Граф коммитов

17415 Коммитов

Автор SHA1 Сообщение Дата
Yunqing Wang e9947a8d2d Remove unused buffer allocation functions
Removed unused buffer allocation functions.

Change-Id: I5aa265a7698a2d2df736ddb89c6d93a5ee34895b
2016-08-29 15:02:09 -07:00
Debargha Mukherjee 162f5f792b Merge "Tiling in loop restoration + cosmetics" into nextgenv2 2016-08-29 20:46:13 +00:00
Debargha Mukherjee 100846a8ac clpf experiment build fix
Change-Id: I729e14916ecb58b5a75756078ab96a2d340bc0d6
2016-08-29 12:28:00 -07:00
Aamir Anis e40e6e576a Tiling in loop restoration + cosmetics
Frame can be split into rectangular tiles for application of separate
bilateral or Wiener filters per tile. Some variable names changed for
better readability.

Change-Id: I13ebc4d0b0baf368e524db5ce276f03ed76af9c8
2016-08-29 11:24:11 -07:00
Debargha Mukherjee 8ee5ab9f13 Fix for supertx with rect-tx
Change-Id: I0cc3523a8992f889f8dd203449ceb55f2a422324
2016-08-29 11:16:17 -07:00
Pascal Massimino 04ed7ad57b fix uint32_t <-> size_t mismatch in tests
Change-Id: Ifde4d57957219560e01ebd1657f1c0721f041054
2016-08-29 09:09:09 +02:00
James Zern f7a865859b Merge "vp10_alloc_context_buffers: clear cm->mi* on failure" into nextgenv2 2016-08-27 16:43:53 +00:00
Jingning Han 003dff6962 Merge "Fix VS build warnings in blend_a64_mask_test.cc" into nextgenv2 2016-08-26 21:47:39 +00:00
Hui Su 976a9b9304 Merge "Remove unnecessary buffer pointers in PICK_MODE_CONTEXT" into nextgenv2 2016-08-26 21:38:14 +00:00
Jingning Han 91ae5d954a Fix VS build warnings in blend_a64_mask_test.cc
Change-Id: Id4c764198549a60d98e5c4a74083972b97da5b81
2016-08-26 11:25:08 -07:00
Debargha Mukherjee 8b7e4dbaf4 Fix compile error in dering
Change-Id: I56890c813de1b366e4ef482d9fc6da81034636ab
2016-08-26 10:48:16 -07:00
Geza Lore d21982c80f Use rectangular transforms for >= 8x8 blocks
For rectangular blocks between 8x8 and 32x32, we can now code the
transform size as one bigger than the largest square that fits in
the block (eg, for 16x8, we can code a transform size of 16x16
rather than the previous maximum of 8x8), when this oversized
transform is coded in the bitstream, the codec will use the full
size rectangular transform for that block (eg 16x8 transform in
the above example).

Also fixes a scaling bug in 16x8/8x16 transforms.

Change-Id: I62ce75f1b01c46fe2fbc727ce4abef695f4fcd43
2016-08-25 17:31:51 -07:00
Sarah Parker a97fd6c43e Merge "Update VP9_PROB_COST_SHIFT to VP10_PROB_COST_SHIFT" into nextgenv2 2016-08-25 18:31:22 +00:00
Wei-ting Lin 4c7e1cd973 Separate EXT_ARFs' frame context index
This commit separate the frame index of EXT_ARFs' from other frame
types in the ext-refs setting.

It improves the average RD performance by

0.206% in the lowres, and
0.173% in the midres.

The overall gains for the ext-refs compared to the baseline are

5.665% in the lowres, and
4.883% in the midres.

Change-Id: I6591ad29120880c1aef0bd0b7cf15238c3f3b8f3
2016-08-25 09:31:00 -07:00
Yunqing Wang 167a4efbb5 Merge "Fix motion vector out of range bugs" into nextgenv2 2016-08-25 15:29:20 +00:00
Sarah Parker 6632915485 Update VP9_PROB_COST_SHIFT to VP10_PROB_COST_SHIFT
Change-Id: Ie1416569e73e66518cdb2765d79a2fb3dd570489
2016-08-24 17:25:00 -07:00
Yunqing Wang 90e12eaecb Fix motion vector out of range bugs
2 bugs were fixed in VP9.
https://chromium-review.googlesource.com/#/c/366873/
https://chromium-review.googlesource.com/#/c/368440/
Fixed them in VP10 as well.

