]>
granicus.if.org Git - libvpx/log
Johann Koenig [Thu, 27 Aug 2015 20:56:32 +0000 (20:56 +0000)]
Merge "Add sse2 versions of halfpix variance"
Johann [Thu, 27 Aug 2015 16:55:42 +0000 (09:55 -0700)]
Add sse2 versions of halfpix variance
These were lost in the great sub pixel variance move of
6a82f0d7fb9ee908c389e8d55444bbaed3d54e9c
Not having these functions caused a ~10% performance regression in
some realtime vp8 encodes.
Change-Id: I50658483d9198391806b27899f2c0d309233c4b5
James Zern [Tue, 18 Aug 2015 01:19:22 +0000 (18:19 -0700)]
vpx_dsp_common: add VPX prefix to MIN/MAX
prevents redeclaration warnings;
vp8 has its own define which will be resolved in a future commit
Change-Id: Ic941fef3dd4262fcdce48b73075fe6b375f11c9c
James Zern [Thu, 27 Aug 2015 03:02:57 +0000 (03:02 +0000)]
Merge changes I56fad85c,I7401a70a
* changes:
vp8e_encode: check validate_config return
vp9_update_layer_...config: delete dead assignment
Marco Paniconi [Wed, 26 Aug 2015 23:37:00 +0000 (23:37 +0000)]
Merge "vp9_spatial_svc_encoder: Enable aq-mode for real-time mode."
Marco [Wed, 26 Aug 2015 22:18:13 +0000 (15:18 -0700)]
vp9_spatial_svc_encoder: Enable aq-mode for real-time mode.
For real-time mode (speeds >=5) enable aq-mode=3.
Change-Id: Ib8b4ef7609bc30ac935742c8d27e8cd89933c6af
Hui Su [Wed, 26 Aug 2015 22:19:01 +0000 (22:19 +0000)]
Merge "Remove vp10_default_scan_orders[] that is unused"
Yunqing Wang [Wed, 26 Aug 2015 20:36:54 +0000 (13:36 -0700)]
vp10 cleanup: remove nonrd and real-time code
Code cleanup.
Change-Id: I668bd5a4e5fbe96969d51d9ee777fc5f1d8abfe6
Marco Paniconi [Wed, 26 Aug 2015 19:47:15 +0000 (19:47 +0000)]
Merge "vp9_spatial_svc_encoder: Add encoding time log."
Marco [Wed, 26 Aug 2015 19:18:46 +0000 (12:18 -0700)]
vp9_spatial_svc_encoder: Add encoding time log.
Change-Id: Iab4e4a13a65c9dc5314e5107b126c30667c7b448
Johann Koenig [Wed, 26 Aug 2015 18:36:10 +0000 (18:36 +0000)]
Merge "Clean up unused function warnings in vp8 decoder"
hui su [Wed, 26 Aug 2015 17:37:29 +0000 (10:37 -0700)]
Remove vp10_default_scan_orders[] that is unused
Change-Id: Ibc3b5cb3799f080808cf22ccb01d733e74ab4461
Johann [Tue, 25 Aug 2015 23:38:59 +0000 (16:38 -0700)]
Remove vp8_priv_sz
This is a leftover of the XMA code which was removed a long time ago.
Found while looking for unused functions.
Change-Id: I07a3d542ae55440af59380dcdcf9a6c11cdfcb75
Johann [Tue, 25 Aug 2015 23:06:36 +0000 (16:06 -0700)]
Clean up unused function warnings in vp8 decoder
Mark functions in dboolhuff.h and treereader.h with INLINE.
