/lossless: workaround gcc-4.8 bug on arm
and all older versions.
force Sub3() to not be inlined, otherwise the code in Select() will be
incorrect.

extends the check add previously in:
637b3888 dsp/lossless: workaround gcc-4.9 bug on arm

BUG=webp:363

Change-Id: I1403b558f8660b764f3a570a3326822d5ef0be29
2017-10-19 13:05:48 -07:00
Pascal Massimino
30042faa9a WebPDemuxGetI: add doc details around WebPFormatFeature
Change-Id: Id5f17686a8dbd6a6a0ba354d9216a8b89a51597c
2017-10-17 19:59:16 +00:00
Pascal Massimino
0a17f4712c Merge "WIP: list includes as descendants of the project dir" 2017-10-11 08:21:42 +00:00
James Zern
a439972175 WIP: list includes as descendants of the project dir
#include "(.|..)/..." -> #include "src/..."

Change-Id: I772880aa097a770722043c8a4393552ba38a89b6
2017-10-10 23:04:05 -07:00
Pascal Massimino
0827570873 Merge "Make sure we reach the full range for alpha blending." 2017-10-09 09:21:21 +00:00
James Zern
d361a6a733 yuv_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: I02a66f7446c75a10c3ce4766235e5767617d0dce
2017-10-08 14:06:34 -07:00
James Zern
6921aa6f0c upsampling_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: I3a02cc717eb7506bd87511d6a17ab1691e84f72c
2017-10-08 14:06:30 -07:00
James Zern
08c67d3ed1 ssim_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: I1282559888118b8cb0a46b7f0aa627d26b8838f5
2017-10-08 14:06:24 -07:00
James Zern
582a1b572a rescaler_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: I978fd826ff90149c0ffd9d7607dcc6f88082d3e6
2017-10-08 14:06:19 -07:00
James Zern
2c1b18ba2f lossless_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: I59d828800c2ab2a36e0ea90f629b74bd57207411
2017-10-08 14:06:14 -07:00
James Zern
0ac46e818b lossless_enc_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: I06c64416103c3f3fc0519dd46d64b0a35f9798e4
2017-10-08 14:06:05 -07:00
James Zern
bc634d57c2 enc_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: Idd2f289fcf99f12bf36494111b07a8906c99c826
2017-10-08 14:05:59 -07:00
James Zern
bcb7347c2b dec_sse2: harmonize function suffixes
BUG=webp:355

Change-Id: Ic0390a4a24a5d8caff5b8af9fc9d59769ec533b1
2017-10-07 15:14:03 -07:00
Vincent Rabaud
e14ad93c0a Make sure we reach the full range for alpha blending.
255*255*257>>16 == 254 while we want 255.

BUG=webp:360

Change-Id: I2b9ac18f8802145f5a3d500c149ad9eceacbd75b
2017-10-05 16:53:57 +02:00
James Zern
7038ca8d52 demux,StoreFrame: restore hdr size check to min req
avoids over reading if the reported ANMF payload is < 8 bytes.

likely broken since:
81b8a741 Design change in ANMF and FRGM chunks:

Change-Id: I3e267bafea348a50545587dea8fafb2199c6b650
2017-10-03 23:18:18 -07:00
James Zern
fb3daad604 cpu: fix ssse3 check
ssse3 is bit #9 in ecx, bit 1 is sse3. this only controls the check for
slow ssse3 and likely had no ill effect.

Change-Id: I84ce73dc480e1cdbd085e37be06f3f402116c201
2017-09-29 16:27:47 -07:00
Vincent Rabaud
a5216efc8c Fix integer overflow warning.
Though the overflow could happen, it does not change the
end results.

Change-Id: I1b84e022a0776d35eab5c5c4fb7d3563f5667bfa
2017-09-25 11:02:22 +02:00
James Zern
a9c8916b87 decode.h,WebPIDecGetRGB: clarify output ptr validity
*last_y, *width, *height, *stride are only valid on non-NULL return

Change-Id: Iee2eeb29dd36392e2e7876d47df182a81dbb41ce
2017-09-22 16:58:33 -07:00
James Zern
f78da3dea6 add LOCAL_CLANG_PREREQ and avoid WORK_AROUND_GCC w/3.8+
this results in a 15-20% speedup for lossy decoding on a N5/S6/CM1

BUG=webp:339

Change-Id: Icdeb84c3e0b8908147ac276b4d8f76c3d565b735
2017-09-19 20:59:49 -07:00
James Zern
01c426f1e7 define WEBP_USE_INTRINSICS w/gcc-4.9+
32-bit builds are neutral to slightly faster using ndk r15c on a
N5/S6/CM1

BUG=webp:339

Change-Id: I94b9442e0ceaf2f5edb2b4026bc8b99cd77c918b
2017-09-19 20:59:43 -07:00
Vincent Rabaud
a80fcc4ae1 ifdef code not used by Chrome/Android.
Change-Id: Id086f6fd602b1fe3dc9034764b6a920a696ff1d2
2017-08-31 14:02:05 +02:00
Vincent Rabaud
3993af127e Fix signed integer overflows.
Change-Id: I62c9949f0edac58d69d991d6be5f85ae9e4d62a9
2017-08-31 11:56:42 +02:00
Pascal Massimino
6eba857b75 Merge "rationalize the Makefile.am" 2017-08-29 21:05:20 +00:00
Vincent Rabaud
c5e34fba66 function definition cleanup
Change-Id: I1f6bb4405f09c0c863d0bd1c77340636abdbae69
2017-08-29 14:09:35 +02:00
Pascal Massimino
3822762a6c rationalize the Makefile.am
one library addition per line, etc...

