]>
granicus.if.org Git - libvpx/log
Wan-Teh Chang [Fri, 19 Jul 2019 15:45:55 +0000 (15:45 +0000)]
Merge "Remove unused fb_cb related fields from VP9_COMMON"
Angie Chiang [Thu, 18 Jul 2019 22:12:12 +0000 (22:12 +0000)]
Merge changes I3216c984,I70d40060
* changes:
Make vp9_prepare_nb_full_mvs only return valid mvs
Let vp9_nb_mvs_inconsistency call log2 just once
Wan-Teh Chang [Thu, 18 Jul 2019 21:37:32 +0000 (14:37 -0700)]
Remove unused fb_cb related fields from VP9_COMMON
Remove the cb_priv, get_fb_cb, release_fb_cb, and int_frame_buffers
fields from the VP9_COMMON struct. They are not being used.
Change-Id: I235194aa8b315cd8ec9405bbba5feb3bee69f7e0
Angie Chiang [Wed, 17 Jul 2019 19:36:14 +0000 (12:36 -0700)]
Make vp9_prepare_nb_full_mvs only return valid mvs
In this case, vp9_nb_mvs_inconsistency doesn't need to check
whether each neighbor mv is valid or not.
non_greedy_mv encoding time is reduced by 1.5%
Change-Id: I3216c98481e777d5e0b917ea20ee39b7ca9c9d23
Angie Chiang [Wed, 17 Jul 2019 00:49:22 +0000 (17:49 -0700)]
Let vp9_nb_mvs_inconsistency call log2 just once
The bahavior of this function is to compute log2 of mv difference,
i.e. min log2(1 + row_diff * row_diff + col_diff * col_diff)
against available neghbor mvs.
Since the log2 is monotonic increasing, we can compute
min row_diff * row_diff + col_diff * col_diff first
then apply log2 in the end
non_greedy_mv encoding time is reduced by 1.5%
Change-Id: I70d40060e2621daec27229f1f6d9fea0286aa04e
Wan-Teh Chang [Wed, 17 Jul 2019 20:13:10 +0000 (20:13 +0000)]
Merge "Fix comment typos."
Wan-Teh Chang [Wed, 17 Jul 2019 18:09:55 +0000 (11:09 -0700)]
Fix comment typos.
Fix comment typos in transpose_s16_4x4q() and transpose_u16_4x4q().
Change-Id: I21bcc1fb3fb880798e5a3927c3dbe81dd518c83b
Angie Chiang [Sun, 14 Jul 2019 16:20:58 +0000 (09:20 -0700)]
Add vpx_sad32x32x8_c/avx2
Change-Id: I4dbb7b6c8979c39eb6ffb97750e3cca0f4b7921f
Angie Chiang [Sun, 14 Jul 2019 15:59:18 +0000 (08:59 -0700)]
Add unit test for vpx_sadMxNx8
Change-Id: Ica85e3738708e2a6cc7388fd2cbf6a8840a540d5
Johann [Tue, 16 Jul 2019 18:35:06 +0000 (11:35 -0700)]
Merge remote-tracking branch 'origin/orpington'
BUG=webm:1624
Change-Id: I62e7154d95b3361d6184f0448430bed951f15044
Paul Wilkins [Tue, 16 Jul 2019 15:06:05 +0000 (15:06 +0000)]
Merge "Limit active best quality of layered ARF frames"
Johann [Mon, 15 Jul 2019 21:55:33 +0000 (14:55 -0700)]
Release v1.8.1 Orpington Duck
BUG=webm:1624
Change-Id: Ibd63b64058e52448e0916939a3f85eb23c8161b6
Angie Chiang [Mon, 15 Jul 2019 18:40:10 +0000 (18:40 +0000)]
Merge changes I9288c88d,Ib1ac6f57,I02fac56a,Id6a8b117
* changes:
Use sdx8f in exhaustive_mesh_search_single_step
Sync the behavior of exhaustive_mesh_search
Refactor exhaustive_mesh_search_new
Simplify code in exhaustive_mesh_search_new
Yunqing Wang [Mon, 15 Jul 2019 18:31:09 +0000 (18:31 +0000)]
Merge "Revert "Set up frame contexts based on frame type""
Yunqing Wang [Mon, 15 Jul 2019 17:15:37 +0000 (17:15 +0000)]
Revert "Set up frame contexts based on frame type"
This reverts commit
affd9921e405063ed7815f412b15e634c8c9cfca .
