Commit Graph

11499 Commits

Author SHA1 Message Date
Marco
4eb8a850e0 Merge "Fixes to adaptive denoiser mode." 2014-08-20 11:39:44 -07:00
Marco
90ccad74d4 Merge "Bugfix for vp8 aggressive denoiser mode." 2014-08-20 11:39:21 -07:00
Dmitry Kovalev
7025d72cd8 Revert "Removing unnecessary casts to int64_t."
My fault, that was a float (not integer) which was converted to int64_t.

This reverts commit a885e1cbf0

Change-Id: Ic50708b959e1c3cb3e37da1429d334fafc3391d6
2014-08-20 11:24:48 -07:00
Dmitry Kovalev
9988b9b164 Merge "Converting MODE to be just BEST, GOOD, and REALTIME." 2014-08-20 11:03:03 -07:00
Dmitry Kovalev
a3c0708a80 Merge "Adding set_mv_search_params()." 2014-08-20 11:02:56 -07:00
Jingning Han
acb11a471a Merge "Fix def pairs in 32x32 2D-DCT sse2" 2014-08-20 10:46:19 -07:00
Marco
d2274ed354 Bugfix for vp8 aggressive denoiser mode.
Change-Id: I6c1ccd88df7c39377d5f98f8533c0cabd53cf86f
2014-08-20 10:38:59 -07:00
Marco
aa88e32017 Fixes to adaptive denoiser mode.
Scale min_consec_zero_last wrt to #temporal layers,
and use full framerate as factor in noise metric.

Change-Id: Id0842b90164ce468d1236173c51965e7620c0e12
2014-08-20 10:34:58 -07:00
Dmitry Kovalev
611a45fb0e Merge "Adding gf_group temp variable." 2014-08-20 09:55:05 -07:00
Johann
911e96a4eb Revert "Revert "VP8 for ARMv8 by using NEON intrinsics 06" This reverts commit 81ad047ee5. Revert "VP8 for ARMv8 by using NEON intrinsics 15" This reverts commit 727af7cebe3698b8493ba6c1360b0a6606c310fb."
This reverts commit 920f803f2e

Change-Id: I410d9036214a1b18427cca70b4bc6d8239740737
2014-08-20 09:41:50 -07:00
Jingning Han
5b21708fd5 Fix def pairs in 32x32 2D-DCT sse2
Properly pair the def/undef order.

Change-Id: I9736a6f8d2efc075b1d72dafc75b9350d055cf65
2014-08-20 09:40:30 -07:00
James Yu
eed005b076 VP8 encoder for ARMv8 by using NEON intrinsics 6
Add shortfdct_neon.c
- vp8_short_fdct4x4_neon
- vp8_short_fdct8x4_neon

Change-Id: I90152c803b484f5fab839473d632c50af0524e68
Signed-off-by: James Yu <james.yu@linaro.org>
2014-08-20 09:25:29 -07:00
James Yu
6d6fdd9c3d VP8 encoder for ARMv8 by using NEON intrinsics 3
Add subtract_neon.c
- vp8_subtract_b_neon
- vp8_subtract_mby_neon
- vp8_subtract_mbuv_neon

Change-Id: If9a17a093478552e3e3276eeaa3f098b9021d08c
Signed-off-by: James Yu <james.yu@linaro.org>
2014-08-20 09:20:55 -07:00
Scott LaVarnway
8013aaa10b VP8 encoder for ARMv8 by using NEON intrinsics 2
Add vp8_shortwalsh4x4_neon.c
- vp8_short_walsh4x4_neon

