]>
granicus.if.org Git - libvpx/log
Johann [Fri, 23 Jun 2017 00:12:27 +0000 (17:12 -0700)]
partial fdct test
Test the _1 variant of the fdct, which simply sums the block and applies
a modifying shift based on the block size.
BUG=webm:1424
Change-Id: Ic80d6008abba0c596b575fa0484d5b5855321468
Johann Koenig [Fri, 23 Jun 2017 01:57:59 +0000 (01:57 +0000)]
Merge "fdct32x32 neon implementation"
Marco Paniconi [Thu, 22 Jun 2017 22:18:47 +0000 (22:18 +0000)]
Merge "vp9: Add high source sad to content state."
Linfeng Zhang [Thu, 22 Jun 2017 21:42:23 +0000 (21:42 +0000)]
Merge changes I783c5f4f,I365f8e53,I5dac0e98
* changes:
Clean vpx_idct16x16_256_add_sse2()
Update vpx_idct{8x8,16x16,32x32}_1_add_sse2()
Clean 32x32 full idct sse2 and ssse3 code
Paul Wilkins [Thu, 22 Jun 2017 16:30:39 +0000 (16:30 +0000)]
Merge "Fix int overflow in rate control for high bit rates."
Johann [Fri, 16 Jun 2017 20:14:01 +0000 (13:14 -0700)]
fdct32x32 neon implementation
Almost 3x faster in constrained loop testing. Over 10x faster in HBD
builds.
BUG=webm:1424
Change-Id: I2b7f8453e1d4ada63cde729d8115d684c4a71ff9
paulwilkins [Wed, 21 Jun 2017 16:26:20 +0000 (17:26 +0100)]
Fix int overflow in rate control for high bit rates.
Fix misplaced cast that caused an overflow and incorrect rate adaptation
behavior for high data rates. This in particular will have affected 4k encodes
but could also have come into play for some higher rate 1080p cases.
In our standard test sets the quality impact is small though several high rate
clips show improved rate accuracy. This can also impact the number of recode
loop hits and on one problem 4k clip the encode time for speeds 0 and 1 was
reduced by >25%
Change-Id: I108da7ca42f3bc95c5825dd33c9d84583227dac1
Marco [Thu, 22 Jun 2017 03:39:59 +0000 (20:39 -0700)]
vp9: Add high source sad to content state.
Use it to limit NEWMV early exit in nonrd pickmode
Small change in RTC metrics, has some improvement
for high motion clips.
Change-Id: I1d89fd955e1b3486d5fb07f4472eeeecd553f67f
Marco Paniconi [Thu, 22 Jun 2017 03:27:46 +0000 (03:27 +0000)]
Merge "vp9: Adjustments for aq-mode and pickmode for speed >= 8."
James Zern [Thu, 22 Jun 2017 02:01:10 +0000 (02:01 +0000)]
Merge "vp8_dx_iface: clear -Wclobbered warnings"
Marco [Thu, 22 Jun 2017 00:30:53 +0000 (17:30 -0700)]
vp9: Put skin detection usage around cpi flag.
Skin detection usage in choose_partitioning should be
around the cpi->use_skin_detection.
Change-Id: I6986179af9ce94c60c0974d66c311fc07cc04cfe
Marco [Wed, 21 Jun 2017 22:24:56 +0000 (15:24 -0700)]
vp9: Adjustments for aq-mode and pickmode for speed >= 8.
Adjust the threshold for turning off cyclic refresh for high motion,
and avoid testing golden in nonrd pickmode for speed >= 8 if
golden refresh was long ago.
No change/neutral on RTC metrics.
Change-Id: I40959b8d9637f3553e7458bbabd8c6024c2c09c0
James Zern [Wed, 21 Jun 2017 01:34:13 +0000 (18:34 -0700)]
vp8_dx_iface: clear -Wclobbered warnings
with gcc 6.x
Change-Id: Ib2070421603a6777892d4ea01f4b0921696f38b3
Johann Koenig [Wed, 21 Jun 2017 21:16:23 +0000 (21:16 +0000)]
Merge "dct tests: align InvAccuracyCheck buffers"
Linfeng Zhang [Tue, 20 Jun 2017 01:10:38 +0000 (18:10 -0700)]
Clean vpx_idct16x16_256_add_sse2()
Remove macro IDCT16 which is redundant with idct16_8col().