Reason for revert: Quality regression
(VP9/EndToEndTestLarge.EndtoEndPSNRTest/195 failed)
BUG=webm:1635
Original change's description:
> Set up frame contexts based on frame type
>
> In single layer ARF case, use different frame
> contexts for KF, ARF/GF, LF, OVERLAY update types.
>
> Change-Id: Iebb7f9bb430e483dea1e75fc122b9b67645ce804
Change-Id: I98a4eaa6ec0ae6616ea5ad35d1580501b7422e1b
Angie Chiang [Fri, 12 Jul 2019 23:15:55 +0000 (16:15 -0700)]
Use sdx8f in exhaustive_mesh_search_single_step
This speed up non_greedy_mv by 4%
Change-Id: I9288c88db56ea4201a7ec4493ca5c567d76af0f1
Angie Chiang [Thu, 11 Jul 2019 23:52:50 +0000 (16:52 -0700)]
Sync the behavior of exhaustive_mesh_search
Change-Id: Ib1ac6f57519eb4da93e7c75b0c26a372ffc5d524
Angie Chiang [Thu, 11 Jul 2019 23:04:13 +0000 (16:04 -0700)]
Refactor exhaustive_mesh_search_new
Add the following two functions:
exhaustive_mesh_search_multi_step
exhaustive_mesh_search_single_step
Change-Id: I02fac56a815b091beab2203afce560d7d29aad44
Angie Chiang [Thu, 11 Jul 2019 22:25:31 +0000 (15:25 -0700)]
Simplify code in exhaustive_mesh_search_new
Change-Id: Id6a8b117b066a56e9312f528ec8f417dd4b2a2d8
Yunqing Wang [Thu, 11 Jul 2019 15:07:55 +0000 (15:07 +0000)]
Merge "Adjust the quality of boosted frames"
Yunqing Wang [Thu, 11 Jul 2019 15:07:35 +0000 (15:07 +0000)]
Merge "Set up frame contexts based on frame type"
Yunqing Wang [Thu, 11 Jul 2019 15:07:17 +0000 (15:07 +0000)]
Merge "Modify frame context index"
Marco Paniconi [Wed, 10 Jul 2019 16:48:19 +0000 (09:48 -0700)]
vp9-rtc: Reduce color artifact for speed 8
Push the reduced chroma check to speed > 8.
Change-Id: I92dd0aa9933bb5417b1dc5eef8f805ee51e04ac9
Jerome Jiang [Tue, 9 Jul 2019 23:46:24 +0000 (16:46 -0700)]
vp9: Use mb_rows/cols from VP9_COMMON in postproc.
When frame height is not divisible by 16, the calculation of mb_rows in
postproc was wrong.
Change-Id: I69d108f1b8facdd5650b5b7928a0033b268530d2
James Zern [Tue, 2 Jul 2019 18:46:30 +0000 (18:46 +0000)]
Merge "Remove android_tools deps"
Yun Liu [Mon, 1 Jul 2019 21:44:28 +0000 (14:44 -0700)]
Remove android_tools deps
Bug: 428426
Change-Id: Ia3c31fe2b513ac995baad15c8376c590fd1104f7
James Zern [Sun, 30 Jun 2019 01:18:15 +0000 (18:18 -0700)]
vp9_cx_iface,encoder_encode: fix -Wclobbered for pts
Change-Id: Ia7fd4fedb0dcbb626d0e7f4951360e2462b518e2
(cherry picked from commit
ae3c6e9ec7dbaad8ee507b1c531c54bdc0141e9f )
James Zern [Mon, 1 Jul 2019 22:17:46 +0000 (22:17 +0000)]
Merge "vp9_cx_iface,encoder_encode: fix -Wclobbered for pts"
Marco Paniconi [Mon, 1 Jul 2019 04:16:40 +0000 (21:16 -0700)]
vp9-rtc: Fix color artifacts for speed >= 8
Fix to avoid color artifacts observed for speed >= 8.
