GitLab

Makes a set of 16 transforms total, adding all 1D
combinations of ADST and FlipADST, and removng all DST
transforms.

lowres, midres both improve by about 0.1% and hdres by
-0.378% in BDRATE but with fewer transforms that are also
simpler.

Further experiments to continue later.

Change-Id: I7348a4c0e12078fdea5ae3a2d36a89a319ffcc6e

Change-Id: Iff04c82b3d3b5cf2c7700717c3c3d678bbbb9f9b

This commit re-designs the probability model for the syntax elements
of the dynamic motion vector referencing system.

Change-Id: Icfb8203c7e8f64e10e99f5890e25e6f6b15fe5d1

Not getting good results for this feature, will try again
when transforms are frozen.

Change-Id: Id12396786cb9369ad34d0bd845f7beba3a037726

Change-Id: I1d5624fb2f34f87a55613036851034ec7c2d0b76

This commit enables the dynamic motion vector predictor for NEWMV
mode. It allows the codec to select the best motion vector predictor
in a rate-distortion optimization framework for motion vector
residual coding. The compression performance is improved:
lowres  0.14%
midres  0.27%
hdres   0.24%

Change-Id: I6a601c74eb6cb0b71a613336d40363359f2edecd

Simple refactoring to 2 dimensional arrays, in preparation for 128
wide superblocks.

Change-Id: I40d447bd9fbd4f755534ea3cc82fc8f4676cea07

This is in preparation of increasing the superblock size.

Change-Id: I9197e397399fbe8aec1178a45ea0337dd90412d7

No performance change observed from borg tests.

Change-Id: I20c232c5dde8cfc84452a4c7185389bd8b812ffd

Change-Id: I52c209a5db1b4a6525c04b70291a08ab5a68c6fe

Change-Id: I71da102550aa7d81961e5f10c71058d5fa8dc6c4

Change-Id: Id52744e140585c08d047fb395b75ac9318a0b4c3

This commit fixes the computation of rate_nocoef for situation when
rate_y is uninitialized at INT_MAX for  x->skip is true.

Change-Id: If3dde4e4ee16667f4408067d3bb3084f916272f1

PSNR test
        lowres  hdres
lowbd   -0.013  0.067
highbd  -0.044  0.039

Change-Id: Iefdb1e966bd004b2027456778185b675e8fb9b81

Instead of testing all interpfilter-BMC/OBMC combinations, we choose
the best interpolation filter based on regular inter prediction.

Reduction in encoding time: ~10%
Drop in performance gain: 0.08% lowres, 0.04% midres

Change-Id: Ifc19097a918ac76b529db9af4c60e2c70e93f7ad

Using the up-sampled reference frames in sub-pixel motion search is
enabled as a speed feature for good-quality mode speed 0 and speed 1.

Change-Id: Ieb454bf8c646ddb99e87bd64c8e74dbd78d84a50

Change-Id: I6fab352eb1f7d9c5dc783a4d4d878b6b42838ca2

Coding gain on screen_content is 12.2% (was 6.6%).

Some features such as frame-level color buffer, adaptive
entropy coding, are coming in future patches.

Change-Id: I2658cf5ec0cbb02cff685475759f3b68c9807697

Filled in prune one and prune two. Prune three is still
being experimented with.

Change-Id: Ic07f828c448e86cacb0369aa3a9a0feb2edae054

Incorporates wedge compound prediction modes.

Change-Id: Ie73b54b629105b9dcc5f3763be87f35b09ad2ec7

Change-Id: I5df744dc6b21ed9dbbf6ddf38004f2a9e88b7d00

Change-Id: I442475e559be2acdc1c2a3e5ca021b3de77adda5

No performance change.

Change-Id: If35125fed909d89235b303514f77a33183bb36b3

Change-Id: Iccaacee9b7a66b016b5747a3902c236888ad4ba1

Pixel domain distortion calculation is enabled for the rd loop of
inter sub8x8 and intra 4x4 cases.

