Commit Graph

6491 Commits

Author SHA1 Message Date
JackyChen
7ba646f7e6 Fix a bug in calculating delta in VP9 denoiser.
When calculating delta in VP8 denoiser, since the block size is fixed to 16x16,
the divisor is 256, which is the number of the pixel.
But in VP9, the block size varies, the divisor should correspond to the block
size.

Change-Id: Ibdc1e5d23ba8c788b0d0dc6d406bcdfc34c1b142
2014-09-29 13:09:18 -07:00
Jingning Han
8b4dd536a5 Merge "Skip certain ALTREF inter modes in ARF coding" 2014-09-29 10:43:45 -07:00
Deb Mukherjee
d4713f1d50 Fix a bug introduced in a previous patch on highbd
Change-Id: Ice692334f75157446a44a6e81503cada977934f4
2014-09-26 15:43:55 -07:00
Jingning Han
ccdb518ff8 Skip certain ALTREF inter modes in ARF coding
This commit enables the encoder to skip checking ALTREF inter modes
in ARF coding, if the predicted motion vectors suggest that the
GOLDEN_FRAME provides higher prediction accuracy than ALTREF_FRAME.

It improves the speed 3 encoding speed by about 5%, at the expense
of compression performance loss -0.041% and -0.225% for derf and
stdhd, respectively.

pedestrian_area 1080p 2000 kbps
66705 b/f, 40.909 dB, 118738 ms ->
66732 b/f, 40.908 dB, 113688 ms

old_town_cross 720p 1500 kbps
14427 b/f, 36.256 dB, 62746 ms ->
14412 b/f, 36.252 dB, 60690 ms

blue_sky 1080p 1500 kbps
51026 b/f, 35.897 dB, 73310 ms ->
50921 b/f, 35.893 dB, 70406 ms

bus CIF 1000 kbps
21301 b/f, 34.841 dB, 7326 ms ->
21248 b/f, 34.837 dB, 7196 ms

Change-Id: I76cf88b4d655e1ee3c0cb03c8a5745493040e8d2
2014-09-26 12:53:43 -07:00
hkuang
62848d2a87 Merge "Avoid calling vp9_is_scaled two times in a function." 2014-09-25 13:58:54 -07:00
Deb Mukherjee
0b3560c7a2 Merge "Adds various high bit-depth encode functions" 2014-09-25 13:48:05 -07:00
hkuang
c53a95ad1d Avoid calling vp9_is_scaled two times in a function.
Use a local variable to hold the result of vp9_is_scaled.

Change-Id: I5e203909805923e20eefef596bc84424da47dbe2
2014-09-25 11:52:16 -07:00
Yaowu Xu
845d4f333d Fix a couple of comments
The first comment is obselete given the way is now normative in VP9
bitstream. The second comment line was too long.

Change-Id: I6546585babf60d466485ddcf2daa6d2fa79e999a
2014-09-25 08:24:16 -07:00
Deb Mukherjee
993d10a217 Adds various high bit-depth encode functions
Change-Id: I6f67b171022bbc8199c6d674190b57f6bab1b62f
2014-09-25 01:50:36 -07:00
Jingning Han
6989e81d61 Remove unused variable in handle_inter_mode
Change-Id: Id757d2c940756ce1b0ead2ea24af9ac0a493de05
2014-09-24 18:27:44 -07:00
Yaowu Xu
22d8c156f9 Merge "Correct the condition for border extension" 2014-09-24 14:09:45 -07:00
Yaowu Xu
d237d483a5 Correct the condition for border extension
As reported in issue #850, the condition for border extension was not
complete. This commit added the case when the scaling is enabled.

This fixes issue #850.

Change-Id: I67768b23f0dcc4ac9a9aa0a0825b0fe8cb85a72e
2014-09-24 11:26:40 -07:00
Paul Wilkins
76035d16d9 Merge "Fix build issue with stats enabled." 2014-09-24 10:32:37 -07:00
Yaowu Xu
148c57d231 Merge "Fix invalid memory access on 2x downscale." 2014-09-24 09:58:05 -07:00
Yaowu Xu
7be8954e34 Merge "Fix incorrect subsampling used in VP9 non420 loopfilter." 2014-09-24 08:59:11 -07:00
Alexander Voronov
eafd842a3e Fix incorrect subsampling used in VP9 non420 loopfilter.
Change-Id: Ia959e24b4676242c80a8867d2c39a6fee90f71a5
2014-09-24 17:01:09 +04:00
Paul Wilkins
5b724fc78e Fix build issue with stats enabled.
Compiler build issue when output stats enabled.