In model_rd_large in non_rd pickmode: always do the
transform skipping test for UV plane.
BUG=b/
136198713
Change-Id: Idd91322fb898fe731846d8581b21010096f87680
(cherry picked from commit
c33c7ca85fa7b28658f3b086d579a7fbbcf41dea )
Marco Paniconi [Mon, 1 Jul 2019 04:16:40 +0000 (21:16 -0700)]
vp9-rtc: Fix color artifacts for speed >= 8
Fix to avoid color artifacts observed for speed >= 8.
In model_rd_large in non_rd pickmode: always do the
transform skipping test for UV plane.
BUG=b/
136198713
Change-Id: Idd91322fb898fe731846d8581b21010096f87680
Ravi Chaudhary [Fri, 24 May 2019 05:57:06 +0000 (11:27 +0530)]
Adjust the quality of boosted frames
As the boosted frames, early in key frame interval,
are used as reference by many subsequent boosted frames,
boosted frames that are closer to the reference key frame
should be allocated with more target bits than the rest.
Similarly, the active best quality should be lower for
boosted frames early in the key interval and vice versa.
Hence, the bits allocation and active best quality are varied
based on their temporal position in the key frame interval.
Change-Id: I1362248560d074b9e209657a23ae73dda0b01d52
James Zern [Sun, 30 Jun 2019 01:18:15 +0000 (18:18 -0700)]
vp9_cx_iface,encoder_encode: fix -Wclobbered for pts
Change-Id: Ia7fd4fedb0dcbb626d0e7f4951360e2462b518e2
Dan Zhu [Fri, 28 Jun 2019 21:07:44 +0000 (14:07 -0700)]
add flags for empty blocks
Change-Id: Iedf3bdd87d203db5163d3cc47fcbef1fd002218f
Angie Chiang [Fri, 28 Jun 2019 17:50:28 +0000 (17:50 +0000)]
Merge changes I833c82fb,I05a39165,Ie044bb01,I565f477f
* changes:
Integerize vp9_full_pixel_diamond_new
Integerize vp9_refining_search_sad_new
Integerize diamond_search_sad_new()
Refactor vp9_full_pixel_diamond_new
James Zern [Fri, 28 Jun 2019 02:52:20 +0000 (02:52 +0000)]
Merge "vp9_encodeframe: quiet a few integer sanitizer warnings"
James Zern [Thu, 27 Jun 2019 22:53:58 +0000 (15:53 -0700)]
vp9_encodeframe: quiet a few integer sanitizer warnings
implicit conversion from type 'int' of value -2 (32-bit, signed) to type
'uint8_t' (aka 'unsigned char') changed the value to 254 (8-bit,
unsigned)
BUG=webm:1615
Change-Id: I9b8f5a9df3211e344e91d67a45d321e7115f5d4a
James Zern [Thu, 27 Jun 2019 22:21:23 +0000 (15:21 -0700)]
timestamp_test: enable TestMicrosecondTimebase
this doesn't cause any overflow issues after:
11de1b838 Fix timestamp overflow issues
BUG=webm:701,webm:1614
Change-Id: I7e1cbfa4264d1661eb9a5baa2b2111a0899360f2
Sai Deng [Thu, 27 Jun 2019 16:57:05 +0000 (16:57 +0000)]
