]> granicus.if.org Git - libvpx/log
libvpx
8 years agoAdd Daala entropy coder.
Nathan E. Egge [Sun, 6 Mar 2016 15:59:29 +0000 (10:59 -0500)]
Add Daala entropy coder.

Change-Id: I2849a50163268d58cc5d80aacfec1fd02299ca43

8 years agoMerge "Switch rANS to 15 bit precision, and adjust L_BASE." into nextgenv2
Alex Converse [Fri, 14 Oct 2016 21:56:34 +0000 (21:56 +0000)]
Merge "Switch rANS to 15 bit precision, and adjust L_BASE." into nextgenv2

8 years agoSwitch rANS to 15 bit precision, and adjust L_BASE.
Alex Converse [Fri, 9 Sep 2016 17:51:48 +0000 (10:51 -0700)]
Switch rANS to 15 bit precision, and adjust L_BASE.

This causes rANS to operate at the same precision as the Daala EC.

aom/master stats: rans10uabs8lbase12 → rans15uabs8lbase15

objective-1-fast
PSNR YCbCr:      0.01%      0.01%      0.01%
   PSNRHVS:      0.01%
      SSIM:      0.01%
    MSSSIM:      0.01%
 CIEDE2000:      0.01%

subset1
PSNR YCbCr:     -0.01%     -0.00%     -0.00%
   PSNRHVS:     -0.01%
      SSIM:     -0.01%
    MSSSIM:     -0.01%
 CIEDE2000:     -0.01%

(cherry picked from aom/master commit ddbc2e2a68bfc997dc61fca5bcaac3a75245e965)

Change-Id: I6ef0a4f6198784b3712a61af9f105d560a22eaea

8 years agoWrap palette code inside CONFIG_PALETTE flag.
Urvang Joshi [Wed, 12 Oct 2016 23:28:56 +0000 (16:28 -0700)]
Wrap palette code inside CONFIG_PALETTE flag.

This flag was already added to aomedia/master, so bringing it back to
webm/nextgenv2, as part of an effort to get the two codebases in sync.

Change-Id: I2b933a6a160e4210d1411a9e7978149eb8553205

8 years agoMerge changes I4850b36e,Ic4d7128a into nextgenv2
James Zern [Fri, 14 Oct 2016 19:10:20 +0000 (19:10 +0000)]
Merge changes I4850b36e,Ic4d7128a into nextgenv2

* changes:
  variance_avx2: sync variance functions with c-code
  Resolve -Wshorten-64-to-32 in variance.

8 years agoUse "av1" as codec name
Yaowu Xu [Fri, 14 Oct 2016 18:05:54 +0000 (11:05 -0700)]
Use "av1" as codec name

Change-Id: I7650f1e96df0bcd53b1733c7967aae52dccf836a

8 years agoMerge "Revert "Revert "Move CLPF block signals from frame to SB level.""" into nextgenv2
Yaowu Xu [Fri, 14 Oct 2016 17:58:20 +0000 (17:58 +0000)]
Merge "Revert "Revert "Move CLPF block signals from frame to SB level.""" into nextgenv2

8 years agoMerge "Delete some redundant function declarations in aom_dsp_rtcd_defs.pl" into...
Yi Luo [Fri, 14 Oct 2016 17:50:37 +0000 (17:50 +0000)]
Merge "Delete some redundant function declarations in aom_dsp_rtcd_defs.pl" into nextgenv2

8 years agoRevert "Revert "Move CLPF block signals from frame to SB level.""
Yaowu Xu [Fri, 14 Oct 2016 15:47:03 +0000 (08:47 -0700)]
Revert "Revert "Move CLPF block signals from frame to SB level.""

This reverts commit 9b25f3067485b32442e13964df098903736c3fd8 to
reinstate the reverted commit with fixes that solved the build issues
when --enalbe-clpf is used in configure.

Change-Id: I15447cae7fa9b3deb27976345dc3db230a4a7a60

8 years agoMerge "Revert "Move CLPF block signals from frame to SB level."" into nextgenv2
Yaowu Xu [Fri, 14 Oct 2016 15:39:36 +0000 (15:39 +0000)]
Merge "Revert "Move CLPF block signals from frame to SB level."" into nextgenv2

8 years agoRevert "Move CLPF block signals from frame to SB level."
Yaowu Xu [Fri, 14 Oct 2016 15:39:03 +0000 (15:39 +0000)]
Revert "Move CLPF block signals from frame to SB level."

This reverts commit 975350387ce0b55bf5af8cb944f6a242b72251ff.