Change-Id: I783c5f4fda038a22d5ee5c2b22e8c2cdfb38432c
Linfeng Zhang [Thu, 15 Jun 2017 23:48:40 +0000 (16:48 -0700)]
Update vpx_idct{8x8,16x16,32x32}_1_add_sse2()
Change-Id: I365f8e53d9ccd028cef0f561d4de9e5916278609
Linfeng Zhang [Thu, 15 Jun 2017 23:06:53 +0000 (16:06 -0700)]
Clean 32x32 full idct sse2 and ssse3 code
vpx_idct32x32_1024_add_ssse3() is actually a sse2 function and faster
than vpx_idct32x32_1024_add_sse2(). Replace the slow one. All are
code relocations, no new code.
Change-Id: I5dac0e98cc411a4ce05660406921118986638d19
Hui Su [Wed, 21 Jun 2017 20:38:44 +0000 (20:38 +0000)]
Merge "VP9 level targeting: properly handle max_gf_interval"
Johann [Wed, 21 Jun 2017 18:46:58 +0000 (11:46 -0700)]
dct tests: align InvAccuracyCheck buffers
'in' is used for the reference fdct. 'coeff' is input to the idct being
tested and 'dst[16]' is output
Fixes a segfault on unaligned memory access on x86.
Change-Id: I3691b1380ed49986897dd89a63ce63a80a0e0962
James Zern [Wed, 21 Jun 2017 02:42:29 +0000 (19:42 -0700)]
fix build, rm ref to vpx_idct8x8_64_add_ssse3
this was deleted in:
98967645a Remove vpx_idct8x8_64_add_ssse3()
but this was merged in:
9e03eedf6 Merge changes Ib26dd515,Ie60dabc3
after:
a92991133 Merge "dct tests: run all possible sizes in one test"
which added a new reference
Change-Id: I8da4a6c80d27b237a378ff15eead1daab89e7e25
Linfeng Zhang [Wed, 21 Jun 2017 00:38:24 +0000 (00:38 +0000)]
Merge changes Ib26dd515,Ie60dabc3
* changes:
Clean 8x8 idct x86 optimization
Remove vpx_idct8x8_64_add_ssse3()
hui su [Tue, 20 Jun 2017 23:26:00 +0000 (16:26 -0700)]
VP9 level targeting: properly handle max_gf_interval
Don't overide max_gf_interval if it's not specified. It will
be assigned with a default value in vp9_rc_set_gf_interval_range().
BUG=b/
62803416
Change-Id: Ide46ce00279ed076865fc54ce98c55a994f0c798
Marco [Tue, 20 Jun 2017 19:20:52 +0000 (12:20 -0700)]
vp9: Adjust key-frame pars in vpx_temporal_svc_encoder.
Sample encoder change: reduce max-intra-rate to 1000 and
buf-initial to 600. Paramaters affect target size of key frame.
Change-Id: I2be6bc2927f5fa74e19e1efa3fb574d23a503300
Marco [Tue, 20 Jun 2017 16:05:56 +0000 (09:05 -0700)]
vp9: Adjust key-frame pars in vpx_temporal_svc_encoder.
Sample encoder change: reduce max-intra-rate to 1500 and
buf-initial to 700. Paramaters affect target size of key frame.
Change-Id: I01e238378b63eeef28dfc2178baadffcd3cc7561
Johann Koenig [Tue, 20 Jun 2017 15:04:24 +0000 (15:04 +0000)]
Merge "dct tests: run all possible sizes in one test"
Marco Paniconi [Tue, 20 Jun 2017 03:08:32 +0000 (03:08 +0000)]
Merge "vp9: SVC: Rework the usage of base_mv for SVC."
Marco [Tue, 20 Jun 2017 01:25:53 +0000 (18:25 -0700)]
vp9: Adjust key-frame pars in vpx_temporal_svc_encoder.
Adjust some parameters in sample encoder: vpx_temporal_svc_encoder.
Parameters adjusted to set lower QP for initial key frame,
and allow for larger target size on subsequent key frames.
