Commit Graph

3371 Commits

Author SHA1 Message Date
Ronald S. Bultje
f72fdf1c7f Fix another typo in 4x4-transform-for-i8x8-intra-pred coeff contexts.
Change-Id: I02e7f64a7c99f6c222b9ef21341b0e801d646810
2012-10-22 11:55:06 -07:00
Ronald S. Bultje
84ea002eb1 8x8 transform support in splitmv.
For splitmv, where partitioning is 8x16, 16x8 or 8x8, this patch
uses the 8x8 transform (instead of the 4x4) if txfm_mode is
ALLOW_8X8 or ALLOW_16X16. For TX_MODE_SELECT, splitmv can indicate
which of the 2 transform sizes (4x4 or 8x8) it wants to use.

Gains (with hybridtx4x4/8x8/16x16 and tx_select experiments
enabled) on derf: +0.9%, HD: +0.4%, STD/HD: +0.8% (SSIM or overall
PSNR, both metrics show similar improvements).

Change-Id: Ide954b000b415548ed92a7ac78e24f36e60fcf06
2012-10-22 11:49:00 -07:00
Scott LaVarnway
372e001d68 Merge "Re-enable specialized sse2 block loop filter functions" into experimental 2012-10-22 11:39:08 -07:00
Scott LaVarnway
e095785820 Merge "sse2 intrinsic version of vp8_mbloop_filter_vertical_edge()" into experimental 2012-10-22 11:38:37 -07:00
Scott LaVarnway
bb550d90f7 Merge "sse2 intrinsic version of vp8_mbloop_filter_horizontal_edge()" into experimental 2012-10-22 11:38:21 -07:00
Ronald S. Bultje
5a80e7a292 Use SPLITMV_PARTITIONING instead of a plain integer type.
This can be used to distinguish between 16x8, 8x16, 8x8 and 4x4
partitioning modes when choosing splitmv as a MB mode.

Change-Id: Idc8b59772e1a80ccc4ad44d63c5c2ec3fc061a3c
2012-10-22 11:25:48 -07:00
Ronald S. Bultje
09fb253b39 Merge "Invert SB probability." into experimental 2012-10-22 11:18:01 -07:00
Jim Bankoski
8fc5774f8b Merge "Remove variance vtable from rtcd" into experimental 2012-10-22 11:12:07 -07:00
Ronald S. Bultje
ef50e48323 Invert SB probability.
It currently counts the probability that the branch is true, but it
should count the probability that the branch is false.

Change-Id: I963825da2e7a7ed3a613eb23ffd085e427dc36e5
2012-10-22 11:05:56 -07:00
Ronald S. Bultje
6e5cb12886 Merge "Fix use of wrong context in i8x8 coefficient cost calculation." into experimental 2012-10-22 11:03:26 -07:00
Jim Bankoski
118b2fe962 Remove variance vtable from rtcd
Change-Id: Idd2722a538423b451e1e3495f89a7141480493d6
2012-10-21 20:47:57 -07:00
Deb Mukherjee
788d7f10d4 Merge "Allow B_VL_PRED & B_LD_PRED modes with Superblocks" into experimental 2012-10-21 15:35:24 -07:00
Deb Mukherjee
57679badf2 Allow B_VL_PRED & B_LD_PRED modes with Superblocks
Allows B_VL_PRED & B_LD_PRED modes to be used for all blocks
within a MB in B_PRED mode. These modes were temporarily
disabled with super-block coding.

Change-Id: I973b9bdb82c3da5f12d7cc963162a28805f25303
2012-10-20 21:28:38 -07:00
Ronald S. Bultje
0d4e82aa4e Fix use of wrong context in i8x8 coefficient cost calculation.
Change-Id: Ica58d0c7d8addb5ddd5d6e1fdea8f0f96e110862
2012-10-19 22:05:50 -07:00
Scott LaVarnway
a1571b8c11 Re-enable specialized sse2 block loop filter functions
vp8_loop_filter_bh and vp8_loop_filter_bv

Change-Id: I0a77d38c701a6cce96cb8371c7f3ab7e2e7f878c
2012-10-19 16:02:50 -07:00
Scott LaVarnway
085433c2d0 sse2 intrinsic version of vp8_mbloop_filter_vertical_edge()
First sse2 version of vp8_mbloop_filter_vertical_edge().  For now,
intrinsics are being used until the bitstream is finalized.  This function
will be revisited later for further performance improvements.

For the test clip used, a 34+% decoder performance improvement
was seen.  This will vary depending on material.