Change-Id: I9f8e891739352ca2bde4b294e37c85a668f416e0

8 years agovariance_avx2: sync variance functions with c-code
James Zern [Mon, 19 Sep 2016 23:17:24 +0000 (16:17 -0700)]
variance_avx2: sync variance functions with c-code

add missing int64 -> uint32 cast; quiets -Wshorten-64-to-32 warnings

Change-Id: I4850b36e18dc8b399108342be4bfe0b684aefb78
(cherry picked from commit 6acd061aad8cf62000cc9117390d0c94581a8591)

8 years agoResolve -Wshorten-64-to-32 in variance.
Alex Converse [Thu, 28 Jul 2016 16:48:50 +0000 (09:48 -0700)]
Resolve -Wshorten-64-to-32 in variance.

The subtrahend is small enough to fit into uint32_t.

Change-Id: Ic4d7128aaa665eaf6b25d562610ba8942c46137f
(cherry picked from commit c0241664aac3a1805db9bd8e09e071ac326531e0)

8 years agoMerge "Simplify 8x16 and 16x8 inverse transform tests" into nextgenv2
Debargha Mukherjee [Fri, 14 Oct 2016 02:53:38 +0000 (02:53 +0000)]
Merge "Simplify 8x16 and 16x8 inverse transform tests" into nextgenv2

8 years agoMerge "Enable test system to detect transforms misusing 'stride' parameter" into...
Debargha Mukherjee [Fri, 14 Oct 2016 02:50:46 +0000 (02:50 +0000)]
Merge "Enable test system to detect transforms misusing 'stride' parameter" into nextgenv2

8 years agoMerge "Add sse2 forward and inverse 16x32 and 32x16 transforms" into nextgenv2
Debargha Mukherjee [Fri, 14 Oct 2016 02:49:20 +0000 (02:49 +0000)]
Merge "Add sse2 forward and inverse 16x32 and 32x16 transforms" into nextgenv2

8 years agoMerge "Renamings for OBMC experiment" into nextgenv2
Yue Chen [Fri, 14 Oct 2016 01:33:00 +0000 (01:33 +0000)]
Merge "Renamings for OBMC experiment" into nextgenv2

8 years agoDelete some redundant function declarations in aom_dsp_rtcd_defs.pl
Yi Luo [Fri, 14 Oct 2016 00:53:45 +0000 (17:53 -0700)]
Delete some redundant function declarations in aom_dsp_rtcd_defs.pl

Change-Id: I4df57a7faba5800c048b2dc469ec31545406f55c

8 years agoMove CLPF block signals from frame to SB level.
Steinar Midtskogen [Wed, 21 Sep 2016 11:38:16 +0000 (13:38 +0200)]
Move CLPF block signals from frame to SB level.

These signals were in the uncompressed frame header (as a temporary
hack), which caused two problems:

* We don't want that header to be duplicated in the slice header
* It was necessary to signal the number of bits to transmit up front

However, the filter size can be 128x128 which is greater than the SB
size, and a decoder wouldn't be able to know whether to read a bit or
not until the final SB of that 128x128 block has been decoded
(depending on whether the 128x128 is all skip or not).  Therefore the
signalling was changed for 128x128 blocks so that every top left SB of
a 128x128 filter block contains a signal regardless of whether the
block is all skip or not.  Also, all the MB's of 128x128 block are
filtered even if they are skip MB's.  This gives the signal a purpose
even when the 128x128 block is all skip, and it also gives a slight
coding gain as it leaves a way to filter skip blocks, which was
previously forbidden.

Low latency:
PSNR YCbCr:     -0.19%     -0.14%     -0.06%
   PSNRHVS:     -0.15%
      SSIM:     -0.13%
    MSSSIM:     -0.15%
 CIEDE2000:     -0.19%

High latency:
PSNR YCbCr:     -0.03%     -0.01%     -0.09%
   PSNRHVS:      0.04%
      SSIM:      0.00%
    MSSSIM:      0.02%
 CIEDE2000:     -0.02%

Change-Id: I69ba7144d07d388b4f0968f6a53558f480979171

8 years agoRenamings for OBMC experiment
Yue Chen [Thu, 13 Oct 2016 22:18:22 +0000 (15:18 -0700)]
Renamings for OBMC experiment

To get ready for pulling AV1 to nextgenv2
Replace the experimental flag by MOTION_VAR. Rename major variables.

