]>
granicus.if.org Git - libvpx/log
Angie Chiang [Thu, 3 Nov 2016 23:01:03 +0000 (16:01 -0700)]
Separate coefficient cost of U/V planes in write_modes_b()
Change-Id: Ie3082db5b0fead8c322b2aeede4eff7cd723ea12
Angie Chiang [Thu, 3 Nov 2016 22:36:18 +0000 (15:36 -0700)]
Add av1_ prefix on ###_rd_stats functions
Change-Id: I3bc782d68bcd9b52b38210eec9eecb21146fde75
Angie Chiang [Thu, 3 Nov 2016 22:32:11 +0000 (15:32 -0700)]
Use init_rd_stats() in encodeframe.c
Change-Id: I8359e8498efd301ff81eea1d7466d0f3fec5e006
Angie Chiang [Thu, 3 Nov 2016 19:20:58 +0000 (12:20 -0700)]
Add transform block coefficient cost in RD_STATS for debugging
Change-Id: Iae50d0b0c4f8f383ab4f91d2c1c2fa4e799c7250
Angie Chiang [Thu, 3 Nov 2016 18:49:15 +0000 (11:49 -0700)]
Add helper functions to modify RD_STATS
Those functions includes
init_rd_stats()
invalid_rd_stats()
merge_rd_stats()
This CL help simplify the code.
Change-Id: Id1704d883bd21a039b0478a940994ca14184ae1c
Angie Chiang [Thu, 3 Nov 2016 18:13:15 +0000 (11:13 -0700)]
Add mi_row and mi_col into mbmi to facilitate rd_debug process
Change-Id: Ia69f13c47f2dd34fabd220652691049166a06a68
Angie Chiang [Thu, 3 Nov 2016 01:30:25 +0000 (18:30 -0700)]
Add token cost comparison in write_modes_b()
This is just partial implementation
Compare token cost of pack_mb_tokens/pack_txb_tokens with token cost
from rate-distortion loop. If there is any difference, dump out mode
info.
Change-Id: I46b373ee2522c5047f799f36baf7cec5fbc06f06
Jingning Han [Fri, 4 Nov 2016 17:53:53 +0000 (17:53 +0000)]
Merge "Properly schedule the transform block recursion order" into nextgenv2
Yue Chen [Fri, 4 Nov 2016 17:11:07 +0000 (17:11 +0000)]
Merge "Remove duplicated variables in EXT_INTER" into nextgenv2
Jingning Han [Thu, 3 Nov 2016 19:47:47 +0000 (12:47 -0700)]
Properly schedule the transform block recursion order
This commit replaces the offset based block index calculation with
incremental based one. It does not change the coding statistics.
Change-Id: I3789294eb45416bd0823e773ec30f05ed41ba0dc
Jingning Han [Fri, 4 Nov 2016 15:31:24 +0000 (08:31 -0700)]
Fix format issue in handle_inter_mode()
Change-Id: I681fd799cf46991de419cc867ccb649a6990c19d
Debargha Mukherjee [Fri, 4 Nov 2016 09:32:07 +0000 (09:32 +0000)]
Merge "Further work on 64x64 fwd/inv transform support" into nextgenv2
Angie Chiang [Fri, 4 Nov 2016 05:42:59 +0000 (05:42 +0000)]
Merge "Refactor: Replace rate dist sse skip by RD_STATS in VAR_TX" into nextgenv2
Debargha Mukherjee [Fri, 4 Nov 2016 05:25:49 +0000 (05:25 +0000)]
Merge "Fix bilateral filter asan error for highbitdepth" into nextgenv2
James Zern [Fri, 4 Nov 2016 00:58:06 +0000 (00:58 +0000)]
Merge "warped_motion: Fix ubsan warning for signed integer overflow" into nextgenv2
James Zern [Fri, 4 Nov 2016 00:56:23 +0000 (00:56 +0000)]
Merge "Fix ubsan divide by zero warning in ransac" into nextgenv2
Sarah Parker [Wed, 26 Oct 2016 19:46:03 +0000 (12:46 -0700)]
warped_motion: Fix ubsan warning for signed integer overflow
Change-Id: Ie698aa02ef56128759c71079e9bfa1af25149644
Angie Chiang [Wed, 2 Nov 2016 23:19:58 +0000 (16:19 -0700)]
Refactor: Replace rate dist sse skip by RD_STATS in VAR_TX
This is to facilitate implementation of rd_debug tool; it doesn't change
coding behavior.