Change-Id: Ica5f584be608c9e636f62db14f563757e94be09b
Signed-off-by: James Yu <james.yu@linaro.org>
2014-08-20 09:19:23 -07:00
Dmitry Kovalev
d7a32e5ae0 Merge "Using functions from vpx_mem.h inside vpx_image.c." 2014-08-19 19:01:47 -07:00
Dmitry Kovalev
a9e38537cd Merge "Cleaning up detect_transition_to_still()." 2014-08-19 18:42:52 -07:00
Alex Converse
b9aae3b95d Merge "Fix valgrind error in pickmode." 2014-08-19 17:46:27 -07:00
Minghai Shang
229f2aa5b8 Merge "[spatial svc]Add a few different encode frame tests." 2014-08-19 17:45:01 -07:00
James Zern
84992a5136 Merge "Removing unused tbd struct member." 2014-08-19 17:35:02 -07:00
James Zern
9a1d2f099c Merge "vp9_ratectrl.c : remove dead assignment to clear warning" 2014-08-19 17:09:46 -07:00
Dmitry Kovalev
b1b568e0b3 Cleaning up detect_transition_to_still().
Change-Id: I70f4bfbfb474ae0f3e34f647b596fc0269b4aa51
2014-08-19 16:01:34 -07:00
Dmitry Kovalev
8c72c79756 Merge "Adding set_rc_buffer_sizes()." 2014-08-19 16:00:38 -07:00
Dmitry Kovalev
d4adf25d96 Removing extraconfig_map from vp9_cx_iface.c.
We have just one default set of extra config values, g_usage is
never used with VP9.

Change-Id: I0f1939b0a2da4522a87a393d85be8c8acf7f364c
2014-08-19 15:59:12 -07:00
Dmitry Kovalev
c1a769d0fc Using functions from vpx_mem.h inside vpx_image.c.
Change-Id: Idfd606bf9d23c898bcdfb98fb90a23a5fdace960
2014-08-19 15:37:25 -07:00
James Zern
7b0faffcf8 Merge changes I13d975d1,I26710359
* changes:
  vpxenc: fix compile with --disable-libyuv
  vpxdec: fix compile with --disable-libyuv
2014-08-19 15:16:29 -07:00
Dmitry Kovalev
6acde6bfcd Removing unused tbd struct member.
Change-Id: I07f7aea60ef7b9904f67e12eb5b2bfeb7348a454
2014-08-19 15:00:11 -07:00
Tom Finegan
7d870bd3ed Merge "Add basic tests for the set_maps example." 2014-08-19 14:59:05 -07:00
Alex Converse
0bb9dec8f7 Fix valgrind error in pickmode.
Don't intra transform blocks past the edge of the image.

Change-Id: Ib8de6901ea339f352f6a7c9cd2af64aa2f55fa8a
2014-08-19 14:09:57 -07:00
Dmitry Kovalev
6c4fecbcf1 Adding set_mv_search_params().
Change-Id: I1d1bd2d636a03e23f7d44cda349053966f99736d
2014-08-19 11:49:05 -07:00
Dmitry Kovalev
87ee096117 Adding set_rc_buffer_sizes().
Change-Id: I76aab1e4f56a790597104b5c9b0d3543d6982eae
2014-08-19 11:33:36 -07:00
Dmitry Kovalev
215151b78d Converting MODE to be just BEST, GOOD, and REALTIME.
After that change oxcf->pass and oxcf->mode become two orthogonal variables.