Change-Id: If6cf4f37b9319c46d8f90df551cc7295d66ca205

8 years agoDon't use _mm_cvtsi128_si64 on 32 bit systems
Steinar Midtskogen [Wed, 5 Oct 2016 21:09:21 +0000 (23:09 +0200)]
Don't use _mm_cvtsi128_si64 on 32 bit systems

Change-Id: I332afb8d9e35cd60f05915160a5b2e1dc8757de5

8 years agoFix formatting in a few files
Yaowu Xu [Thu, 13 Oct 2016 20:04:29 +0000 (13:04 -0700)]
Fix formatting in a few files

Change-Id: Ia5175afe82b142d9e18c01c546610202c630588e

8 years agoRemoving some useless loops in deringing filter
Jean-Marc Valin [Fri, 7 Oct 2016 22:10:19 +0000 (18:10 -0400)]
Removing some useless loops in deringing filter

No change in the output

Change-Id: I1627feaa163d65da0df90e9dacbc5e39ee755de8

8 years agoFix deringing level choice for 10-bit and 12-bit
Jean-Marc Valin [Sat, 8 Oct 2016 18:10:39 +0000 (14:10 -0400)]
Fix deringing level choice for 10-bit and 12-bit

Making sure we never exceed a base level of 63

Change-Id: I821254b8d970446bd40fdd6e4d7073c69760a86d

8 years agoDon't dering superblocks that have deringing disabled
Jean-Marc Valin [Fri, 7 Oct 2016 19:03:39 +0000 (15:03 -0400)]
Don't dering superblocks that have deringing disabled

Doesn't change the output, but avoids useless deringing with threshold=0

Change-Id: I69f3e54abad2d2493cfbc76c188ad7d190f0aeff

8 years agoMerge "Add SSE4.1 code for deringing functions." into nextgenv2
Yaowu Xu [Thu, 13 Oct 2016 18:02:59 +0000 (18:02 +0000)]
Merge "Add SSE4.1 code for deringing functions." into nextgenv2

8 years agoAdd SSE4.1 code for deringing functions.
Michael Bebenita [Thu, 6 Oct 2016 21:27:34 +0000 (14:27 -0700)]
Add SSE4.1 code for deringing functions.

Change-Id: I363f7fb610a5c86ea9f417e34b57c6373af877e5

8 years agoMerge "Simpler threshold calculation for the second filter" into nextgenv2
Yaowu Xu [Thu, 13 Oct 2016 18:01:45 +0000 (18:01 +0000)]
Merge "Simpler threshold calculation for the second filter" into nextgenv2

8 years agoMerge "Make 4x4 deringing (chroma) use shorter filters" into nextgenv2
Yaowu Xu [Thu, 13 Oct 2016 18:01:23 +0000 (18:01 +0000)]
Merge "Make 4x4 deringing (chroma) use shorter filters" into nextgenv2

8 years agoMerge "Removing Daala-specific deringing code" into nextgenv2
Yaowu Xu [Thu, 13 Oct 2016 18:01:11 +0000 (18:01 +0000)]
Merge "Removing Daala-specific deringing code" into nextgenv2

8 years agoMerge "Clean code a bit and fix a couple of small bugs in ext-refs" into nextgenv2
Zoe Liu [Thu, 13 Oct 2016 16:47:03 +0000 (16:47 +0000)]
Merge "Clean code a bit and fix a couple of small bugs in ext-refs" into nextgenv2

8 years agoMerge "Use a quantizer-based threshold rather than full search for deringing" into...
Yaowu Xu [Thu, 13 Oct 2016 16:35:07 +0000 (16:35 +0000)]
Merge "Use a quantizer-based threshold rather than full search for deringing" into nextgenv2

8 years agoMerge changes Ic3a68557,Ib1dbe41a,I0da09270,Ibdbd720d into nextgenv2
Yaowu Xu [Thu, 13 Oct 2016 15:54:32 +0000 (15:54 +0000)]
Merge changes Ic3a68557,Ib1dbe41a,I0da09270,Ibdbd720d into nextgenv2

* changes:
  Deringing cleanup: remove DERING_REFINEMENT (always on now)
  Don't run the deringing filter on skipped blocks within a superblock
  Don't dering skipped superblocks
  On x86 use _mm_set_epi32 when _mm_cvtsi64_si128 isn't available

8 years agoClean code a bit and fix a couple of small bugs in ext-refs
Zoe Liu [Wed, 12 Oct 2016 19:31:43 +0000 (12:31 -0700)]
Clean code a bit and fix a couple of small bugs in ext-refs

Currently the patch does not have any impact on the RD performance. The
fix could however potentially help on the next step of work, especially
when the extra altref frames allow non-zero temporal filtering strength
and their corresponding OVERLAY frames, i.e. the INTNL_OVERLAY frames
are being added.

