Commit Graph

10187 Commits

Author SHA1 Message Date
Jingning Han
2c1cdf69b6 Fix a potential overflow issue in inverse 16x16 full 2D-DCT
An overflow issue could potentially happen in the second round 1-D
transform of the SSSE3 full inverse 16x16 2D-DCT. This commit fixes
this issue.

Change-Id: Ia19e4888fda1cc929a28a5f89a5beec612d628dc
2014-05-29 11:46:32 -07:00
Yaowu Xu
2e6040daca Merge "Fixing -Wextra warnings in vp9_{cx, dx}_iface.c." 2014-05-29 09:09:58 -07:00
Yaowu Xu
d553cc10dc Merge "Fixed a crash windows build" 2014-05-29 08:16:19 -07:00
Yaowu Xu
43414f3f7b Fixed a crash windows build
Change-Id: I58baa1da1f3bfc8a6da454399139fe6a7473ff10
2014-05-28 15:50:50 -07:00
Marco Paniconi
688c5ac2a8 Merge "vp8 denoiser: fix to zero_mv mode selection." 2014-05-28 12:32:44 -07:00
Marco Paniconi
609e91f9b7 vp8 denoiser: fix to zero_mv mode selection.
In the current logic, if the sse for zero motion is smaller
than the sse for new_mv (i.e., best_sse), we may still end up
using the non-zero mv for denoising (if the magnitude of new_mv is above threshold).
This can happen for very noisy content, and can lead to artifacts.

This change ensures that we always use zero_mv (over new_mv) for
denoisng if sse_zero_mv <= best_sse.

Change-Id: I8ef9294d837b077013b77a46c9a71d17c648b48a
2014-05-28 11:23:52 -07:00
Dmitry Kovalev
852fcbcc68 Fixing -Wextra warnings in vp9_{cx, dx}_iface.c.
Change-Id: I0abad32551dc534d3db27424c118e4b2f6b50f37
2014-05-28 11:15:43 -07:00
Dmitry Kovalev
39b9731876 Merge "Using 2 instead of 3 elements for last_q array." 2014-05-28 10:57:40 -07:00
Dmitry Kovalev
377950f111 Merge "Removing redundant vp9_zero() call." 2014-05-28 10:55:12 -07:00
Dmitry Kovalev
5023627cb4 Merge "Cleaning up vp9_variance_sse2.c." 2014-05-28 10:50:46 -07:00
Alex Converse
f9501295c9 Merge "Always allow ZEROMV when SEG_LVL_SKIP is on." 2014-05-28 10:19:49 -07:00
Alex Converse
8a69cef042 Merge "Fix the all intra modes mask constant." 2014-05-28 10:19:18 -07:00
Paul Wilkins
15600eb8b8 Merge "Removing this_frame_stats member from TWO_PASS struct." 2014-05-28 08:07:50 -07:00
Paul Wilkins
8df1b869a2 Merge "Remove brightness weighting in two pass." 2014-05-28 02:04:29 -07:00
Deb Mukherjee
5c93c580f8 Removing undeclared identifier - build fix
Fixes build with --enable-internal-stats

Change-Id: I137169c859f561478e45891defe976d595454166
2014-05-27 23:24:06 -07:00
Dmitry Kovalev
edccfcebb2 Using 2 instead of 3 elements for last_q array.
Change-Id: I2c6950e7d79fc89c6f97e6dcf47317ef66c453a5
2014-05-27 18:19:19 -07:00
Alex Converse
6fbbb33aaf Always allow ZEROMV when SEG_LVL_SKIP is on.
Change-Id: I6db1dc82f66438ac48f571d2f1a2ac7c39a97a1a
2014-05-27 18:17:17 -07:00
Alex Converse
75d77e36db Fix the all intra modes mask constant.
The new constant expands to 0x3fc00808.

Change-Id: Ib5109e4faf035fe0402b59f8a8d2e412628b9276
2014-05-27 18:17:17 -07:00
Dmitry Kovalev
0becfe42bb Merge "Removing ctrl_id parameter from vpx_codec_control_fn_t." 2014-05-27 17:35:38 -07:00
Dmitry Kovalev
3f95a230c7 Removing this_frame_stats member from TWO_PASS struct.
Change-Id: Id8877fad1f1e88b145e7c40c43174109b9c4f373
2014-05-27 17:09:28 -07:00
Jingning Han
d5bcef5242 Merge "Fix compiling error in MSVS" 2014-05-27 16:58:00 -07:00
Dmitry Kovalev
8a8b662eaa Removing ctrl_id parameter from vpx_codec_control_fn_t.
Change-Id: I2b61c8c17ded1074dea92b4f6ad9be84d128b52a
2014-05-27 16:45:58 -07:00
Dmitry Kovalev
df6f618079 Removing redundant vp9_zero() call.
rd.tx_select_threshes is cleared in encode_frame_internal().

