]>
granicus.if.org Git - libvpx/log
Jingning Han [Tue, 18 Dec 2018 00:49:09 +0000 (16:49 -0800)]
Localize x->encode_breakout setup to non-rd mode search
The breakout speed feature is currently only used by the non-rd
mode search path. Localize it to simplify set_offset() logic.
Change-Id: I27e7519c987a7caac2e4bd6be0ede1b9c8320e55
Jingning Han [Mon, 17 Dec 2018 22:24:35 +0000 (14:24 -0800)]
Add frame header control to turn on PSNR_AQ mode
Change-Id: I46f695b15153c8c508f525a5673db24326371977
Jingning Han [Mon, 17 Dec 2018 18:30:17 +0000 (10:30 -0800)]
Add PSNR_AQ mode
Placeholder to support adaptive quantizer for PSNR and SSIM coding
quality improvement.
Change-Id: Id967c9914bb1d72a6f480ef1ba9d6650914dd658
Jingning Han [Thu, 13 Dec 2018 18:13:42 +0000 (18:13 +0000)]
Merge "Make the use of tpl model controlled by the encoder params"
Jingning Han [Thu, 13 Dec 2018 17:21:28 +0000 (09:21 -0800)]
Make the use of tpl model controlled by the encoder params
The control has been exposed to the vpxenc input parameter. Remove
the internal hard coded control that disables it at speed 1 and
above settings.
Change-Id: Ib17772cb895f24da5a7d0487e748cc1a9c6740b3
James Zern [Thu, 13 Dec 2018 00:03:03 +0000 (00:03 +0000)]
James Zern [Wed, 12 Dec 2018 02:06:20 +0000 (18:06 -0800)]
update libwebm to
libwebm-1.0.0.27-352-g6ab9fcf
https://chromium.googlesource.com/webm/libwebm/+log/
af81f26 ..
6ab9fcf
Change-Id: I9d56e1fbaba9b96404b4fbabefddc1a85b79c25d
Angie Chiang [Wed, 12 Dec 2018 22:43:44 +0000 (22:43 +0000)]
Merge "Replace mv_arr by pyramid_mv_arr"
Angie Chiang [Wed, 12 Dec 2018 00:06:38 +0000 (16:06 -0800)]
Replace mv_arr by pyramid_mv_arr
We plan to compute mv field in different scale.
Change-Id: I49a92d948f8b5dbab78e38c61f5f4f879bbe269f
Angie Chiang [Wed, 12 Dec 2018 19:25:53 +0000 (19:25 +0000)]
Merge changes I44da4884,I36e3bcae
* changes:
Change interface of motion_compensated_prediction
Move prepare_nb_full_mvs to vp9_mcomp.c
Marco Paniconi [Wed, 12 Dec 2018 05:57:57 +0000 (21:57 -0800)]
vp8: Fix to enabling MFQE
Remove the unused *_DEBUG_* enum values in vpx/vp8.h
This fixes issue with enabling MFQE, which was
caused in
4807f15 , where the unused DEBUG flags
were removed from common/ppflags.h but not in vp8.h.
BUG=913246
Change-Id: I47f114ef20adc084cb4883add5ac3ebf58ae9f1d
Deepa K G [Wed, 12 Dec 2018 04:47:56 +0000 (04:47 +0000)]
Merge "Rescale arf bit budget calculation"
Deepa K G [Wed, 12 Dec 2018 04:47:30 +0000 (04:47 +0000)]
Merge "Use undamped adjustment for rate correction factors"
James Zern [Wed, 12 Dec 2018 01:59:42 +0000 (01:59 +0000)]
Merge "test/svc_end_to_end_test: fix SetConfig() signature"
Angie Chiang [Wed, 12 Dec 2018 00:19:50 +0000 (16:19 -0800)]
Change interface of motion_compensated_prediction
Change-Id: I44da4884eea26f0feb7b17f4100db7e5bddd14b4
Angie Chiang [Mon, 10 Dec 2018 22:22:53 +0000 (14:22 -0800)]
Move prepare_nb_full_mvs to vp9_mcomp.c
Change-Id: I36e3bcae60751a9caeac03a3c94cb752b73a010b
Jerome Jiang [Tue, 11 Dec 2018 18:39:41 +0000 (18:39 +0000)]
Merge "Refactor svc_*_test.cc"
Paul Wilkins [Tue, 11 Dec 2018 13:33:56 +0000 (13:33 +0000)]
Merge "Fix intra_count_low calculation in first pass"
Deepa K G [Tue, 11 Dec 2018 12:08:06 +0000 (17:38 +0530)]
Use undamped adjustment for rate correction factors
Undamped adjustment is used for the first frame
of each frame type while updating the rate
correction factors.
