webm/libvpx.git
111 min agoMerge "Remove 'local' from configure and configure.sh" master
James Zern [Wed, 30 Jul 2014 03:07:05 +0000 (20:07 -0700)]
Merge "Remove 'local' from configure and configure.sh"

2 hours agoMerge "vpx_once: implement once() for OS/2"
James Zern [Wed, 30 Jul 2014 02:36:11 +0000 (19:36 -0700)]
Merge "vpx_once: implement once() for OS/2"

2 hours agoMerge "Makefile: add -mstackrealign to CFLAGS on OS/2\r \r This prevents SIGSEGV of...
James Zern [Wed, 30 Jul 2014 02:21:24 +0000 (19:21 -0700)]
Merge "Makefile: add -mstackrealign to CFLAGS on OS/2\r \r This prevents SIGSEGV of test_libvpx.\r \r Change-Id: I788743841469f4141bc8d29b1d1a8683cb00655c\r"

3 hours agoRemove 'local' from configure and configure.sh 41/70941/5
KO Myung-Hun [Tue, 22 Jul 2014 02:09:27 +0000 (11:09 +0900)]
Remove 'local' from configure and configure.sh

'local' is not posix.

Change-Id: I46d7acfa574cbef686cee1b35022f5781485c3fa

4 hours agovpx_once: implement once() for OS/2 12/71012/2
KO Myung-Hun [Sat, 26 Jul 2014 05:53:59 +0000 (14:53 +0900)]
vpx_once: implement once() for OS/2

Change-Id: I9f736f299490464bbdbb6cd24ee6f5b46ad45ec6

7 hours agoMerge "Remove the redundant index computation in the first pass"
Pengchong Jin [Tue, 29 Jul 2014 21:27:24 +0000 (14:27 -0700)]
Merge "Remove the redundant index computation in the first pass"

10 hours agoMerge "vp8: Add an aggressive denoising mode."
Marco Paniconi [Tue, 29 Jul 2014 18:23:25 +0000 (11:23 -0700)]
Merge "vp8: Add an aggressive denoising mode."

10 hours agovp8: Add an aggressive denoising mode. 85/70985/20
Marco Paniconi [Thu, 24 Jul 2014 19:54:38 +0000 (12:54 -0700)]
vp8: Add an aggressive denoising mode.

Change-Id: Ie4686e1b15af6bcc8d59d585bbeb996f38224522

13 hours agoAdded vp9_fdct8x8_neon(), vp9_fdct8x8_1_neon() 44/70944/4
Scott LaVarnway [Mon, 28 Jul 2014 22:21:10 +0000 (15:21 -0700)]
Added vp9_fdct8x8_neon(), vp9_fdct8x8_1_neon()

On a Nexus 7, vpxenc (in realtime mode, speed -12)
reported a performance improvement of ~3.7%.

Change-Id: I428c72c40df82c6d537955e320a8debf99343004

15 hours agoMakefile: add -mstackrealign to CFLAGS on OS/2 40/70940/3
KO Myung-Hun [Tue, 22 Jul 2014 01:48:36 +0000 (10:48 +0900)]
Makefile: add -mstackrealign to CFLAGS on OS/2

This prevents SIGSEGV of test_libvpx.

Change-Id: I788743841469f4141bc8d29b1d1a8683cb00655c

27 hours agoRemove the redundant index computation in the first pass 10/71010/2
Pengchong Jin [Mon, 28 Jul 2014 23:04:36 +0000 (16:04 -0700)]
Remove the redundant index computation in the first pass

Remove the redundant index computation when store the first
pass block-wise statistics. Currently, a single byte is
allocated for a 16x16 blocks, and all the frame statistics
saved during the first pass will be kept in memory for use
in the second pass. For a 1920x1080 300-frame clip, it will
take about 2.3 MB memory. This feature is off in current
setting.

Change-Id: I135a95b348ec093d54c6a07e1e8237626909e3bd

30 hours agoFix bug 805 79/70979/3
levytamar82 [Thu, 24 Jul 2014 07:42:13 +0000 (00:42 -0700)]
Fix bug 805

Remove all the redundant dct functions (dct4x4, dct8x8)
in avx2 except dct32x32 those functions were copied originally from dct_sse2

Change-Id: I742576fbf5175f3ac09f2076976a9247b259323e

31 hours agoMerge "Store block-wise statistics obtained in the first pass"
Pengchong Jin [Mon, 28 Jul 2014 21:49:05 +0000 (14:49 -0700)]
Merge "Store block-wise statistics obtained in the first pass"

36 hours agoMerge "Fix reference frame size restrictions."
Jim Bankoski [Mon, 28 Jul 2014 16:47:57 +0000 (09:47 -0700)]
Merge "Fix reference frame size restrictions."