Change-Id: I0eb82b31473883ba6652ed11dca09b9ec4530183
Debargha Mukherjee [Thu, 3 Nov 2016 22:14:27 +0000 (15:14 -0700)]
Fix bilateral filter asan error for highbitdepth
BUG=webm:1334
Change-Id: I5886eec0a22a8cc056e1bdb493d2faf183816656
James Zern [Thu, 3 Nov 2016 22:59:33 +0000 (22:59 +0000)]
Merge "rdopt,global-motion: Fix -1 indexing ubsan warning" into nextgenv2
Sarah Parker [Wed, 26 Oct 2016 19:48:55 +0000 (12:48 -0700)]
rdopt,global-motion: Fix -1 indexing ubsan warning
Change-Id: I1b3caf3543ab385f39f5f253c9949ad89ea5af7d
Alex Converse [Thu, 3 Nov 2016 22:13:33 +0000 (22:13 +0000)]
Merge "Use TX_SIZES in intra_high_pred_fn declarations" into nextgenv2
Yue Chen [Thu, 3 Nov 2016 22:12:42 +0000 (15:12 -0700)]
Remove duplicated variables in EXT_INTER
Introduced by merge commit
141f7a9
Change-Id: Idd68e09a6cd925d97466eabebe0e4905b5031340
Sarah Parker [Thu, 3 Nov 2016 22:09:46 +0000 (22:09 +0000)]
Merge "Make inline function static to fix clang compile error" into nextgenv2
Alex Converse [Thu, 3 Nov 2016 21:44:20 +0000 (21:44 +0000)]
Merge "Don't use a TX_SIZE as a TX_TYPE" into nextgenv2
Debargha Mukherjee [Wed, 2 Nov 2016 21:57:42 +0000 (14:57 -0700)]
Further work on 64x64 fwd/inv transform support
For higher level fwd and inv transform functions.
Change-Id: I91518250a0be7d94aada7519f6c9e7ed024574fb
Sarah Parker [Wed, 26 Oct 2016 19:48:01 +0000 (12:48 -0700)]
Fix ubsan divide by zero warning in ransac
Change-Id: I8c736ff665a27ce8307fd62571b9728333756d7e
Debargha Mukherjee [Thu, 3 Nov 2016 19:59:09 +0000 (19:59 +0000)]
Merge "Replace hard coded numbers with TX_SIZES macro" into nextgenv2
Jingning Han [Thu, 3 Nov 2016 19:31:48 +0000 (19:31 +0000)]
Merge "Fix txb_w/h use case in av1_tx_block_rd_b()" into nextgenv2
Yaowu Xu [Thu, 3 Nov 2016 18:45:57 +0000 (18:45 +0000)]
Merge "fix build issue with --enable-delta-q" into nextgenv2
Sarah Parker [Wed, 26 Oct 2016 19:50:52 +0000 (12:50 -0700)]
Make inline function static to fix clang compile error
Change-Id: I0432b8274a2764ba978dd6c4ed532fb7e4b7b519
Alex Converse [Thu, 3 Nov 2016 17:27:40 +0000 (10:27 -0700)]
Use TX_SIZES in intra_high_pred_fn declarations
Change-Id: I078bb5244dbff153bcfab226206540ca6cebdad0
Alex Converse [Thu, 3 Nov 2016 17:25:20 +0000 (10:25 -0700)]
Don't use a TX_SIZE as a TX_TYPE
Change-Id: I26b02e6578ad2d82aadfe1df2aeb84e6c11a747b
Angie Chiang [Tue, 1 Nov 2016 22:50:43 +0000 (15:50 -0700)]
Add rd_debug flag
rd_debug is a debug tool aim at finding discrepancy between rate-distortion
loop and bitstream packing.
Change-Id: I751c4121516c5e6368668229c77778880a9dcb9d
Jingning Han [Thu, 3 Nov 2016 16:20:08 +0000 (09:20 -0700)]
Fix txb_w/h use case in av1_tx_block_rd_b()
Match them with block_row/col index.