Change-Id: I092ad968e5b51b9f495dadb6ee96e810663c910e
Marco Paniconi [Mon, 19 Jun 2017 23:45:34 +0000 (23:45 +0000)]
Merge "vp9: Speed >= 8: Adjust resolution threshold for subpel."
Johann [Tue, 9 May 2017 17:08:24 +0000 (10:08 -0700)]
dct tests: run all possible sizes in one test
Modify fdct4x4_test.cc to support all size combinations. This does not
add any new tests and in fact fails a few. There were minimal changes
made to the tests so it's not entirely surprising that some of the
larger 12 bit transforms are failing since it was initially only used
for 4x4.
In follow up patches the tests in fdct8x8_test.cc, dct16x16_test.cc and
dct32x32_test.cc will be evaluated and moved to dct_test.cc.
BUG=webm:1424
Change-Id: I72a23430f457d7fae8c91e706adc0e77c25abc8f
James Zern [Mon, 19 Jun 2017 22:15:06 +0000 (22:15 +0000)]
Merge "libs.mk: retry partial testdata download"
James Zern [Fri, 16 Jun 2017 06:33:11 +0000 (23:33 -0700)]
libs.mk: retry partial testdata download
attempt retry on transient failures uncaught by --retry
Change-Id: I7cd8846ff88daf0f521af9ee182e30bfd79f51f3
Marco [Mon, 19 Jun 2017 20:56:04 +0000 (13:56 -0700)]
vp9: Speed >= 8: Adjust resolution threshold for subpel.
Get some quality gain on RTC metrics (~7%), with
~5-8% speed slowdown.
Change-Id: I0d02942a77074424ee0326b6e110ddff09f2df5e
Jerome Jiang [Mon, 19 Jun 2017 16:42:10 +0000 (16:42 +0000)]
Merge "Enable 8x8 skin detection for vp8."
Marco [Thu, 15 Jun 2017 23:23:36 +0000 (16:23 -0700)]
vp9: SVC: Rework the usage of base_mv for SVC.
Set the base_mv_aggressive for temporal enhancement layers (TL > 0).
Under the aggressive mode, skip the NEWMV depending on the
SSE of the base_mv. Also reduce the subpel motion to 1/2 under
aggressive mode if base_mv is good.
Speedup ~3% with small/negligible loss in quality on RTC.
Affects speed >= 6.
Change-Id: I89341b279cad6da2a04b76d5e726016191dacdb8
James Zern [Fri, 16 Jun 2017 20:07:51 +0000 (20:07 +0000)]
Merge "configure: test for -Wparentheses-equality"
Jerome Jiang [Wed, 14 Jun 2017 00:22:26 +0000 (17:22 -0700)]
Enable 8x8 skin detection for vp8.
If 2 or more 8x8 blocks are identified as skin, the macroblock will be
labeled as skin.
Change-Id: I596542c81a2df9e96270cab39d920bbfeb02bc6e
James Zern [Thu, 15 Jun 2017 23:05:20 +0000 (16:05 -0700)]
configure: test for -Wparentheses-equality
Change-Id: I36de79c58461907deaea70d6131da9119bc0bc69
Linfeng Zhang [Thu, 15 Jun 2017 18:29:07 +0000 (11:29 -0700)]
Clean 8x8 idct x86 optimization
Create load_buffer_8x8() and write_buffer_8x8().
Change-Id: Ib26dd515d734a5402971c91de336ab481b213fdf
Linfeng Zhang [Thu, 15 Jun 2017 21:09:33 +0000 (14:09 -0700)]
Remove vpx_idct8x8_64_add_ssse3()
It's almost identical with vpx_idct8x8_64_add_sse2(), except little
difference in instructions order.
Change-Id: Ie60dabc35eaa6ebae7c755e6cff00a710aad284f
Urvang Joshi [Thu, 8 Jun 2017 21:51:01 +0000 (14:51 -0700)]
VP9: Add greedy version of av1_optimize_b().
This was ported from the greedy version in AV1, written by Dake He
(dkhe@google.com).
See:
https://aomedia.googlesource.com/aom/+/master/av1/encoder/encodemb.c#137
Greedy version is disabled by default, but can be picked by setting
USE_GREEDY_OPTIMIZE_B to 1.
To be enabled by default later.