BUG=webp:355

Change-Id: I95761dea598a382db5632c5187210937e129ff75
2017-08-29 00:00:14 -07:00
James Zern
f324b7f9ba cosmetics: normalize fn proto & decl param names
BUG=webp:355

Change-Id: If020eb3b064cdb11853d1fa47058acae34752ce3
2017-08-26 14:17:31 -07:00
Vincent Rabaud
289e62a313 Remove declaration of unimplemented VP8ApplyNearLosslessPredict
Change-Id: I1bebea099a06bccfdd103ba0557320d6ad68af38
2017-08-23 17:16:29 +02:00
James Zern
34130afe8b anim_encode: fix integer overflow
calculate the file duration using unsigned math. this could still result
in an incorrect average duration calculation if there were multiple
rollovers. caching the duration is an option if it was desirable to
support such an extreme case.

Change-Id: I3875d94d081fec947c03a857055df6e27ff5351d
2017-08-18 16:52:57 -07:00
Pascal Massimino
42c79aa66b Merge "Encoder: harmonize function suffixes" 2017-08-09 18:13:57 +00:00
skal
b09307dcde Encoder: harmonize function suffixes
BUG=webp:355

Change-Id: Ia2fe95db7dfb303f3f64e390d43bc41b8933256c
2017-08-09 02:41:01 +00:00
James Zern
bed0456d58 Merge "SSIM: harmonize the function suffix" 2017-08-09 02:37:39 +00:00
skal
54f6a3cf3a lossless_sse2.c: fix some missed suffix changes
BUG=webp:355

Change-Id: If830e3169a4021899ed850aa7edfd94b81fa2cf9
2017-08-08 14:19:05 -07:00
skal
088f1dcce8 SSIM: harmonize the function suffix
BUG=webp:355

Change-Id: I751852ddb2abb7319e41e6c7d022ac4f288b4d08
2017-08-08 08:52:06 -07:00
skal
a0f72a4fe0 VP8LTransformColorFunc: drop an non-respected 'const' from the signature.
BUG=webp:355

Change-Id: Ie99bf377a55db2950bfbac9423bfe0967623ea5d
2017-08-07 19:05:01 -07:00
Pascal Massimino
8c934902cd Merge "Lossess dec: harmonize the function suffixes" 2017-08-08 02:04:10 +00:00
skal
622242aaba Lossess dec: harmonize the function suffixes
BUG=webp:355

Change-Id: I445d64df6aa2e347f41e7af306be12a77e2ac6a5
2017-08-07 18:22:41 -07:00
skal
1411f02761 Lossless Enc: harmonize the function suffixes
BUG=webp:355

Change-Id: I8baf506bd2a27095b956ef22a862b071f60c0d72
2017-08-07 18:02:07 -07:00
Vincent Rabaud
24ad2e3c99 add const to two variables
Change-Id: I97374ccbf118baa59425346ffc439036a4482bf4
2017-08-07 23:02:15 +02:00
Pascal Massimino
46efe062b8 Merge "Allow the lossless cruncher to work for alpha." 2017-08-07 21:00:02 +00:00
Vincent Rabaud
8c3f9a4706 Speed-up LZ77.
No need to look for a bigger maximum reach if we reach the end
of the image.

Change-Id: I56b166f9266f15cdff5aa59a75559972db19858f
2017-08-07 21:01:27 +02:00
Vincent Rabaud
1aef4c710b Allow the lossless cruncher to work for alpha.
Change-Id: I7185e75404fae8c739e0536026f4687d25decad7
2017-08-07 20:58:05 +02:00
Vincent Rabaud
b8821dbd81 Improve the box LZ77 speed.
For a pixel, we look for the longest match starting in a window around it.
For the following pixel, the previous result can be used and smaller
search window is used.

Change-Id: Ice16f9a7c8754099d068380848f0d77de3f756ac
2017-08-06 20:15:20 +02:00
James Zern
7beed2807b add missing ()s to macro parameters
BUG=webp:355

Change-Id: I616c6d3540d6551edd1b1cfdb5bffcf0a044c90f
2017-08-04 17:02:53 -07:00
James Zern
6473d20b3e Merge "fix Android standalone toolchain build" 2017-08-04 18:25:21 +00:00
James Zern
0c83a8bc69 Merge "yuv: harmonize suffix naming" 2017-08-02 06:35:36 +00:00
James Zern
c6d1db4b36 fix Android standalone toolchain build
add a check for cpu-features.h and rework some of the ifdef's around
android + neon. for android builds with cpu-features enabled the
*_neon.c files will still need to be flagged correctly (with e.g.,
.c.neon in Android.mk) to properly build them.

BUG=webp:353

Change-Id: I905ce305af0a204e560b915d8665093a3edaceb9
2017-08-01 22:59:03 -07:00
skal
663a6d9d2e unify the ALTERNATE_CODE flag usage
Pattern is now:
 #if !defined(FLAG)
 #define FLAG 0   // ALTERNATE_CODE
 #endif
...
 #if (FLAG == 1)
 ...
 #else
  ...
 #endif    // FLAG
...