Change-Id: I2e07fb3d0aa547a0b5dd05bb4ba865cd46309076

8 years agoMerge "Sync 2x2 intra predictors" into nextgenv2
Yaowu Xu [Thu, 13 Oct 2016 15:20:52 +0000 (15:20 +0000)]
Merge "Sync 2x2 intra predictors" into nextgenv2

8 years agoSimplify 8x16 and 16x8 inverse transform tests
David Barker [Thu, 13 Oct 2016 15:02:59 +0000 (16:02 +0100)]
Simplify 8x16 and 16x8 inverse transform tests

Change-Id: Ie86aedfb1f3e0d9c0cf58d7183861a0ed0e8ccc8

8 years agoEnable test system to detect transforms misusing 'stride' parameter
David Barker [Thu, 13 Oct 2016 14:10:14 +0000 (15:10 +0100)]
Enable test system to detect transforms misusing 'stride' parameter

This would have caught the bug introduced in patch set 1 of
https://chromium-review.googlesource.com/#/c/397378/

Change-Id: I9c6d5d9c4c98aed5ac48c4fb1c4ff4131b0df1d5

8 years agoAnsTest: Replace the dummy distribution
Alex Converse [Fri, 9 Sep 2016 17:30:36 +0000 (10:30 -0700)]
AnsTest: Replace the dummy distribution

Use constrained token table row 65/256 instead.

Change-Id: I8b442d4c82af8fa9d36ac2de0d73179ed040478d
(cherry picked from commit 47eb9a2ca46821b468903514cd34eaaca2533d45)

8 years agoMerge changes Ic74d9d88,Ie93b474e,I544989ea,Ic273f7d9,Idfd2d2b3, ... into nextgenv2
Alex Converse [Thu, 13 Oct 2016 14:03:15 +0000 (14:03 +0000)]
Merge changes Ic74d9d88,Ie93b474e,I544989ea,Ic273f7d9,Idfd2d2b3, ... into nextgenv2

* changes:
  Remove custom rans types
  Remove add_token_no_extra.
  Remove unused aom_rans_build_cdf_from_pdf
  Add the tool used to generate the constrained tokenset.
  Remove the starting zero from ANS CDFs.
  Import the aom_read/write_symbol abstractions from aom/master

8 years agoAdd sse2 forward and inverse 16x32 and 32x16 transforms
David Barker [Thu, 6 Oct 2016 16:25:40 +0000 (17:25 +0100)]
Add sse2 forward and inverse 16x32 and 32x16 transforms

Change-Id: I1241257430f1e08ead1ce0f31db8272b50783102

8 years agoMerge "Fix a bug in inverse halfright 32x32 transform" into nextgenv2
Debargha Mukherjee [Thu, 13 Oct 2016 08:16:47 +0000 (08:16 +0000)]
Merge "Fix a bug in inverse halfright 32x32 transform" into nextgenv2

8 years agoRemove custom rans types
Alex Converse [Sun, 4 Sep 2016 11:30:43 +0000 (13:30 +0200)]
Remove custom rans types

(cherry picked from aom/master commit 11206c60d930be9d29100567aa67f2a65463852a)

Includes renames in a bunch of places not handled by the original
due to differing tree states.

Change-Id: Ic74d9d8850b8c80a51e55e425bbf472a67e2653f

8 years agoSync 2x2 intra predictors
Jingning Han [Thu, 13 Oct 2016 04:03:18 +0000 (21:03 -0700)]
Sync 2x2 intra predictors

Add 2x2 DC, V, H, TM intra predictors.

Change-Id: I2a614adde553f821c45bc5a9bf09800a9f0aaa26

8 years agoSimpler threshold calculation for the second filter
Jean-Marc Valin [Fri, 16 Sep 2016 15:06:50 +0000 (11:06 -0400)]
Simpler threshold calculation for the second filter

PSNR YCbCr:      0.03%     -0.00%      0.07%
   PSNRHVS:      0.06%
      SSIM:      0.12%
    MSSSIM:      0.09%
 CIEDE2000:      0.05%

Change-Id: I15ef9598a08f6713bc28ab98b0182310433e97ef

8 years agoMake 4x4 deringing (chroma) use shorter filters
Jean-Marc Valin [Thu, 15 Sep 2016 20:23:12 +0000 (16:23 -0400)]
Make 4x4 deringing (chroma) use shorter filters

Avoids blurring chroma for 4:2:0

PSNR YCbCr:      0.03%     -0.31%     -0.29%
   PSNRHVS:      0.02%
      SSIM:      0.03%
    MSSSIM:      0.02%
 CIEDE2000:      0.01%

Change-Id: If744fb902b5f24404479def22b9ca8a19baec722

8 years agoRemoving Daala-specific deringing code
Jean-Marc Valin [Tue, 23 Aug 2016 06:45:43 +0000 (02:45 -0400)]
Removing Daala-specific deringing code

No point in keeping them in sync now that all the code is reformatted

Change-Id: I8a062253ed6a5f86028cd5a2a922b3c760def6fb

8 years agoUse a quantizer-based threshold rather than full search for deringing
Jean-Marc Valin [Mon, 22 Aug 2016 19:47:09 +0000 (15:47 -0400)]
Use a quantizer-based threshold rather than full search for deringing

objective-1-short results (with deringing enabled):
PSNR YCbCr:      0.08%      0.03%      0.11%
   PSNRHVS:      0.06%
      SSIM:      0.12%
    MSSSIM:      0.08%
 CIEDE2000:      0.05%

Change-Id: Ifcfc42c14c33650dcf879c4d0ddd8688d4d07da1

8 years agoRemove add_token_no_extra.
Alex Converse [Sun, 4 Sep 2016 09:16:34 +0000 (11:16 +0200)]
Remove add_token_no_extra.