Change-Id: I42f80daa123c4cd4e45c18c6960cc7a67e7df7e6
James Zern [Tue, 11 Dec 2018 07:01:01 +0000 (23:01 -0800)]
test/svc_end_to_end_test: fix SetConfig() signature
make the parameter constant to match the base class and mark the
function virtual. virtual is used to match the rest of the code base,
but now that c++11 is required all such functions could be changed to
override.
since:
bb3a82ec3 vp9 svc: add test for scaling partition on 1080p crash.
Change-Id: I4717f0116a231ea954b34da9cfec69c462c21699
Jingning Han [Tue, 11 Dec 2018 06:42:18 +0000 (06:42 +0000)]
Merge "Clean up condition logics in rc_pick_q_and_bounds_two_pass()"
Jingning Han [Mon, 10 Dec 2018 22:02:51 +0000 (14:02 -0800)]
Clean up condition logics in rc_pick_q_and_bounds_two_pass()
Factor out common conditions for better readability.
Change-Id: I2a2b576e7d3e5cf036e9e355fc7ce0509ecb3d7e
Hui Su [Tue, 11 Dec 2018 00:14:32 +0000 (00:14 +0000)]
Merge "Remove redundant code about motion vector test"
Jerome Jiang [Mon, 10 Dec 2018 20:53:20 +0000 (12:53 -0800)]
Refactor svc_*_test.cc
Put test classes into svc_test namespace.
Make num_nonref_frames_ and mismatched_nframes private, as they're
computed by encoder/decoder hooks which shouldn't be modified outside
the class.
Add accessor to num_nonref_frames_.
Change-Id: I3836a45426796ba6a8c98dd31e21b5aec4b8abf4
James Zern [Mon, 10 Dec 2018 19:49:14 +0000 (19:49 +0000)]
Merge "test/svc_*_test: fix SetConfig() signature"
Hui Su [Thu, 6 Dec 2018 23:03:12 +0000 (15:03 -0800)]
Remove redundant code about motion vector test
Only need to set find_fractional_mv_step once.
Change-Id: Ib59dd1e3bb8bc973f2e0f3fc436738bfaf2fad81
Hui Su [Mon, 10 Dec 2018 18:17:57 +0000 (18:17 +0000)]
Merge "Add enum definition for subpel search precision"
Johann Koenig [Mon, 10 Dec 2018 16:47:03 +0000 (16:47 +0000)]
Merge "apply -Wextra to third_party/"
Jerome Jiang [Mon, 10 Dec 2018 01:44:43 +0000 (01:44 +0000)]
Merge "vp9 screen: Update motion search offset when set to NSTEP."
Jerome Jiang [Mon, 10 Dec 2018 01:44:20 +0000 (01:44 +0000)]
Merge "vp9 svc: add test for scaling partition on 1080p crash."
James Zern [Sat, 8 Dec 2018 08:43:31 +0000 (00:43 -0800)]
test/svc_*_test: fix SetConfig() signature
make the parameter constant to match the base class and mark the
function virtual. virtual is used to match the rest of the code base,
but now that c++11 is required all such functions could be changed to
override.
Change-Id: I551a05bbd9d05a9eddb653f42eaad68880c88141
Sai Deng [Sat, 8 Dec 2018 18:43:55 +0000 (18:43 +0000)]
Merge "Add satd avx2 implementation"
Jerome Jiang [Fri, 7 Dec 2018 02:24:13 +0000 (18:24 -0800)]
vp9 svc: add test for scaling partition on 1080p crash.