Change-Id: I2e53fabc6131ff80ba6dcfd4c73eb76c59b4c474
2016-08-24 17:11:14 -07:00
Urvang Joshi c691864423 Merge "gitignore: add some entries" into nextgenv2 2016-08-24 22:12:03 +00:00
Urvang Joshi 4b3f980828 Merge "test_intra_pred_speed fix: use dspr2 version when HAVE_DSPR2" into nextgenv2 2016-08-24 22:10:01 +00:00
hui su 71c625d758 Remove unnecessary buffer pointers in PICK_MODE_CONTEXT
Change-Id: I600af6a66dc0e1310c8bfc7c16efa8a82e90856d
2016-08-24 14:18:56 -07:00
Yue Chen 35d4524b5b Merge "Make rectangular txfm in EXT_TX work with VAR_TX" into nextgenv2 2016-08-23 23:54:40 +00:00
Urvang Joshi f1906e966a Palette code: remove the use of same if condition twice.
rd_pick_palette_intra_sby() method is called only when,
cpi->common.allow_screen_content_tools is on. So, no need to check that
again. We just use an assert() instead to still be safe.

Change-Id: I19785c2aac016798c8d331bbe91971b3806b73a8
2016-08-23 15:01:41 -07:00
Urvang Joshi 7e5aa9e7a5 Merge "Rename CONFIG_VPX_HIGHBITDEPTH -> CONFIG_VP9_HIGHBITDEPTH" into nextgenv2 2016-08-23 19:33:53 +00:00
Yue Chen e57b1a5ea5 Make rectangular txfm in EXT_TX work with VAR_TX
Adapt rectangular txfm experiment to syntax/tokenization/loopfilter
framework of VAR_TX

Change-Id: Idcb005ecf5b3712de3e1cccb0d811ca16d87af24
2016-08-23 12:11:23 -07:00
Urvang Joshi e4e63b63c0 Rename CONFIG_VPX_HIGHBITDEPTH -> CONFIG_VP9_HIGHBITDEPTH
"vpx-highbitdepth" config doesn't exist.

Change-Id: Ib6d3691454299bb381ecc75b80657fbebf9f59b2
2016-08-23 12:04:18 -07:00
Urvang Joshi 3bcf3f07ac test_intra_pred_speed fix: use dspr2 version when HAVE_DSPR2
Change-Id: Ie7c78e19e077516615c71669022f505f8b3c80ca
2016-08-23 11:29:44 -07:00
Urvang Joshi 3ea2c234fa gitignore: add some entries
Change-Id: I65507c3d132b2b3ba90cf0a7b1c729da7d3de15f
2016-08-23 11:19:17 -07:00
Wei-ting Lin 7fed5044ca Merge "Allow LF_UPDATE type of frames to use BWDREF" into nextgenv2 2016-08-23 18:06:56 +00:00
Debargha Mukherjee 49b85d3965 Missing fixes for rect-tx
Reintroducing some fixes that were dropped inadvertently in
course of rebasing.

Change-Id: I5f51160c586010590d4bfd5cf225fb21347b0a40
2016-08-23 07:12:51 -07:00
Yaowu Xu 9a89ec5447 Merge "Make type conversion explicit" into nextgenv2 2016-08-23 01:28:05 +00:00
Debargha Mukherjee ccbefec3d8 Merge "Various rect-tx fixes" into nextgenv2 2016-08-23 01:00:39 +00:00
Yaowu Xu 04fe3499a4 Make type conversion explicit
This fixes two MSVC compiler warnings.

Change-Id: I55ad8833676e20c2c4a55885b99a7a9293d9623f
2016-08-23 00:01:00 +00:00
Yaowu Xu 88849e1395 Merge "Apply clang-format" into nextgenv2 2016-08-23 00:00:48 +00:00
Wei-ting Lin 4e8acca925 Allow LF_UPDATE type of frames to use BWDREF
Originally, only bi-pred type of frames can use BWDREF. When
extra alt-refs are inserted in a gf group, the closest alt-ref
serves as ALTREF for the frames within the corresponding
subgroup. Therefore, the original alt-ref can be used as BWDREF
for the LF_UPDATE type of frames.

This patch further swaps the virtual indices of BWDREF and ALTREF
for those frames whose BWDREF is farther than ALTREF. As a result,
the BWDREF is always the closet backward reference frame, and the
ALTREF is the farther one.

