GitLab

The parameters are optimized only on Y, so disable chrominance
filtering for now. Later we can extend the syntax to have
separate parameters for the chrominance, or optimize the
parameters jointly over luminance and chrominance components.

lowres: -0.676% (from -0.759%) becomes a little worse
midres: -1.837% (from -1.520%) substantial improvement
hdres: pending

Change-Id: I98d71f48de98394b05fd9036de259cb43d007614

1. The ethread test can only be ran when both the encoder and decoder are
available. This is also true for ext_tile test.
2. Modify the ethread speed test range to speed 0 to 3 for AV1.

BUG=aomedia:117

Change-Id: I3c3e96fb33a7b825503b228fe43e8089e1265dd2

Replace hard coded assumptions of mode_info size with table fetch.
The codebase coding statistics remain identical. This prepares to
introduce 4x4 mode_info array.

Change-Id: I5d75b86b7e7aa5bc7104c5f6f0b326d6ad0f67bb

Replace hard coded numbers with mi_size table access.

Change-Id: Ia5b8802de9f0128edf2c71542c703f317c7f61da

The control is deprecated and is a noop.

Change-Id: I66f248485a906c26205a1b26e61e15d976207fd8

Experiment ext_tile is not compatible with daala_ec yet, this commit
disables daala_ec and enable ans if ext_tile is enabled.

A follow-up commit should resolve the underline conflicts within
ext-tile experiment.

BUG=aomedia:116

Change-Id: If60cb170e908a844dc8d22efce82f2a2c01b391b

There is currently only one codec supported by this library.

BUG=aomedia:5

Change-Id: Ic83f09b8ca783fcdf0117d338858f3fcee55e038

Change-Id: I724c87ade420bb475ccaecb933ec47513f37d482

Make the loop filter operation to support scalable mode_info size.

Change-Id: Iad30b1b6176f29c840ec38ef777a6e2533ced04f

Change-Id: If2bb7ab2b16ba44e2d6e43eeb8713aa6c05d9d7c

ext_tile is not compatible with reference_buffer, this commit changes
to disable refrence_buffer when ext_tile is enabled.

BUG=aomedia:114

Change-Id: I6416f4c731bec1c4cfeab3cdb539f359627ed2e2

A crash was reported while encoding >1080p file using best mode.
(aom issue 107 and issue 108). The cause was that the speed features
were initialized using good mode in AV1. The patch(commit f330ca)
disabled the upsampled reference usage only for good mode, but not for
best mode. So, for 1st key frame, the encoder incorrectly called
reset_use_upsampled_references() that caused the segfault.

The patch(commit e4e93b) disabled the upsampled reference usage for all
encoding modes and resolved the issue. This patch fixed issue 108 and
further made sure the upsampled references would be used correctly.

BUG=aomedia:108

Change-Id: I8362cc7cd82be324fd77f0bec315aeba509903bd

this requires _POSIX_C_SOURCE >= 1 for glibc, but its only use was with
stdout; use the file number directly

BUG=aomedia:111

Change-Id: Iebfc339c894abbf77b08c7744a4f66b34948b33f

Leave gaps for ABI compatibility for the time being.

BUG=aomedia:5

Change-Id: I8b3fd7ce667af019b81a830aeadc28781190a657

Change-Id: Ic836798f6e582ee82964fa36ac778eef0c301eb3

Accept a small compression loss is in exchange for a fixed sized encoder
side buffering requirement.

subset1:
rans_base@2016-12-02T22:55:56.809Z -> rans_nsym@2016-12-02T22:58:19.859Z

    PSNR | PSNR Cb | PSNR Cr | PSNR HVS |   SSIM | MS SSIM | CIEDE 2000
  0.0304 |  0.0303 |  0.0305 |   0.0317 | 0.0312 |  0.0309 |     0.0301

Change-Id: I09dd143e4f1638b97dc9bba7023efa837a7d48c7

entropy_stats: to generate stats of various symbols encoded
masked_tx: new experiemnt for coding a smaller transform in a
           prediction block

Change-Id: I48baf0f1c37245c43031a9eda879ef49eee5e76b

The main objective is to reduce computational complexity.
The domain transform filter has an effect of edge preserving smoothing
at a lower computational cost than the bilateral filter, and can be
readily paralelized.

A little drop in coding efficiency about 0.06% for lowres, 0.16% for
midres.

Change-Id: Id949406b7e5afe9b64588d130065c63a76e4f3f9

Uses number of 4x4 blocks using global motion as metric to
decide whether to recode or not.