BUG=webm:1578
Change-Id: Ie03ed454394933fa89f751edc6928651393f3f12
Jerome Jiang [Wed, 5 Dec 2018 02:37:02 +0000 (18:37 -0800)]
vp9 screen: Update motion search offset when set to NSTEP.
Search method and step parameter might be changed in speed settings.
In this case, we should update the search area offset due to the change
of search method.
Change-Id: I51dc584bbf35e998757da326355dd4b8a4d0093f
James Zern [Sat, 8 Dec 2018 04:10:38 +0000 (04:10 +0000)]
Merge "test/*: use std::*tuple"
James Zern [Fri, 7 Dec 2018 07:58:37 +0000 (23:58 -0800)]
test/*: use std::*tuple
since:
77fa51003 Replace deprecated scoped_ptr with unique_ptr
c++11 has been required so <tuple> is safe to use
Change-Id: I873cb953104b361a8503b5839a3372ce2b99e73c
Hui Su [Fri, 7 Dec 2018 00:53:26 +0000 (16:53 -0800)]
Add enum definition for subpel search precision
To improve readability.
Change-Id: Idc08b2068c7d8ba9dadc0d559a3b4d61c2a88c94
Yaowu Xu [Fri, 7 Dec 2018 22:53:43 +0000 (22:53 +0000)]
Merge "Optimize RDMult"
Angie Chiang [Fri, 7 Dec 2018 22:27:11 +0000 (22:27 +0000)]
Merge changes Id10f72b3,Icde1b01e,I391aa322
* changes:
Implement find_prev_nb_full_mvs
Implement get_full_mv()
Pass mv_num into vp9_nb_mvs_inconsistency()
Johann [Fri, 7 Dec 2018 17:28:02 +0000 (09:28 -0800)]
apply -Wextra to third_party/
googletest builds cleanly with -Wextra
Remove comments about webm:1069. The vp8 issue is tracked in webm:1246.
Change-Id: I8bbb01d34503cc9c342f5c3aa78e9476f72b94c2
sdeng [Wed, 7 Nov 2018 00:20:41 +0000 (16:20 -0800)]
Add high bit Hadamard 32x32 avx2 implementation
Speed test:
[ RUN ] C/HadamardHighbdTest.DISABLED_Speed/2
Hadamard32x32[ 10 runs]: 9 us
Hadamard32x32[ 10000 runs]: 8914 us
Hadamard32x32[
10000000 runs]:
8991776 us
[ RUN ] AVX2/HadamardHighbdTest.DISABLED_Speed/2
Hadamard32x32[ 10 runs]: 5 us
Hadamard32x32[ 10000 runs]: 4582 us
Hadamard32x32[
10000000 runs]:
4548203 us
Change-Id: Ied1b38b510bd033299f05869216d394e3b7f70f1
Sai Deng [Fri, 7 Dec 2018 17:00:03 +0000 (17:00 +0000)]
Merge "Add high bit Hadamard 16x16 avx2 implementation"
sdeng [Tue, 30 Oct 2018 21:08:07 +0000 (14:08 -0700)]
Add satd avx2 implementation
Speed Test:
C/SatdHighbdTest
blocksize: 16 time: 138 us
blocksize: 64 time: 315 us
blocksize: 256 time: 1120 us
blocksize: 1024 time: 3955 us
AVX2/SatdHighbdTest
blocksize: 16 time: 89 us
blocksize: 64 time: 189 us
blocksize: 256 time: 590 us
blocksize: 1024 time: 1912 us
Change-Id: I6357174462fccd589a475b13d8114b853cab5383
Jerome Jiang [Fri, 7 Dec 2018 02:51:24 +0000 (02:51 +0000)]
Merge "vp9 decoder: cleanup on exit if no available frame buffer."
Jerome Jiang [Wed, 31 Oct 2018 20:45:17 +0000 (13:45 -0700)]
vp9 decoder: cleanup on exit if no available frame buffer.