36 hours agoStore block-wise statistics obtained in the first pass 05/71005/1
Pengchong Jin [Fri, 25 Jul 2014 18:07:44 +0000 (11:07 -0700)]
Store block-wise statistics obtained in the first pass

Change-Id: I9956db2ba2f7d28f484daaf5022d8d1ef5db473c

37 hours agoFix reference frame size restrictions. 43/70843/5
Jim Bankoski [Mon, 28 Jul 2014 15:37:25 +0000 (08:37 -0700)]
Fix reference frame size restrictions.

The issue was introduced by commit g9f37d14 with adding explicit
restrictions on reference-frame scale factors. The restriction
is checked against aligned-by-8 frame dimensions, not against
original ones. So, for example, frame of 35×35 actually can refer
to frame of 70×70, but the new check won't allow this. It will
compare 35 vs 72 (not 70), so 2x downscale limit will be exceeded.

Change-Id: Ic663693034440f64ac8312cbff9e1e773a921060

38 hours agoMerge "This removes a test that fails on 32 bit"
Jim Bankoski [Mon, 28 Jul 2014 14:17:27 +0000 (07:17 -0700)]
Merge "This removes a test that fails on 32 bit"

4 days agoMerge "Fix rd_pick_partition search loop for 4x4 blocks"
Jingning Han [Fri, 25 Jul 2014 22:57:35 +0000 (15:57 -0700)]
Merge "Fix rd_pick_partition search loop for 4x4 blocks"

4 days agoMerge "Fix potential ioc issue in vp9_get_prob for 4K above sizes"
Jingning Han [Fri, 25 Jul 2014 22:56:53 +0000 (15:56 -0700)]
Merge "Fix potential ioc issue in vp9_get_prob for 4K above sizes"

4 days agoMerge "Remove unnecessary conditional assignment"
Jingning Han [Fri, 25 Jul 2014 22:56:31 +0000 (15:56 -0700)]
Merge "Remove unnecessary conditional assignment"

4 days agoMerge "[spatial svc]Fix reference issues"
Minghai Shang [Fri, 25 Jul 2014 20:24:27 +0000 (13:24 -0700)]
Merge "[spatial svc]Fix reference issues"

4 days agoMerge "Encode perf test"
Joshua Litt [Fri, 25 Jul 2014 20:16:36 +0000 (13:16 -0700)]
Merge "Encode perf test"

4 days agoEncode perf test 32/70932/8
Joshua Litt [Mon, 21 Jul 2014 17:57:16 +0000 (10:57 -0700)]
Encode perf test

Change-Id: I61993946fe5cbcd7ebeb6302efd9538bff022e5a

4 days agoMerge "Allocate aligned source in variance test"
Yunqing Wang [Fri, 25 Jul 2014 17:53:13 +0000 (10:53 -0700)]
Merge "Allocate aligned source in variance test"

4 days agoMerge "Refactor inter/intra_suberblock_yrd."
Alex Converse [Fri, 25 Jul 2014 17:51:48 +0000 (10:51 -0700)]
Merge "Refactor inter/intra_suberblock_yrd."

4 days agoMerge "Fix allocation of context buffers on frame resize"
Yaowu Xu [Fri, 25 Jul 2014 15:49:39 +0000 (08:49 -0700)]
Merge "Fix allocation of context buffers on frame resize"

4 days agoMerge "Changed validation of reference frame size"
Yaowu Xu [Fri, 25 Jul 2014 15:48:48 +0000 (08:48 -0700)]
Merge "Changed validation of reference frame size"

4 days agoMerge "vp8: reduce refresh_max_mbs_perframe with fewer layers."
Marco Paniconi [Fri, 25 Jul 2014 15:42:41 +0000 (08:42 -0700)]
Merge "vp8: reduce refresh_max_mbs_perframe with fewer layers."

4 days agoMerge "s/CONFIG_DENOISING/CONFIG_VP9_TEMPORAL_DENOISING"
Tim Kopp [Fri, 25 Jul 2014 14:53:51 +0000 (07:53 -0700)]
Merge "s/CONFIG_DENOISING/CONFIG_VP9_TEMPORAL_DENOISING"

4 days agoFix rd_pick_partition search loop for 4x4 blocks 97/70997/1
Jingning Han [Fri, 25 Jul 2014 14:08:23 +0000 (07:08 -0700)]
Fix rd_pick_partition search loop for 4x4 blocks

The partition search for 4x4 blocks takes unnecessary steps to
reconstruct pixels and an extra partition type update. This commit
removes such operations. No visible compression/speed difference.
Thanks to Yue (yuec@) for finding this issue.