Change-Id: Idf0f924a093e5312b0a36b765d295e52d033eb5a
Yaowu Xu [Wed, 2 Nov 2016 22:31:29 +0000 (15:31 -0700)]
fix build issue with --enable-delta-q
BUG=webm:1330
Change-Id: I120ce8ea3581018b232b19ca7ffbb07d3e99d8d0
Debargha Mukherjee [Thu, 3 Nov 2016 15:51:42 +0000 (08:51 -0700)]
Replace hard coded numbers with TX_SIZES macro
Replaces a couple of hard coded numbers with TX_SIZES macro
in common/reconiontra.c
Change-Id: I8a2a53ca16bc3ab51409cec340bea55292ff2dee
Jingning Han [Thu, 3 Nov 2016 15:41:56 +0000 (15:41 +0000)]
Merge "Refactor recursive transform block partition search" into nextgenv2
Jingning Han [Thu, 3 Nov 2016 15:41:45 +0000 (15:41 +0000)]
Merge "Make bit-stream support rectangular tx_size" into nextgenv2
Yaowu Xu [Thu, 3 Nov 2016 14:03:24 +0000 (14:03 +0000)]
Merge changes from topic 'update_dering' into nextgenv2
* changes:
Reformatting the deringing code
Introducing OD_DERING_SIZE_LOG2 constant (3)
Renaming deringing blockwise write-back functions to make code clearer
Deringing refactoring: replace last_sbc with simpler dering_left flag
Getting rid of the od_dering_in type
Yaowu Xu [Thu, 3 Nov 2016 13:48:58 +0000 (13:48 +0000)]
Merge "Refactoring deringed block list code" into nextgenv2
Yaowu Xu [Thu, 3 Nov 2016 05:02:32 +0000 (05:02 +0000)]
Merge "Deringing line buffer no longer depends on holding OD_DERING_VERY_LARGE" into nextgenv2
Jingning Han [Thu, 3 Nov 2016 04:36:55 +0000 (04:36 +0000)]
Merge "Make recursive txfm encoding process support rectangular tx_size" into nextgenv2
Jingning Han [Thu, 3 Nov 2016 01:06:38 +0000 (01:06 +0000)]
Merge "Fix a merge bug between dual_filter and sub8x8mc" into nextgenv2
Jingning Han [Thu, 3 Nov 2016 01:04:45 +0000 (01:04 +0000)]
Merge "Remove redundant experimental flags from common_data.h" into nextgenv2
Jingning Han [Thu, 3 Nov 2016 00:01:58 +0000 (17:01 -0700)]
Refactor recursive transform block partition search
This commit refactors the recursive transform block partition
search process to make it support rectangular transform block size
coding.
Change-Id: I0207ae40d83c7eae3cb5d460e403f470747590d3
Jingning Han [Wed, 2 Nov 2016 23:22:18 +0000 (16:22 -0700)]
Make bit-stream support rectangular tx_size
Allow the transform size writing, reading, and the reconstruction
process to support rectangular transform block size coding.
Change-Id: I57393c73ec60835a088d785ca838d7e3d7eb29a4
Jean-Marc Valin [Wed, 2 Nov 2016 06:33:46 +0000 (02:33 -0400)]
Reformatting the deringing code
Manally removed the "clang-format off" lines. The rest is done by clang
Change-Id: I88a2028b55a541729b4e8896cdf66b544e9898bb
Jean-Marc Valin [Tue, 18 Oct 2016 19:56:37 +0000 (15:56 -0400)]
Refactoring deringed block list code
Using a struct named dlist rather than an array named bskip. Simplified some
code.
No change in output
Change-Id: Id40d40b19b5d8f2ebafe347590fa1bb8cb80e6e1
Jean-Marc Valin [Tue, 18 Oct 2016 07:32:52 +0000 (03:32 -0400)]
Deringing line buffer no longer depends on holding OD_DERING_VERY_LARGE
The OD_DERING_VERY_LARGE values are now explicitly copied to the buffer instead
of being read from the line buffer when we're on the edge of the frame. This
will make it possible to make the line buffer 8-bit for non-high-bitdepth.