There was no setjmp on vpx_internal_error when there is no available
frame buffer, ready_for_new_data is not reset to 1.
BUG=webm:1571
Change-Id: I4f8efffb7d6fed3085b1f0229d0d1071a056b6c6
Angie Chiang [Thu, 6 Dec 2018 01:58:58 +0000 (17:58 -0800)]
Implement find_prev_nb_full_mvs
In single_motion_search, we use prev coded nb full mvs to compute
mv inconsistency.
lambda is set to block_area / 4.
This is a draft. Will to experiments to figure out the impact on
coding efficiency and visual quality.
Change-Id: Id10f72b3c7e6085bfbe1a6156b9fd6917843d001
Angie Chiang [Thu, 6 Dec 2018 02:17:31 +0000 (18:17 -0800)]
Implement get_full_mv()
Change-Id: Icde1b01ea7f64e2c43dcd039cc37fd306e43030f
Angie Chiang [Wed, 5 Dec 2018 23:19:00 +0000 (15:19 -0800)]
Pass mv_num into vp9_nb_mvs_inconsistency()
This allow av1_nb_mvs_inconsistency to cope with variant number of
motion vectors.
Change-Id: I391aa322d458cfefaf640e7b07d5ad5ce2d3375c
James Zern [Thu, 6 Dec 2018 22:34:14 +0000 (22:34 +0000)]
Merge "configure: test -std=c++11 before enabling unit tests"
Harish Mahendrakar [Thu, 6 Dec 2018 05:53:43 +0000 (05:53 +0000)]
Merge "Fix DoS in Error Streams"
Yaowu Xu [Thu, 29 Nov 2018 17:44:16 +0000 (09:44 -0800)]
Optimize RDMult
This commit introduces the optimized RDMult values for both key
and non-key frames. For key frames, the commit gets values back
from commit#
b13f6154df9c0834d74f7e3d41e41c4208f56d18 . For impact
on key frame only encodings, see commit message for that commit.
For inter frames, the values get optimzied by running encoding tests
in Q mode with the following range using 150 frames:
2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62
The impact of current set of RDMULT values:
PSNR SSIM PSNR-HVS
lowres: -0.325% 0.422% -0.228%
midres: -0.377% 0.158% -0.376%
hdres: -0.309% 0.522% -0.322%
Test baseline is on commit#
35617458
Overall, the values help PSNR based metrics, but hurt SSIM metric
slightly.
Change-Id: I7eba37a6524cb36b8498a1d104d2667781bc2089
sdeng [Wed, 7 Nov 2018 00:20:41 +0000 (16:20 -0800)]
Add high bit Hadamard 16x16 avx2 implementation
Speed test:
[ RUN ] C/HadamardHighbdTest.DISABLED_Speed/1
Hadamard16x16[ 10 runs]: 2 us
Hadamard16x16[ 10000 runs]: 1836 us
Hadamard16x16[
10000000 runs]:
1829451 us
[ RUN ] AVX2/HadamardHighbdTest.DISABLED_Speed/1
Hadamard16x16[ 10 runs]: 1 us
Hadamard16x16[ 10000 runs]: 1009 us
Hadamard16x16[
10000000 runs]: 984856 us
Change-Id: I89b9cdbe19350815576d66e627df87e5025ed0a4
Johann Koenig [Wed, 5 Dec 2018 23:40:57 +0000 (23:40 +0000)]
Merge "remove old visual studio support"
Jerome Jiang [Wed, 5 Dec 2018 18:48:15 +0000 (18:48 +0000)]
Merge "Refactor datarate svc test."
Johann Koenig [Wed, 5 Dec 2018 18:20:29 +0000 (18:20 +0000)]
Merge "quantize neon: fix hbd builds"
Johann [Tue, 4 Dec 2018 23:52:08 +0000 (15:52 -0800)]
remove old visual studio support
Change-Id: I86682ef1aac1991e1ef6965e7aa298f6619bee13
Sai Deng [Wed, 5 Dec 2018 17:55:58 +0000 (17:55 +0000)]
Merge "Fix overflow in calculating highbd SSIM"
Harish Mahendrakar [Wed, 5 Dec 2018 17:31:00 +0000 (17:31 +0000)]
Merge "Add Parse and Recon Split functions"
Deepa K G [Wed, 5 Dec 2018 11:22:49 +0000 (16:52 +0530)]
Rescale arf bit budget calculation
To compute the total budget for a depth layer, exclude the count of
frames in the current layer.
