GitLab

Change-Id: Ie836a113978028f3bde2acd31061d9a663547087

Adds the arguments struct HandleInterModeArgs to hold arguments that
are conditional on compiled features. This means that there are no
longer #if's in the function's argument list.

Some of the array pointers that were optional arguments have been
made array members in the new struct, but not all. This is due to the
function being called with either references to arrays that are
maintained between trying different modes OR with references to
"dummy" arrays initialized to zero. The arrays that are always used
are now members of the HandleInterModeArgs struct.

Change-Id: I3076fd53c3cddf5a6d14bbe7d23a889465ed716d

Offsets for the least-squares for affine motion computation
are now set at the top left corner of the current block.

Improves stability and performance a little.

Change-Id: I68ca7e74c6102502daa8ca3373af2b2dd59400c3

Disable the support of compound prediction modes for sub8x8 codking
blocks. Make the rate-distortion optimizations process account for
such constraints.

With the use 2x2 chroma prediction block, this makes the wrost case
number of inter predictors same as vp9. It affects the coding
gains by 0.35% for lowres, 0.17% for midres, and 0.08% for hdres.

The encoding speed is up by 10%.

Change-Id: Ieb2a83030676911baa403e586f1f800cbf485d81

Segmment based lossless flag is used in select transform size, this
commit fixes a bug where wrong segment_id is used in such selection.

BUG=aomedia:350

Change-Id: Ibc981c779739849bac00447155180abbd319eb28

Also supports homography models for future experiments.

Change-Id: I4510540f54133e063891ed491c95c087222f7810

USE_TXTYPE_SEARCH_FOR_SUB8X8_IN_CB4X4 macro added to turn
tx_type search on/off for sub8x8 in cb4x4 mode.

The purpose is mainly to analyze the coding gains from cb4x4
but this later can be made into a speed feature as well.

Change-Id: Ic22026c373eebba87f324689ac5686a2844315b6

Integerizes computation of the least squares for warped motion.
The model is restricted to only Affine. Affine seems easiest
to compute and integerize since it can be split into two 3-dim
least squares problems, as opposed to rotation-zoom which needs
a 4-dim least-squares problem to be solved.
The current implementation requires only one division per block.

BDRATE impact is mminimal. The upgrade to the affine model improves
coding efficiency but integerization also degrades efficiency a
little. Overall there is a net gain of about -0.07% BDRATE on
the lowres set.
BDRATE lowres: -1.113% with ----enable-warped-motion vs. without
(up from -1.044%).

Change-Id: I6b9216ac0737d76f59054293eabee48e17739ec4

Change-Id: Ifcdd3ce2953c1ecb1d0962da412a4b5ba2cda912

Adds the static inline function get_plane_type to convert a plane number
to the corresponding PLANE_TYPE.

There's no change to the bitstream, it only encapsulates the logic to
get the PLANE_TYPE.

Change-Id: I1199db3a32c89437d9c029ab5b2b2e62582a13a2

Change-Id: Icd4cd42b1b9fa9066cda3e33e487298c8d802546

Skip redundant computation of recon_variance value.

Change-Id: I0ea1a3528f8bdc7095d9cc458ce0bd3a958989b4

Support the use of 2x2 prediction and transform block size for
chroma components. This experiment builds on top of cb4x4. It
brings back the loss in performance gains due to 4x4 chroma
component constrain. The coding gains for lowres are up from 1.5%
to 2.5%.

Change-Id: I8e461e230c2d920f6df1af9d935ebe9dfa3820d9

Since gm_get_motion_vector is trying to give a motion vector for
"the frame as a whole", it makes more sense for it to calculate
the motion of the *center* of the block rather than the top-left
corner of the frame.

In theory, this change should also help the encoder make better
decisions on when to use global motion. It avoids an issue
where, early in the frame, NEARESTMV looks like a good way to use
the global motion vector *without* paying the rate cost applied
to the first few global motion blocks in each frame. This seems
to lead to a better overall result.