It was a fairly small production optimization for VP9.

Change-Id: Ie93b474ea5b7e63384a7c0b3a56b135462d1471b
(cherry picked from aom/master commit df9bb76b1330de42fe13827df4c72010adb51429)

8 years agoRemove unused aom_rans_build_cdf_from_pdf
Alex Converse [Sun, 4 Sep 2016 09:07:53 +0000 (11:07 +0200)]
Remove unused aom_rans_build_cdf_from_pdf

Change-Id: I544989eae45b7dda04250365c3de99f50110a76b
(cherry picked from aom/master commit 06cce842caa5212826d51c2a317de0bdfae74349)

8 years agoAdd the tool used to generate the constrained tokenset.
Alex Converse [Wed, 6 Jul 2016 17:47:27 +0000 (10:47 -0700)]
Add the tool used to generate the constrained tokenset.

The code that generates the raw distribution is based on a MATLAB
program by Debargha Mukherjee, and the algorithm used to quantize the
distribution comes from the ANS Toolkit by Jarek Duda.

Change-Id: Ic273f7d9e43e3ecd999e9e7e04cde57e8559375a
(cherry picked from aom/master commit ef446026aeafa318f9bee182b8c80eb4f1ef5a0a)

8 years agoRemove the starting zero from ANS CDFs.
Alex Converse [Mon, 22 Aug 2016 19:44:16 +0000 (12:44 -0700)]
Remove the starting zero from ANS CDFs.

This brings it in line with the Daala CDFs and will make it easier to
share code.

Change-Id: Idfd2d2b33c3b9b2c4e72ce72fb3d8039013448b9
(cherry picked from aom/master commit af98507ca928afe33e9f88fdd2ca168379528d6a)

8 years agoImport the aom_read/write_symbol abstractions from aom/master
Alex Converse [Wed, 12 Oct 2016 22:59:58 +0000 (15:59 -0700)]
Import the aom_read/write_symbol abstractions from aom/master

Change-Id: I0b255c05108c3b97e74df1b59c34111c9e9a5770

8 years agoDeringing cleanup: remove DERING_REFINEMENT (always on now)
Jean-Marc Valin [Mon, 22 Aug 2016 19:27:46 +0000 (15:27 -0400)]
Deringing cleanup: remove DERING_REFINEMENT (always on now)

Change-Id: Ic3a6855799be010e69aeab924b013679282ab191

8 years agoDon't run the deringing filter on skipped blocks within a superblock
Jean-Marc Valin [Mon, 18 Jul 2016 19:40:42 +0000 (15:40 -0400)]
Don't run the deringing filter on skipped blocks within a superblock

No change in metrics

Change-Id: Ib1dbe41a9e1a564dd9a63a33e2a5315ad6bca70c

8 years agoDon't dering skipped superblocks
Jean-Marc Valin [Mon, 18 Jul 2016 15:48:59 +0000 (11:48 -0400)]
Don't dering skipped superblocks

No change in metrics

Change-Id: I0da09270d78c3caf78a32a3157f02c87f2232e3e

8 years agoMerge "Hybrid forward transform 32x32 AVX2 optimization" into nextgenv2
Yi Luo [Thu, 13 Oct 2016 00:08:48 +0000 (00:08 +0000)]
Merge "Hybrid forward transform 32x32 AVX2 optimization" into nextgenv2

8 years agoOn x86 use _mm_set_epi32 when _mm_cvtsi64_si128 isn't available
Steinar Midtskogen [Tue, 4 Oct 2016 07:55:53 +0000 (09:55 +0200)]
On x86 use _mm_set_epi32 when _mm_cvtsi64_si128 isn't available

Change-Id: Ibdbd720d4f68892da6164a9849e212e759305005

8 years agoMerge changes I3ca2b674,I78afc587,I3ae62181,I5ed91556 into nextgenv2
Alex Converse [Wed, 12 Oct 2016 22:25:27 +0000 (22:25 +0000)]
Merge changes I3ca2b674,I78afc587,I3ae62181,I5ed91556 into nextgenv2

* changes:
  Unfork ANS decode_coefs
  Remove ZERO_TOKEN from the ANS tokenset
  Drop costing ANS tokens from derived probabilities
  Unfork ANS pack_mb_tokens

8 years agoFix a bug in inverse halfright 32x32 transform
Debargha Mukherjee [Wed, 12 Oct 2016 17:49:29 +0000 (10:49 -0700)]
Fix a bug in inverse halfright 32x32 transform

Fix a bug in the C implementation of the ihalfright32
transform, in the case that its input and output buffers are the same.
This occurs when it is called by av1_iht32x16_512_add_c.