Improves performance a little.
lowres: -0.658%

Change-Id: I6b1913b7bd31d7a25ca8ca5c75c9b6bb87151a78

BUG=aomedia:110

Change-Id: Icd64aebf61e945ae3ee5f95e463599c77fa55e8a

Change-Id: Id3936ffd8146c6806fbe93a5bebc23bf5cd8b5f5

Change-Id: I2e3315209559320bbae0ab7e88c9ee5df9e53ef8

Change-Id: Iaf11f439eb4baf2e0c7bf2a19c7c1c012ae7f57b

BUG=aomedia:109

Change-Id: Iaada41806771392478d3e09af249b9701e07beec

Change-Id: I76dd7365454206229e86cf5d942e89cff4a4a519

Change-Id: Id80f76ea2598daa93e693c5efe29b782abe33957

intra-interp experiment allows intra prediction to use different
interpolation filters. It was part of the ext-intra in the nextgenv2
branch.

Change-Id: I27ab692494dc79bb92e457dbf9a72988577f1c6d

Make the reference motion vector search range and ranking weight
scalable with the mode_info size. It retains the same coding
statistics.

Change-Id: I238e0ff6032fea842c3b6094287398f7ef7cb261

This reverts commit 3e882f29.

This causes build and link issues under visual studio in both x86 and
x64 configs

Change-Id: Id8b1e7e2f313817f034afddc2c627db8aa741d14

- For all blocks with width >= 16.
- Add test_count to make the unit tests harder to pass.
- Speed testing on 1080p, 100 frames, 5 Mbps, CPU, i7-6700
  User level time reduction:
   baseline:                  3.68%
   baseline + ext-partition: 36.12%

Change-Id: I78c5d9ca216f0fd91f1a360dca2190b11fd54a08

Change-Id: I846cd0c4f5e3d60c76b969183d17ce11e6ba2999

This means that all shared PVQ code is now in fixed-point.
Applies the following patches from daala:

963d3a9 pvq: fix artificial chroma boost with --enable-float-pvq
7522712 pvq: cosmetics: drop OD_ROUND32 where it is a noop
3395f61 pvq: in float pvq case, mag must default to 1.0
29193bb pvq: fix typo
c774e25 pvq: convert od_init_qm to int
b5607d6 pvq: use small LUT for integer sqrt((n+3)/2) sqrt((n+2)/2)
a989c43 pvq: use od_beta_rcp in od_gain_compand
12c31ea pvq: use od_rcp_beta in od_compute_k

Change-Id: I0ad6a896c9f9fdd752f3b65a0df313efc0db5e63

Use the proper scaling factor to map transform block offset to the
pixel offset in the reconstruction process.

Change-Id: I9af434fbc3a0a32749dec42eebb3461567943b0f

1) Not every transform's internal signal is designed to fit in 16 bits.
2) If overflow happens in this function, it indicates that we need to
adjust the txfm's scaling. We shouldn't mute the overflow signal.
3) Saturation might be handy when all of our transform design are stable,
but I don't think we are at the stable point yet.
4) This will fix C/Trans16x16DCT.AccuracyCheck/1 failure in highbd mode.

Change-Id: I5ef5d130c22adb4b8c3b608ffcb0f2c99dc7523f

The previous refactoring CLs resolve all the potential issue
related to these two conditions. It is safe to remove them now
from the cb4x4 mode.

Change-Id: Ieeb1ca2804318c1f46b60c041538f8ca5f6ba0e7

Change the table content to support 4x4 mode_info unit. Refactor
partition_plane_context() to reflect the fact that the minimum
partition block size is 8x8.

Change-Id: Id26fcfba8d74efdb9ae316c9d066cd65358dd7e6

Replace the 4x4 block counting with minimum transform block size
counting. This retains identical coding statistics.

Change-Id: I13ce6b5b62641304b0ff1c6efd34c6a590195792

With this patch affine or homography models can be enabled
by simply changing the value of the GLOBAL_TRANS_TYPES
macro in common/mv.h to 4 and 5 respectively. Currently
it is left at supporting only rotzoom. There is a small
gain with enabling affine.

Also refactors costing to change based on the model type.

Change-Id: I46c1759de06c42c176c64ec21307ff347ddcc259

Change-Id: I3d0605a314fbd469133c96c76340396810242ada

Change-Id: I5935448b15e5a6f14932ed84451874f4acbaf95a