Merge "Change parameters for highbd tune=ssim"
Angie Chiang [Wed, 26 Jun 2019 23:28:46 +0000 (16:28 -0700)]
Integerize vp9_full_pixel_diamond_new
Change-Id: I833c82fb910c8274b5a237e26fe0dcda7def9796
Angie Chiang [Wed, 26 Jun 2019 23:18:01 +0000 (16:18 -0700)]
Integerize vp9_refining_search_sad_new
Change-Id: I05a39165b9910262eca8fdf644ae982b80d309b4
Angie Chiang [Wed, 26 Jun 2019 23:02:00 +0000 (16:02 -0700)]
Integerize diamond_search_sad_new()
Change-Id: Ie044bb01e26d871bace309ae1f45aa880ea1de62
Angie Chiang [Wed, 26 Jun 2019 22:46:43 +0000 (15:46 -0700)]
Refactor vp9_full_pixel_diamond_new
Remove redundant bestsme assignments
Change-Id: I565f477f51c2a13369ebd1532eed05115e774238
Angie Chiang [Tue, 25 Jun 2019 22:43:43 +0000 (15:43 -0700)]
Remove mv_dist/mv_cost from new mv search funcs
The functions are
diamond_search_sad_new()
vp9_full_pixel_diamond_new()
vp9_refining_search_sad_new()
Change-Id: Ied6fe98b8a1401c95f0488faf781c5cd5e8e0db6
Angie Chiang [Tue, 25 Jun 2019 22:21:17 +0000 (15:21 -0700)]
Speed up diamond_search_sad_new
The percentage of encoding time spent on diamond_search_sad_new
reduces from 8% to 6%
Change-Id: I1be55b957475d780974cc2e721f8c2d4d266e916
Angie Chiang [Tue, 25 Jun 2019 21:35:24 +0000 (14:35 -0700)]
Let full_pixel_exhaustive_new return int64_t
Change-Id: I2c7cd7363a1b61b7aa7c35fd9f4e6b926b67418f
Dan Zhu [Thu, 20 Jun 2019 23:05:30 +0000 (16:05 -0700)]
script to compact frames to y4m video
Change-Id: I2d8c3ccf49c172a54181aeb2e2b8169bf5402456
Dan Zhu [Thu, 20 Jun 2019 04:34:22 +0000 (21:34 -0700)]
add output of frame info
Change-Id: I70d750be13d9a654d1f21d7809d8d44c491ae477
Dan Zhu [Wed, 19 Jun 2019 17:54:23 +0000 (10:54 -0700)]
Add Ray Tracing
Add braces
Change-Id: I5355ccd8f745dfbd4fe3923a81aa3c9f8fda07b3
sdeng [Wed, 12 Jun 2019 23:05:33 +0000 (16:05 -0700)]
Change parameters for highbd tune=ssim
With this CL:
PSNR SSIM MS-SSIM
lowres_10bd 2.8 -5.6 -6.5
midres_10bd 2.6 -5.6 -6.3
Before this CL:
PSNR SSIM MS-SSIM
lowres_10bd 6.1 -6.5 -7.7
midres_10bd 6.2 -6.0 -7.2
Change-Id: Iad0ad96d55ad140db00ce86c34ab85461cd963eb
Deepa K G [Thu, 20 Jun 2019 13:00:41 +0000 (18:30 +0530)]
Set up frame contexts based on frame type
In single layer ARF case, use different frame
contexts for KF, ARF/GF, LF, OVERLAY update types.
Change-Id: Iebb7f9bb430e483dea1e75fc122b9b67645ce804
Deepa K G [Mon, 24 Jun 2019 07:04:28 +0000 (12:34 +0530)]
Modify frame context index
Used separate frame contexts for non-boosted frames.
Adjusted the frame context index grouping for boosted
frames.