Change-Id: I3f83824aa3fd3717d63be0b280fa57258939a70a

4 days agoMerge "rtcd.pl: check for auto_help availability"
James Zern [Fri, 25 Jul 2014 06:09:35 +0000 (23:09 -0700)]
Merge "rtcd.pl: check for auto_help availability"

5 days agoRemove neon version of vp8 extend borders 76/70876/14
Johann [Thu, 24 Jul 2014 16:32:01 +0000 (09:32 -0700)]
Remove neon version of vp8 extend borders

The code fails the unit test. Speed comparisons to the C are invalid
because the code frequently didn't correctly extend the right and
bottom portions of the frame.

Reduce maximum frame size on ARM devices to avoid OOM

Change-Id: Ia664c86406f0bb8120fd7ad401f32d0bd44994fb

5 days agoAdd test for vp8 extend_borders 75/70875/14
Johann [Wed, 16 Jul 2014 10:47:27 +0000 (03:47 -0700)]
Add test for vp8 extend_borders

Change-Id: I5bfc965019815e631a2300eb29aa065fb2bc94aa

5 days agoAllocate aligned source in variance test 95/70995/1
Yunqing Wang [Fri, 25 Jul 2014 00:11:58 +0000 (17:11 -0700)]
Allocate aligned source in variance test

The source buffer is an aligned buffer in VP9. Added the alignment
to make it consistent with libvpx.

Change-Id: I3ebb9d2e8555ed532951da479dd5cbbb8812e02d

5 days agovp8: reduce refresh_max_mbs_perframe with fewer layers. 94/70994/1
Marco Paniconi [Thu, 24 Jul 2014 22:44:29 +0000 (15:44 -0700)]
vp8: reduce refresh_max_mbs_perframe with fewer layers.

Change-Id: Ib39cc0218a1f232c0cde8e99ca3319c17274bd48

5 days agoFix potential ioc issue in vp9_get_prob for 4K above sizes 92/70992/1
Jingning Han [Thu, 24 Jul 2014 22:31:32 +0000 (15:31 -0700)]
Fix potential ioc issue in vp9_get_prob for 4K above sizes

This commit turns on the existing vp9_get_prob function using
64 bit in the intermediate step. It fixes the ioc issue for 4K
above frame sizes (issue 828).

Change-Id: I9f627f3beca2c522f73b38fd2a3e7eefdff01a7c

5 days agoRemove unnecessary conditional assignment 90/70990/3
Jingning Han [Thu, 24 Jul 2014 21:43:06 +0000 (14:43 -0700)]
Remove unnecessary conditional assignment

The assignment of the variable mode_excluded in
vp9_rd_pick_inter_mode_sub8x8 takes redundant conditional jump.
This commit removes it.

Change-Id: Ie195fbe6e54ec2ade7093d562c456a2e93143704

5 days agoSet and use uv_crop_[width|height] 74/70874/7
Johann [Wed, 16 Jul 2014 09:30:21 +0000 (02:30 -0700)]
Set and use uv_crop_[width|height]

Ensure consistent border extension by rounding uv_crop_* at image
creation time. Where it was rounded problems could arise with the right
and bottom extensions.
When padding = 32, y_width = 64, and y_crop_width = 63:
 (padding + width - crop_width + 1) / 2
 32 + 64 - 63 + 1 should equal 32 *but*
 32 + 1 + 1 equals 34 giving a right buffer of 17 instead of 16.
By calculating uv_crop_* earlier we round up at the appropriate time and
for the same values:
 (y_crop_width + 1) / 2
 63 + 1 / 2
 64
 (padding / 2) + uv_width - uv_crop_width
 16 + 16 - 16
 16

Change-Id: If866cd1b63444771440edb1432280ac83875969b

5 days agoChanged validation of reference frame size 82/70982/4
Yaowu Xu [Thu, 24 Jul 2014 16:12:46 +0000 (09:12 -0700)]
Changed validation of reference frame size

A previous change, https://gerrit.chromium.org/gerrit/#/c/70632,
introduced a size validation for reference frames to insuare the
input stream is a valid VP9 stream. However, the logic requiring
all reference frames have valid size turned out to be too strict.