Change-Id: I455b438bc8d8af76cf7533ac42eda5f689b21f7c
2012-10-19 15:52:12 -07:00
John Koleszar
91dc8ce70d calculate probs consistently
There were several different methods for calculating bitstream
probabilities in use. Consolodate these into a pair of functions,
get_prob() and get_binary_prob().

Change-Id: I5534f517f74027fee16d89c9baefaafea8156b2f
2012-10-19 12:31:16 -07:00
John Koleszar
6e364a57c0 lint-hunks: better support for working tree
When run with no arguments, report warnings in the diff between the
working tree and HEAD. With arguments, report warnings in the diff
between the named commit and its parents.

Change-Id: Ie10dcdecb303edf8af51bad645cc11206a1fc26b
2012-10-19 12:31:16 -07:00
John Koleszar
e9fd1eace1 Remove bc, bc2 from pbi,cpi,xd
Pass the bool coder to be used explicitly. This avoids cases where two
different bool coders can be addressed from the same function. Also be
more consistent with bool coder variable naming, start to standardize
on 'bc'.

Change-Id: I1c95e2fdbe24ebe8c0f84924daa1728e3b054a31
2012-10-19 12:31:15 -07:00
Deb Mukherjee
f3208f362b Some cleanups and fixes.
Separates the logic on transform type selection previously spread out
over a number of files into a separate function. Currently the tx_type
field in b_mode_info is not used, but still left in there to eventually
use for signaling the transform type in the bitstream.

Also, now for tx_type = DCT_DCT, the regular integer DCT is used, as
opposed to the floating point DCT used in conjuction with hybrid
transform.

Results change somewhat due to the transform change, but are within
reasonable limits. The hd/std-hd sets are slightly up, while derf/yt
are slightly down.

Change-Id: I5776840c2239ca2da31ca6cfd7fd1148dc5f9e0f
2012-10-19 06:58:15 -07:00
Ronald S. Bultje
b44397ebcd Merge "Fix encode breakout skip handling." into experimental 2012-10-18 15:21:30 -07:00
Ronald S. Bultje
1f11d13e5e Merge "Detect skippable blocks in splitmv RD loop case." into experimental 2012-10-18 15:21:14 -07:00
Scott LaVarnway
992b5e2d95 sse2 intrinsic version of vp8_mbloop_filter_horizontal_edge()
First sse2 version of vp8_mbloop_filter_horizontal_edge().  For now,
intrinsics are being used until the bitstream is finalized.  This function
will be revisited later for further performance improvements.
For the test clip used, a 31+% decoder performance improvement
was seen.  This will vary depending on material.

Change-Id: I03ed3a7182478bdd1f094644ff3e0442625600e7
2012-10-18 14:29:26 -07:00
Ronald S. Bultje
637c4d2ec8 Merge "Use correct context for optimize_b() in vp8_optimize_mby_8x8()." into experimental 2012-10-18 12:39:10 -07:00
Deb Mukherjee
96276f97ec A fix for crashes when #macroblock rows is odd.
Change-Id: If9fc2df4051ccfb8d804dae9d1be08d12d0f81ce
2012-10-18 12:00:23 -07:00
John Koleszar
13d69d40da lint-hunks: exit status for only affected lines
Prior to this patch, if there were any lint errors, this script would
exit with an error, even if those errors were not in the hunks being
tested by this script. This change makes it so that if any lint lines
are printed, an error is returned.

Change-Id: I69c8bef4367ccf25d287508f29e587b1f4426143
2012-10-17 21:43:29 -07:00
Ronald S. Bultje
87d2b2549c Fix encode breakout skip handling.
Change-Id: Ib42a5dbded27abf2a01fd656a47a1e96b1fe2912
2012-10-17 18:00:56 -07:00
John Koleszar
2b6fd28dfe Merge "lint-hunks: support operating on arbirary revs" into experimental 2012-10-17 16:44:46 -07:00
Ronald S. Bultje
b276efd002 Detect skippable blocks in splitmv RD loop case.
Change-Id: Id8a7265dc721e5cdcaa144e0041beeb32a98f51c
2012-10-17 15:45:00 -07:00
pascal massimino
07c03b3f23 Merge "cosmetic: fix '= & ' construct" into experimental 2012-10-17 15:17:21 -07:00
John Koleszar
e17bb9e39b Merge "rtcd: merge rtcd.c from master" into experimental 2012-10-17 15:16:20 -07:00
John Koleszar
cdd0ed1352 Merge "consolidate update_mb_segmentation_map data" into experimental 2012-10-17 15:15:59 -07:00
John Koleszar
c6f13f3278 Merge changes Ic711a378,I43a46c36,If19ee74a into experimental
* changes:
  Move remaining per-frame data into partition 0
  Interleave modes/residual per macroblock
  Force interleaved decoding
2012-10-17 15:15:10 -07:00
John Koleszar
3883dab0a2 cosmetic: fix '= & ' construct
remove useless space after address-of operator.