Change-Id: I7f6f83f53d46f66a83a6806c2b568bd833ce940d
Dan Zhu [Fri, 14 Jun 2019 18:42:01 +0000 (11:42 -0700)]
Add Scene module to manage other objects
and calculation
Add interpolation in the Scene
Delete Color interpolation
Build triangle mesh
Reconstruct the code of depth interpolation
Add new data structure Node for back linking
Change-Id: Ibb1e896a2e3623d4549d628539d81d79827ba684
Angie Chiang [Fri, 21 Jun 2019 00:06:14 +0000 (17:06 -0700)]
Integerize exhaustive_mesh_search_new()
Change-Id: Ia87ed60f46384e7bb7c5f55e9e28c406562a6f19
Angie Chiang [Thu, 20 Jun 2019 23:06:56 +0000 (16:06 -0700)]
Make vp9_nb_mvs_inconsistency return int64_t
Change-Id: I925156ed45e13a06c449c2fbff8a3c26baf8d835
Angie Chiang [Wed, 19 Jun 2019 23:25:00 +0000 (16:25 -0700)]
Make type of lambda int in TplDepFrame
Change-Id: I8fdf1ad4790201b1624c8408d92983aeb0b08302
Angie Chiang [Wed, 19 Jun 2019 21:51:26 +0000 (14:51 -0700)]
Integerize log2_approximation()
Change-Id: If645bf6a90f4bfb5a51ca0a78b88d1eb5bedbec2
Johann [Thu, 20 Jun 2019 17:43:09 +0000 (10:43 -0700)]
vsx: disable on all builds
The previous change to disable some vsx functions did not clear
the test failures. Disable vsx by default until it is investigated
and fixed.
BUG=webm:1522
Change-Id: I8ba2e7261ea3eee5022832da7e4a22bf8daa0996
Jerome Jiang [Thu, 20 Jun 2019 17:20:54 +0000 (17:20 +0000)]
Merge "vp8: Allow higher resolution to get periodic keyframe."
Angie Chiang [Thu, 20 Jun 2019 17:14:15 +0000 (17:14 +0000)]
Merge "Change log2_fast to log2_approximation"
Ravi Chaudhary [Thu, 20 Jun 2019 08:57:43 +0000 (08:57 +0000)]
Merge "Start with q=active_best_quality for non-forced key frames"
Ravi Chaudhary [Wed, 22 May 2019 04:46:30 +0000 (10:16 +0530)]
Start with q=active_best_quality for non-forced key frames
Change-Id: I435d247ab4d1d160f12f5a3710e6cafb5cfd6610
Jerome Jiang [Wed, 19 Jun 2019 22:26:05 +0000 (15:26 -0700)]
vp8: Allow higher resolution to get periodic keyframe.
BUG=webm:1632
Change-Id: Ib05a010245e77f9d502c3e7b8f488fca280ea544
Angie Chiang [Tue, 18 Jun 2019 22:33:23 +0000 (15:33 -0700)]
Change log2_fast to log2_approximation
This reduce non_greedy_mv encoding time by 8.9%
Use linear approximation for value >= 1024
BDRate increases slightly on hdres
lowres: -0.002
midres: 0.007
hdres: 0.057
Change-Id: I55fd5e0bf0ab2206a286e11974f701cc48084be8
Dan Zhu [Wed, 19 Jun 2019 17:13:50 +0000 (17:13 +0000)]
Merge "3D reconstruction tool build by Processing"
Angie Chiang [Wed, 19 Jun 2019 16:45:15 +0000 (16:45 +0000)]
Merge "Implement log2_fast for vp9_nb_mvs_inconsistency"
Yue Chen [Tue, 18 Jun 2019 23:39:09 +0000 (23:39 +0000)]
Merge "Fix timestamp overflow issues"
Angie Chiang [Tue, 18 Jun 2019 16:53:25 +0000 (09:53 -0700)]
Implement log2_fast for vp9_nb_mvs_inconsistency
This speed up non_greedy_mv by 8.7%
Change-Id: Ia46e3e7c4d32ec364091fad26cc953c62963e526
Yue Chen [Fri, 12 Apr 2019 22:30:05 +0000 (15:30 -0700)]
Fix timestamp overflow issues
- Save the initial user-specified timestamp and rebase all further
timestamps by this value. This makes libvpx internal timestamps to
always start from zero, regardless of the user's timestamps.
- Calculate reduced timestamp conversion ratio and use it to convert
user's timestamps to libvpx internal timestamps and back. The effect
of this is that integer overflow due to multiplication doesn't
happen for a much longer time.