In this commit, we modify the validation to require one of the
reference frame has valid dimension. In addition, the decoder
reports error whenever it detects the use of reference frame
with invalid scalig ratio.

Change-Id: If8efc312244087556cfe00f1fcbdff811268ebad

5 days agortcd.pl: check for auto_help availability 91/70991/1
James Zern [Thu, 24 Jul 2014 21:55:19 +0000 (14:55 -0700)]
rtcd.pl: check for auto_help availability

'auto_help' was added to Getopt::Long in 2.33
this isn't strictly necessary as an unrecognized option (--help) will
issue a warning and then print the usage

Change-Id: Ia757553a4e19d22a8eb70768a8866ab1a76a0eec

5 days agoFix allocation of context buffers on frame resize 86/70986/3
Adrian Grange [Thu, 24 Jul 2014 20:37:47 +0000 (13:37 -0700)]
Fix allocation of context buffers on frame resize

The patch:
https://gerrit.chromium.org/gerrit/#/c/70814/
changed the test that determined whether the context
frame buffers needed to be reallocated or not.

The code checked for a change in total frame area
to signal the need to reallocate context buffers.
However, the above_context buffer needs to be
resized i:xf only the width of the frame has increased.

Change-Id: Ib89d75651af252908144cf662578d84f16cf30e6

5 days agos/CONFIG_DENOISING/CONFIG_VP9_TEMPORAL_DENOISING 88/70988/1
Tim Kopp [Thu, 24 Jul 2014 20:30:00 +0000 (13:30 -0700)]
s/CONFIG_DENOISING/CONFIG_VP9_TEMPORAL_DENOISING

This should prevent confusion with the VP8 CONFIG_TEMPORAL_DENOISING and other
flags.

Change-Id: I1fe4e2977895b7966841d861ab74317ad875b6c8

5 days agoRefactor inter/intra_suberblock_yrd. 93/70993/1
Alex Converse [Wed, 23 Jul 2014 00:57:40 +0000 (17:57 -0700)]
Refactor inter/intra_suberblock_yrd.

Move txfm_rd_in_plane into choose_tx_size_from_rd and cleanup callers.

Change-Id: I1df2d7dc984802bd5e204cbe881ada0d75fbb3f7

5 days agoMerge "x86inc: fix compilation with NASM\r \r Change-Id: I5978921ab1ccad6648a5bde6ad023...
James Zern [Thu, 24 Jul 2014 05:45:51 +0000 (22:45 -0700)]
Merge "x86inc: fix compilation with NASM\r \r Change-Id: I5978921ab1ccad6648a5bde6ad023c3e3430f5f4\r"

6 days agox86inc: fix compilation with NASM 39/70939/2
KO Myung-Hun [Tue, 22 Jul 2014 01:47:20 +0000 (10:47 +0900)]
x86inc: fix compilation with NASM

Change-Id: I5978921ab1ccad6648a5bde6ad023c3e3430f5f4

6 days agoMakefile: strip debugging informations only 42/70942/3
KO Myung-Hun [Tue, 22 Jul 2014 11:19:52 +0000 (20:19 +0900)]
Makefile: strip debugging informations only

--strip-unneeded causes SIGSEGV when accessing g_executable_path. So
test_libvpx crashes due to SIGSEGV in ::testing::InitGoogleTest().

OS/2, aout, strip v2.23.2

Change-Id: I2718d082447ee0d9ad0c021b9156c50e1ac085a6

6 days ago[spatial svc]Fix reference issues 68/70968/2
Minghai Shang [Wed, 23 Jul 2014 20:52:34 +0000 (13:52 -0700)]
[spatial svc]Fix reference issues

1. Remove last reference flag for first frame upper layers in one pass mode.
2. Disable refresh golden frame flag for key frames.

Change-Id: I44ac1bd2c795169e4fbfdd078ea79a1d33a204d6

6 days agoMerge "Remove redundant argument entry in handle_inter_mode"
Jingning Han [Wed, 23 Jul 2014 22:07:01 +0000 (15:07 -0700)]
Merge "Remove redundant argument entry in handle_inter_mode"

6 days agoMerge "Use the chessboard pattern pred search in newmv mode"
Jingning Han [Wed, 23 Jul 2014 22:06:52 +0000 (15:06 -0700)]
Merge "Use the chessboard pattern pred search in newmv mode"

6 days agoMerge "Fix clang compiler warning in denoising_neon."
Marco Paniconi [Wed, 23 Jul 2014 20:58:45 +0000 (13:58 -0700)]
Merge "Fix clang compiler warning in denoising_neon."