Change-Id: Ifc1008de0713c8d43f4fec83a8c2b836622be365
James Zern [Tue, 25 Aug 2015 03:07:48 +0000 (20:07 -0700)]
vp8e_encode: check validate_config return
Change-Id: I56fad85cae4537f9d68febf9b9264c3e1cf82ce1
James Zern [Tue, 25 Aug 2015 03:03:19 +0000 (20:03 -0700)]
vp9_update_layer_...config: delete dead assignment
Change-Id: I7401a70a3f33d13fa1ab0393e59fceb5ef083534
Johann [Tue, 25 Aug 2015 21:53:09 +0000 (14:53 -0700)]
Mark Scale2Ratio INLINE
WebRTC builds with -Wunused-function do not like all the uses of onyx.h
Change-Id: Ic57cb143b48df9c9895f94b5f8f395c260fa7025
Yaowu Xu [Tue, 25 Aug 2015 17:07:34 +0000 (17:07 +0000)]
Merge "Limit new motion vector size"
Yaowu Xu [Tue, 25 Aug 2015 02:09:03 +0000 (19:09 -0700)]
Limit new motion vector size
This commit adds clamp of new vectors similar to the logic in RD loop.
Such clamp is not necessary from the perspective of VP8 bitstream, but
is added to improve ChromeCast mirroring's robustness.
Change-Id: I42f6adbc60ffce283b994869364230858632d6fa
Shunyao Li [Tue, 25 Aug 2015 16:51:04 +0000 (16:51 +0000)]
Merge "Add transform size rate for intra skip mode in rdopt"
Debargha Mukherjee [Tue, 25 Aug 2015 16:31:54 +0000 (16:31 +0000)]
Merge "Prevent an ASAN failure for svc tests"
Shunyao Li [Wed, 19 Aug 2015 19:04:56 +0000 (12:04 -0700)]
Add transform size rate for intra skip mode in rdopt
stdhd +0.226 hevchr +0.091 hevcmr +0.052 derflr +0.033
Change-Id: I84034209c5760609a99bd6e0ce55e02534b72cac
hui su [Mon, 24 Aug 2015 20:46:05 +0000 (13:46 -0700)]
Fix a bug induced in
f1d090e2f56e28b3218fb99f648bd5d59ae2f4b1
Change-Id: I5f7ab8ee75192a4c6c7aa723d2e113400d6ca9a5
Hui Su [Mon, 24 Aug 2015 19:57:22 +0000 (19:57 +0000)]
Merge "Refactoring on transform types"
hui su [Mon, 24 Aug 2015 18:40:04 +0000 (11:40 -0700)]
Avoid setting rate multiplier as 0
In high bitdepth setting, the rate multipier may be set as 0. In
lossless mode, the RD cost would always be 0, resulting in bad
partition and prediction mode choices.
Change-Id: I297014dd8bfa8a07ff0ab480119f75678300ff68
hui su [Thu, 13 Aug 2015 23:27:19 +0000 (16:27 -0700)]
Refactoring on transform types
Prepare for adding more transform varieties (EXT_TX and TX_SKIP in nextgen).
Change-Id: I2dfe024f6be7a92078775917092ed62abc2e7d1e
Debargha Mukherjee [Mon, 24 Aug 2015 17:28:02 +0000 (10:28 -0700)]
Prevent an ASAN failure for svc tests
This patch just fixes the test for the time being, but does not
actually solve the underlying issue, which still needs investigation.
Change-Id: I54a35de839723f5b499b57e38dd2bdd400adc427
Jingning Han [Wed, 19 Aug 2015 21:13:18 +0000 (14:13 -0700)]
Turn on codec behavior unit tests for vp10
This commit adds codec behavior unit tests for vp10.
Change-Id: Ieb49cb66f0b29679ff2a3e2c0804d1ebbf48f986
Jingning Han [Sat, 22 Aug 2015 00:00:27 +0000 (00:00 +0000)]
Merge "vpxenc: fix DECODER ifdef"
Hui Su [Fri, 21 Aug 2015 23:21:04 +0000 (23:21 +0000)]
Merge "Refactor get_tx_type and get_scan"
James Zern [Fri, 21 Aug 2015 23:08:36 +0000 (16:08 -0700)]
vpxenc: fix DECODER ifdef
CONFIG_VP10_ENCODER -> CONFIG_VP10_DECODER
Change-Id: I6f1a798835acc4ef66bd2c5d6ffb02247c74b1d6
hui su [Tue, 18 Aug 2015 23:57:07 +0000 (16:57 -0700)]
Refactor get_tx_type and get_scan
This makes it easier to add new transform types and scan orders
to VP10 in the future.