Removed some unused code / flags:
  WEBP_YUV_USE_TABLE, WEBP_REFERENCE_IMPLEMENTATION,
  experimental code,  VP8YUVInit(), ...

BUG=webp:355

Change-Id: I98deb9189446a4cfd665c13ea8aa1ce6a308c63f
2017-08-01 20:49:29 -07:00
skal
73ea9f2702 yuv: harmonize suffix naming
BUG=webp:355

Change-Id: I403c4b3cdfc55b3b1648f98a1d189326a3e660a3
2017-08-01 20:40:00 -07:00
skal
c4568b47fd Rescaler: harmonize the suffix naming
BUG=webp:355

Change-Id: I7720502c62f96c780793d3d881eac7b3afae1418
2017-08-01 23:49:44 +00:00
Pascal Massimino
6cb13b0532 Merge "alpha_processing: harmonize the naming suffixes to be _C()" 2017-08-01 03:38:03 +00:00
James Zern
83a3e69a20 Merge "simplify WEBP_EXTERN macro" 2017-08-01 03:29:12 +00:00
Pascal Massimino
7295fde2e6 Merge "filters: harmonize the suffixes naming to _SSE2(), _C(), etc." 2017-08-01 01:55:48 +00:00
James Zern
8e42ba4c80 simplify WEBP_EXTERN macro
including the type in the macro doesn't bring much benefit to ordering,
current platforms work with a prefix, this would be insufficient if the
attribute needed to follow the function prototype. this form makes it
easier to override on the command line.

BUG=webp:355

Change-Id: Iba41ec0bb319403054be0e899c4cc472dd932fd9
2017-07-31 18:27:52 -07:00
skal
331ab34bcd cost*.c: harmonize the suffix namings
BUG=webp:355

Change-Id: Ic2e60eaab71cdffe1ebf93fc36aaa3eb25bbf08d
2017-07-31 17:18:32 -07:00
skal
b161f670f8 filters: harmonize the suffixes naming to _SSE2(), _C(), etc.
BUG=webp:355

Change-Id: I28f464eb13444c3046332cdda3c547f81700ecf4
2017-08-01 00:09:05 +00:00
skal
dec5e4d330 alpha_processing: harmonize the naming suffixes to be _C()
BUG=webp:355

Change-Id: Iae8221cd34957764ead21aa46abfc320e5514a4b
2017-07-31 23:34:24 +00:00
James Zern
92982609bc dsp.h: fix -Wundef w/__mips_dsp_rev
Change-Id: I552a543c7b039774041b43ace75b0cbea566b119
2017-07-11 16:12:32 -07:00
James Zern
88c73d8a7a backward_references_enc.h: fix WINDOW_SIZE_BITS check
this check was relocated in:
b903b80c Split cost-based backward references in its own file.

quiets -Wundef

Change-Id: I7f7a4773fb8cc77ca9f671b11f50d5db2275d415
2017-07-11 15:36:14 -07:00
James Zern
4ea49f6b82 rescaler_sse2.c: fix WEBP_RESCALER_FIX -> _RFIX typo
quiets -Wundef

Change-Id: I8f1facf401b6f1ab393005c93086ac3e2ae354d5
2017-07-11 15:35:27 -07:00
James Zern
b34a9db1a1 cosmetics,dec_sse2: remove some redundant comments
Change-Id: I5a59d6dde9b6638b318f36d51d0d53870a3de273
2017-07-06 23:19:18 -07:00
James Zern
66ad84f0f9 Merge "muxread: remove unreachable code" 2017-07-05 23:48:45 +00:00
Pascal Massimino
50ec3ab790 muxread: remove unreachable code
IDX_UNKNOWN chunks are handled separately in WebPMuxGetChunk()

Change-Id: Iaa0d3a1cd1080264c27671637551f52f71d73b07
2017-07-05 15:47:50 -07:00
hui su
7d67a1646d Lossy encoding: smoothen transparent areas to improve compression
If "exact" is false, we can modify the luma samples in fully transparent
areas to facilitate lossy compression. Experiments on some PNG images
show compression improvement of more than 20%.

Change-Id: I1a728cfa920a6652bc1f600d87c01f7f648c4942
2017-07-05 10:03:01 -07:00
James Zern
e50650c77f Merge "fix signature for DISABLE_TOKEN_BUFFER compilation" 2017-07-02 07:33:52 +00:00
Pascal Massimino
671d2567d4 fix signature for DISABLE_TOKEN_BUFFER compilation
Change-Id: Idb7aa2503c7870ef3b15a23325832428d47ae7c6
2017-07-01 14:46:28 -07:00
Vincent Rabaud
8acb4942f7 Remove the argb* files.
Half of the functionality was duplicated.
The rest is about the alpha channel handling so we
might as well put it in the appropriate file.

Change-Id: I8d5ef0afce82cc4842ab7132fd97995c42e6140a
2017-06-25 14:44:33 +02:00
Vincent Rabaud
7ca0df1363 Have the SSE2 version of PackARGB use common code.
The common code actually got sped-up by 25% by using the code
from PackARGB.

