7950 Commits

Author SHA1 Message Date
Jingning Han
0fe7ee6df5 Merge "Fix sub8x8 motion search on scaled reference frame" into sandbox/jingning@google.com/decoder_test_suite 2015-12-11 20:28:21 +00:00
Jingning Han
35b121b0d8 Merge "Fix high bit depth with scaled reference frame" into sandbox/jingning@google.com/decoder_test_suite 2015-12-11 20:27:56 +00:00
Jingning Han
e50c37ba82 Fix sub8x8 motion search on scaled reference frame
This commit fixes a buffer overflow issue related to sub8x8 motion
search on scaled reference frame.

Change-Id: Iffeebc8787c1ed2b0cb3a7821349e028639d9eb5
2015-12-10 22:26:41 -08:00
Jingning Han
db7312483f Fix high bit depth with scaled reference frame
This commit fixes an encoder issue in high bit depth implementation
that causes enc/dec mismatch in scaled reference frame.

Change-Id: I97655aa213f830d8e2dcd98253c3008abda30eea
2015-12-10 15:47:33 -08:00
hui su
e38e2b8343 Add moving-average bit rate stats
Change-Id: Id764e573776d4d0ee2c400a4eca0832268e1e2b1
2015-11-17 17:58:42 -08:00
Jingning Han
fb27ffbb06 Exercise internal frame resizing
This commit enables the encoder to exercise internal frame resizing
and to use scaled reference frame for sub8x8 block motion compensated
prediction.

Change-Id: I42703da4a4b075c6aefe9a9f687374af65c3c73f
2015-09-08 11:20:39 -07:00
Debargha Mukherjee
29a325f78e Expose params min-gf-interval/max-gf-interval
Adds two new vp9 parameters --min-gf-interval and --max-gf-interval
to enable testing based on frequency of alt-ref frames.

Also adds a unit-test to test enforcement of min-gf-interval.

For both these parameters the default value is 0, which indicates
they are picked by the encoder, based on resolution and framerate
considerations. If they are greater than zero, the specified
parameter is honored.

(Additional note by paulwilkins)
Note that there is a slight oddity in that key frames are also GFs and
considered part of  GF only group. However they are treated as not
being part of an arf group because for arf groups the previous GF is
assumed to be the terminal or overlay frame for the previous group.

(end note)

Change-Id: Ibf0c30b72074b3f71918ab278ccccc02a95a70a0
(cherry picked from commit 98526433737e388ebe576e9f7631b6763e58a2f9)
2015-09-01 17:50:59 +00:00
Jingning Han
cb8b7d0dc2 VP9 decoder performance test suite - buffer process
This commit hacks the vp9 encoder to produce bit-streams that
exercise maximum reference frame buffers at decoder, as part of
the VP9 decoder performance test suite. It fullfills all the 8
reference frame buffers first and then temporarily suspends the
buffer update. It selects the frames from this static buffer pool
as reference frames for the next a few coding frames. When all
the frames in the reference frame buffer are covered, the codec
resumes the buffer update process. Such pattern repeats every
64 frames.

It verifies the decoder capability to handle up to 8 reference
frames in buffer.

Change-Id: I796701eba53ed19ae73351d23d676311f12c43a1
2015-08-26 15:33:20 -07:00
Jingning Han
9cec9cb26f Add more coding staticstics tracker
Count the average sub-pixel motion vector, sub8x8 block, intra
prediction mode use case.

Change-Id: Idbccc67a7eb4d2447b02b1fc158fdc8a344c2f21
2015-07-29 10:31:47 -07:00
Jingning Han
b75e4d0f13 Set up decoder end coding statistics tracker
This commit allows the decoder to track the percentage of intra
coding mode and sub pixel filter usage, when it is configured. This
provides a measurable approach to generate the VP9 decoder
performance test suite.

Change-Id: I26d40b991f41b1408de2b206ecb0a322cdb561b8
2015-06-29 12:26:54 -07:00
Parag Salasakar
b92cc27b76 mips msa vp9 temporal filter optimization
average improvement ~4x-5x

Change-Id: Iad9c0a296dbc2ea96d000bd009077999ed58a3c5
2015-06-26 12:00:24 +05:30
Parag Salasakar
c040f96e4b mips msa vp9 subtract block optimization
average improvement ~3x-4x

Change-Id: Idbe4d13a00d05ff8be6559b116f416e42c3b4097
2015-06-26 09:23:56 +05:30
Parag Salasakar
d017f5ba38 Merge "mips msa vp9 block error optimization" 2015-06-26 03:42:31 +00:00
Parag Salasakar
1543f2b60e mips msa vp9 block error optimization
average improvement ~3x-4x

Change-Id: If0fdcc34b17437a7e3e7fb4caaf1067bc175f291
2015-06-26 09:04:00 +05:30
Marco
1c7b1f9aec Update to dynamic resize logic for 1pass CBR.
Only do the check for resizing if the feature is selected
(i.e., resize_mode = RESIZE_DYNAMIC).