Change-Id: I1501f83cd2805480e8118135e6d4045fb41e26d5
2014-08-19 10:37:12 -07:00
Dmitry Kovalev
f617889be7 Moving frame_is_boosted() to vp9_speed_features.c.
Change-Id: I9261ded5fbba7a625d8224d91be296265a932410
2014-08-19 10:31:29 -07:00
Dmitry Kovalev
86533dc72d Merge "Removing unnecessary casts to int64_t." 2014-08-19 10:28:21 -07:00
Jim Bankoski
64988cee43 Merge "vpxdec.c : resolve uninitialized member warning" 2014-08-19 09:59:23 -07:00
Jim Bankoski
83eff950ad Merge "vp9_rdopt.c : signed unsigned mismatch in choose_tx_size" 2014-08-19 09:59:10 -07:00
Jim Bankoski
0cb31c7270 Merge "vp9_firstpass.c : remove unused parm in get_zero_motion" 2014-08-19 09:58:57 -07:00
Frank Galligan
c92cb1ceb4 Merge "Improved vp9_quantize_fp_neon()" 2014-08-19 09:47:39 -07:00
Jim Bankoski
7ae326923e vpxdec.c : resolve uninitialized member warning
Change-Id: Icd07ee241b1e36bb4c9e40e592bae259e65f2ce4
2014-08-19 06:44:10 -07:00
Jim Bankoski
f274ac7df5 vp9_rdopt.c : signed unsigned mismatch in choose_tx_size
Change-Id: Id68736ffa71843b02ee3634edc525219d6e54d4f
2014-08-19 06:40:29 -07:00
Jim Bankoski
b29d517344 vp9_firstpass.c : remove unused parm in get_zero_motion
Change-Id: I803bfc0b26804912ccd088d8a90a0f02fc55eed9
2014-08-19 06:39:12 -07:00
Jim Bankoski
8f4b357d99 vp9_pickmode.c: check value that can be null to avoid warning
Change-Id: Ia023a49ddb606899171ff19f9e271ec8ed0b2fb3
2014-08-19 06:16:34 -07:00
Jim Bankoski
76e9992b9b vp9_ratectrl.c : remove dead assignment to clear warning
Change-Id: Iacc0b5c58032bde372ed9f99b87712547c4d8454
2014-08-19 06:16:25 -07:00
Jim Bankoski
d735cd5e8f Merge "vp9_firstpass.c : clear warning re: use of null lc" 2014-08-19 06:16:13 -07:00
Yunqing Wang
4d98b50be5 Merge "Add early termination in transform size search" 2014-08-18 19:00:24 -07:00
Jingning Han
af2a736953 Merge "Speed up mode search depending on relative ref frame position" 2014-08-18 17:58:43 -07:00
Dmitry Kovalev
a8cd30cb5d Removing gold_is_last, alt_is_last, gold_is_alt flags.
Change-Id: I061d3bc5b13e3819710fc29504f01f01f065032e
2014-08-18 17:11:20 -07:00
Jim Bankoski
b78f8f6700 vp9_firstpass.c : clear warning re: use of null lc
Change-Id: I80f81f6f4f15766f504c15f8c20258a84bbaa9ac
2014-08-18 16:44:48 -07:00
Yunqing Wang
ba70f16011 Add early termination in transform size search
In the full-rd transform size search, we go through all transform
sizes to choose the one with best rd score. In this patch, an
early termination is added to stop the search once we see that the
smaller size won't give better rd score than the larger size. Also,
the search starts from largest transform size, then goes down to
smallest size.

A speed feature tx_size_search_breakout is added, which is turned off
at speed 0, and on for other speeds. The transform size search is
turned on at speed 1.

Borg test results:
1. At speed 1,
   derf set: psnr gain: 0.618%, ssim gain: 0.377%;
   stdhd set: psnr gain: 0.594%, ssim gain: 0.162%;
   No noticeable speed change.
3. At speed 2,
   derf set: psnr loss: 0.157%, ssim loss: 0.175%;
   stdhd set: psnr loss: 0.090%, ssim loss: 0.101%;
   speed gain: ~4%.

Change-Id: I22535cd2017b5e54f2a62bb6a38231aea4268b3f
2014-08-18 16:27:04 -07:00
Jingning Han
6a464eca05 Speed up mode search depending on relative ref frame position
This commit enables the encoder to record the location of the
center frame to generate alter reference frame. It then allows to
skip checking prediction modes of other reference frame types when
it comes to encode this frame.

The speed 3 runtime is reduced for the test sequences:
bus at CIF 1000 kbps, 9791 ms -> 9446 ms, i.e., 3.5% speed-up,
pedestrian at 1080p 2000 kbps, 184043 ms -> 175730 ms, i.e., 4.5%
speed-up.

No compression performance change observed.

Change-Id: Iacfde3bcc1445964e7a241f239bd6ea11cb94bd1
2014-08-18 16:06:54 -07:00
Dmitry Kovalev
4590c5e508 Adding gf_group temp variable.
Change-Id: Ifc33b1e893480b36ccf88794ea15e87e5cc97823
2014-08-18 15:56:26 -07:00