Change-Id: I94be6ccff2bfe02fff13c8e2698669e6a0d8fc74
2017-06-20 17:41:14 +02:00
Pascal Massimino
7b250459d6 Merge "Re-use the transformed image when trying several LZ77 in lossless." 2017-06-20 11:53:01 +00:00
Vincent Rabaud
e132072f0b Re-use the transformed image when trying several LZ77 in lossless.
Change-Id: If879b960131ba26859f8e4719a4c74cec1e0aaa8
2017-06-19 17:30:01 +02:00
Vincent Rabaud
5d7a50efee Get code to compile in C++.
Change-Id: I2f56e6b71e33ffecdba9e4fa9ef8f891c88f850f
2017-06-19 17:12:29 +02:00
Pascal Massimino
f1d5a397db multithread cruncher: only copy stats when picture->stats != NULL
BUG=webp:336

Change-Id: I6dfbbdcf61a6cb455f6cbf3dcd7f4c46578f42aa
2017-06-14 21:49:13 +02:00
Vincent Rabaud
f8c2ac15af Multi-thread the lossless cruncher.
BUG=webp:336

Change-Id: I8e861d6a61d51a5cdc4bbd00cd4f17d4ff006d2f
2017-06-14 16:50:36 +02:00
Vincent Rabaud
a88c6522f6 Merge "Integrate a new LZ77 looking for matches in the neighborhood of a pixel only." 2017-06-13 13:15:48 +00:00
Vincent Rabaud
8f6df1d0b9 Unroll Predictors 10, 11 and 12.
We see the following speed-ups:
10 -> 13%
11 -> 13%
12 -> 13%

Change-Id: I4734fd388d0f4e508884d0b123976bf2cbe69d2f
2017-06-08 20:37:47 +02:00
Vincent Rabaud
355c3d1bc7 Integrate a new LZ77 looking for matches in the neighborhood of a pixel only.
Change-Id: Ie2bbfee0a8d154b58f4a3068f3f634b7dad4c12d
2017-06-08 15:19:40 +02:00
Vincent Rabaud
a1779a017b Refactor LZ77 handling in preparation for a new method.
Change-Id: If305c328c8f508bd778d1af108e4eb979fbd2eca
2017-06-07 16:43:04 +02:00
Vincent Rabaud
f209a5481e Use the plane code and not the distance when computing statistics.
As backward references use the plane code when checking the cost
of a distance, statistics used to compute the cost should use it too.
This provides a small compression improvement at no speed cost.

Change-Id: Icade150929ee39ef6dc0d8b1fc85973086ecf41d
2017-06-01 17:08:43 +02:00
Vincent Rabaud
b903b80c30 Split cost-based backward references in its own file.
Change-Id: I4d8281e69b0e41f7c90337e5be70a6c65b044086
2017-06-01 16:22:31 +02:00
Vincent Rabaud
498cad34be Cosmetic changes in backward reference.
Change-Id: Ieb3dd65a647c034f67cf029e7b03f843b4650b0f
2017-06-01 14:20:16 +02:00
Vincent Rabaud
e4eb458741 lossless, VP8LTransformColor_C: make sure no overflow happens with colors.
Change-Id: Iec0d07cf1188ba96391cdb1b62131fc1469dfac6
2017-05-24 11:34:40 +02:00
Vincent Rabaud
7caef29b86 Fix typo that creeped in.
Change-Id: I600725721d9df80d6deddc65e00663649a47c941
2017-05-23 17:10:34 +02:00
Vincent Rabaud
39e19f927c Merge "near lossless: fix unsigned int overflow warnings." 2017-05-23 14:47:47 +00:00
Vincent Rabaud
9bbc0891c6 near lossless: fix unsigned int overflow warnings.
Change-Id: Ic1111a66761b5821cbbea1c91b038b2327dd20b5
2017-05-23 13:43:08 +02:00
James Zern
b588729740 cosmetics,FindClosestDiscretized: use uint in mask creation
Change-Id: Ib1edd74be1c051c10cfe2defadbea48cabe15135
2017-05-22 08:21:44 -07:00
James Zern
f1784aee04 near_lossless,FindClosestDiscretized: use unsigned ops
quiets undefined sanitizer warnings of the form:
left shift of 128 by 24 places cannot be represented in type 'int'

Change-Id: I8a389f2ac9238513517180f302f759425eeb7262
2017-05-22 03:11:03 -07:00
hui su
972104b34b webpmux: tolerate false positive Alpha flag
Change-Id: I390aebdda31f5320e83b7d7012735947a234650b
2017-05-19 11:26:37 -07:00
Vincent Rabaud
defc98d72c Doc: clarify the role of quality in WebPConfig.
Change-Id: I13706847f6263b9bb91eaf3c9284dd2d42ea7a85
2017-05-15 18:01:35 +02:00
Pascal Massimino
faf42213f4 NEON: implement ConvertRGB24ToY/BGR24/ARGB/RGBA32ToUV/ARGBToUV
Change-Id: Ie68aaed36d17f56d998c1b284514860cf5d28b8a
2017-05-09 15:57:20 +02:00
Vincent Rabaud
e644c556c5 Fix bad bit writer initialization.
When re-initializing a bit writer, we could set invalid values because
the bit writer was not big enough.