Change-Id: I94874ddc9b19928d7820d57e94e2af04adf51efe
Jingning Han [Fri, 21 Aug 2015 15:31:26 +0000 (15:31 +0000)]
Merge "Replace MAX_REF_LF_DELTAS with MAX_REF_FRAMES"
Marco Paniconi [Fri, 21 Aug 2015 15:28:23 +0000 (15:28 +0000)]
Merge "Update to dynamic resize for 1 pass CBR: source scaling."
Scott LaVarnway [Fri, 21 Aug 2015 12:06:34 +0000 (12:06 +0000)]
Merge "VPX: scaled convolve : fix windows build errors"
Marco [Thu, 20 Aug 2015 23:29:24 +0000 (16:29 -0700)]
Update to dynamic resize for 1 pass CBR: source scaling.
Switch to use the normative (convolve8) filter for source scaling,
only for 1/2x1/2 scaling for now. This is faster and has better
quality than either the vpx_scale_frame or the nonnormative scaler.
Remove the vp9_scale_if_required_fast, which is now not used.
Change-Id: I2f7d73950589d19baafb1fa650eac987d531bcc8
Scott LaVarnway [Thu, 20 Aug 2015 20:09:27 +0000 (13:09 -0700)]
VPX: scaled convolve : fix windows build errors
Change-Id: Ic81d435ea928183197040cdf64b6afd7dbaf57e4
Jingning Han [Thu, 20 Aug 2015 18:01:06 +0000 (18:01 +0000)]
Merge "Refactor REFS_PER_FRAME definition"
Jingning Han [Thu, 20 Aug 2015 16:51:08 +0000 (09:51 -0700)]
Replace MAX_REF_LF_DELTAS with MAX_REF_FRAMES
The macro MAX_REF_LF_DELTAS is repeatedly defined. Merge its use
cases with MAX_REF_FRAMES.
Change-Id: I71c4634fb5feae78ce6f9bf99074f45ed7f68b73
Jingning Han [Thu, 20 Aug 2015 16:26:17 +0000 (09:26 -0700)]
Refactor REFS_PER_FRAME definition
Define it as a function of reference frame types to provide
scalability for multiple reference frames.
Change-Id: I77b856c96916f352bc31004b9266b3f24e19bd0f
James Zern [Wed, 19 Aug 2015 23:24:31 +0000 (16:24 -0700)]
x86inc.asm: only set visibility for chromium builds
this restores the previous version's behavior avoiding issues with
builds that may split sources on directory boundaries; protected
visibility may work in this case.
Change-Id: If37c70d9bd81de85a8e112457b9819a5cac6129d
Scott LaVarnway [Wed, 19 Aug 2015 22:12:21 +0000 (22:12 +0000)]
Merge "VPX ssse3 scaled convolve"
Jingning Han [Wed, 19 Aug 2015 18:26:30 +0000 (18:26 +0000)]
Merge "Rename inv_txfm_sse2.asm to inv_wht_sse2.asm"
Jingning Han [Wed, 19 Aug 2015 16:14:21 +0000 (09:14 -0700)]
Rename inv_txfm_sse2.asm to inv_wht_sse2.asm
Change-Id: I43bcc70680503e4c18d8f021097307778cf9ea70
Marco Paniconi [Wed, 19 Aug 2015 15:47:40 +0000 (15:47 +0000)]
Merge "Fix to non-rd variance partition selection."
Marco Paniconi [Tue, 18 Aug 2015 23:28:05 +0000 (23:28 +0000)]
Merge "Allow for re-encoding frame if high overshoot."
Johann Koenig [Tue, 18 Aug 2015 23:09:03 +0000 (23:09 +0000)]
Merge "Rename vp8 loopfilter_filters_dspr2.c"
Johann Koenig [Tue, 18 Aug 2015 23:08:51 +0000 (23:08 +0000)]
Merge "Rename vp8 loopfilter[_neon.c]"
Johann Koenig [Tue, 18 Aug 2015 23:08:44 +0000 (23:08 +0000)]
Merge "Rename vp8 quantize.c"
Marco [Mon, 6 Jul 2015 22:55:57 +0000 (15:55 -0700)]
Allow for re-encoding frame if high overshoot.