6 days agoMerge "Moved call to vp9_clear_system_state() to a proper location"
Yaowu Xu [Wed, 23 Jul 2014 19:46:11 +0000 (12:46 -0700)]
Merge "Moved call to vp9_clear_system_state() to a proper location"

6 days agoRemove redundant argument entry in handle_inter_mode 67/70967/1
Jingning Han [Wed, 23 Jul 2014 19:02:52 +0000 (12:02 -0700)]
Remove redundant argument entry in handle_inter_mode

The value of mode_excluded has been properly set in
vp9_rd_pick_inter_mode_sb(). It is redundant to send it in
handle_inter_mode() and re-set the value again.

Change-Id: I408d4731f2f42e0bcf3ae62e85757717bb410471

6 days agoUse the chessboard pattern pred search in newmv mode 66/70966/1
Jingning Han [Wed, 23 Jul 2014 18:47:56 +0000 (11:47 -0700)]
Use the chessboard pattern pred search in newmv mode

This commit extends the chessboard pattern prediction filter search.
If the above and left blocks have the same prediction filter type,
the encoder will skip the prediction filter type search and use the
reference one.

The overall chessboard pattern prediction filter type search reduces
speed 3 runtime for hard clips. Experiments on park joy at 1080p
and 15000 kbps show that the runtime goes from 723265 ms to 65832 ms,
i.e., about 10% speed-up. Compression performance wise, it affects
the coding quality by

Change-Id: I880975497c7ad166532e9eea9bf46684d77ff327
derf:    -0.326%
yt:      -0.257%
hd:      -0.241%
stdhd:   -0.417%

6 days agoMerge "Remove redundant num_refs definition"
Jingning Han [Wed, 23 Jul 2014 17:34:54 +0000 (10:34 -0700)]
Merge "Remove redundant num_refs definition"

6 days agoFix clang compiler warning in denoising_neon. 61/70961/3
Marco Paniconi [Wed, 23 Jul 2014 16:42:09 +0000 (09:42 -0700)]
Fix clang compiler warning in denoising_neon.

Issue: https://code.google.com/p/webm/issues/detail?id=829

Change-Id: I580308f8aa4af194b5d8990a9692ebd18db68ee8

6 days agoRemove redundant num_refs definition 60/70960/1
Jingning Han [Wed, 23 Jul 2014 16:29:04 +0000 (09:29 -0700)]
Remove redundant num_refs definition

Use is_comp_pred to replace the use case of num_refs.

Change-Id: I4d0c1e14d5f728428a2ae3d293cd2b4a8b2f31d8

6 days agoMerge "Enable chessboard inter prediction filter type search"
Jingning Han [Wed, 23 Jul 2014 16:12:56 +0000 (09:12 -0700)]
Merge "Enable chessboard inter prediction filter type search"

7 days agoMerge "frame_size_tests: reduce default 'large' frame size"
James Zern [Wed, 23 Jul 2014 01:40:31 +0000 (18:40 -0700)]
Merge "frame_size_tests: reduce default 'large' frame size"

7 days agoEnable chessboard inter prediction filter type search 54/70954/1
Jingning Han [Tue, 22 Jul 2014 23:32:20 +0000 (16:32 -0700)]
Enable chessboard inter prediction filter type search

This commit enables a chessboard pattern prediction filter type
search scheme for rate-distortion optimization speed-up. For the
inferred motion vector modes, the encoder can re-use its above/left
neighbor blocks' prediction filter type and skip a full test on
all possible filter types. Such operation is turned on/off
alternatively in a chessboard manner.

It is turned on in speed 3. For test clip pedestrian 1080p, the
runtime is reduced from 231500 ms -> 221700 ms. The compression
performance is changed:
derf:  -0.147%
yt:    -0.134%
hd:    -0.079%
stdhd: -0.220%

Change-Id: I1912f278e7576c2dc632688e3ad7a257410c605a

7 days agoframe_size_tests: reduce default 'large' frame size 53/70953/1
James Zern [Tue, 22 Jul 2014 22:48:25 +0000 (15:48 -0700)]
frame_size_tests: reduce default 'large' frame size

VP9FrameSizeTestsLarge exposed an integer overflow in the VP9 encoder,
for now reduce the size to allow the tests to clear and prevent further
regressions.
4096x4096 -> 4096x2160

this should be restored after the bug is fixed:
https://code.google.com/p/webm/issues/detail?id=828

Change-Id: I47fdf0648f1d9a3951f731bbf0b727f85ada4fa1

7 days agoMerge "vp8: Set default denoiser_decision to copy for UV channel."
Marco Paniconi [Tue, 22 Jul 2014 22:25:06 +0000 (15:25 -0700)]
Merge "vp8: Set default denoiser_decision to copy for UV channel."