Change-Id: I7b5409108f3f27ba61b0241b9340b412683eff45
2014-09-24 11:48:58 +01:00
Deb Mukherjee
e1d3c36525 Adds high bit-depth frame resize functions
Change-Id: I35b015a759325d72d0da427c61a09f19f8e69697
2014-09-23 22:55:33 -07:00
Yaowu Xu
8751e49a6f Merge "Adapt mode based rd_threshold for similar block size" 2014-09-23 22:28:08 -07:00
Yaowu Xu
60737c9fc8 Merge "Fix an IOC" 2014-09-23 20:44:35 -07:00
Deb Mukherjee
4109372af3 Adds high bit-depth psnr/sse functions
Also adds some miscellaneous high bit-depth setup functions.

Change-Id: I66488b08a5a2a8cb9518ca10497cf1c1501ceded
2014-09-23 17:28:05 -07:00
Deb Mukherjee
e2a90c0b21 Merge "High bit-depth loop/arf/postproc filter functions" 2014-09-23 17:26:32 -07:00
Deb Mukherjee
6c6213d960 Merge "Pruned subpel search for speed 3." 2014-09-23 17:12:03 -07:00
Deb Mukherjee
931ed516ba High bit-depth loop/arf/postproc filter functions
Adds high-bitdepth loopfilter, temporal filter and postproc functions

Change-Id: I81c8a9176890784686bc4f2af0d550d243b3b2d3
2014-09-23 16:20:43 -07:00
Yaowu Xu
4a101310e8 Adapt mode based rd_threshold for similar block size
The rd_thresholds are adaptively changed based on best mode tested.
It was only changed for the same block size, this commit makes the
adaptation for similar block sizes too. The commit also made minor
adjustment and code cleanups.

The impact on encoding time for _ped:
118089 ms -> 111927 ms

The impact on compression:
derf:  -0.339%
stdhd: -0.303%

Change-Id: I8817fed1102350497f2ec631849e43f753878e5d
2014-09-23 16:10:59 -07:00
Yaowu Xu
56032b471d Fix an IOC
Change-Id: I0ca6746696d81657c035b0f6523c9af370da3c95
2014-09-23 16:07:22 -07:00
Deb Mukherjee
c94b17f4b2 Pruned subpel search for speed 3.
Adds code to return an integer cost list for NSTEP search. Then
uses it for pruned subpel search in speed 3.

derf: -0.06%
Speed on mobcal 720p increaes from 10.28 fps to 10.65 fps.
[Subject to further testing].

Change-Id: Ib591382d25b2c11bcaba9d3a27a93a9d1ab27a96
2014-09-23 11:27:58 -07:00
Yaowu Xu
7feede9869 Merge "Remove code duplication" 2014-09-22 17:13:59 -07:00
Yaowu Xu
052bc8ea6a Merge "Simplify rd_pick_intra_sby_mode()" 2014-09-22 17:13:55 -07:00
Yaowu Xu
c7ab18fe56 Remove code duplication
Change-Id: I453b3e0d946951665d5919248445fc4f3222d2ad
2014-09-22 15:22:51 -07:00
Yaowu Xu
f46326c7a2 Simplify rd_pick_intra_sby_mode()
Change-Id: Ifb0915c94c2db48827ddbd446314cb6e3155b99c
2014-09-22 14:58:51 -07:00
Minghai Shang
38b6aed8fd Merge "[spatial svc] Remove vpx_svc_parameters_t and the loop that sets it for each layer" 2014-09-22 14:01:24 -07:00
Jingning Han
f7023ea014 Remove unnecessary local variable declaration
This commit removes a repetitive local variable declaration in
vp9_rd_pick_inter_mode_sb.

Change-Id: I1b0afa98ff1ecbfb46e17d3d1cee95d32c4309db
2014-09-22 09:29:28 -07:00
Jingning Han
eee904c9b9 Adaptive mode search scheduling
This commit enables an adaptive mode search order scheduling scheme
in the rate-distortion optimization. It changes the compression
performance by -0.433% and -0.420% for derf and stdhd respectively.
It provides speed improvement for speed 3:

bus CIF 1000 kbps
24590 b/f, 35.513 dB, 7864 ms ->
24696 b/f, 35.491 dB, 7408 ms (6% speed-up)

stockholm 720p 1000 kbps
8983 b/f, 35.078 dB, 65698 ms ->
8962 b/f, 35.054 dB, 60298 ms (8%)

old_town_cross 720p 1000 kbps
11804 b/f, 35.666 dB, 62492 ms ->
11778 b/f, 35.609 dB, 56040 ms (10%)

blue_sky 1080p 1500 kbps
57173 b/f, 36.179 dB, 77879 ms ->
57199 b/f, 36.131 dB, 69821 ms (10%)

pedestrian_area 1080p 2000 kbps
74241 b/f, 41.105 dB, 144031 ms ->
74271 b/f, 41.091 dB, 133614 ms (8%)

Change-Id: Iaad28cbc99399030fc5f9951eb5aa7fa633f320e
2014-09-22 09:28:16 -07:00
hkuang
c70cea97ac Remove mi_grid_* structures.
mi_grid_* are arrays of pointer to pointer. They save the pointers that point
to the MIs in cm->mi. But they are unnecessary and complicated. The original
goal was to remove MODE_INFO_t copy. But with an extra MODE_INFO_t pointer
inside MODE_INFO_t, same goal could be achieved.