For 1 pass CBR mode under screen content mode:
if pre-analysis (source temporal-sad) indicates significant
change in content, then check the projected frame size after
encode_frame(), and if size is above threshold, force re-encode
of that frame at max QP.
Change-Id: I91e66d9f3167aff2ffcc6f16f47f19f1c21dc688
Scott LaVarnway [Tue, 18 Aug 2015 22:13:02 +0000 (15:13 -0700)]
VPX ssse3 scaled convolve
Change-Id: I71d5994e21813554a927d35ebcc26bf7a68984fd
Johann [Tue, 18 Aug 2015 22:10:37 +0000 (15:10 -0700)]
Rename vp8 loopfilter_filters_dspr2.c
Change-Id: I914b456558edbdee5eefdfba731bc70d3d5f5d53
Marco [Tue, 18 Aug 2015 17:55:33 +0000 (10:55 -0700)]
Updates to skin detection
-Use 8x8 or 16x16 blocks.
-Option to use average of center pixel(s).
Change-Id: I7d962da30039d41385843e11606bc6482c7906f9
Johann [Tue, 18 Aug 2015 18:47:00 +0000 (11:47 -0700)]
Rename vp8 loopfilter[_neon.c]
Avoid conflict with vpx_dsp version
Change-Id: I041b1532a9276400a5547de8dfed1de43ad4e83d
Johann [Tue, 18 Aug 2015 18:43:40 +0000 (11:43 -0700)]
Rename vp8 quantize.c
Move it to vp8_quantize.c and make sure to
use the full path for all vp8 includes of quantize.h
Change-Id: I284651ff681707385f4924ea7db1541905c1624a
Jingning Han [Tue, 18 Aug 2015 02:37:45 +0000 (02:37 +0000)]
Merge "Turn on dspr2 loop filter functions in vpx_dsp"
Hui Su [Tue, 18 Aug 2015 00:49:24 +0000 (00:49 +0000)]
Merge "create vp10_arg_ctrl_map"
Hui Su [Tue, 18 Aug 2015 00:49:13 +0000 (00:49 +0000)]
Merge "remove unused high-bit-depth args from arg list"
Jingning Han [Mon, 17 Aug 2015 23:14:24 +0000 (16:14 -0700)]
Turn on dspr2 loop filter functions in vpx_dsp
Add the dspr2 files to vpx_dsp.mk and enable these functions in
vpx_dsp_rtcd_defs.pl file.
Change-Id: I79feb5af24f174f4a0788dc6f3b6df7f4e1fa467
Marco [Mon, 17 Aug 2015 19:25:58 +0000 (12:25 -0700)]
Fix to non-rd variance partition selection.
Only test for using golden as reference for variance partition
selection if it is used as a reference for that frame.
For temporal layers, golden may not be a reference on a given frame,
even though it was for some previous frame. If it is not a reference
for current frame, don't check/use it for partition selection.
Change-Id: I6b0f2bd36aebbb5903077c9a0a66d80f1de9a7b1
hui su [Mon, 17 Aug 2015 01:21:56 +0000 (18:21 -0700)]
create vp10_arg_ctrl_map
Change-Id: Ibe49acb022f1d557ffccb91e7bd2558870ff1a0b
Jingning Han [Mon, 17 Aug 2015 16:57:29 +0000 (16:57 +0000)]
Merge "Configure vp10 disabled by default"
Hui Su [Mon, 17 Aug 2015 16:46:06 +0000 (16:46 +0000)]
Merge "VP9EncoderConfig -> VP10EncoderConfig"
hui su [Mon, 17 Aug 2015 16:44:40 +0000 (09:44 -0700)]
remove unused high-bit-depth args from arg list
Change-Id: I8373a0e980aaaa0d0cab8b2e69d1a897ff5aec2d
Jingning Han [Sat, 15 Aug 2015 23:04:11 +0000 (16:04 -0700)]
Configure vp10 disabled by default
To turn on vp10, one needs to explicitly set --enable-vp10.