No change in output
Change-Id: I1a4134d67ac7f8c239f08d73941405c56f01050b
Jean-Marc Valin [Wed, 26 Oct 2016 05:26:25 +0000 (01:26 -0400)]
Introducing OD_DERING_SIZE_LOG2 constant (3)
Also cleans up the size of the deringing destination buffer.
No change in output.
Change-Id: I7fc50d862d3906ce809c1031bf0789acdf39cf34
Jean-Marc Valin [Wed, 26 Oct 2016 04:11:16 +0000 (00:11 -0400)]
Renaming deringing blockwise write-back functions to make code clearer
No change in output.
Change-Id: Ifa5df3adce9f24ef6dcd89a5f33a744bfb57194d
Jean-Marc Valin [Wed, 26 Oct 2016 03:51:26 +0000 (23:51 -0400)]
Deringing refactoring: replace last_sbc with simpler dering_left flag
No change in output.
Change-Id: I1cc2e14b2bb6c343baa7f88348c875085e5863af
Jean-Marc Valin [Tue, 25 Oct 2016 23:30:49 +0000 (19:30 -0400)]
Getting rid of the od_dering_in type
We no longer need the deringing code to be generic wrt the input depth.
No change in output.
Change-Id: I2db2beb82f1816e611cd2c0438dff217d363de33
Jingning Han [Wed, 2 Nov 2016 22:50:58 +0000 (22:50 +0000)]
Merge "Remove unused get_intra/inter_scan() from scan.h" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 22:50:49 +0000 (22:50 +0000)]
Merge "Remove redundant config flags from get_entropy_context" into nextgenv2
Jean-Marc Valin [Tue, 18 Oct 2016 00:51:53 +0000 (20:51 -0400)]
Now using a single line buffer
No change in output.
Change-Id: I4701a5517fb97889f970acfb0b44cee51c34fd95
Jean-Marc Valin [Mon, 17 Oct 2016 21:48:12 +0000 (17:48 -0400)]
Only copy data from deringed blocks to the line buffer
No change in output
Change-Id: I6ec4a8c635337562170153585e427afd6f9d9a0f
Jean-Marc Valin [Mon, 17 Oct 2016 20:44:46 +0000 (16:44 -0400)]
Splitting out 8->16 block copy code into copy_sb8_16()
No change in output.
Change-Id: I4f0e37a879432e2647b3debe6a2c0c670a79dd6f
Jean-Marc Valin [Mon, 17 Oct 2016 19:38:36 +0000 (15:38 -0400)]
Eliminate the big superblock row buffer.
Now only buffering three lines across the entire frame and four lines
over the height of one superblock.
No change in output.
Change-Id: I6b99399974e197dc02f2e4ff2e60cdd7fdaa2e43
Jean-Marc Valin [Fri, 14 Oct 2016 17:05:14 +0000 (13:05 -0400)]
Making deringing buffer only one row of superblocks at a time
This introduces a line buffer that hold the last three lines of each original
row so that the next row can be deringed with the original input of the upper
row.
No change in output
Change-Id: I8fad3bc48745e9ce3e440289f453477a0c5442c0
Jingning Han [Wed, 2 Nov 2016 22:45:07 +0000 (15:45 -0700)]
Make recursive txfm encoding process support rectangular tx_size
This commit makes the encoding process of the recursive transform
block partition support both rectangular and square transform block
sizes as the starting point. If the coding block size is rectangular,
it would allow the transform block size to start from the largest
rectangular transform size, and recursive parse to the selected
coding sizes.
Change-Id: I576628b9166565bada6a918f0a1e67849dfef4cd
Jean-Marc Valin [Thu, 13 Oct 2016 19:24:39 +0000 (15:24 -0400)]
Duplicating deringing input superblock copy to make upcoming changes easier
No change in output
Change-Id: Iaa06043dcc31308c83f667424e5a83c2db50ed24
Jean-Marc Valin [Thu, 13 Oct 2016 18:44:22 +0000 (14:44 -0400)]
Using a uniform definition for "bsize" in deringing filter
No change in output
Change-Id: Ia3a1679aa75cb58f4bc6459791e061176eeafd52
Jean-Marc Valin [Thu, 13 Oct 2016 18:18:49 +0000 (14:18 -0400)]
Using the copied input for od_dir_find8()
No change in output
Change-Id: Iec1411c35bf175a462eade34e89a4c60eb2a1da4
Yaowu Xu [Wed, 2 Nov 2016 22:37:38 +0000 (22:37 +0000)]
Merge "Increase deringing horizontal padding to 4 pixels on each side" into nextgenv2
Jean-Marc Valin [Thu, 13 Oct 2016 18:14:39 +0000 (14:14 -0400)]
Increase deringing horizontal padding to 4 pixels on each side
This makes vectorization easier by having buffer lines be a multiple of 4.