Change-Id: I9ffd1f63ea597de3ea95e0832b13f5b1f35cb086
Jerome Jiang [Wed, 5 Dec 2018 00:16:43 +0000 (16:16 -0800)]
Refactor datarate svc test.
Bring some repeated test set up into a function.
Change-Id: I6acc545a349dc16581a23baf848c91ec36a2e83f
sdeng [Wed, 5 Dec 2018 00:35:34 +0000 (16:35 -0800)]
Fix overflow in calculating highbd SSIM
Example internal stats
Before the fix:
Bitrate AVGPsnr GLBPsnr AVPsnrP GLPsnrP VPXSSIM VPSSIMP FASTSIM PSNRHVS WstPsnr WstSsim WstFast WstHVS AVPsnrY APsnrCb APsnrCr Block WstBlck Consist WstCons Time RcErr AbsErr
153.39 37.131 36.420 37.151 36.437 716.077 817.445 10.422 34.347 32.980 0.916 9.281 30.208 36.024 41.830 40.581 0.000 0.000 100.000 100.000 55006 2.26 2.26
No mismatch detected in recon buffers
After the fix:
Bitrate AVGPsnr GLBPsnr AVPsnrP GLPsnrP VPXSSIM VPSSIMP FASTSIM PSNRHVS WstPsnr WstSsim WstFast WstHVS AVPsnrY APsnrCb APsnrCr Block WstBlck Consist WstCons Time RcErr AbsErr
153.39 37.131 36.420 37.151 36.437 69.808 70.023 10.422 34.347 32.980 0.910 9.281 30.208 36.024 41.830 40.581 0.000 0.000 100.000 100.000 55067 2.26 2.26
No mismatch detected in recon buffers
Change-Id: I820abc498c1543548f193874046582b50afd0238
James Zern [Wed, 5 Dec 2018 00:38:08 +0000 (16:38 -0800)]
configure: test -std=c++11 before enabling unit tests
since:
77fa51003 Replace deprecated scoped_ptr with unique_ptr
the unit tests require a c++11 capable compiler; future versions of
googletest (1.9.x) will as well, so this change was inevitable if we
wanted to keep the snapshot up to date.
Change-Id: Id5c646bd10fae09e7b705b7d5fad1344f2216282
Angie Chiang [Tue, 4 Dec 2018 23:44:56 +0000 (15:44 -0800)]
Fix the parameter of vp9_full_pixel_diamond_new
Change-Id: I36b970953c960fde65d7b7705ccfa575c8741c43
Johann Koenig [Tue, 4 Dec 2018 23:40:41 +0000 (23:40 +0000)]
Merge "Reland "third_party/googletest: update to v1.8.1""
Jerome Jiang [Tue, 4 Dec 2018 23:34:20 +0000 (23:34 +0000)]
Merge "vp9: force refresh of long term ref when denoiser reset."
Jerome Jiang [Thu, 29 Nov 2018 23:53:41 +0000 (15:53 -0800)]
vp9: force refresh of long term ref when denoiser reset.
This will allocate extra frame buffer if long term temporal reference is
used and denoiser is enabled on non-key frame.
Add test.
Change-Id: I0e8d1fdb9a2d697a8eed7fe6206bcb362e69f1c8
Jingning Han [Tue, 4 Dec 2018 17:33:42 +0000 (17:33 +0000)]
Merge "Clean up rc_pick_q_and_bounds_two_pass()"
Deepa K G [Mon, 3 Dec 2018 09:33:18 +0000 (15:03 +0530)]
Fix intra_count_low calculation in first pass
In first pass, scaled_low_intra_thresh should not be
compared with motion_error, as scaled_low_intra_thresh
accounts for bit-depth, whereas motion_error does not.