7 days agoMerge "Fix get_frame_type function"
Adrian Grange [Tue, 22 Jul 2014 22:17:27 +0000 (15:17 -0700)]
Merge "Fix get_frame_type function"

7 days agoMerge "VP9 denoiser bugfix in debugging code."
Tim Kopp [Tue, 22 Jul 2014 21:48:42 +0000 (14:48 -0700)]
Merge "VP9 denoiser bugfix in debugging code."

7 days agoMerge "USE local best_filter variable in handle_inter_mode"
Jingning Han [Tue, 22 Jul 2014 21:21:34 +0000 (14:21 -0700)]
Merge "USE local best_filter variable in handle_inter_mode"

7 days agoVP9 denoiser bugfix in debugging code. 52/70952/1
Tim Kopp [Tue, 22 Jul 2014 21:00:11 +0000 (14:00 -0700)]
VP9 denoiser bugfix in debugging code.

When OUTPUT_YUV_DENOISED is enabled the encoder outputs the uncompressed,
denoised video to a separate file. Moved the point at which the file is
written to in order to avoid an extra blank frame at the beginning of the video.

Change-Id: I805f6a912b18b3d9cae59b13c5b8108279439ce3

7 days agoMerge "vp9_bitstream.c: cosmetics"
James Zern [Tue, 22 Jul 2014 20:17:40 +0000 (13:17 -0700)]
Merge "vp9_bitstream.c: cosmetics"

7 days agoMerge "[spatial svc]Use #if instead of #ifdef on macro CONFIG_SPATIAL_SVC"
Minghai Shang [Tue, 22 Jul 2014 19:43:40 +0000 (12:43 -0700)]
Merge "[spatial svc]Use #if instead of #ifdef on macro CONFIG_SPATIAL_SVC"

7 days agoFix get_frame_type function 51/70951/1
Adrian Grange [Tue, 22 Jul 2014 19:12:16 +0000 (12:12 -0700)]
Fix get_frame_type function

Fixed the function get_frame_type to return the correct
frame type for golden and last frames.

Change-Id: I8edddd9aa26cbe7a1de8ff211389410b22b1bd14

7 days agovp9_bitstream.c: cosmetics 48/70948/1
James Zern [Tue, 22 Jul 2014 01:27:06 +0000 (18:27 -0700)]
vp9_bitstream.c: cosmetics

fix indent, spelling and drop some vertical whitespace

Change-Id: I722671381a374a24763b07a02805ab1d149ab3f4

7 days agoUSE local best_filter variable in handle_inter_mode 46/70946/1
Jingning Han [Tue, 22 Jul 2014 18:32:17 +0000 (11:32 -0700)]
USE local best_filter variable in handle_inter_mode

This should be a local variable. Move the definition from
vp9_rd_pick_inter_mode_sb to handle_inter_mode.

Change-Id: I14f4168bb1c896ed04e8f6d4cd89fbf4c9839944

7 days agoMerge "vp8 denoiser fix: Update denoised altref on key frame."
Marco Paniconi [Tue, 22 Jul 2014 18:15:08 +0000 (11:15 -0700)]
Merge "vp8 denoiser fix: Update denoised altref on key frame."

7 days agovp8: Set default denoiser_decision to copy for UV channel. 45/70945/1
Marco Paniconi [Tue, 22 Jul 2014 18:06:00 +0000 (11:06 -0700)]
vp8: Set default denoiser_decision to copy for UV channel.

Since the UV decision to denoise is based on Y, we need to set
the default/initial denoiser decision_u/v to COPY_BLOCK,
to make sure if no uv_denoiser is applied we still update
(uv)running_avg with source.

Change-Id: I5af1c2afbd40c498cd3de208bea88c837099b24d

7 days ago[spatial svc]Use #if instead of #ifdef on macro CONFIG_SPATIAL_SVC 43/70943/2
Minghai Shang [Tue, 22 Jul 2014 17:05:40 +0000 (10:05 -0700)]
[spatial svc]Use #if instead of #ifdef on macro CONFIG_SPATIAL_SVC

Change-Id: Ifc94377a0d05d66e3d21b007893a985b66db6082

7 days agovp8 denoiser fix: Update denoised altref on key frame. 29/70929/4
Marco Paniconi [Mon, 21 Jul 2014 16:58:54 +0000 (09:58 -0700)]
vp8 denoiser fix: Update denoised altref on key frame.