Change-Id: Id25ab6712603245a5a12d5f4a86fe35a9a799a5d
2017-05-02 18:59:01 +02:00
Pascal Massimino
b62cdad287 Merge "Implement a cruncher for lossless at method 6." 2017-04-28 14:37:57 +00:00
skal
da3e4dfb6f use the exact constant for the gamma transfer function
As found in BT2020 reference:
https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.2020-2-201510-I!!PDF-E.pdf

The difference in output-size/PSNR are under the noise level.

Change-Id: I42c12000b61dca791226f3af772c5bd9d58201b8
2017-04-27 11:53:15 -07:00
Vincent Rabaud
adab8ce020 Implement a cruncher for lossless at method 6.
Go over the whole compression step for each of the
transforms and pick the best one.

Change-Id: I3a1b1458348c468558be0fcf491038a5724c9364
2017-04-27 18:16:04 +02:00
Vincent Rabaud
0dcd85b6c4 Fix VP8ApplyNearLossless to respect const and stride.
Change-Id: I94a90220b5a175228fc48a4b299847b60cf5081b
2017-04-25 19:04:52 +02:00
Pascal Massimino
f768218966 yuv: rationalize the C/SSE2 function naming
+ implement some easy missing targets in SSE2 (565/4444)

Change-Id: Ib575f7ada2a0ed7309cddd238f8bfc0e8999f145
2017-04-21 13:52:25 +02:00
Pascal Massimino
52245424b0 NEON implementation of some Sharp-YUV420 functions
Change-Id: I449ef9c76b06f971f6e2ad7f9db96bf906d8fe1f
new-file: dsp/yuv_neon.c
2017-04-18 19:22:37 +02:00
Vincent Rabaud
690efd82f8 Avoid several backward reference copies.
An extra VP8LBackwardRefs struct is used but it was used internally anyway.

Change-Id: Ifcb36ce42b8c21ef3c7a1daf38cb7f714687fcb2
2017-04-14 15:34:50 +02:00
hui su
4bb1f607d7 src/dec/vp8_dec.h, cosmetics: fix comments
Change-Id: If076cdf63965bad9b00cbe19c922320638b65f46
2017-04-07 15:29:57 -07:00
James Zern
78fd199c3b backward_references_enc.c: clear -Wshadow warnings
index -> position

similar to: 2f5e8934 remove -Wshadow warnings

Change-Id: Ieaf8363d726a8fdf4e799845a81364e1dfae438a
2017-03-22 23:52:19 -07:00
James Zern
ae836410bb WebPLog2FloorC: clear -Wshadow warning
log -> log_value

Change-Id: Iae7cea7e0875e59b0806de41eb0b58815307495e
2017-03-22 23:50:31 -07:00
Pascal Massimino
c08adb6fbc Merge "VP8LEnc: remove use of BitsLog2Ceiling()" 2017-03-20 13:12:19 +00:00
Pascal Massimino
28c37ebd5a VP8LEnc: remove use of BitsLog2Ceiling()
was only used once. Better fall back for Log2Floor.

Change-Id: Ibcc26505440971bffe62ba6aca3d179ca85791d4
2017-03-20 02:58:16 -07:00
Pascal Massimino
bb175a935e Merge "rename some symbols clashing with MSVC headers" 2017-03-16 17:53:37 +00:00
Vincent Rabaud
39eda6584f Remove a duplicated pixel hash implementation.
Change-Id: If0df61add2fdf404f9baf0820ca83faa50f2791c
2017-03-16 16:15:40 +01:00
Pascal Massimino
36b8274deb rename some symbols clashing with MSVC headers
This is to prepare the inclusion of <windows.h>

FrameRect => FrameRectangle
CLIP_MASK => CLIP_8b_MASK

Change-Id: Ia4b1fa4ac06137b4102c91e232206a1fb7159ce0
2017-03-16 10:42:00 +01:00
James Zern
976adac1ce Merge "lossless incremental decoding: fix missing eos_ test" 2017-03-01 02:54:13 +00:00
Pascal Massimino
f8fad4fac1 lossless incremental decoding: fix missing eos_ test
The patch  21735e0 introduced a bug where a goto path was not testing
the eos_ state.  If this happened just before a row_sync, a SaveState()
would be called that would store the eos_ state as '1' till the end
of the loop.  This usually was not a problem, except for the very last
chunk where we disable the incremental decoding altogether (we have all
the data). The termination tests were then going wrong.
The fix is to add a proper eos_ test and avoid falling in this inconsistent
state.

(21735e06f7)

BUG=webp:332

Change-Id: Ib16773aee26bfd068fbf4e9db3d2313bd978b269
2017-02-28 15:17:33 -08:00
Vincent Rabaud
6f75a51b58 Analyze the transform entropy on the whole image.
The first column was skipped before.

Change-Id: I11521a6a48288d771e8fb83869fda08753f0fbcb
2017-02-28 15:48:59 +01:00
Vincent Rabaud
a5e4e3aff1 Use palette only if we can in entropy analysis.
Change-Id: I2be74b4d5fe48e059d5ced619d76fd030db79f74
2017-02-28 11:43:13 +01:00
Vincent Rabaud
75a9c3c452 Improve compression by better entropy analysis.
Change-Id: I6b56ca8d7d9a046a581baa0e85504136685d1161
2017-02-27 21:58:22 +01:00
Vincent Rabaud
7d985bd1d0 Fix small entropy analysis bug.
Change-Id: I70554b5a898b38930dfd03ca1ed78a075261e84b
2017-02-24 13:38:41 +01:00
Vincent Rabaud
6e7caf06ee Optimize the color cache size.
Before, the color cache size was chosen optimally for LZ77 and
the same value was used for RLE. Now, we optimize its value
taking both LZ77 and RLE into account.