In addition, mv_cost is excluded for comparison.
Change-Id: Id2fa02d364c086876c71ffebb2dd763eaa647e4a
Sai Deng [Tue, 4 Dec 2018 00:53:42 +0000 (00:53 +0000)]
Merge "Add high bit Hadamard 8x8 avx2 implementation"
Marco Paniconi [Mon, 3 Dec 2018 22:21:04 +0000 (14:21 -0800)]
vp9-svc: Fix to postencode drop for layers.
Postencode drop is only checked on base spatial
layers, and if set, whole superframe is dropped and
and next superframe is encoded at max-q.
Fix here is to make sure all layers are encoded at
max-q on a postencode dropped frame.
Change-Id: I2313d83ee29a382465bcec1085d8c73c37ce26d6
Marco Paniconi [Mon, 3 Dec 2018 22:09:09 +0000 (22:09 +0000)]
Merge "vp9: Rename post_encode drop function."
Marco Paniconi [Mon, 3 Dec 2018 18:49:41 +0000 (10:49 -0800)]
vp9: Overshoot detection for skipped base layer.
If scene/slide change is detected on current
superframe and max-q set because of high overshoot:
then if the lower/base spatial layer are skipped on
the current superframe, max-q is forced on the
next encoded base/lower spatial layers.
Change-Id: Id61efda86ee545395012e19476d19845e3932678
Marco Paniconi [Mon, 3 Dec 2018 18:32:06 +0000 (10:32 -0800)]
vp9: Rename post_encode drop function.
Feature works also for non-screen content mode,
so rename it.
Change-Id: I665362d50cf9a4017f114973586ad0eead066ddd
Johann [Sat, 1 Dec 2018 00:27:33 +0000 (16:27 -0800)]
quantize neon: fix hbd builds
BUG=webm:1448
Change-Id: I2140fb9b6ce92716d2d9509f3031244088a62127
sdeng [Tue, 6 Nov 2018 23:58:16 +0000 (15:58 -0800)]
Add high bit Hadamard 8x8 avx2 implementation
Speed tests:
[ RUN ] C/HadamardHighbdTest.DISABLED_Speed/0
Hadamard8x8[ 10 runs]: 0 us
Hadamard8x8[ 10000 runs]: 316 us
Hadamard8x8[
10000000 runs]: 311749 us
[ OK ] C/HadamardHighbdTest.DISABLED_Speed/0 (371 ms)
[ RUN ] AVX2/HadamardHighbdTest.DISABLED_Speed/0
Hadamard8x8[ 10 runs]: 0 us
Hadamard8x8[ 10000 runs]: 161 us
Hadamard8x8[
10000000 runs]: 156910 us
[ OK ] AVX2/HadamardHighbdTest.DISABLED_Speed/0 (160 ms)
Change-Id: I94f7324be20405ff55f8a02ad4651c4ab4c10202
Venkatarama NG. Avadhani [Wed, 24 Oct 2018 11:47:58 +0000 (17:17 +0530)]
Fix DoS in Error Streams
This fixes an issue where, in very rare error cases, one row of LPF
could be waiting infinitely for its previous row's LPF to complete.
With LPF optimization, the second row's LPF could be triggered before
the first row's LPF. In this case, the second row's LPF will wait for
LPF of n-sync number of SBs of the first row to finish. In error
streams, depending on when the error was detected, the LPF job of the
first row may then never be triggered. This puts the thread doing the
second row's LPF in an infinite wait.
The issue is reproduceable once in approximately 500 runs of the clip in
bug 1562.
BUG=webm:1562
Change-Id: I265d7df5ceeff0410334f5b9a4181f895bb54cab
Shubham Tandle [Fri, 30 Nov 2018 11:00:15 +0000 (16:30 +0530)]
Add Parse and Recon Split functions
Add functions that will do only parse or only recon. These are
duplicated and modified from decode_partition and decode_block.