This is both faster and better in terms of compression.
Compression Improvement:
------------------------
lowres: -0.119
midres: -0.064
hdres: -0.405
Speed Improvement:
------------------
(Based on encode time of 3 videos of different difficulties at
3 different target bitrates)
With --cpu-used=0: 0.38% to 5.55% faster
With --cpu-used=1: 0.24% to 2.79% faster
With --cpu-used=2: 0.29% to 1.46% faster
Change-Id: Ia7a23b3b244ad8eb253ac9e43cd03c5e021d2635
Linfeng Zhang [Thu, 15 Jun 2017 17:57:50 +0000 (17:57 +0000)]
Merge changes Ibf9d120b,I341399ec,Iaa5dd63b,Id59865fd
* changes:
Update high bitdepth load_input_data() in x86
Clean array_transpose_{4X8,16x16,16x16_2) in x86
Remove array_transpose_8x8() in x86
Convert 8x8 idct x86 macros to inline functions
Marco Paniconi [Wed, 14 Jun 2017 15:59:54 +0000 (15:59 +0000)]
Merge "vp8: Adjust the pred_err threhsold for drop on overshoot."
Linfeng Zhang [Tue, 13 Jun 2017 23:53:53 +0000 (16:53 -0700)]
Update high bitdepth load_input_data() in x86
BUG=webm:1412
Change-Id: Ibf9d120b80c7d3a7637e79e123cf2f0aae6dd78c
Linfeng Zhang [Mon, 12 Jun 2017 23:23:53 +0000 (16:23 -0700)]
Clean array_transpose_{4X8,16x16,16x16_2) in x86
Change-Id: I341399ecbde37065375ea7e63511a26bfc285ea0
Linfeng Zhang [Mon, 12 Jun 2017 22:45:50 +0000 (15:45 -0700)]
Remove array_transpose_8x8() in x86
Duplicate of transpose_16bit_8x8()
Change-Id: Iaa5dd63b5cccb044974a65af22c90e13418e311f
Linfeng Zhang [Mon, 12 Jun 2017 21:03:37 +0000 (14:03 -0700)]
Convert 8x8 idct x86 macros to inline functions
Change-Id: Id59865fd6c453a24121ce7160048d67875fc67ce
James Zern [Tue, 13 Jun 2017 01:08:20 +0000 (18:08 -0700)]
vp8_skin_detection: add 'vp8_' prefix to public fns
BUG=webm:1438
Change-Id: I5feb31c254d02e116e624cfe702e73ba5a1f7aca
James Zern [Tue, 13 Jun 2017 00:54:21 +0000 (17:54 -0700)]
rename vp8/common/skin_detection.[hc] -> vp8_*
some build systems have trouble with duplicate basenames.
vpx_dsp/skin_detection.[hc] were added in:
658e85425 Merge skin detection code in vp8/9.
BUG=webm:1438
Change-Id: Ieaa70b40bda409ec23e6d179b47a930ac6243b05
Marco [Mon, 12 Jun 2017 16:51:47 +0000 (09:51 -0700)]
vp8: Adjust the pred_err threhsold for drop on overshoot.
Change-Id: Ica2a09ac87160936b6f7bd01f167f464ea3ac41c
Hui Su [Mon, 12 Jun 2017 16:38:02 +0000 (16:38 +0000)]
Merge "vp9 level targeting: more strict constraint on min_gf_interval"
Jerome Jiang [Sat, 10 Jun 2017 04:50:19 +0000 (04:50 +0000)]
Merge "Remove duplication on vp8/9_write_yuv_frame."
Marco [Sat, 10 Jun 2017 00:48:03 +0000 (17:48 -0700)]
vp9: SVC: Use prune_evenemore only for non_reference.
Set subpel prune_evenmore only for non_reference frames,
instead of all TL > 0 frames. Gain some quality back at
cost of small speed loss (~1-2%).
Change only effects SVC encoding at speed >= 7.
Change-Id: I5b9f51e51dccfd7050521a66996176b0415ca3f9
Jerome Jiang [Thu, 8 Jun 2017 23:07:02 +0000 (16:07 -0700)]
Remove duplication on vp8/9_write_yuv_frame.