Unfortunately, that comes with a small CPU hit.

Change-Id: I6261f04af78cf0784bb8e8fc4b4af5f566a0e071
2017-02-23 18:20:00 +01:00
Vincent Rabaud
833c92198c More efficient stochastic histogram merge.
Between each iteration we keep track of the previously found
potential merge hence less work to do.

Change-Id: I2b6237447e79443516a6111727d96c24f10bd98a
2017-02-23 14:35:33 +01:00
Vincent Rabaud
5183326ba8 Refactor the greedy histogram merge.
This is preparatory work for an upcoming commit.
No impact on speed or compression.

Change-Id: I62488ae6d2dbb5398f1604068c212ab8e26e82bc
2017-02-22 15:47:47 +01:00
Pascal Massimino
99f6f462f5 Merge "histogram_enc.c,MyRand: s/ul/u/ for unsigned constants" 2017-02-22 06:29:40 +00:00
James Zern
80a2218668 ssim.c: remove dead include
Change-Id: Ia4be534b3b95d5d9f712ff53e530c98b942df860
2017-02-21 20:17:19 -08:00
James Zern
a128dfff51 histogram_enc.c,MyRand: s/ul/u/ for unsigned constants
this is more consistent with the rest of the code base

Change-Id: Ifcf5d2729b2ebf32ffa12017db0106166829c77e
2017-02-21 20:12:17 -08:00
Pascal Massimino
693bf74ec0 move the SSIM calculation code in ssim.c / ssim_sse2.c
Change-Id: I63a63fa7f44f257f2e17e45358b206c23069c448
2017-02-21 12:53:35 +01:00
Vincent Rabaud
10d791ca70 Merge "Fix the random generator in HistogramCombineStochastic." 2017-02-19 10:11:48 +00:00
Vincent Rabaud
fa63a96603 Fix the random generator in HistogramCombineStochastic.
It was a bad implementation of a Lehmer random number generator
(the saturation was done wrong and mostly & was used instead of % .....).

That lead to "for" loop stuck with the same values given a specific seed,
hence wasted "for" loops (e.g. seed getting at 374988608 and modulo of 64
later leads to 0 even when updating the seed with the old formula).

As the "for" loops now always return a proper pair of histograms, their
number can greatly be reduced, hence a speedup.

Change-Id: I9f5b44d66cc96fd4824189d92276c3756c8ead5b
2017-02-19 10:49:16 +01:00
skal
16be192f47 VP8LSetBitPos: remove the eos_ setting
This code is ultra-critical for lossless decoding, especially on ARM.
The extra call VP8LIsEndOfStream() was causing unnecessary slow-down.

Now, we check for bitstream-end separately in the main loop.

Change-Id: I739b5d74cc29578e2b712ba99b544fd995ef0e0d
2017-02-11 02:35:02 -08:00
Pascal Massimino
4105d565d3 disable WEBP_USE_XXX optimisations when EMSCRIPTEN is defined
Currently, none are available. If WEBP_HAVE_SSE2 eventually works,
we'll have to refine this conditionals.

BUG=webp:261

Change-Id: Ibc63ee1c013f2a4169eeb85cc8b6317b6420c2ad
2017-02-08 15:44:20 +00:00
Vincent Rabaud
868aa6901f Perform greedy histogram merge in a unified way.
Previously, the stochastic method for histogram
combination could finish in a greedy way
if the number of iterations to perform so was smaller.

Except that another greedy combination was performed
afterwards ... hence wasted CPU in some cases.

Change-Id: Ic0f26873e6dc746679486b91cb35d73efee91931
2017-02-07 11:57:20 +01:00
Pascal Massimino
b494fdec45 optimize the ARGB->ARGB Import to use memcpy
(instead of the generic VP8PackARGB call)

Change-Id: I86edeb5934e7c062593f0248de7607cca5f1027c
2017-02-03 16:54:52 +01:00
Vincent Rabaud
5cfd4ebc5e LZ77 interval speedups. Faster, smaller, simpler.
The initial re-writing of this part of the code with intervals
had to be done with a complex logic (mostly intervals with a
lower and upper bound, not a constant value like now) to properly
deal with the inefficiencies of the then LZ77 algorithm.
The improvements made to LZ77 since, now allow for a simpler logic.

There were also small errors in the interval insertion logic
that lead to small inefficiencies (hence a slightly better
compression rate).

Change-Id: If079a0cafaae7be8e3f253485d9015a7177cf973
2017-02-02 11:51:30 +01:00
Pascal Massimino
be73378684 Merge "Add clang build fix for MSA" 2017-02-01 12:43:09 +00:00
Parag Salasakar
aa893914fc Add clang build fix for MSA
Change-Id: If139f4ecbdce756c69ba4ae032a70f81179683f8
2017-02-01 17:45:17 +05:30
Jehan
32ed856f60 Fix "all|no frames are keyframes" settings.
Documentation says: "if kmin == 0, then key-frame insertion is disabled;
and if kmax == 0, then all frames will be key-frames."
Reading this, you'd expect that if kmax == 0, then with any kmin <= 0
all frames will be key-frames. But actually the kmin <= 0 test is caught
first and you get the opposite (no keyframes but the first). You'd have
instead to set kmax == 0 and any value kmin > 0, which is absolutely
counter-intuitive (reversing order).
Moreover kmax == 1 has no valid kmin (kmin == 1 conflicts with the
`kmax > kmin` rule and kmin == 0 conflicts with `kmin >= kmax / 2 + 1`).
So it should be considered an exception too.