BUG=webm:701
Change-Id: Ic6f5eacd9a7c21b95707d31ee2da77dc8ac7dccf
Jerome Jiang [Sat, 15 Jun 2019 00:34:40 +0000 (00:34 +0000)]
Merge "Fix memory leak for vp8 multi-res encoder."
Harish Mahendrakar [Fri, 14 Jun 2019 18:08:36 +0000 (11:08 -0700)]
vpx_dec_fuzzer: Remove fmemopen dependency
fmemopen is not preferred during fuzzing.
Removed all file operations.
Removed need for allocating a different input buffer.
data buffer is appropriately incremented and passed directly to decoder
This will also test input being sent in an unaligned buffer to the library.
Removed read_frame function and did the required parsing inline.
Change-Id: I32829b0149dba9339f2e8bb4c0249a4987a630c7
Dan Zhu [Fri, 7 Jun 2019 21:54:30 +0000 (14:54 -0700)]
3D reconstruction tool build by Processing
(a java based language for data visualization)
add MotionField module
reformat the code by using newest clang-format version
add necessary comments
add new functions
move basic settings to setup
Change-Id: I64a6b2daec06037daa9e54c6b8d1eebe58aa6de0
Jerome Jiang [Fri, 14 Jun 2019 16:50:30 +0000 (09:50 -0700)]
Fix memory leak for vp8 multi-res encoder.
BUG=webm:1630
Change-Id: I03e74e78aa0ead66eda7506e921b1774b5442ed5
Deepa K G [Fri, 14 Jun 2019 06:49:43 +0000 (06:49 +0000)]
Merge "Use previous ARF as GOLDEN frame for the next GOP"
Johann Koenig [Thu, 13 Jun 2019 22:06:22 +0000 (22:06 +0000)]
Merge "ppc: disable vsx for small predictors"
Johann [Thu, 13 Jun 2019 18:29:09 +0000 (11:29 -0700)]
ppc: disable vsx for small predictors
These functions cause test failures when running the entire suite.
BUG=webm:1522
Change-Id: I2c1dc4923e9f149464f365ef63dc59621cfabf5a
Harish Mahendrakar [Thu, 13 Jun 2019 18:45:02 +0000 (18:45 +0000)]
Merge "vpx_dec_fuzzer: Add -fsanitize=fuzzer-no-link"
Johann Koenig [Thu, 13 Jun 2019 18:40:42 +0000 (18:40 +0000)]
Merge "ppc: disable vsx optimizations with hbd"
Harish Mahendrakar [Wed, 12 Jun 2019 22:14:15 +0000 (15:14 -0700)]
vpx_dec_fuzzer: Add -fsanitize=fuzzer-no-link
Updated build instructions for vpx_dec_fuzzer to include
-fsanitize=fuzzer-no-link while configuring library
Change-Id: Id158256aa1cfe3d847720e8558cb5998ad4fd777
Deepa K G [Tue, 9 Apr 2019 12:25:36 +0000 (17:55 +0530)]
Use previous ARF as GOLDEN frame for the next GOP
This patch uses ARF itself as the GOLDEN frame for the
next gf group instead of replacing it with the overlay
frame. By doing so, bits consumed by the overlay frame
will be reduced.
Change-Id: I909ceaa6d501c267d315614075913d45ad426c15
Johann [Mon, 10 Jun 2019 22:08:40 +0000 (15:08 -0700)]
sse: remove unused HAVE_SSE files
There are no sse functions which use these files. Cleans up spurious
warnings when building with --disable-sse2
Change-Id: I04d84b8b7ecfe6da7d5d4df63840796c7b04c085
Johann [Mon, 10 Jun 2019 21:35:03 +0000 (14:35 -0700)]
ppc: disable vsx optimizations with hbd
vsx optimizations do not support 32 bit tran_low_t values.