Change-Id: Ib3546032a27c715bf509c0e24d26a189bc829da8
Johann Koenig [Fri, 9 Jun 2017 20:42:39 +0000 (20:42 +0000)]
Merge "idct_test: don't use std::nothrow anymore"
Johann Koenig [Fri, 9 Jun 2017 20:42:21 +0000 (20:42 +0000)]
Merge "buffer.h: allow declaring an alignment"
Johann Koenig [Fri, 9 Jun 2017 20:41:56 +0000 (20:41 +0000)]
Merge "Remove some dead code. Coverity CID
1310058 "
Johann [Thu, 8 Jun 2017 17:09:23 +0000 (10:09 -0700)]
idct_test: don't use std::nothrow anymore
But still check for NULL before calling Init()
Change-Id: I2bf2887e1064c9103d29c542d20365c0aea75d76
Johann [Tue, 6 Jun 2017 22:15:47 +0000 (15:15 -0700)]
buffer.h: allow declaring an alignment
x86 simd register operations generally prefer and may require 16 byte
alignment.
Change-Id: I73ce577a90dc66af60743c5727c36f23200950ba
Sylvestre Ledru [Fri, 9 Jun 2017 12:08:18 +0000 (14:08 +0200)]
Remove some dead code. Coverity CID
1310058
Change-Id: I1186cf1dd8cde42f5970928f43edfc852298289d
James Zern [Thu, 8 Jun 2017 23:17:49 +0000 (23:17 +0000)]
Merge "vp8_decode_frame: fix oob read on truncated key frame"
James Zern [Thu, 8 Jun 2017 03:46:13 +0000 (20:46 -0700)]
vp8_decode_frame: fix oob read on truncated key frame
the check for error correction being disabled was overriding the data
length checks. this avoids returning incorrect information (width /
height) for the decoded frame which could result in inconsistent sizes
returned in to an application causing it to read beyond the bounds of
the frame allocation.
BUG=webm:1443
BUG=b/
62458770
Change-Id: I063459674e01b57c0990cb29372e0eb9a1fbf342
Johann [Thu, 8 Jun 2017 17:20:21 +0000 (10:20 -0700)]
Revert "buffer.h: use size_t"
This reverts commit
f08581c1d010ea95b8cfae686b5c0a64b32519f9 .
type conversion warnings abound.
Change-Id: I41d4c0e7a388e1008bdbc55fefda4bbca3f89f00
Jerome Jiang [Thu, 8 Jun 2017 16:35:59 +0000 (16:35 +0000)]
Merge "Merge skin detection code in vp8/9."
Johann Koenig [Thu, 8 Jun 2017 15:19:35 +0000 (15:19 +0000)]
Merge "fdct16x16 neon optimization"
Jerome Jiang [Mon, 5 Jun 2017 18:09:05 +0000 (11:09 -0700)]
Merge skin detection code in vp8/9.
BUG=webm:1438
Change-Id: Ie3dc034c7dbb498a0b088a767b1936ddeed4df14
hui su [Wed, 7 Jun 2017 22:50:36 +0000 (15:50 -0700)]
vp9 level targeting: more strict constraint on min_gf_interval
min_gf_interval should be no less than min_altref_distance + 1,
as the encoder may produce bitstream with alt-ref distance being
min_gf_interval - 1.
BUG=b/
38450599
Change-Id: Ifb733daa643ebc668d1b23e1ce92db94b66dabe8
Johann [Thu, 25 May 2017 17:02:34 +0000 (10:02 -0700)]
fdct16x16 neon optimization
Roughly 2x speedup. Since the only change for HBD is to store(), the
improvement appears to hold there as well.
BUG=webm:1424
Change-Id: I15b813d50deb2e47b49a6b0705945de748e83c19
Marco Paniconi [Wed, 7 Jun 2017 21:12:13 +0000 (21:12 +0000)]
Merge "vp9: SVC: Enable simple_block_yrd for temporal layers."
Johann Koenig [Wed, 7 Jun 2017 19:20:15 +0000 (19:20 +0000)]
Merge changes Iade45f69,I18d90658,Ieca3f1ef
* changes:
buffer.h: add num_elements_
buffer.h: zero-init all values
buffer.h: use size_t
Marco [Wed, 7 Jun 2017 18:33:40 +0000 (11:33 -0700)]
vp9: SVC: Enable simple_block_yrd for temporal layers.