And modify condition for checking to be resize_count >= window,
(since framerate can change).

Change-Id: Idceb4e50956bb965a1492b4993b0dcb393c9be4d
2015-06-25 12:28:43 -07:00
Marco
3dd9cde2a5 Fix to unstable build from commit 517a66.
Change-Id: I123db2d20ae65a10e2dec95eec61150e2f69546d
2015-06-24 17:28:57 -07:00
James Zern
d219f2b9d2 Merge "vp9_reconintra_neon: add d45 16x16" 2015-06-24 21:23:15 +00:00
Marco
517a662005 aq-mode=3: Reduce boost for segment#2 at low bitrates/low res.
Reduce boost for segment#2 for low bitrates and low-res.

This change is to reduce the rate overshoot at low bitrates.
No change in behavior, except at the very low bitrates.

Change-Id: I0dbd9d3b6356da5804de94adf10fca6a7a8f8948
2015-06-23 16:50:43 -07:00
James Zern
9db1f24c47 vp9_reconintra_neon: add d45 16x16
~90% faster over 20M pixels

Change-Id: I92d80f66e91e0a870a672cfb5dd29bf1a17cb11a
2015-06-22 21:00:07 -07:00
Parag Salasakar
7555e2b822 mips msa vp9 avg optimization
average improvement ~2x-3x

Change-Id: I76f7fc00c0ffdf2b4ba41bf3819f3b6044bcdeff
2015-06-23 07:32:25 +05:30
Parag Salasakar
7b71cdb0b4 Merge "mips msa vp9 fdct 4x4 optimization" 2015-06-23 01:46:54 +00:00
Marco
fb2a89b1fb Fixes for key frame coding at speed 5.
Keep the same transform cutoff and partition selection
for speed 5 as in speeds >=6 (non-rd speed settings).

Existing setting for key frame at speed 5 allowed transform size
up to 32x32 on key frames, and did not allow for 4x4 block partition size.
This created more visual artifacts on first few frames.

avgPSNR/overallPSNR/SSIM gains of 0.2/0.7/0.8 for rtc_derf(low-res) set,
and 0/0.7/1.1 gains for rtc set.

Change-Id: I8c139ec6c9bb74e14b4ffbad5f12e94f18a59c0b
2015-06-22 16:57:35 -07:00
James Zern
c8b9658ecc Merge "vp9_reconintra_neon: add d45 8x8" 2015-06-22 22:27:57 +00:00
Marco
8e029fcfa9 Merge "Reduce max_partition_size for low resolutions at speed 5." 2015-06-22 16:59:48 +00:00
Parag Salasakar
bc94999148 mips msa vp9 fdct 4x4 optimization
average improvement ~2x-3x

Change-Id: Idf8be780b8b4228fc91f110a94e4ee1fd9af0163
2015-06-22 14:30:24 +05:30
Parag Salasakar
b6131a733d Merge "mips msa vp9 fdct 8x8 optimization" 2015-06-20 02:58:10 +00:00
James Zern
12c6688e31 vp9_reconintra_neon: add d45 8x8
based on ssse3 implementation

~91% faster over 20M pixels

Change-Id: I6d743a53352c2d6de0efe7899d7996e8b0f7fa29
2015-06-19 19:19:22 -07:00
Parag Salasakar
7ca84888c2 mips msa vp9 fdct 8x8 optimization
average improvement ~4x-5x

Change-Id: I37582efc2622bc20b2bf99617a76110ab24e9f6a
2015-06-20 07:48:35 +05:30
James Zern
7b480ee90c Merge "Add dynamic range comment to vp9_int_pro_row" 2015-06-20 01:43:54 +00:00
Jingning Han
922af194bc Add dynamic range comment to vp9_int_pro_row
Change-Id: Icaa643568159c4e2db24eef42090b002ae02a45e
2015-06-19 17:39:33 -07:00
Jingning Han
c539ec022f Merge "Add dynamic range comment to vp9_int_pro_col" 2015-06-20 00:35:07 +00:00
Marco
debe4e920f Reduce max_partition_size for low resolutions at speed 5.
For speed 5 real-time mode, the selection of the partition size for
superblocks on the segment (aq-mode=3) uses the non-rd recursive
pick partition search, and can sometimes select 64x64.