Change-Id: I61c652e2662178520c0639a2879ae128a9c7ec3f

8 years agoHybrid forward transform 32x32 AVX2 optimization
Yi Luo [Fri, 7 Oct 2016 16:46:05 +0000 (09:46 -0700)]
Hybrid forward transform 32x32 AVX2 optimization

- av1_fht32x32 AVX2 function level time reduction ~89% compared to C.

- av1_fht32x32_avx2() on DCT_DCT improves 42.62% over aom_fdct32x32_avx2()
  But function replacement must go with the corresponding inverse txfm.

- No obvious user level time reduction due to 32x32 TX_TYPE selection.

- Zero high 128b YMM to avoid AVX-SSE transition penalties
  (fix 16x16 case).

- Added 32x32 AVX2 unit tests to verify bitexact.

- AVX2 optimization summary:
  On CPU i7-6700, based on 16x16/32x32 fwd txfm optimization results:
  C to AVX2: function level time reduction, ~86-89%.
  SSE2 to AVX2: function level time reduction, ~51%.

Change-Id: Idd0cd8bf066a61c7117140ef15ab6c1f8eb4b036

8 years agoMerge "Send allow_screen_content flag for both key and intra only frames" into nextgenv2
Hui Su [Wed, 12 Oct 2016 21:13:24 +0000 (21:13 +0000)]
Merge "Send allow_screen_content flag for both key and intra only frames" into nextgenv2

8 years agoMerge "Refactor expand dry_run types to return coef rate" into nextgenv2
Debargha Mukherjee [Wed, 12 Oct 2016 21:06:41 +0000 (21:06 +0000)]
Merge "Refactor expand dry_run types to return coef rate" into nextgenv2

8 years agoUnfork ANS decode_coefs
Alex Converse [Wed, 12 Oct 2016 20:23:33 +0000 (13:23 -0700)]
Unfork ANS decode_coefs

This is less code and more like what we have in aom/master.

Change-Id: I3ca2b674e4ad9e2e211d08bb51d78549e8b63a54

8 years agoRemove ZERO_TOKEN from the ANS tokenset
Alex Converse [Wed, 12 Oct 2016 19:53:40 +0000 (12:53 -0700)]
Remove ZERO_TOKEN from the ANS tokenset

This can be re-added after aligning AOM's ANS with nextgenv2's ANS.

This partially reverts commit 3829cd2f2f9904572019aa047d068baeee843767.

Change-Id: I78afc587f1abfe33ffcd53b3262910cfae135534

8 years agoDrop costing ANS tokens from derived probabilities
Alex Converse [Wed, 12 Oct 2016 20:03:55 +0000 (13:03 -0700)]
Drop costing ANS tokens from derived probabilities

This mimics what's currently done in aom/master. This can be re-added
after aligning AOM's ANS with nextgenv2's ANS.

Change-Id: I3ae62181dd4803694204a234c717a86a15ca8a40

8 years agoUnfork ANS pack_mb_tokens
Alex Converse [Tue, 11 Oct 2016 23:50:56 +0000 (16:50 -0700)]
Unfork ANS pack_mb_tokens

This is less code and more like what we have in aom/master.

Change-Id: I5ed915563cbfbc6281113c1eb31455f50710ba9f

8 years agoAUTHORS regenerated
Jim Bankoski [Tue, 29 Mar 2016 21:21:56 +0000 (14:21 -0700)]
AUTHORS regenerated

script changed to remove extra entities and clang-format bot.

Change-Id: I102cd80fdf4b240e6e4d5172943e49146a601a72

8 years agoMerge "minor updates" into nextgenv2
Yaowu Xu [Wed, 12 Oct 2016 19:25:47 +0000 (19:25 +0000)]
Merge "minor updates" into nextgenv2

8 years agoSend allow_screen_content flag for both key and intra only frames
hui su [Wed, 12 Oct 2016 18:36:24 +0000 (11:36 -0700)]
Send allow_screen_content flag for both key and intra only frames

BUG=webm:1311

Change-Id: I03c1043d17ed4e4ea22002473779a9612884c6c6

8 years agoMerge "Include fix: use aom_integer.h" into nextgenv2
Yaowu Xu [Wed, 12 Oct 2016 18:26:30 +0000 (18:26 +0000)]
Merge "Include fix: use aom_integer.h" into nextgenv2