Enable simple_block_yrd for temporal enhancement layers (TL > 0).
And remove block size condiiton for SVC mode.
Only affects speed >= 7 SVC.
Speedup ~3-4%.
avgPSNR regression on RTC for (3 spatial, 3 temporal) layers: ~1%.
Change-Id: Iff4fc191623b71c69cd373e7c0823385e7ac67ed
Johann [Tue, 6 Jun 2017 20:17:07 +0000 (13:17 -0700)]
buffer.h: add num_elements_
raw_size_ was being incorrectly computed and used
Change-Id: Iade45f69964c567ffb258880f26006a96ae5a30d
Johann [Wed, 7 Jun 2017 18:27:26 +0000 (11:27 -0700)]
buffer.h: zero-init all values
Change-Id: I18d90658bcd4365d49adcadd6954090b3b399aa8
Johann [Tue, 6 Jun 2017 19:58:15 +0000 (12:58 -0700)]
buffer.h: use size_t
Change-Id: Ieca3f1ef23cd1d7b844ea3ecb054007ed280b04f
Marco [Wed, 7 Jun 2017 17:27:19 +0000 (10:27 -0700)]
vp9: SVC: Enable row-mt in sample encoder.
Change-Id: I4b51043cb3f5955efe947fe4685aed4a21adb8bd
James Zern [Tue, 6 Jun 2017 23:52:39 +0000 (23:52 +0000)]
Merge "ppc: Add vpx_sadnxmx4d_vsx for n,m = {8, 16, 32 ,64}"
Marco Paniconi [Tue, 6 Jun 2017 23:07:44 +0000 (23:07 +0000)]
Merge "vp9: SVC: Adjust some speed settings for SVC speed >= 7."
Marco [Tue, 6 Jun 2017 22:28:06 +0000 (15:28 -0700)]
vp9: SVC: Adjust some speed settings for SVC speed >= 7.
Keep the 1/4subpel for all frames, use SUBPEL_TREE_PRUNED_EVENMORE
for all temporal enhancement layer frames.
Change-Id: Ibc681acbb6fc75b7b3c57fc483fcb11d591dfc9a
Johann [Tue, 6 Jun 2017 19:48:01 +0000 (12:48 -0700)]
buffer.h: split out init
Change-Id: Idfbd2e01714ca9d00525c5aeba78678b43fb0287
Johann [Tue, 6 Jun 2017 19:05:14 +0000 (12:05 -0700)]
buffer.h: Use T for values
Change-Id: I2da4110e843b6e361028b921c24b6ca2ea9077d9
Jerome Jiang [Tue, 6 Jun 2017 17:13:34 +0000 (10:13 -0700)]
Initialize cost_list all to INT_MAX.
It is initialized to be { INT_MAX, 0, ... } in
ffe0f9b .
No effect on encoders.
Make it consistent with other initializations.
BUG=webm:1440
Change-Id: Ie2a180d93626b55914c8c4255e466a1986d2b922
James Zern [Tue, 6 Jun 2017 05:52:58 +0000 (22:52 -0700)]
vp9_mcomp,get_cost_surf_min: quiet conversion warning
visual studio will warn if a 32-bit shift is implicitly converted to 64.
in this case integer storage is enough for the result.
since:
f3a9ae5ba Fix ubsan failure in vp9_mcomp.c.
Change-Id: I7e0e199ef8d3c64e07b780c8905da8c53c1d09fc
Jerome Jiang [Tue, 6 Jun 2017 03:47:30 +0000 (03:47 +0000)]
Merge "Fix valgrind failure on uninitialized variables."
James Zern [Tue, 6 Jun 2017 02:19:41 +0000 (02:19 +0000)]
Merge "ppc: Add vpx_sad64/32/16x64/32/16_avg_vsx"
Jerome Jiang [Mon, 5 Jun 2017 18:41:02 +0000 (11:41 -0700)]
Fix valgrind failure on uninitialized variables.
BUG=webm:1440
Change-Id: I7074e42bdfa8dd25f11bbb3f2ab1b41d6f4c12e4
Jerome Jiang [Fri, 2 Jun 2017 20:50:08 +0000 (13:50 -0700)]
Fix ubsan failure in vp9_mcomp.c.