It improves the average RD performance by

0.132% in lowres, and
0.030% in midres.

The overall gains for the ext-refs compared to the baseline are

5.486% in lowres, and
4.666% in midres.

Change-Id: I22e4e5f378f19c4c89196a0a5e9214adb46c3428
2016-08-22 17:00:41 -07:00
Yaowu Xu c3cc46d8c2 Apply clang-format
Change-Id: Ie283af5f30324f54b4f749becdb48f937584707d
2016-08-22 16:22:10 -07:00
Debargha Mukherjee 44026851c3 Various rect-tx fixes
Change-Id: I02f44713b99284092ecfc50ce7ab268e91d2c6f8
2016-08-22 14:18:40 -07:00
Sarah Parker 3464aff41f Add integerize function back in warped_motion.c
This function was previously unused and removed in
I6bc740e778658d6f81ca54888fc6fa822d3b5ee0. I am adding it back in
with previously suggested fixes.

Change-Id: Iee0afb39170d25895b11d07e71843eae6913efd1
2016-08-22 12:29:26 -07:00
Urvang Joshi 3c7aa7ce2d Merge "Palette: count Y colors only for screen content." into nextgenv2 2016-08-19 23:18:39 +00:00
Urvang Joshi 28ca8554c5 Merge "Handle centroid rounding inside palette.c itself." into nextgenv2 2016-08-19 22:22:52 +00:00
Wei-ting Lin 7417932401 Merge "Insert extra ARFs' in a gf group" into nextgenv2 2016-08-19 22:10:41 +00:00
James Zern 7b2537b5e9 Merge "Fix compiler warnings in rdopt when warped motion is enabled" into nextgenv2 2016-08-19 21:39:59 +00:00
Urvang Joshi d68c7b6d6d Palette: count Y colors only for screen content.
Change-Id: Id4e12708598100df54bdfcf8cdb248161ab6ef88
2016-08-19 13:02:02 -07:00
Urvang Joshi f746c103a7 Handle centroid rounding inside palette.c itself.
Mostly refactoring, but a very tiny functional change:
Do all rounding in calc_centroids() itself, instead of rounding in two
places inside palette.c

This gives a slight performance improvement for screen content:
0.078% on average.

Change-Id: I7a0e007d30ebf4e59839483a167123f31a222dd4
2016-08-19 12:23:41 -07:00
Sarah Parker 984f073b8a Fix compiler warnings in rdopt when warped motion is enabled
The previous code was giving:
 unused variable ‘tmp_rate’ [-Wunused-variable]
 unused variable ‘tmp_dist’ [-Wunused-variable]
 ‘rate2_nocoeff’ may be used uninitialized in this function [-Wmaybe-uninitialized]

Change-Id: I26326d0e5ffc141ad548654356a877cd3627cea6
2016-08-19 11:10:44 -07:00
James Zern e0ab852f0b vp10_alloc_context_buffers: clear cm->mi* on failure
based on:
8b4c315 vp9_alloc_context_buffers: clear cm->mi* on failure

Change-Id: I3438a052721b960ff178cb647780f11bc33571fe
2016-08-19 10:39:46 -07:00
Alex Converse 32c92c97ea Merge "Don't send segment probability updates when the map isn't updated." into nextgenv2 2016-08-19 16:52:11 +00:00
James Zern b360168783 Merge "apply clang-format" into nextgenv2 2016-08-19 07:31:50 +00:00
Wei-ting Lin 41d5d52d78 Insert extra ARFs' in a gf group
Insert multiple arfs in a gf group to emulate multi-layer backward
reference frames structure. At maximum, two extra ARF's are inserted
in a gf group.

It improves the RD performance by 0.317% in Avg in lowres dataset.

Change-Id: I62c32e1b0f25b978484dd113b319bebcd959bf60
2016-08-18 18:21:13 -07:00
Sarah Parker daa4ba8d19 Disable global motion experiment when incompatible experiments are enabled
This is temporary until the global motion experiment is made to work
with ext_inter and dual_filter.

Change-Id: I73624ca6f536fd98218d7e07bcd7a2c1e6f5aebd
2016-08-18 16:00:38 -07:00
clang-format 21a0c2c9d7 apply clang-format
after:
253c001 Port dering experiment from aom
7208145 Adding 8x16/16x8/32x16/16x32 transforms

Change-Id: Id93e0d7b72a128701d8dec35fc2fac473944d0c1
2016-08-18 15:10:22 -07:00