The motion vector storage location in the reference frame is set
to be dependent on the relative frame position, instead of the
reference frame index.

Change-Id: Id8f7d1e6ba32a5ed9e182194178fd2a70094a309

Small related fix in the test itself.

BUG=aomedia:1114

Change-Id: I4a2d9c096e691e6b1a7aa9e4df93752aed8c0935

Also, updated the reference 2D transform function for this to zero out
3/4th of the coeffs and repack the 1/4th non-zeros.

BUG=aomedia:1114

Change-Id: I3dd091689a080e45db0ac4bb2595e705b63f9649

BUG=aomedia:1114

Change-Id: I4f9ae1bbc44517313494aeba68d7aab8596e92d0

At both callsites, the "rsi" parameter is the rst_info field from cm,
which is already passed.

Change-Id: I837ac655a03ebf0de6fbdaece4f4910f750e4898

This is done at the two call sites (where it's needed to avoid saving
lines unnecessarily for striped loop restoration context), so there's
no need to repeat it here.

Change-Id: I11e1ed5f50711fe1b4e8cb2101d3bfb4d16cda57

This is always called with all components and no destination buffer.

Change-Id: I76d1a16a87e05b8ecec387288139e846e9894384

This patch fixes up av1_tx_type_cost to match the code in
av1_write_tx_type. Beforehand, we wrongly assumed a 32x16 block needed
to signal its transform size (with rect-tx-ext & rect-tx-ext-intra)
because we were passing 16x16 to get_ext_tx_types.

I've also changed av1_write_tx_type to use get_min_tx_size rather than
inlining its body. No functional change, but it's probably better to
use the same helper function both times.

Change-Id: Iff6ee0bff2d332d5270fe0219db88c95e0b051d0

Make lpf_sb compatible with loopfilter_level, when USE_GUESS_LEVEL = 1.

Filter levels will be selected based on q index and applied for
filtering on Y, U, V planes separately.

Current model only allows to guess one filter level.
Now Y_vert = Y_horz = U = V. In the future, we need to retrain the
model and get filter levels for Y_vert, Y_horz, U and V separately.

When USE_GUESS_LEVEL = 0, lpf_sb can't work with loopfilter_level yet.

Change-Id: Icd774a147c07a4035cf8204a8754b2a99668bbfd

Remove unnecessary inter / intra condition check.

Change-Id: I51b776a3df6ce45b31fe3d2f43fbf6ca7b1d14fb

We should be selecting a transform size when tx_mode is TX_MODE_SELECT
and we're not lossless. (In lossless mode, the transform size is
forced to be TX_4X4).

Change-Id: I65b346fb2116689598e2251e97897be91f481ceb

The change in encodeframe.c has no effect (it's the same test), but
the change in rdopt.c fixes a miscalculated rate for (say) 8x4 blocks
which must signal whether to use TX_8X4 or TX_4X4.

Change-Id: I1dd2556bb90e86e690dd1fabaad278813ba73ea6

Change-Id: Ifa3b55d9c8022a66b6c0b1a15d3e8943cd7c69e6

This experiment is to explore the use of a forward-coded KEY_FRAME
as a backward prediction reference frame, for the coding of all other
frames in the current Golden-Frame (GF) group.

Change-Id: I9ebef9ffef4191fbb24a6cd92b819db390537471

Intrabc search area can't use pixels of the last 8 rows in the current
superblock row. Add a delay so that lpf_sb can work with intrabc.

Let lpf_sb on with intrabc while other loop filtering off (cdef, loop
restoration).

Change-Id: I98ab093a17096387b6233182da2656f90dfa09eb

Change-Id: Ia069373be864cae7a283bce3d12ee10270e6beb3

This patch enables the skip prediction feature for blocks larger than
16x16; but the tx size used in this module is still limited to 16x16.

Also makes it exit early upon large quantized coeff.

Test results:
-encoding speed up by 0~17%, on average 8.6%, based on several lowres
test sequences;
-no quality loss observed, encoding 30 frames on lowres and midres.

BUG=aomedia:1106

Change-Id: Ic2051a978eda5c8ce6a795a9d97d8d114c6edfe2

Change-Id: I47a5eda017fe0d2592411eab621934eabeae7aa4

Change-Id: Ia7d69d8582f8c37ad4b413ccd7e24711b8c3e005

There are too much if branches in get_nz_map_ctx_from_stats().
Look up the table to get the ctx offset when tx_class is TX_CLASS_2D.