Instead I propose this new logic:
- kmax == 1 means that all frames are keyframes (you are explicitly
  requesting a keyframe every 1 frame at most, i.e. all frames).
- kmax == 0 means no keyframes (you ask for a keyframe every 0 frames,
  i.e. never).
This is more "logical" language-wise, and also does not involve any
conflicts about what if both kmax and kmin are 0, since now a single
property value is meaningful for the 2 exceptional cases.

Change-Id: Ia90fb963bc26904ff078d2e4ef9f74b22b13a0fd
(cherry picked from commit 2dc0bdcaee)
2017-01-26 22:31:16 -08:00
James Zern
1c3190b6ed Merge "Fix "all|no frames are keyframes" settings." 2017-01-27 00:02:04 +00:00
Pascal Massimino
f4dc56fd77 disable GradientUnfilter_NEON
Compile with XCode, it appears quite slower than the C-version,
especially for arm64.

Change-Id: Ic46dba184a36be454fef674129d2f909003788fc
(cherry picked from commit 4f3e3bbd44)
2017-01-25 20:30:15 -08:00
Pascal Massimino
4f3e3bbd44 disable GradientUnfilter_NEON
Compile with XCode, it appears quite slower than the C-version,
especially for arm64.

Change-Id: Ic46dba184a36be454fef674129d2f909003788fc
2017-01-25 16:33:26 -08:00
Jehan
2dc0bdcaee Fix "all|no frames are keyframes" settings.
Documentation says: "if kmin == 0, then key-frame insertion is disabled;
and if kmax == 0, then all frames will be key-frames."
Reading this, you'd expect that if kmax == 0, then with any kmin <= 0
all frames will be key-frames. But actually the kmin <= 0 test is caught
first and you get the opposite (no keyframes but the first). You'd have
instead to set kmax == 0 and any value kmin > 0, which is absolutely
counter-intuitive (reversing order).
Moreover kmax == 1 has no valid kmin (kmin == 1 conflicts with the
`kmax > kmin` rule and kmin == 0 conflicts with `kmin >= kmax / 2 + 1`).
So it should be considered an exception too.

Instead I propose this new logic:
- kmax == 1 means that all frames are keyframes (you are explicitly
  requesting a keyframe every 1 frame at most, i.e. all frames).
- kmax == 0 means no keyframes (you ask for a keyframe every 0 frames,
  i.e. never).
This is more "logical" language-wise, and also does not involve any
conflicts about what if both kmax and kmin are 0, since now a single
property value is meaningful for the 2 exceptional cases.

Change-Id: Ia90fb963bc26904ff078d2e4ef9f74b22b13a0fd
2017-01-25 13:12:52 -08:00
James Zern
36c42ea415 bump version to 0.6.0
libwebp{,decoder} - 0.6.0
libwebp libtool - 7.0.0
libwebpdecoder libtool - 3.0.0

mux - 0.4.0
libtool - 3.0.0

demux - 0.3.2
libtool - 2.2.0

Change-Id: Ie46dc70df1e283df0ccef6eb07c5694feb4d4a2b
2017-01-23 18:07:00 -08:00
James Zern
919f9e2fd6 Merge "add .rc files for windows dll versioning" 2017-01-20 19:29:50 +00:00
Pascal Massimino
4689ce1635 cwebp: add a -sharp_yuv option for 'sharp' RGB->YUV conversion
Change-Id: I6edd5b44d693da50f702fa8218f14872874d91ba
2017-01-20 16:54:54 +01:00
Pascal Massimino
79bf46f120 rename the pretentious SmartYUV into SharpYUV
Change-Id: Ifeeb9cb85896c5f3ba0cc1c2c821f8d00295f69e
2017-01-20 14:36:21 +01:00
Pascal Massimino
eb1dc89a5f silently expose use_delta_palette in the WebPConfig API
is just a placeholder for now, unless WEBP_USE_EXPERIMENTAL_FEATURES
is defined.

Change-Id: I087cb49781560bc1a7fbb01b136d36115c97ef72
2017-01-20 10:25:19 +01:00
James Zern
43d3f01a2f add .rc files for windows dll versioning
BUG=webp:323

Change-Id: Id415a32b63618d39af2e599cec0d40f64c35bbce
2017-01-20 00:35:15 -08:00
James Zern
668e1dd44f src/{dec,enc,utils}: give filenames a unique suffix
this avoids duplicates between these trees and dsp/, e.g., enc/tree.c,
dec/tree.c, making pulling the whole library source tree into one target
possible

BUG=webp:279

Change-Id: I060a614833c7c24ddd37bf641702ae6a5eef1775
2017-01-19 19:09:48 -08:00
Pascal Massimino
71c53f1aeb NEON: speed-up strong filtering
The sub-expression trick removes two constants and
two vmlal_s8 instructions.