Change-Id: Ifc7e294aea9a5bdf5d08f0001ae26fba5a837b64
hui su [Mon, 17 Aug 2015 00:24:35 +0000 (17:24 -0700)]
VP9EncoderConfig -> VP10EncoderConfig
Change-Id: Id691d8a9d4aa137e8de73fe680016fd4fd070e6e
Jingning Han [Sat, 15 Aug 2015 22:57:54 +0000 (15:57 -0700)]
Set VP9 as vpxenc default encoder
Change-Id: Idb8a133364fc29ac762ad344262bcc47c568a55c
Jingning Han [Sat, 15 Aug 2015 22:40:11 +0000 (22:40 +0000)]
Merge "Change vp9_ prefix function names in vpx_scale to vpx_"
Jingning Han [Fri, 14 Aug 2015 22:27:02 +0000 (15:27 -0700)]
Change vp9_ prefix function names in vpx_scale to vpx_
Change-Id: Iac85902cbbb3e752801dc85de9a3c778e47304aa
Yaowu Xu [Fri, 14 Aug 2015 21:52:29 +0000 (21:52 +0000)]
Merge "change vp10 to use a different sync code than vp9"
Yaowu Xu [Fri, 14 Aug 2015 21:52:18 +0000 (21:52 +0000)]
Merge "write correct codecID for vp10 webm file"
Yaowu Xu [Fri, 14 Aug 2015 21:52:10 +0000 (21:52 +0000)]
Merge "Remove vp10's build dependency on vp9"
Yaowu Xu [Fri, 14 Aug 2015 20:06:56 +0000 (13:06 -0700)]
change vp10 to use a different sync code than vp9
Change-Id: If5b5beb38c938bd7a6016dedc9bd0c8e06153d73
Yaowu Xu [Fri, 14 Aug 2015 19:30:10 +0000 (12:30 -0700)]
write correct codecID for vp10 webm file
Change-Id: Ifeb6073c4a687a73540a0f8aef0d834f4712c193
James Zern [Fri, 14 Aug 2015 19:45:27 +0000 (19:45 +0000)]
Merge changes I2fe52bfb,I5e5084eb
* changes:
VPX: removed filter == 128 checks from mips convolve code
VPX: removed step checks from mips convolve code
James Zern [Fri, 14 Aug 2015 19:23:46 +0000 (19:23 +0000)]
Merge "VPX: removed step checks from neon convolve code"
Yaowu Xu [Fri, 14 Aug 2015 19:16:07 +0000 (12:16 -0700)]
Remove vp10's build dependency on vp9
CONFIG_VP9_HIGHBITDEPTH is currently used by both vp9 and vp10, but in
many place outside vp9/vp10, the macro was used in conjunction of
CONFIG_VP9. This created a dependency on vp9 for vp10 to build. This
commit removes the dependency by use CONFIG_VP9_HIGHBITDEPTH only in
these places.
Change-Id: I8cc007fc9cf132394c6498ce6759e606b64a6ad0
Yaowu Xu [Fri, 14 Aug 2015 16:18:07 +0000 (09:18 -0700)]
vpx_highbd_ssim_parms_8x8: make parameter types consistent
Change-Id: Ie1fe6603232adc22dbe4d51bd1008c856a6d40ca
Yunqing Wang [Thu, 13 Aug 2015 23:51:27 +0000 (23:51 +0000)]
Merge "Copy VP9 second-level sub-pixel search change to VP10"
Yaowu Xu [Thu, 13 Aug 2015 22:51:51 +0000 (22:51 +0000)]
Merge "VP9Common -> VP10Common"
Yaowu Xu [Thu, 13 Aug 2015 22:32:42 +0000 (22:32 +0000)]
Merge "VP9Decoder -> VP10Decoder"
Yunqing Wang [Thu, 13 Aug 2015 22:00:32 +0000 (15:00 -0700)]
Copy VP9 second-level sub-pixel search change to VP10
Merged patch "Improve the second-level sub-pixel motion search"
(commit:
7418b176ce1c985637c639280d70b12bdc850e57 ) to VP10.