8 years agoMerge "Add compiler flag -Wsign-compare" into nextgenv2
Yaowu Xu [Wed, 12 Oct 2016 18:26:21 +0000 (18:26 +0000)]
Merge "Add compiler flag -Wsign-compare" into nextgenv2

8 years agoMerge "LIBVPX_TEST_DATA_PATH -> LIBAOM_TEST_DATA_PATH" into nextgenv2
Yaowu Xu [Wed, 12 Oct 2016 17:56:26 +0000 (17:56 +0000)]
Merge "LIBVPX_TEST_DATA_PATH -> LIBAOM_TEST_DATA_PATH" into nextgenv2

8 years agominor updates
Yaowu Xu [Wed, 12 Oct 2016 17:50:08 +0000 (10:50 -0700)]
minor updates

1. vp8->aom
2. removed no-effect statements and spaces

Change-Id: I367d05ff9bf1b9f3c71c517c45d8049d9d4236ec

8 years agoMerge "Fix inconsistency in gm parameter write to bitstream" into nextgenv2
Sarah Parker [Wed, 12 Oct 2016 17:32:21 +0000 (17:32 +0000)]
Merge "Fix inconsistency in gm parameter write to bitstream" into nextgenv2

8 years agoInclude fix: use aom_integer.h
Urvang Joshi [Mon, 11 Jul 2016 22:51:21 +0000 (15:51 -0700)]
Include fix: use aom_integer.h

Change-Id: I98919a04bead417379e555461f67978501f922e7

8 years agoAdd compiler flag -Wsign-compare
Urvang Joshi [Fri, 8 Jul 2016 23:09:36 +0000 (16:09 -0700)]
Add compiler flag -Wsign-compare

Also, fix the warnings generated by this flag.

Conflicts:
examples/aom_cx_set_ref.c

Change-Id: I0451e119c52000aa7c1c55027d53f1da5a02a11f

8 years agoLIBVPX_TEST_DATA_PATH -> LIBAOM_TEST_DATA_PATH
Yaowu Xu [Wed, 12 Oct 2016 15:25:39 +0000 (08:25 -0700)]
LIBVPX_TEST_DATA_PATH -> LIBAOM_TEST_DATA_PATH

This commit renames LIBVPX_TEST_DATA_PATH to LIBAOM_TEST_DATA_PATH,
with a work around for working with jenkins environmnet variables.

Change-Id: If664ce57e25ad2af8121d1b578bf64043f0baa2a

8 years agoMerge "y4m_test: fix segfault if test files are missing" into nextgenv2
Yaowu Xu [Wed, 12 Oct 2016 04:26:51 +0000 (04:26 +0000)]
Merge "y4m_test: fix segfault if test files are missing" into nextgenv2

8 years agoMerge "Remove two files not in use" into nextgenv2
Yaowu Xu [Wed, 12 Oct 2016 04:26:36 +0000 (04:26 +0000)]
Merge "Remove two files not in use" into nextgenv2

8 years agoFix inconsistency in gm parameter write to bitstream
Sarah Parker [Tue, 11 Oct 2016 19:06:33 +0000 (12:06 -0700)]
Fix inconsistency in gm parameter write to bitstream

Before this change, gm parameters were being written to the
bitstream for all frames, but only read for inter only frames,
causing a bitstream error.

Change-Id: I63b8e2fdf6358e07cc00718de04cc399809bde37

8 years agoy4m_test: fix segfault if test files are missing
Tristan Matthews [Fri, 22 Jan 2016 23:05:48 +0000 (18:05 -0500)]
y4m_test: fix segfault if test files are missing

Change-Id: I7a04beb83095e5c0821048909f81f45be8b5eee3

8 years agoMerge "Remove -fno-strict-aliasing flag" into nextgenv2
Alex Converse [Tue, 11 Oct 2016 23:24:39 +0000 (23:24 +0000)]
Merge "Remove -fno-strict-aliasing flag" into nextgenv2

8 years agoRemove two files not in use
Yaowu Xu [Tue, 11 Oct 2016 22:04:47 +0000 (15:04 -0700)]
Remove two files not in use

test/cx_set_ref.sh: replaced by test/aomcx_set_ref.sh
test/vpxdec.sh: replaced by aomdec.sh

Change-Id: I74136d311eee7666e08ed8f573a17f810992fc52

8 years agoMerge "change to use aomedia copyright notice" into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 22:11:09 +0000 (22:11 +0000)]
Merge "change to use aomedia copyright notice" into nextgenv2

8 years agoMerge "Fix missing parentheses in v64_align()" into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 22:10:08 +0000 (22:10 +0000)]
Merge "Fix missing parentheses in v64_align()" into nextgenv2