BUG=webm:1563
Change-Id: I9e6348078f6e4855acfd381133eb840a435b7f81
James Zern [Fri, 7 Jun 2019 22:06:29 +0000 (15:06 -0700)]
update libwebm to
libwebm-1.0.0.27-361-g81de00c
81de00c Check there is only one settings per ContentCompression
5623013 Fixes a double free in ContentEncoding
93b2ba0 mkvparser: quiet static analysis warnings
Change-Id: Ieaa562ef2f10075381bd856388e6b29f97ca2746
Jerome Jiang [Thu, 6 Jun 2019 22:38:13 +0000 (15:38 -0700)]
vp8: fix leak in vp8e_mr_alloc_mem
BUG=webm:1596
Change-Id: I09ba00a7b7ad331671a7a285a2ac5630d8b62199
Sai Deng [Thu, 6 Jun 2019 15:50:14 +0000 (15:50 +0000)]
Merge "Update performance test results for tune=SSIM"
Sai Deng [Thu, 6 Jun 2019 06:13:08 +0000 (06:13 +0000)]
Merge "Fix a bug in best RD cost updating"
sdeng [Thu, 6 Jun 2019 03:40:16 +0000 (20:40 -0700)]
Update performance test results for tune=SSIM
I made a mistake (used the outdated baseline) in the CL I
submitted earlier this week:
https://chromium-review.googlesource.com/c/webm/libvpx/+/
1638854
The corrected results are following:
The additional gains/loss on top of the tune=ssim are:
Data Set Overall PSNR SSIM MS-SSIM
Lowres 3.490 -3.164 -2.267
Midres 2.245 -2.270 -2.287
HDres 2.562 -1.804 -1.681
Lowres_10bd 3.477 -2.399 -2.689
Midres_10bd 3.467 -1.534 -1.636
The overall gains/loss comparing to tune=psnr are:
Data Set Overall PSNR SSIM MS-SSIM
Lowres 6.127 -5.818 -4.783
Midres 4.574 -5.383 -6.242
HDres 4.908 -6.218 -7.106
Lowres_10bd 6.115 -6.212 -7.790
Midres_10bd 6.238 -6.064 -7.249
Change-Id: Iae72482f7b30f200e5021a98c920eed841d0972a
sdeng [Tue, 4 Jun 2019 20:54:18 +0000 (13:54 -0700)]
Fix a bug in best RD cost updating
This CL fixed a bug that sometimes we calculate the best rd cost using
uninitialized rd_div. This CL also includes a small refactoring of
rd_pick_partition().
Speed change: (the smaller the better)
Performance counter stats for './vpxenc park_joy_480p.y4m --limit=50
-o output.webm':
with this CL: 297,086,181,136 instructions:u
without this CL: 299,285,835,104 instructions:u
Quality change: (negative is better)
avg_psnr ovr_psnr ssim
(low_res) 0.007 0.005 -0.002
(mid_res) 0.022 0.028 0.007
(hd_res) -0.008 -0.003 -0.014
Change-Id: I8924d8426364304212bcef3aba13346783e6f1a8
James Zern [Tue, 4 Jun 2019 21:45:36 +0000 (14:45 -0700)]
configure: test -Wno-* flags used with libyuv
with g++ this avoids:
command line option ‘-Wno-missing-prototypes’ is valid for C/ObjC but
not for C++
the flag is necessary with clang.
BUG=webm:1584
Change-Id: I250c76483302d913999e5f9e0d09ee6449b052df
James Zern [Tue, 4 Jun 2019 21:12:50 +0000 (21:12 +0000)]
Merge changes Ib73136b2,Ie514f663
* changes:
configure: enable -Wmissing-declarations for more files
vp9_thread_test: quiet -Wmissing-prototypes
Marco Paniconi [Tue, 4 Jun 2019 18:35:15 +0000 (18:35 +0000)]
Merge "vp9-rtc: Use speed 5 for postencode drop tests."
Marco Paniconi [Tue, 4 Jun 2019 17:07:11 +0000 (10:07 -0700)]
vp9-rtc: Use speed 5 for postencode drop tests.