On a key frame, the denoised-running_avg for all references
frames should be updated with the source.
The altref denoised-running_avg was not being updated on key frame,
this fixes that.

Change-Id: Ie02cd0ba5383e013af59240e6df7e185d11703f6

7 days agoMerge "Turn on adaptive pred filter scheme for sub8x8 below 720p"
Jingning Han [Tue, 22 Jul 2014 16:11:52 +0000 (09:11 -0700)]
Merge "Turn on adaptive pred filter scheme for sub8x8 below 720p"

8 days agoTurn on adaptive pred filter scheme for sub8x8 below 720p 37/70937/1
Jingning Han [Mon, 21 Jul 2014 23:22:56 +0000 (16:22 -0700)]
Turn on adaptive pred filter scheme for sub8x8 below 720p

For sequences of resolution below 720p, the encoder will check
intra prediction modes and inter prediction modes from LAST_FRAME.
This commit turns on adaptive prediction filter scheme for sub8x8
blocks, where inter prediction modes are enabled. For the test
sequence bus at CIF, the speed 2 runtime goes down from 17879 ms
to 16783 ms, i.e., 6% speed up. The compression performance of
derf set is down by -0.128%.

Change-Id: I01d5321a5ceab4e0666ac5be56c52d896c7a8d45

8 days agoRemove unfinished VP9 alpha channel. 05/70905/2
Alex Converse [Thu, 17 Jul 2014 18:29:59 +0000 (11:29 -0700)]
Remove unfinished VP9 alpha channel.

Change-Id: Ic5d3a3a0dac10b49495771886a31e793bb78b5ca

8 days agoMoved call to vp9_clear_system_state() to a proper location 36/70936/1
Yaowu Xu [Mon, 21 Jul 2014 22:55:12 +0000 (15:55 -0700)]
Moved call to vp9_clear_system_state() to a proper location

The commit moved a call to vp9_clear_system_state() to a correct
location, i.e. prior function calls using floating point numbers.
This was to fix a mismatch mmx code and sse2 version, where a
floating point number used in adjust_frame_rate(cpi) gets NAN due
to mmx registers being in wrong state.

Change-Id: I40e0a6de98812000ccee6a729badb630604fd7e6

8 days agoAdd -DNDEBUG when config option debug is disabled 66/70866/9
Yunqing Wang [Tue, 15 Jul 2014 23:18:34 +0000 (16:18 -0700)]
Add -DNDEBUG when config option debug is disabled

For gcc, when libvpx config option debug is disabled, added the
flag -DNDEBUG to disable the assertions in libvpx for some speedup.

Change-Id: Ifcb7b9e8ef5cbe5d07a24407b53b9a2923f596ee

8 days agoMerge "VP9 denoiser fix: ref frames now updated properly"
Tim Kopp [Mon, 21 Jul 2014 15:28:41 +0000 (08:28 -0700)]
Merge "VP9 denoiser fix: ref frames now updated properly"

11 days agoy4m_test: drop '_t' from local typenames 01/70901/4
James Zern [Thu, 17 Jul 2014 02:03:08 +0000 (19:03 -0700)]
y4m_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I2ede16e0903a96225a342609545281fdbb83d638

11 days agovp9_subtract_test: drop '_t' from local typenames 00/70900/4
James Zern [Thu, 17 Jul 2014 02:02:37 +0000 (19:02 -0700)]
vp9_subtract_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I636012c74f5de91d594d7398e92343f5efe08327

11 days agosubtract_test: drop '_t' from local typenames 99/70899/4
James Zern [Thu, 17 Jul 2014 02:02:16 +0000 (19:02 -0700)]
subtract_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: Ic738b97d492f244c58989a0f73a4d763340ec274

11 days agosixtap_predict_test: drop '_t' from local typenames 98/70898/4
James Zern [Thu, 17 Jul 2014 02:01:46 +0000 (19:01 -0700)]
sixtap_predict_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: Ie59ddc080530d538e4ffb1d862844a84de926d92

11 days agosad_test: drop '_t' from local typenames 97/70897/4
James Zern [Thu, 17 Jul 2014 01:59:28 +0000 (18:59 -0700)]
sad_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I60746bc93ba2446c1458a1f09fd1e49cc2e68534

11 days agopp_filter_test: drop '_t' from local typenames 96/70896/3
James Zern [Thu, 17 Jul 2014 01:58:53 +0000 (18:58 -0700)]
pp_filter_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: If6cf006a089b0fb42d6bf050570d928d955d658a