8 years agoMerge "Improve v128 and v64 8 bit shifts for x86" into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 22:09:53 +0000 (22:09 +0000)]
Merge "Improve v128 and v64 8 bit shifts for x86" into nextgenv2

8 years agoMerge "Clean up and speed up CLPF clipping" into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 22:09:30 +0000 (22:09 +0000)]
Merge "Clean up and speed up CLPF clipping" into nextgenv2

8 years agoMerge "Fix typos in CLPF unit test" into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 22:06:59 +0000 (22:06 +0000)]
Merge "Fix typos in CLPF unit test" into nextgenv2

8 years agoMerge "Make generic SIMD code compile if no native support" into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 22:06:43 +0000 (22:06 +0000)]
Merge "Make generic SIMD code compile if no native support" into nextgenv2

8 years agoRefactor expand dry_run types to return coef rate
Debargha Mukherjee [Tue, 11 Oct 2016 12:26:50 +0000 (05:26 -0700)]
Refactor expand dry_run types to return coef rate

Adds the functionality to return the rate cost due to
coefficients without doing full search of all modes.
This will be subsequently used in various experiments,
including in new_quant experiment to search quantization
profiles at the superblock level without repeating the
full mode/partition search.

Change-Id: I4aad3f3f0c8b8dfdea38f8f4f094a98283f47f08

8 years agoMerge "Bugfix in CLPF RDO. Prevented selection of enable_fb_flag=0." into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 21:54:12 +0000 (21:54 +0000)]
Merge "Bugfix in CLPF RDO.  Prevented selection of enable_fb_flag=0." into nextgenv2

8 years agoMerge "Bugfix in the CLPF RDO." into nextgenv2
Yaowu Xu [Tue, 11 Oct 2016 21:53:56 +0000 (21:53 +0000)]
Merge "Bugfix in the CLPF RDO." into nextgenv2

8 years agoMerge "Read mode to mi->bmi for sub 8x8 blocks" into nextgenv2
Sarah Parker [Tue, 11 Oct 2016 21:48:01 +0000 (21:48 +0000)]
Merge "Read mode to mi->bmi for sub 8x8 blocks" into nextgenv2

8 years agochange to use aomedia copyright notice
Yaowu Xu [Mon, 10 Oct 2016 23:21:45 +0000 (16:21 -0700)]
change to use aomedia copyright notice

Change-Id: Idb2cf2555bcbe04a6650c492a3a714d7d5836b67

8 years agoFix missing parentheses in v64_align()
Steinar Midtskogen [Wed, 28 Sep 2016 18:30:24 +0000 (20:30 +0200)]
Fix missing parentheses in v64_align()

Change-Id: I16469062853c101965f56002be30ebc5823975b1

8 years agoImprove v128 and v64 8 bit shifts for x86
Steinar Midtskogen [Wed, 28 Sep 2016 15:38:46 +0000 (17:38 +0200)]
Improve v128 and v64 8 bit shifts for x86

Change-Id: I25dc61bab46895d425ce49f89fceb164bee36906

8 years agoClean up and speed up CLPF clipping
Steinar Midtskogen [Mon, 26 Sep 2016 10:51:25 +0000 (12:51 +0200)]
Clean up and speed up CLPF clipping

* Move clipping tests from inside to outside loops
* Let sizex and sizey to clpf_block() be the clipped block size rather
  than both just bs
* Make fallback tests to C more accurate

Change-Id: Icdc57540ce21b41a95403fdcc37988a4ebf546c7

8 years agoFix typos in CLPF unit test
Steinar Midtskogen [Mon, 26 Sep 2016 19:48:09 +0000 (21:48 +0200)]
Fix typos in CLPF unit test

Change-Id: Ia69bad44e47509208e3b9d306165d0872d4e92f3

8 years agoMake generic SIMD code compile if no native support
Steinar Midtskogen [Mon, 26 Sep 2016 19:05:51 +0000 (21:05 +0200)]
Make generic SIMD code compile if no native support

Change-Id: I7f691a0ae27f06ef3d727764829a60a8ffc509eb

8 years agoBugfix in CLPF RDO. Prevented selection of enable_fb_flag=0.
Steinar Midtskogen [Fri, 23 Sep 2016 10:30:50 +0000 (12:30 +0200)]
Bugfix in CLPF RDO.  Prevented selection of enable_fb_flag=0.

PSNR YCbCr:     -0.01%     -0.06%     -0.17%
   PSNRHVS:      0.01%
      SSIM:      0.03%
    MSSSIM:      0.00%
 CIEDE2000:     -0.05%

Change-Id: I1205c021bfc5cee6f80344fec92aabb529af9bd1