Change-Id: I1fb9e82e8d6cf87558fbd454fb5c0f87599ca2ab
2012-10-17 14:51:27 -07:00
John Koleszar
47a2154c0e Move remaining per-frame data into partition 0
This commit moves a bit of data that ended up packed with the
modes/mv/residual partition during the change to interleaved encoding
into partition 0 where it belongs.

Change-Id: Ic711a378c58d9d6a17254384f492c213a15bad92
2012-10-17 14:48:13 -07:00
John Koleszar
b1cb0077f2 lint-hunks: support operating on arbirary revs
Rather than diffing only the index, support checking arbitrary commits.

Change-Id: Ia135a487990d8293d1e0799dc062b9f49e020b25
2012-10-17 14:23:08 -07:00
Ronald S. Bultje
3182ab619d Use correct context for optimize_b() in vp8_optimize_mby_8x8().
Change-Id: Ic975ab85a9924adc7b3d421f64155cc79f40ffd0
2012-10-17 14:12:59 -07:00
John Koleszar
6ef5a00538 Interleave modes/residual per macroblock
Packs the bitstream with each mb's residual following its mode/mv
information.

TODO: There are still a few fields that should be packed into partition
0 but are included in partition 1, due to them being serialized from
write_kfmodes/pack_inter_mode_mvs, which execute after the first
partition is finalized. These need to be separated out into a separate
function, similar to mb_mode_mv_init() in decodemv.c.

Change-Id: I43a46c363601ab36954d07ebe498760e1e2e3af4
2012-10-17 13:19:41 -07:00
Ronald S. Bultje
eab804558b Use 8x8 block bases for non-4x4 splitmv RD.
Change-Id: I76095d5a02edcc5d404ea830157780b24fd509d7
2012-10-17 11:40:00 -07:00
Scott LaVarnway
edf099d272 Merge "Removed the loopfilter rtcd invoke macro code" into experimental 2012-10-17 09:57:23 -07:00
Jim Bankoski
ffff213463 removed obselete build dependency
this commit fixes the build on windows with visual studio 2008.

Change-Id: I0baa4044e9e54237da29f2e17332ea6f766dbbec
2012-10-17 09:22:05 -07:00
Scott LaVarnway
15ce6bd62e Removed the loopfilter rtcd invoke macro code
Change-Id: I446b2ffcbe732ffb112dbd97a4799272d4c01a84
2012-10-16 16:19:35 -07:00
John Koleszar
c073e5ca96 Force interleaved decoding
Rather than decoding all modes/mvs separately, decode them per MB. This
forces the mode which was already used form the CONFIG_NEWBESTREFMV and
CONFIG_SUPERBLOCKS experiments, and is a precursor to changing to
interleaved encoding.

Change-Id: If19ee74ac8a987846d1cd0cf2b2e02a82f1a43ad
2012-10-16 14:08:40 -07:00
Jim Bankoski
9a1962210d Merge "removed the recon rtcd invoke macro code (unrevert)" into experimental 2012-10-16 13:49:22 -07:00
Marco Paniconi
4cf4c94ad1 Fix to rd cost computation for mv bias.
This prevents the rd cost from wrapping around and going negative.

Change-Id: Id61390537d2ad47762e410918bb4e18f6a0912d4
2012-10-16 13:37:38 -07:00
Ronald S. Bultje
9443f05e6b Fix a mismatch if a keyframe contains skip-blocks.
Change-Id: I0681d3183f51627be8c2bb76f343b7270f9116d8
2012-10-16 12:41:56 -07:00
Jim Bankoski
7c15c18c5e removed the recon rtcd invoke macro code (unrevert)
This reinstates reverted commit 2113a83157

Change-Id: I9a9af13497d1e58d4f467e3e083fddf06b1b786c
2012-10-16 12:02:31 -07:00
Ronald S. Bultje
d5955a4231 Merge "Remove test from B_MODE_INFO." into experimental 2012-10-15 19:50:20 -07:00
Ronald S. Bultje
dd3646466e Merge "Remove mode_rdopt from MB_MODE_INFO." into experimental 2012-10-15 19:50:06 -07:00
Ronald S. Bultje
bcaba2e5cb Merge "Properly use chroma block object for chroma coefficient costing." into experimental 2012-10-15 19:49:37 -07:00