Change-Id: Ia5c6259ceb8b4ff3d00a5d553e1d18bdb802da59

If there no sub-pixel in both direction of MV,
the interp_filter is set to EIGHTTAP_REGULAR

Change-Id: I208c34906544f1937c57d78017999be6eddd9e87

Change-Id: If5d4f6d261276d9cfdd59fd7779467d92f56c395

Change-Id: I1d599b70fb338de4b0d2076fdc7060361a46c21f

Skip coding the intra mode for chroma component for sub8x8 blocks
that are not in the reference position.

Change-Id: Ic3daab2668ab41c3b743664faf195e34124ead33

Make the rate-distortion optimization account for the special order
for sub8x8 block chroma component coding.

Change-Id: I050abc50dda7b9e8e32f7d0959aa954df1d69398

Skip the 2x2 uv block rate-distortion optimization process in the
non-reference positions in key frame encoding.

Change-Id: I51112492fd79f9a1c205fba151f8658cea9ee85f

Here we create a default set for the ext-tx experiment which
includes:

Intra: {DCT, ADST} X {DCT, ADST} + IDTX (identity in both directions)
Inter: DCT_DCT + IDTX

This set is used if indicated in the uncompressed header.

Change-Id: Ic907ada12b4e0c98cabb7c9650ab99942d0c0859

Use the top-left 4x4 luma block's coding information for chroma
component encoding at size 4x4.

Change-Id: I4bcfbc2bf8b71f7fc30094553503468460a56f9b

Adds functinoality for least-squares, RANSAC as well as encoding and
decoding with new constrained homographies that warp blocks to horizontal
and/or vertical trapezoids. This is for future experimentation. None
of the models are actually enabled in the code.

Change-Id: I1936018c6b11587d6fd83c3a2c63548cb641b33f

Which was a workaround to balance the quality between
luma and chroma when PVQ is enabled, by
1) skipping PVQ DC RDO for chroma and
2) using smaller lambda value (x0.5) for intra prediction of chroma,
intended to spend more bits and thus improve quality for chroma.

The change in metric for 5 frames of objective-1-fast is:
PSNR  PSNR-HVS  SSIM  CIEDE-2000  PSNR Cb PSNR Cr  MS-SSIM
-1.45 -1.54     -1.46 4.62        13.71   14.87     -1.47

For still images (subset1)
PSNR  PSNR-HVS  SSIM  CIEDE-2000  PSNR Cb PSNR Cr  MS-SSIM VMAF
-1.52 -1.54     -1.60 2.56        11.92   12.17     -1.52  -1.66

Change-Id: I31c4d7a7de2939f6004c112649f7303f0e944156

Since token coding has changed the relationship
between EOB values and non-zero values, coeff costs
and trellis quant need to change to be more accurate.

Change-Id: I27ef400e8290db4c5faa9c21a253575eea7955c4

Change-Id: Ib1b5dde5d4c6479c968cd7d7a4a1d01abae1fcf6

Change-Id: I1bf9002c159563d0e5238b26d1515ef305d416dc

This CL allows us to use different interpolation filters for
YUV planes.

Change-Id: I9446d43ae2be5a9a48b8a4cb6efcac43df6196d5

The only compound mode used with sub 8x8 blocks is COMPOUND_AVERAGE, so
we don't have to send anything in this case

Change-Id: I90d0162e5f7f1ad205e65094293cde2a48eb77b1

BDRATE results:
lowres: -0.880% (up from -0.844%)

Change-Id: I017c0beddcc687148fed33c1e9963e05f1eaf6ea

Piecewise linear fit without activity masking on subset3 intra
and objective-1-fast inter, by simple linear regression.