Change-Id: I89b5d76624290c47dd84d9c5332b4909238e268d
Yaowu Xu [Thu, 13 Aug 2015 21:55:28 +0000 (14:55 -0700)]
VP9Common -> VP10Common
Change-Id: Ifddc68a8ab020f35a266ffc74429363301fe7d9e
Yaowu Xu [Thu, 13 Aug 2015 21:56:33 +0000 (21:56 +0000)]
Merge "VP9_COMP -> VP10_COMP"
Yaowu Xu [Thu, 13 Aug 2015 21:43:55 +0000 (14:43 -0700)]
VP9Decoder -> VP10Decoder
Change-Id: I427b56e92aa2d1381192cb394ac1bb95becb32a1
Scott LaVarnway [Thu, 13 Aug 2015 19:57:01 +0000 (12:57 -0700)]
VPX: removed filter == 128 checks from mips convolve code
The check is handled by the predictor table.
Change-Id: I2fe52bfbbfccb2edd13ba250986e3a4b4b589459
Alex Converse [Thu, 13 Aug 2015 19:08:36 +0000 (12:08 -0700)]
opsnr.stt: Add a missing tab after PSNRHVS
Logfile parsers expect it to be there.
Change-Id: I7c96a42c8be570a28f51cc4edfa619d9cf9af73b
Yaowu Xu [Thu, 13 Aug 2015 16:42:27 +0000 (09:42 -0700)]
VP9_COMP -> VP10_COMP
Change-Id: I83b5c69621f9f28b742e5b13517d4e5b99c6cd26
Scott LaVarnway [Thu, 13 Aug 2015 18:27:04 +0000 (11:27 -0700)]
VPX: removed step checks from mips convolve code
The check is handled by the predictor table.
Change-Id: I5e5084ebb46be8087c8c9d80b5f76e919a1cd05b
Yaowu Xu [Thu, 13 Aug 2015 18:00:27 +0000 (18:00 +0000)]
Merge "VP9_COMMON -> VP10->COMMON"
Yaowu Xu [Thu, 13 Aug 2015 18:00:02 +0000 (18:00 +0000)]
Merge "Correct guard macros in header files"
Yaowu Xu [Thu, 13 Aug 2015 16:36:53 +0000 (09:36 -0700)]
VP9_COMMON -> VP10->COMMON
Change-Id: I651b7bee90f33581368853da81f9622805ccc0ea
Jingning Han [Thu, 13 Aug 2015 15:38:50 +0000 (08:38 -0700)]
Remove common_data.c file from vp10 folder
This file is empty in vp10. Remove it to clear out visual studio
warning.
Change-Id: I741e88adcf3c9fc6b73b13ea08a2094e2f3a7eba
Yaowu Xu [Thu, 13 Aug 2015 16:25:39 +0000 (09:25 -0700)]
Correct guard macros in header files
Change-Id: Ifce12a95c1cdc36dc6ac5a72759249a17407da9e
Yaowu Xu [Thu, 13 Aug 2015 15:41:10 +0000 (15:41 +0000)]
Merge "Add vpxenc help info"
Scott LaVarnway [Wed, 12 Aug 2015 23:14:53 +0000 (16:14 -0700)]
VPX: removed step checks from neon convolve code
The check is handled by the predictor table.
Change-Id: I42479f843e77a2d40cdcdfc9e2e6c48a05a36561
Jingning Han [Wed, 12 Aug 2015 23:33:46 +0000 (23:33 +0000)]
Merge "vp10_encoder: add missing vpx_clear_system_state"
Jingning Han [Wed, 12 Aug 2015 22:42:43 +0000 (15:42 -0700)]
vp10_encoder: add missing vpx_clear_system_state
Mirror the fix in vp9_encoder for sse-only build.
Change-Id: I9f272874e52f4cc295fcaa71920f8a5d3a80045b
James Zern [Wed, 12 Aug 2015 22:41:51 +0000 (22:41 +0000)]
Merge "vp9_encoder: add missing vpx_clear_system_state"
Scott LaVarnway [Wed, 12 Aug 2015 20:13:33 +0000 (20:13 +0000)]
Merge "VPX: remove step == 16 and filter[3] != 128 checks"