Change-Id: I200022573b4880871b528b13a11a8f3d95def113
2017-01-19 20:46:48 +00:00
Pascal Massimino
a345068aba ARM: speed up bitreader by avoiding tables
(and using BitsLog2Floor() from utils.h instead)

9-10% speed-up, apparently

Change-Id: I9acae4a4dceb1ddcc99306f99b722079bb06f6f8
2017-01-17 23:52:37 -08:00
Pascal Massimino
1dc82a6bba Merge "introduce a generic GetCoeffs() function pointer" 2017-01-18 07:44:36 +00:00
Pascal Massimino
8074b89eb3 introduce a generic GetCoeffs() function pointer
We can switch at run-time between the standard GetCoeffs() critical
function, that uses a fast variant of VP8GetBit().
However, some platforms have slow instructions that make standard
VP8GetBit() slow. GetCoeffs() is the right level of branching to
switch to GetCoeffsAlt() that avoids these slow instructions in some
not-frequent cases.

Next patch will upgrade VP8GetBit() to use clz, after this one
is proved to be neutral speed-wise.

Change-Id: Ia6cef5de9de6131574d2202bbc0bea8559c9b693
2017-01-17 16:24:00 +01:00
Pascal Massimino
749a45a520 Merge "NEON: implement alpha-filters (horizontal/vertical/gradient)" 2017-01-17 15:13:08 +00:00
Pascal Massimino
74c053b57d Merge "NEON: fix overflow in SSE NxN calculation" 2017-01-17 15:10:54 +00:00
Pascal Massimino
0a3aeff75b Merge "dsp: WebPExtractGreen function for alpha decompression" 2017-01-17 15:08:20 +00:00
Pascal Massimino
1de931c669 NEON: implement alpha-filters (horizontal/vertical/gradient)
gradient-filter code is not much faster, but maybe improvable in the future.

Change-Id: Ia16070e409fe8703b02276166f19526917df6b35
2017-01-17 15:44:46 +01:00
Pascal Massimino
9b3aca404d NEON: fix overflow in SSE NxN calculation
vmlal_u8() is prone to overflow during the accumulation.
There was a mismatch happening at low q mostly. Because in this
case the distortion is important and the accumulated sum was
later than 16bit-unsigned.

Change-Id: I1a08a2f744bcdf0b26647e61b9ee92a0c2e28fe8
2017-01-17 11:47:36 +01:00
Pascal Massimino
1c07a3c639 dsp: WebPExtractGreen function for alpha decompression
+ NEON implementation

Change-Id: I67204f99d6e4c5974718bdf21dad30381978f72c
2017-01-17 09:33:25 +00:00
Pascal Massimino
9ed5e3e5dd use pointers for WebPRescaler's in WebPDecParams
This makes the structure more generic, without the hard-coded
internal structure.

This is a borderline incompatible ABI change, even if WebPIDecoder structure
is opaque.

Change-Id: I518765c3f76fc17a136cef045a5a8aa70ed70e85
2017-01-16 22:30:29 -08:00
James Zern
db013a8d5c Merge "ARM: don't use USE_GENERIC_TREE" 2017-01-13 22:15:04 +00:00
Pascal Massimino
fcd4784dcd use a 8b table for C-version for clz()
30% faster on x86, 5% faster on N5.

New generic function: WebPLog2FloorC()
This function is called as fallback for BitsLog2Floor() when there's
no clz() available.

Change-Id: Ica15c6092112e514c0e200fab89c434de48d4b19
2017-01-13 15:36:26 +01:00
Pascal Massimino
fbb5c473b4 ARM: don't use USE_GENERIC_TREE
It's 1-2% faster to use hard-coded tree on ARM

Change-Id: I54403a70f6c692e50148c33f36833588957c20ee
2017-01-13 10:05:21 +01:00
Pascal Massimino
8fda56126e Merge "add a kSlowSSSE3 feature for CPUInfo" 2017-01-13 07:01:48 +00:00
Pascal Massimino
86bbd24552 add a kSlowSSSE3 feature for CPUInfo
This is meant to be used for run-time detection of slow platforms
regarding instructions like pshufb and bsr.

Adapted from libvpx patch: https://chromium-review.googlesource.com/#/c/367731

Change-Id: I2c22fbb9aae699d87a041393ba1ad5f1f21ff640
2017-01-13 06:19:27 +00:00
Vincent Rabaud
7c2779e95a Get code to fully compile in C++.
Change-Id: I6d8490c8c9b955d90dcc89ee8a9cf29ca0f93b08
2017-01-12 18:03:55 +01:00
Vincent Rabaud
250c358662 Merge "When compiling as C++, avoid narrowing warnings." 2017-01-12 13:00:56 +00:00
Vincent Rabaud
c0648ac2ae When compiling as C++, avoid narrowing warnings.
The gcc compilation warning was: narrowing conversion from ‘int’ to ‘int8_t’

Change-Id: I4803dd60ad04060cdb5d61a1aa98b25215b9d4eb
2017-01-12 13:39:22 +01:00
Pascal Massimino
0d55f60c91 40% faster ApplyAlphaMultiply_SSE2
process four pixels at a time

Change-Id: I1dee7f70772be4915654fc6638ef4729a1a239d4
2017-01-12 02:33:09 -08:00