This commit totally removes the mi_grid_* structures. But there are still
many dummy MODE_INFO_t inside cm->mi which are a waste of memory. Next commit
will do on-demand MODE_INFO_t allocation in order to save these memories.

Change-Id: I3a05cf1610679fed26e0b2eadd315a9ae91afdd6
2014-09-19 21:27:11 -07:00
Deb Mukherjee
822b51609b High bit-depth coefficient coding functions
Tokenization and Detokenization enhancements for 10/12 bit

Change-Id: I3c269ec30f8eb160ee024905638a193975237559
2014-09-19 15:21:24 -07:00
Minghai Shang
209ee12110 [spatial svc] Remove vpx_svc_parameters_t and the loop that sets it for each layer
vpx_svc_parameters_t contains id, resolution and min/max qp for each spatial layer.

In this change we will use extra config to send min/max qp and scaling factors, then calculate layer resolution inside encoder.

Change-Id: Ib673303266605fe803c3b067284aae5f7a25514a
2014-09-18 18:05:07 -07:00
Frank Galligan
49dc7b05d0 Merge "FIX: vp9_loopfilter_intrin_sse2.c" 2014-09-18 15:10:16 -07:00
Scott LaVarnway
13284311eb FIX: vp9_loopfilter_intrin_sse2.c
Fixes Visual Studio build failures

Change-Id: I233719cd63b3ad0db16e2834bf1d7ea1df805880
2014-09-18 13:09:13 -07:00
Minghai Shang
f66be91f61 Merge "[spatial svc] Use same golden frame for all temporal layers" 2014-09-18 12:29:40 -07:00
Minghai Shang
f780b16bb8 [spatial svc] Use same golden frame for all temporal layers
Overhead goes down from 8% to 3% for 1080 60p

Change-Id: Idf3e5ca8712402a914a8cb79df17d3cdab63b163
2014-09-18 11:16:29 -07:00
Deb Mukherjee
6d0ee9860e Merge "Adds high bitdepth convolve, interpred & scaling" 2014-09-18 10:52:23 -07:00
Deb Mukherjee
0d3c3d3ce7 Adds high bitdepth convolve, interpred & scaling
Change-Id: Ie51c352a6b250547207cbc1ebba833a01ed053e3
2014-09-18 07:26:17 -07:00
Paul Wilkins
c389b37bb4 Substantial reworking of code for arf and kf groups.
Substantial restructuring of the way we estimate
the rate of decay in prediction quality and determine
the arf interval and amount of boost used.

Also other changes to support moving to a lower first pass
Q which exposes some new features and allows us to better
distinguish genuinely static blocks from low motion or noisy
blocks.

Net gains now visible on all the test sets with std-hd PSNR up
1.87%. There are still some bad outlier cases but most of these
are low motion or slide show type content where the metrics
are already high at any given rate. The best + case is up by
more than 10%.

Change-Id: I18e25170053bdf3188f493ff8062f48a74515815
2014-09-18 12:53:48 +01:00
Frank Galligan
4e066299d9 Merge "Improved mb_lpf_horizontal_edge_w_sse2_16() #2" 2014-09-17 18:52:30 -07:00
Scott LaVarnway
217e3cb1fb Improved mb_lpf_horizontal_edge_w_sse2_16() #2
The decoder performance improved up to 1% for the
test clips used.

Change-Id: I4621112bdccfba01640322facfa4ba8da8290ea5
2014-09-17 17:25:20 -07:00
Deb Mukherjee
7d0e4f9ad1 Resolves a few gcc warnings
clang is fine.

Change-Id: Ia4e9ff17ea3b86bc87dca35828ee7ce45bea6994
2014-09-16 22:44:40 -07:00
Deb Mukherjee
f7cf05cfe0 Merge "Adding high-bitdepth intra prediction functions" 2014-09-16 17:10:24 -07:00
Frank Galligan
ecd7e3d2b7 Merge "Remove memset of every external frame buffer." 2014-09-16 15:17:26 -07:00
Dmitry Kovalev
4f506358c4 Merge "Speeding up decode_coeffs()." 2014-09-16 15:13:14 -07:00