In combination with 79c0f32c "Remove DCT from od_compute_dist_8x8",
this gives a even trade between perceptual and non-perceptual metrics.

av1_daala_dist_30f@2017-01-23T21:32:27.902Z
 -> daala_dist_scale10_30f@2017-01-31T21:52:07.635Z

  PSNR | PSNR Cb | PSNR Cr | PSNR HVS |   SSIM | MS SSIM | CIEDE 2000
2.1080 |  2.9645 |  3.4697 |  -2.2086 | 0.2541 | -2.5232 |     2.1645

Piecewise linear-quadratic fit with activity masking, by same method.

The total effect of activity masking and daala-dist changes, with PVQ:

av1_pvq_5f@2017-01-31T01:05:24.219Z
 -> av1_float_pvq_dist_scale_AM_5f_Jan31_crfix@2017-02-02T15:14:40.477Z

   PSNR | PSNR Cb | PSNR Cr | PSNR HVS |    SSIM |  MS SSIM | CIEDE 2000
22.5041 | 42.6349 | 40.8516 | -10.3510 | -6.8030 | -16.9057 |    21.2613

Change-Id: I9b513509a03aa058dc5c1479c01d62c8fc363a34

rd_pick_palette_intra_sby() was being passed a dummy 'best_mbmi' in
one case, and later that dummy value could be assigned to 'mbmi'
causing invalid values in 'mbmi'.

This bug was introduced by:
de0c70a2: Refactor rd_pick_intra_sby_mode()

See the referenced bug for a repro case.

BUG=aomedia:337

Change-Id: If6b0aa329f4e4902cd4de8a5e09bc5d9196492a1

Allow the above combination of experiments to work together
correctly, fixing an encode/decode mismatch bug when they
were all enabled.

This change causes build_masked_compound(_highbd) to only
ever be called if CONFIG_SUPERTX is off, so wrap these functions
in an '#if !CONFIG_SUPERTX' block.

BUG=aomedia:313

Change-Id: Ic3886bc69ba9624b8fcb0a4c2d71fc64d2c0f22c

Change-Id: I2a490e144099d7692296992528192c1f11d2c06f

- Added comments for some tables and #defines for clarity.
- Renamed some variables to ensure we use "color_index" instead of
"color" for palette color index related variables.

Change-Id: Ica95a26e0f171a41a3259c8e6b3b891b8cd10151

This reduces the complexity in a number of ways:
- We need just 3 neighbors instead of 4.
- Possible contexts reduce from 16 to 5.
- On hardware side, getting the contexts for a whole block will be more
parallelizable.

At the same time, compression performance improves very slightly:
- Screen-content set (videos) (Google): BDRate improved by 0.32
- screenshots set (images) (AWCY): PSNR improved by 0.62:
https://arewecompressedyet.com/?job=palette_withTR2%402017-01-27T21%3A30%3A28.890Z&job=palette_noTR2%402017-01-27T21%3A41%3A34.312Z

Change-Id: Ie84ca32f05d55ad481a51c2d3abc579468597189

Cherry-pick Daala e248823a
 Getting rid of the DCT in od_compute_dist_8x8()
Replacing the DCT and frequency weighting by a filter

Change-Id: Icc3a46e5dbb561e4e3b00fa6c2290d54299c05cb

BUG=aomedia:314

Change-Id: I66af7f69ca0b97b9d840918a6b9ec34708a7f4e5

Change-Id: Id86b48ad34059668beb9464200dd9e03fc1b8a48

If part of a block falls outside right and/or bottom image boundary,
then only store tokens for the part of it within the boundary.

Also, consider only the part of the block within the boundary when
calculating the number of colors in the image, deciding the base
colors for palette, RD calculation etc.

The part of color map corresponding to pixels outside the image
boundary is padded with color indices copied from same row/column.
This behavior is similar to how pixels outside the boundary are padded.

For screen_content set, this is improves compression performance by
0.038 overall. One clip, in particular, has a significant gain of 0.8.

Change-Id: I745ca032f313c5041aacc98c03ae4bfc33d840de