Coding gain: 0.124% derflr, 0.122% derfhd

Change-Id: I43b47fe81b4f5ccc1c66bc626bd310c413a1ed87

This sets up the interface for 3 speed features that progressively
eliminate a greater number of transforms in ext tx using
pre-trained support vector machines.
Each speed feature still needs to be implemented.

Change-Id: Ia508aeadc0cffdc080fb227f357a5d1dfbca08e2

This commit fixes an encoding issue related to var-tx and ref-mv
experiments that causes the codec to use random values for transform
block skip flag.

Change-Id: I8daa6d6b88ea45b5bbeb81b43dd0eeff545c8e5a

The sum of squared value of a block can overflow 32bit, this commit
changes to use int64_t to avoid the overflow issue.

Change-Id: I78fcd6999634f186f86d649cfce85d97a993d040

Up-sampled the reference frames to 8 times in each dimension using
the 8-tap interpolation filter. In sub-pixel motion search, use the
up-sampled reference frames to find the best matching blocks. This
largely improved the motion search precision, and thus, improved
the compression quality. There was no change in decoder side.

Borg test and speed test results:
1. On derflr set,
Overall PSNR gain: 1.306%, and SSIM gain: 1.512%.
Average speed loss on derf set was 6.0%.
2. On stdhd set,
Overall PSNR gain: 0.754%, and SSIM gain: 0.814%.
On hevchd set,
Overall PSNR gain: 0.465%, and SSIM gain: 0.527%.
Speed loss on HD clips was 3.5%.

Change-Id: I300ebaafff57e88914f3dedc8784cb21d316b04f

Includes various cosmetic changes and refactoring including
naming the sharp filters differently (since they are no longer
8-tap).

Change-Id: Ida5a19ca0daa9f6a64a6734394c685b2a4a2564a

The interintra experiment, which combines an inter prediction and an
inter prediction have been ported from the nextgen branch. The
experiment is merged into ext_inter, so there is no separate configure
option to enable it.

Change-Id: I0cc20cefd29e9b77ab7bbbb709abc11512320325

Adds hooks to use 32x32 ext-tx. Also adds scan orders for the masked
transforms for 32x32.
Make macro USE_MSKTX_FOR_32X32 1 in blockd.h to support 32x32 masked
transforms for ext-tx.

Change-Id: Ie6564830266651fcafae2d536c274dafd664ce17

This macro works for the shift parameter being 0.
The ROUND_POWER_OF_TWO macro does not.

Change-Id: I8434d2933892e09bbc0d2dafc934d0c3637df347

Instead of using model_rd_for_sb() to estimate the cost and make the
decision on bmc/obmc, we use super_block_yrd/uvrd() to calculate and
compare the real rd costs of bmc and obmc.

Average bit-rate reduction(%) of obmc experiment:
derflr/derfhd/hevcmr/hevchd
2.353/TBD/TBD/TBD
Before the optimization, the coding gain was:
1.582/1.109/1.600/1.164

Note: there is still some mysterious bug because that compared to
the previous version, the performance at low bit rate drops a lot.

Change-Id: I8dbee04a272190f10516a3953c1ae690f8136766

Change-Id: Ib29f510289716b5ab5c7d74d32a450c190308a83

This commit accounts for the context based probability model for
motion vector cost estimate in rate-distortion optimization.

Change-Id: Ia068a9395dcb4ecc348f128b17b8d24734660b83

This commit converts the scalar motion vector probability model
into vector format for later precise estimate.

Change-Id: I7008d047ecc1b9577aa8442b4db2df312be869dc

Change-Id: I51307a217eeba14dbdaa2522be474530316a4faa

This fixes a bug in HBD sum of squared error computation introduced
in  #abd00505.

Change-Id: I9d4e8627eb8ea491bac44794c40c7f1e6ba135dc

Preliminary tests indicated that these changes make cost_coeffs
approximately 20% faster which is a 2% improvement overall

Change-Id: Iaf013ba75884415cd824e98349f654ffb1c3ef33