Change-Id: Id08d450add5d17aeff8ab27c699d7e97ff88993d

When !USE_CAUSAL_BASE_CTX, stats (or count) <= 5, so ctx <= 3.
Remove redundant AOMMIN(ctx, 4).

Change-Id: I6a5c9201c949573ab4e96ac8ae03bde0c000e4af

This is an adopted experiment.

Change-Id: I2377d581d192ac66f45b42ddc6b58dc2d57df87b

Change-Id: I2212609532783849ac5d8c91aec6e34af48f01f7

Now it supports inter blocks only

Change-Id: I6307664452588e654853b4ff18b83cce0343268b

Fix the block context model used in the hash rd for lv-map.

Change-Id: Id590413bf8282b820ceca8b61132094df93be892

Change-Id: I5fc45fa9fe6a354ae34001f48850eb68364a5a79

Change-Id: I1e71848e9d6bc592a6d4a12051d513eb07b4acd1

Change-Id: I09b7e06ab678b4e3158525c299a8a3d4bce0bd37

Fix for the following warning:
redefinition of typedef 'Accounting' is a C11 feature

Change-Id: I1ac074e2f0bb1694d4069a20defaed4706c89c5c

Added "cm->cdef_strengths[0] == 0" in CDEF checking.

Change-Id: I5a63b4b9835665cc90966ad1ade8c3e16f0a2d79

(1) USE_GUESS_LEVEL = 0: each superblock search and apply best
deblocking filter level. Filter level is transmitted each superblock.

(2) USE_GUESS_LEVEL = 1: use q to guess a filter level for the whole
frame. Only one filter level is transmitted.

Change-Id: I585f6ca0621f5c88fc9d0d95c4fc671114b384c4

I had done only a partial fix before. But updating the costs seems to
make an even bigger improvement to the baseline.

BUG=aomedia:1094

Change-Id: Id427819507143c05b4dc3d4c05e6b2a66d17899f

subset-1:

new_dct8@2017-11-27T23:57:04.520Z -> new_dst8@2017-11-29T13:49:40.011Z

  PSNR | PSNR Cb | PSNR Cr | PSNR HVS |   SSIM | MS SSIM | CIEDE 2000
0.0034 |  0.0257 |  0.0831 |   0.0270 | 0.0023 |  0.0135 |    -0.0235

Change-Id: Ia4873465e7c70e9a0e663d4185cba3ae619c77a0

Instead of applying different offsets for each of quadri-split, horizontal split,
vertical split, store the start addresses of src and dst 8x8 blocks in each of
YUV palnes then pass it to distortion compute function for dist-8x8, dist_8x8_yuv().

Also, don't save decoded pixels for 8x8 block in additinal buffer but directly
access the xd->dst buffer. Thus removed the decoded-8x8[8x8] buffer in struct macroblock.

However, this patch does NOT fix the case where dist-8x8's assert fails with
CfL enabled.

Fix part of issue (i.e. 4:4:4 chroma iput) in
BUG=aomedia:1094

Change-Id: I6399d0eef3e17cacc9ec6467480a7959c85d419b

This patch updates also the default cdfs for skip mode. It also remove
the unneeded checking on whether the best obtained RD mode aligned
with the skip mode.

For 30 frames, on both AWCY and Google lowres/midres test sets, with
the following config, the coding gain ranges from ~-0.15% to -0.23%:

--disable-ext-partition --disable-ext-partition-types --disable-txmg
--enable-jnt-comp --enable-mfmv

Change-Id: I39aeeb9e7b27ca679c1b3f9c0eea67b5e1f694e2

This change was missing before.

Change-Id: Ie8aecb8d51284b32a02d76f6f3de5f1c77a5f692

Inter scan orders for 4x4, 8x8, 16x16 are modified to use
row/col scans for DCT_ADST combinations.
New 4:1 transforms 4x16/16x4 use row col scans
for DCT_ADST combinations.
The h2, v2, qtr scans are removed for 32x32, since they are
not used.

Change-Id: I4dab6b111858e14b3c5b92b5c46124c69e06808d

The monochrome bit has been removed from the end
of the sequence header.
Monochrome is now coded as a type of colorspace.

Encode in monochrome by using
  aomenc --color-space=monochrome ...

Change-Id: I9550de58eb3d67dae0eb651697cb63dc8b137931

This should give a slight speed-up for streams with CDEF disabled,
since it avoids having to test the same for every filter block.

BUG=aomedia:1108

Change-Id: I7f3f2ddfe5414ed1b4b6aafb5e28ae61df2cb79e

BUG=aomedia:1093

Change-Id: I0871c67d9a2ddc6aa9d3084f645305f6b74a999f