Change-Id: I2201e235bf491e823ae63d27b2586bbb43b48929
Johann [Fri, 30 Nov 2018 23:42:57 +0000 (15:42 -0800)]
quantize 32x32: saturate dqcoeff on x86
This slows down low bitdepth builds but is necessary to obtain correct
values.
BUG=webm:1448
Change-Id: I4ca9145f576089bb8496fcfeedeb556dc8fe6574
Jingning Han [Fri, 30 Nov 2018 21:12:51 +0000 (21:12 +0000)]
Merge "Simplify constant q mode qp selection"
Jingning Han [Fri, 30 Nov 2018 19:01:14 +0000 (11:01 -0800)]
Clean up rc_pick_q_and_bounds_two_pass()
Remove unneeded VPX_Q condition check.
Change-Id: I46b09ae522caa47fa7ea4441b6a6ac2840315d1c
Sai Deng [Fri, 30 Nov 2018 19:09:28 +0000 (19:09 +0000)]
Merge "Use 16 bit ints in Hadamard highbd col8 first pass"
Johann Koenig [Fri, 30 Nov 2018 18:58:25 +0000 (18:58 +0000)]
Merge changes Ic80def57,I61a2f8bf
* changes:
quantize 32x32: fix dqcoeff
quantize: fix x86 hbd builds
Angie Chiang [Fri, 30 Nov 2018 18:25:50 +0000 (18:25 +0000)]
Merge changes I18680413,Iebe38092
* changes:
Consider mv inconsistency in single_motion_search
Change the interface of vp9_full_pixel_diamond_new
Jingning Han [Fri, 30 Nov 2018 18:23:33 +0000 (10:23 -0800)]
Simplify constant q mode qp selection
Decouple the constant q mode qp selection from vbr/cbr/cq modes.
Skip vp9_frame_type_qdelta() adjustment for non-ARF inter frames,
instead keep using the cq-level. It improves the compresson
performance:
avg PSNR overall PSNR SSIM
lowres -0.17% -0.20% -0.1%
midres -0.21% -0.24% -0.08%
hdres -0.15% -0.19% -0.04%
Change-Id: I52fd5f8edbd3fdcbeda31ee3a6d6eb016091a7e3
Jingning Han [Thu, 29 Nov 2018 19:37:00 +0000 (11:37 -0800)]
Factor key frame qp selection from two-pass qp and bound decision
Factor out this common code needed all rate control modes.
Change-Id: If17850fbebcdce7ff24afb211aa2e6054486b814
Angie Chiang [Fri, 30 Nov 2018 01:36:57 +0000 (17:36 -0800)]
Consider mv inconsistency in single_motion_search
This is still a work-in-process.
nb_full_mvs and lambda are set to zero for now, which means
mv inconsistency penalty is zero while doing the mv search.
Change-Id: I18680413d748fbdb9a33621f92f83e021036a3ab
Angie Chiang [Thu, 29 Nov 2018 22:54:17 +0000 (14:54 -0800)]
Change the interface of vp9_full_pixel_diamond_new
Avoid passing in tpl_stats because this function will be called in
motion search, where tpl_stats should be fixed at the point.
Let further_steps becomes internal variable in the function.
Change-Id: Iebe380925eb1891c19e0b78163dab8e6bfafccdb
sdeng [Thu, 29 Nov 2018 23:20:35 +0000 (15:20 -0800)]
Use 16 bit ints in Hadamard highbd col8 first pass
Change-Id: I2f04937d8a4e171d42b25ee6c6555ccad29eb192
Jingning Han [Thu, 29 Nov 2018 17:24:13 +0000 (17:24 +0000)]
Revert "Optimize RDMULT values for key frames"
This reverts commit
b13f6154df9c0834d74f7e3d41e41c4208f56d18 .
Temporarily revert this change due to interactions with rate control at very low target bit-rate.