Change-Id: Iff1dea1fe9d4ea1d3fc95ea736ddf12f30e6f48d
Marco [Fri, 26 May 2017 18:36:45 +0000 (11:36 -0700)]
vp9: SVC: Force subpel search off under certain conditions.
For SVC 1 pass non-rd mode:
Force subpel seach off for SVC for non-reference frames
under motion threshold.
Add flag to svc context to indicate if the frame is not used
as a reference.
Little/no quaity loss, ~2% speedup.
Change-Id: Ic433c44b514d19d08b28f80ff05231dc943b28e9
Marco Paniconi [Thu, 1 Jun 2017 23:56:53 +0000 (23:56 +0000)]
Merge "vp9: Speed >8: Set subpel_search_method for low motion."
Marco [Thu, 1 Jun 2017 23:07:15 +0000 (16:07 -0700)]
vp9: Speed >8: Set subpel_search_method for low motion.
Speed >=8: for resolutions above CIF, and for low motion content,
set subpel_search_method to SUBPEL_TREE_PRUNED_EVENMORE.
Small speed gain (~2%) on vga clips,
RTC metrics up by ~2-3% on average.
Change-Id: Ie26ba0264589652f92dfe74308740debf94cf0cc
Jerome Jiang [Thu, 1 Jun 2017 20:46:46 +0000 (13:46 -0700)]
vp8 skin detection: Fix visual studio build failure.
Change-Id: I510b755550ebbfa2aaf9b974920d7f1c6454a845
Jerome Jiang [Wed, 31 May 2017 22:56:29 +0000 (15:56 -0700)]
Fix corruption in skin map debugging output yuv.
For both vp8 and vp9.
BUG=webm:1437
Change-Id: Ifd06f68a876ade91cc2cc27c574c4641b77cce28
Jerome Jiang [Wed, 31 May 2017 21:57:10 +0000 (14:57 -0700)]
vp8: Clean up skin detection.
Use only the average of center 2x2 pixels in vp8.
Change-Id: I2b23ff19a90827226273e0fca49e90c734eda59b
Johann Koenig [Wed, 31 May 2017 18:17:27 +0000 (18:17 +0000)]
Merge "comp_avg_pred neon: used by sub pixel avg variance"
Jerome Jiang [Wed, 31 May 2017 16:37:07 +0000 (16:37 +0000)]
Merge "Write skin map of vp8 skin detection for debug."
Linfeng Zhang [Wed, 31 May 2017 15:56:19 +0000 (15:56 +0000)]
Merge "Update vpx_highbd_idct4x4_16_add_sse2()"
Johann [Wed, 3 May 2017 21:58:52 +0000 (14:58 -0700)]
comp_avg_pred neon: used by sub pixel avg variance
BUG=webm:1423
Change-Id: I33de537f238f58f89b7a6c1c2d6e8110de4b8804
Jerome Jiang [Fri, 26 May 2017 23:20:49 +0000 (16:20 -0700)]
Write skin map of vp8 skin detection for debug.
Change-Id: Ica1b4e918aa759cd0ce65920f9d88452bbf9e3b4
Linfeng Zhang [Mon, 22 May 2017 23:04:05 +0000 (16:04 -0700)]
Update vpx_highbd_idct4x4_16_add_sse2()
BUG=webm:1412
Change-Id: I26e4b34ae9bc1ae80c24f56d740d737a95f1ab84
Johann Koenig [Tue, 30 May 2017 16:21:05 +0000 (16:21 +0000)]
Merge "comp_avg_pred: alignment"
Johann Koenig [Tue, 30 May 2017 15:58:16 +0000 (15:58 +0000)]
Merge "remove DECLARE_ALIGNED from neon code"
Johann [Tue, 30 May 2017 14:46:43 +0000 (07:46 -0700)]
comp_avg_pred: alignment
x86 requires 16 byte alignment for some vector loads/stores.
arm does not have the same requirement.
The asserts are still in avg_pred_sse2.c. This just removes them from
the common code.
Change-Id: Ic5175c607a94d2abf0b80d431c4e30c8a6f731b6