No change in output
Change-Id: I7ec06e03a49554206af0a55aab03daccc411b50f
Yaowu Xu [Wed, 2 Nov 2016 22:35:04 +0000 (22:35 +0000)]
Merge "De-sparsifying the deringing output buffer" into nextgenv2
Jean-Marc Valin [Wed, 12 Oct 2016 22:09:48 +0000 (18:09 -0400)]
De-sparsifying the deringing output buffer
No change in output
Change-Id: I940203975564aedca8734d6f74b013edb513f517
Yaowu Xu [Wed, 2 Nov 2016 22:34:45 +0000 (22:34 +0000)]
Merge "No need to store the deringing filter direction variance in an array" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 21:57:11 +0000 (14:57 -0700)]
Fix a merge bug between dual_filter and sub8x8mc
The function module in inter_predictor() has been changed to
universally support arbitrary block size inter prediction. Hence
sub8x8mc can be a standalone experiment now.
Change-Id: Ie9d87f61fc317b1d114edb4e0bf5544f918ed08e
Jingning Han [Wed, 2 Nov 2016 19:21:01 +0000 (12:21 -0700)]
Remove redundant experimental flags from common_data.h
No coding statistics change.
Change-Id: I88cbb828308b5796a2e87079c2f1bf0dabd99a11
Jingning Han [Wed, 2 Nov 2016 21:49:41 +0000 (21:49 +0000)]
Merge "Support rectangular tx_size in the common lib" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 21:49:17 +0000 (21:49 +0000)]
Merge "Support rectangular transform block units in the codebase" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 21:49:07 +0000 (21:49 +0000)]
Merge "Make highbd rectangular transform block available in the common lib" into nextgenv2
Jean-Marc Valin [Wed, 12 Oct 2016 21:45:07 +0000 (17:45 -0400)]
No need to store the deringing filter direction variance in an array
No change in output
Change-Id: Ifa5c5d4ed33ff11ea3c56ee5d559c7a40599b3dc
Sarah Parker [Wed, 2 Nov 2016 19:45:44 +0000 (19:45 +0000)]
Merge "Fix ubsan left shift warnings in warped motion library" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 19:00:51 +0000 (12:00 -0700)]
Remove unused get_intra/inter_scan() from scan.h
Change-Id: I96fc1da1ce56593ae35ebbc93a668e4ba241234a
Jingning Han [Wed, 2 Nov 2016 16:52:36 +0000 (09:52 -0700)]
Remove redundant config flags from get_entropy_context
The rectangular transform syntax is by default supported, hence
no need to put it under the experimental flag. This does not change
the coding statistics.