11 days agopartial_idct_test: drop '_t' from local typenames 95/70895/3
James Zern [Thu, 17 Jul 2014 01:58:16 +0000 (18:58 -0700)]
partial_idct_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: Ied220e09fceef53039f234ebbb7e51c8e081c84e

11 days agointrapred_test: drop '_t' from local typenames 94/70894/3
James Zern [Thu, 17 Jul 2014 01:57:42 +0000 (18:57 -0700)]
intrapred_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I74d7489ea35f988fcfec3651bd16f02e688e406a

11 days agoidct_test: drop '_t' from local typenames 93/70893/3
James Zern [Thu, 17 Jul 2014 01:57:11 +0000 (18:57 -0700)]
idct_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I660a77c0c58ca64075c1051ea440fdf4d7c7e6af

11 days agofdct8x8_test: drop '_t' from local typenames 92/70892/3
James Zern [Thu, 17 Jul 2014 01:56:22 +0000 (18:56 -0700)]
fdct8x8_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: Id2d43931e9b6e16aa4b8e5e346eee08505bdccfb

11 days agofdct4x4_test: drop '_t' from local typenames 91/70891/3
James Zern [Thu, 17 Jul 2014 01:55:40 +0000 (18:55 -0700)]
fdct4x4_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I796252a6f7f749b4faa969bbe7fe83eae64fa393

11 days agodecode_perf_test: drop '_t' from local typenames 90/70890/3
James Zern [Thu, 17 Jul 2014 01:55:08 +0000 (18:55 -0700)]
decode_perf_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I5fdd6319449364f2e930667d3bd54f68ba3cbb2e

11 days agodct32x32_test: drop '_t' from local typenames 89/70889/3
James Zern [Thu, 17 Jul 2014 01:54:31 +0000 (18:54 -0700)]
dct32x32_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: Ie1918d93c0a0d7a854e88b5a77b8c21b4cdc28b7

11 days agodct16x16_test: drop '_t' from local typenames 88/70888/3
James Zern [Thu, 17 Jul 2014 01:53:33 +0000 (18:53 -0700)]
dct16x16_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I4e6f293790922684402bfa9c22ec1ac2b9981a0e

11 days agoconvolve_test: drop '_t' from local typenames 87/70887/3
James Zern [Thu, 17 Jul 2014 01:48:20 +0000 (18:48 -0700)]
convolve_test: drop '_t' from local typenames

_t is reserved by posix

+ switch to camelcase
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Type_Names

Change-Id: I2a22ffc36e9f88781bc7db0d5a28a7ed924bab1a

11 days agoMerge "Re-introduce frame size check inadvertantly deleted"
Adrian Grange [Fri, 18 Jul 2014 23:33:50 +0000 (16:33 -0700)]
Merge "Re-introduce frame size check inadvertantly deleted"

11 days agoMerge "Use custom mkstemp() to fix Win issue in y4m_test"
Deb Mukherjee [Fri, 18 Jul 2014 23:30:06 +0000 (16:30 -0700)]
Merge "Use custom mkstemp() to fix Win issue in y4m_test"

11 days agoRe-introduce frame size check inadvertantly deleted 22/70922/1
Adrian Grange [Fri, 18 Jul 2014 22:40:39 +0000 (15:40 -0700)]
Re-introduce frame size check inadvertantly deleted

This patch adds back in code that checks that the frame
size lies within defined bounds was inadvertantly removed
by a previous patch:
https://gerrit.chromium.org/gerrit/#/c/70814/

Change-Id: If526570ba559260c4b7e98098bc75f7700ae7f97

11 days agoUse custom mkstemp() to fix Win issue in y4m_test 52/70852/17
Deb Mukherjee [Mon, 14 Jul 2014 20:31:29 +0000 (13:31 -0700)]
Use custom mkstemp() to fix Win issue in y4m_test

Uses mkstmp() with directory being the same as the test data
directory to create temporary output file. For Windows
GetTempFileNameA() function is used.

Change-Id: Ie4681b2b4f44f8c22d3b3faf134c44087b484f94

11 days agoVP9 denoiser fix: ref frames now updated properly 20/70920/2
Tim Kopp [Fri, 18 Jul 2014 21:15:30 +0000 (14:15 -0700)]
VP9 denoiser fix: ref frames now updated properly

The ALT_REF_FRAME is now updated in the case of a KEY_FRAME in the VP9 denoiser.

Change-Id: Idf9a9772706f50e774fb240afcc01db38841043c