For low resolutions, visually better to limit this to 32x32.

Change-Id: I69657a7ed8899f8b3cf8c9c318a2509c5c72c565
2015-06-19 16:48:16 -07:00
Alex Converse
90c9ede8e6 Limit cyclic refresh revisitng blocks at the same quantizer.
For screen content don't refresh a block at a quantizer higher than
it was last coded at. PReviosuly at realtime speeds the encoder had a
tendency to recode a block from GOLDEN with a higher Q than it was last
coded at.

Change-Id: Iacd561806c769dcce1a81b9827ffc70090f5ba18
2015-06-19 15:23:02 -07:00
Jingning Han
8e8bc5f28b Add dynamic range comment to vp9_int_pro_col
Change-Id: If14d9f874bd0bf2c5a455982088fd70591f5ea5a
2015-06-19 09:43:57 -07:00
Yaowu Xu
d8428ae35d Fix a msvc compiler warning
Change-Id: Ida8a04370895ed14bd118324ec2577da926e4648
2015-06-19 09:04:29 -07:00
James Zern
714a46a63c Merge "vp9_filter: make all filter tables static" 2015-06-19 03:32:24 +00:00
James Zern
c5d779d266 Merge changes I2552d810,I51952c0a,Ib82e4247,I9c8d16cb
* changes:
  vp9_mcomp: make search_step_table static
  vp9_encodeframe: delete auto_partition_range()
  vp9_mcomp: don't mark setup_center_error() inline
  vp9_encoder: hide adjust_image_stat()
2015-06-19 03:31:38 +00:00
James Zern
a2c69af50e Merge "vp9_reconintra_neon: add d45 4x4" 2015-06-19 03:27:23 +00:00
James Zern
5d1d72df16 Merge changes from topic 'vp9-intra-pred'
* changes:
  vp9_reconintra_neon: add d135 4x4
  vp9_reconintra: correct d135 4x4 signature
2015-06-19 03:24:58 +00:00
Marco
4b45088fc9 Merge "Add dynamic resize logic for 1 pass CBR." 2015-06-19 00:56:06 +00:00
Marco
d77f51ba9e Add dynamic resize logic for 1 pass CBR.
Decision to scale down/up is based on buffer state and average QP
over previous time window. Limit the total amount of down-scaling
to be at most one scale down for now.

Reset certain quantities after resize (buffer level, cyclic refresh,
rate correction factor).

Feature is enable via the setting rc_resize_allowed = 1.

Change-Id: I9b1a53024e1e1e953fb8a1e1f75d21d160280dc7
2015-06-18 17:13:37 -07:00
James Zern
ce88d74d34 vp9_reconintra_neon: add d45 4x4
based on webp's LD4()

~59% faster over 20M pixels

Change-Id: I371eaed9ce8f470451046997e130b0ba1a2f7a9c
2015-06-18 15:25:07 -07:00
James Zern
337b221e00 vp9_reconintra_neon: add d135 4x4
based on webp's RD4()

~50% faster over 20M pixels

Change-Id: Ifcb7bf7f7fc8eabf79d9e3b219ce1be67abc524a
2015-06-18 15:25:06 -07:00
James Zern
e8e3583fc7 vp9_reconintra: correct d135 4x4 signature
add missing '_c' suffix

Change-Id: I928d6cf8f90db0b8ca0b1f3bbf10b3d792062cec
2015-06-18 15:25:06 -07:00
James Zern
41d8545ab6 Merge "vp9_reconintra_neon: add DC 4x4 predictors" 2015-06-18 22:24:55 +00:00
James Zern
6e44bf20f7 vp9_reconintra_neon: add DC 4x4 predictors
~85-89% faster over 20M pixels

Change-Id: I3812e8adfffe5255034da88dfe6546e12f4d10ee
2015-06-18 15:22:43 -07:00
James Zern
e77f859d72 Merge "vp9_reconintra_neon: add DC 32x32 predictors" 2015-06-18 22:17:51 +00:00
Jingning Han
d1398e9f13 Merge "Add dynamic range comment to vp9_satd" 2015-06-18 19:36:53 +00:00
Jingning Han
4f1f510f16 Add dynamic range comment to vp9_satd
Change-Id: I75873846e6fdafbe7597a1bd0192115d2d1e9987
2015-06-18 09:18:22 -07:00
Parag Salasakar
b6ea0c4c57 Merge "mips msa vp9 fdct 32x32 optimization" 2015-06-18 04:30:53 +00:00