Change-Id: I3a147503d973a03400f8a86e11f07c7d754e6234
Jingning Han [Wed, 2 Nov 2016 16:37:14 +0000 (09:37 -0700)]
Support rectangular tx_size in the common lib
Change-Id: I4128ab932a967a3d657bb1f95f0fa2af20a06469
Jingning Han [Wed, 2 Nov 2016 16:05:43 +0000 (09:05 -0700)]
Support rectangular transform block units in the codebase
Change-Id: I9183851258478a36dc5a4ad2d4faa3d3c8b18bd3
Jingning Han [Wed, 2 Nov 2016 15:53:12 +0000 (08:53 -0700)]
Make highbd rectangular transform block available in the common lib
Change-Id: Ief08b23b30b78d640f6d7c702145e5bcf1b37b57
Debargha Mukherjee [Wed, 2 Nov 2016 18:40:54 +0000 (18:40 +0000)]
Merge "Adding 64x64 forward and inverse transforms" into nextgenv2
Yaowu Xu [Wed, 2 Nov 2016 18:02:56 +0000 (18:02 +0000)]
Merge changes I313bde67,I2ddc2d70,Ifb9094c3,I9051ed6e,I5681e332, ... into nextgenv2
* changes:
Avoid the "initial copy" in the deringing filter
Only copy the deringed blocks back into the buffer
Reducing copies in deringing filter
sb_all_skip_out() now computes a list of deringed blocks
compute bskip as we go
Revert "Fix dering filter when using 4:2:2 or 4:4:0 subsampling"
Debargha Mukherjee [Tue, 1 Nov 2016 21:37:39 +0000 (14:37 -0700)]
Adding 64x64 forward and inverse transforms
Change-Id: I213f3111fc0656aecd1303a8b871ecded2b92bc2
Zoe Liu [Wed, 2 Nov 2016 16:39:36 +0000 (16:39 +0000)]
Merge "Make a small code clean on handle_inter_mode()" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 16:17:00 +0000 (16:17 +0000)]
Merge "Make rectangular transform block available in the common lib" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 16:16:50 +0000 (16:16 +0000)]
Merge "Simplify tx_size enums" into nextgenv2
Jean-Marc Valin [Wed, 12 Oct 2016 21:39:54 +0000 (17:39 -0400)]
Avoid the "initial copy" in the deringing filter
No change in output
Change-Id: I313bde67e59835f88e3b2e6079b0df2d7ed1a903
Jean-Marc Valin [Wed, 12 Oct 2016 05:56:43 +0000 (01:56 -0400)]
Only copy the deringed blocks back into the buffer
No change in output
Change-Id: I2ddc2d70c6534e7cfd315d66e838410677f91356
Jean-Marc Valin [Tue, 11 Oct 2016 21:47:36 +0000 (17:47 -0400)]
Reducing copies in deringing filter
Only copy the modified pixels from the first filter back into the input of the
second filter.
Change-Id: Ifb9094c33c876a8c6caa0f68771fc7ef59c78b53
Jean-Marc Valin [Tue, 11 Oct 2016 20:53:59 +0000 (16:53 -0400)]
sb_all_skip_out() now computes a list of deringed blocks
No change in output
Change-Id: I9051ed6e1fbca7d80412ba2b53f7aacbc3ef70eb
Jean-Marc Valin [Sun, 9 Oct 2016 17:36:39 +0000 (13:36 -0400)]
compute bskip as we go
Change-Id: I5681e3329ad3677296161de59f5ff1236a14f086
Yaowu Xu [Wed, 2 Nov 2016 15:22:02 +0000 (08:22 -0700)]
Revert "Fix dering filter when using 4:2:2 or 4:4:0 subsampling"
This reverts commit
401204a50bea8b21e5fcd721cd5db513b8f70e72 .
Change-Id: Id27eadf679b0df2d2ccfab61155be29979b0b6ba
Jingning Han [Wed, 2 Nov 2016 01:19:30 +0000 (18:19 -0700)]
Make rectangular transform block available in the common lib
This prepares the integration of rectangular transform block size
with recursive transform block partition system.
Change-Id: Id96aa3790dace15619c665f438241938992d1730
Yaowu Xu [Wed, 2 Nov 2016 04:28:50 +0000 (04:28 +0000)]
Merge "Disable upsampled references for resolutions above 1080p." into nextgenv2
Jingning Han [Tue, 1 Nov 2016 23:09:57 +0000 (16:09 -0700)]
Simplify tx_size enums
Remove redundant experimental flag. This does not change the coding
statistics.
Change-Id: I35b3cb04025c5c2d2744312e5efc00d0473c990d
Yi Luo [Wed, 2 Nov 2016 01:43:19 +0000 (01:43 +0000)]
Merge "Remove unused copies of transform related source code" into nextgenv2
Yi Luo [Wed, 2 Nov 2016 01:43:01 +0000 (01:43 +0000)]
Merge "Hybrid inverse transforms 16x16 AVX2 optimization" into nextgenv2
Jingning Han [Wed, 2 Nov 2016 01:18:19 +0000 (01:18 +0000)]
Merge "Change TXFM_CONTEXT from TX_SIZE to uint8_t" into nextgenv2