Original change's description:
> Optimize RDMULT values for key frames
>
> Encoding of 5 frames of each sequence using key frames only, the new
> values help the metrics by:
>
> PSNR SSIM PSNR-HVS
> lowres: -0.870% -0.140% -0.892%
> midres: -0.899% -0.146% -1.052%
> hdres: -0.944% -0.115% -1.028%
>
> Tested q range:
> 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58 62
>
> Change-Id: I5b0dda366d589f52987c5bad11a1f95c4e6dc1a5
TBR=yaowu@google.com,paulwilkins@google.com,jingning@google.com,builds@webmproject.org
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: Ie1e4cf21308d69699a65a393a83884882682ea8e
Marco Paniconi [Thu, 29 Nov 2018 17:27:52 +0000 (09:27 -0800)]
vp9-svc: Add num_encoded_top layer counter
Useful for noise estimation when top layer
is skipped encoded.
Change-Id: I18cbe6119bac6c21514941b1e3b530a05a42df14
Marco Paniconi [Thu, 29 Nov 2018 07:57:44 +0000 (07:57 +0000)]
Merge "vp9: Fix condition for disabling noise estimation"
Marco Paniconi [Thu, 29 Nov 2018 06:08:08 +0000 (22:08 -0800)]
vp9: Fix condition for disabling noise estimation
Fix condition for turning off denoiser due to high
motion: use proper superframe counter and
frames_since_key counter so this condition won't
take effect on key (super)frame.
Change-Id: Ic502bf5ebfa32a921f611a78e8e963eb62b5bc79
Jerome Jiang [Thu, 29 Nov 2018 01:26:18 +0000 (17:26 -0800)]
vp9 denoiser: force copy block when last not a reference.
Last reference doesn't always exist when SVC layers changed dynamically.
When last is not a reference for current layer, copy block directly on
denoiser.
Change-Id: I9d98c4d6fdcfa25ba707db3333712761b5cf9ab8
Jingning Han [Wed, 28 Nov 2018 18:35:35 +0000 (18:35 +0000)]
Merge "Remove ineffective condition from rc_pick_q_and_bounds"
Johann [Wed, 21 Nov 2018 18:30:45 +0000 (13:30 -0500)]
quantize 32x32: fix dqcoeff
Calculate the high bits of dqcoeff and store them appropriately in high
bit depth builds.
Low bit depth builds still do not pass. C truncates the results after
division. X86 only supports packing with saturation at this step.
BUG=webm:1448
Change-Id: Ic80def575136c7ca37edf18d21e26925b475da98
Johann [Wed, 21 Nov 2018 18:30:45 +0000 (13:30 -0500)]
quantize: fix x86 hbd builds
Calculate the high bits of dqcoeff in high bit depth builds and store
them appropriately.
BUG=webm:1448
Change-Id: I61a2f8bfcf2e30765f10a94073c4d58321d2fa24
Marco Paniconi [Wed, 28 Nov 2018 02:58:59 +0000 (02:58 +0000)]
Merge "vp9-svc: Add check/reset for long term reference."
Marco Paniconi [Wed, 28 Nov 2018 00:16:04 +0000 (16:16 -0800)]
vp9-svc: Add check/reset for long term reference.
Add check and reset (turn off) usage of long term
reference if some conditons (layer id of reference vs
current frame) are not met.
Change-Id: Ie3a84e3618f4fc4d5f8da4e67316cfbefb8bae78
Jerome Jiang [Tue, 27 Nov 2018 20:29:47 +0000 (12:29 -0800)]
vp9 svc: copy block if ref buffer in denoiser is NULL.
BUG=b/
119097707
Change-Id: I6569306e897da46a44f9d8f2fb28a2a355dd4c2c
Johann [Mon, 17 Sep 2018 20:57:30 +0000 (13:57 -0700)]
Reland "third_party/googletest: update to v1.8.1"
This is a reland of
7d777ce6136d7325a102524b6d80c19bd420e4b6
Previous attempt was reverted due to build issues with older
versions of Visual Studio. We no longer support VS <= 13.
Original change's description:
> third_party/googletest: update to v1.8.1
>
> BUG=webm:1559
>
> Change-Id: I7a0b16c7bf3f97db2d8650a190b93aae7e12a948
Bug: webm:1559
Change-Id: I9cb39988286cc56125879222ef0bd952d61b7c1d