Test was running at speed 4, which is not used for real-time.
With this change all Datarate tests are now running at
(speed >= 5, 1 pass, real-time mode), which is what they
were intended for.
BUG=webm:1512
Change-Id: I47a721dadd24b73df722c44419df7cfc06c44226
Sai Deng [Tue, 4 Jun 2019 16:10:51 +0000 (16:10 +0000)]
Merge "Hierarchical rdmult scaling when tune=ssim"
James Zern [Mon, 3 Jun 2019 21:43:24 +0000 (14:43 -0700)]
configure: enable -Wmissing-declarations for more files
avoid using it with third_party/libyuv as that still requires some work.
BUG=webm:1584
Change-Id: Ib73136b22c89d927b112364e19d725c51768bbb7
sdeng [Sat, 1 Jun 2019 00:25:09 +0000 (17:25 -0700)]
Hierarchical rdmult scaling when tune=ssim
Use different lagrangian multiplier scaling factor for different block
size. The blocks whose sizes are less than 16x16 share the same multiplier
of their parent block.
The additional gains/loss on top of the tune=ssim are:
Data Set Overall PSNR SSIM MS-SSIM
Lowres 2.918 -3.691 -2.596
Midres 1.708 -2.656 -2.624
HDres 1.619 -2.496 -2.391
Midres_10bd 1.518 -3.263 -3.561
The overall gains/loss comparing to tune=psnr are:
Data Set Overall PSNR SSIM MS-SSIM
Lowres 5.583 -6.208 -4.978
Midres 4.024 -5.610 -6.411
HDres 4.102 -6.614 -7.457
Midres_10bd 4.647 -7.181 -8.614
Change-Id: I0e6c5008488734e979b2dacde9fc2a17f3aa620f
James Zern [Tue, 4 Jun 2019 00:08:23 +0000 (17:08 -0700)]
vp9_thread_test: quiet -Wmissing-prototypes
BUG=webm:1584
Change-Id: Ie514f6630acfb018a3ac4a05758c8b4119ae28fa
Jerome Jiang [Mon, 3 Jun 2019 19:54:28 +0000 (19:54 +0000)]
Merge "Remove unused func for CONFIG_REALTIME_ONLY"
Jerome Jiang [Mon, 3 Jun 2019 17:11:34 +0000 (10:11 -0700)]
Remove unused func for CONFIG_REALTIME_ONLY
Change-Id: I503e147e20e5b69b910c425d169e59821874f627
Sai Deng [Mon, 3 Jun 2019 16:19:06 +0000 (16:19 +0000)]
Merge changes I7b1d1482,I01588758,I6f17864e
* changes:
Update rdcost using the rd_mult in current block
Use distortion and rate of best_rd as the params
Use distortion and rate recursively in rd_pick_partition()
sdeng [Fri, 31 May 2019 22:11:53 +0000 (15:11 -0700)]
Update rdcost using the rd_mult in current block
This CL is a preparation for implementing hierarchical SSIM rdmult scaling.
There is very little impact on metrics and speed:
avg_psnr ovr_psnr ssim
midres 0.009 0.009 0.015
perf stat -e instructions:u ./vpxenc park_joy_480p.y4m --limit=50
with this cl: 317,722,808,461
before: 317,700,108,619
Change-Id: I7b1d1482ac69f7bc87065a93223a0274bcbe8ce3
sdeng [Fri, 31 May 2019 21:35:02 +0000 (14:35 -0700)]
Use distortion and rate of best_rd as the params
Also added rd calculation for negative rates and distortions.
This CL is a preparation for implementing hierarchical SSIM rdmult scaling.
Little impact on quality and speed:
avg_psnr ovr_psnr ssim
(mid_res) -0.015 -0.009 -0.018
perf stat -e instructions:u ./vpxenc park_joy_480p.y4m --limit=50
with this cl: 317,700,108,619
before: 317,669,279,763
Change-Id: I01588758b7be2aab32236440ec0e57d7af56e920