Change-Id: Ie03776a41c585f13b392a9b62d4e91ef26ebeaf0
2014-05-27 16:24:01 -07:00
Jingning Han
239e68ddbf Fix compiling error in MSVS
Need to include math.h before tmmintrin.h in some versions of MSVS.

Change-Id: Ia6b83ae599316887ecf30c4e4b9e4355fb8a4219
2014-05-27 15:58:47 -07:00
Yaowu Xu
32228ac13a Merge "vp9_rdopt.c: Removed 2 unused parameters" 2014-05-27 15:52:50 -07:00
Marco Paniconi
567746d3f0 Merge "Fix to reduce block artifacts from vp8 temporal denoiser." 2014-05-27 15:32:16 -07:00
Dmitry Kovalev
1349e8634c Merge "Converting target_bandwidth to Bit/s at very beginning." 2014-05-27 15:02:21 -07:00
Yaowu Xu
2d6bd24b96 Merge "Revert "Making vp9_get_sse_sum_{8x8, 16x16} static."" 2014-05-27 14:53:49 -07:00
Yaowu Xu
4c9843cbef vp9_rdopt.c: Removed 2 unused parameters
Change-Id: I935ec0e78570ce3d3585f972350e39043eefa30a
2014-05-27 14:45:19 -07:00
Erik Niemeyer
c92d776c32 Merge "Turn on unit tests for AVX2 convolve functions" 2014-05-27 14:00:39 -07:00
Dmitry Kovalev
a789bfec87 Cleaning up vp9_variance_sse2.c.
Change-Id: I5ec336848f6489c31cf2b645026fa2025db07466
2014-05-27 13:53:19 -07:00
Yunqing Wang
1f2200080b Revert "Making vp9_get_sse_sum_{8x8, 16x16} static."
This reverts commit e8bbb3d9db.

Change-Id: Ie368d36fd249d323d859d208609c711f04537bbc
2014-05-27 13:37:08 -07:00
Deb Mukherjee
444f93945b Merge "Remove Wextra warnings from vp9_sad.c" 2014-05-27 11:54:05 -07:00
Yunqing Wang
a591ac9e5a Merge "Fix decoder mismatch in sub-pixel AVX2 intrinsic filters" 2014-05-27 10:52:16 -07:00
Dmitry Kovalev
bf503e5236 Merge "Reusing rd_less_than_thresh() function." 2014-05-27 10:50:55 -07:00
Dmitry Kovalev
ed784fd169 Merge "Removing vp8/common/pragmas.h." 2014-05-27 10:50:31 -07:00
Yunqing Wang
4f0943b996 Turn on unit tests for AVX2 convolve functions
This patch turned on unit tests for AVX2 convolve functions.

Change-Id: I51b8bfdaa290fb22862c68af61abf2394d00d47c
2014-05-27 10:36:56 -07:00
Scott LaVarnway
e767cdeda3 Merge "neon matches "C" when using increase_denoising" 2014-05-27 09:01:48 -07:00
Paul Wilkins
f085d128f7 Remove brightness weighting in two pass.
This code dates from the ancient past and
applied an error score weighting based on pixel
brightness. This not seem to be providing any
benefit metrics wise and could be making some
visual issues in dark frames worse.

The field is left in place in the FIRSTPASS_STATS data
structure in this patch, pending changes to unit tests that
use a pre-defined first pass file.

Change-Id: Id50f04205230234858e7548ce523f11acaf3567d
2014-05-27 13:27:49 +01:00
Scott LaVarnway
03de5a38e2 neon matches "C" when using increase_denoising
If increase_denoising is set,
vp8_denoiser_filter_neon() produced incorrect results.

Change-Id: I645f78e48b8f6657fa8a4b69d2c4d3488a0581dc
2014-05-26 08:06:25 -07:00
Paul Wilkins
debd048531 Merge "Further first pass allocation changes." 2014-05-25 14:48:36 -07:00
Paul Wilkins
620ce56154 Merge "Re-factor bit allocation in first pass." 2014-05-25 14:47:35 -07:00
Dmitry Kovalev
3fff4bd2df Converting target_bandwidth to Bit/s at very beginning.
Change-Id: I1d8c9fe4228e2f1ef67a66883694842a9545e7b9
2014-05-23 18:11:07 -07:00
levytamar82
773596050f Fix decoder mismatch in sub-pixel AVX2 intrinsic filters
The subpixel SSSE3 was fixed in this patch:
https://gerrit.chromium.org/gerrit/#/c/70283/
So the equivalent AVX2 is fixed accordingly.

Change-Id: Ieebbc1949c99d34b12b8b47692df71aca5001f3a
2014-05-23 16:48:40 -07:00
Jingning Han
26a76dc3f8 Merge changes I4f0f4378,I02250d13
* changes:
  Re-enable unit test on neon implementation of 16x16 idct
  Turn on unit test for inverse 16x16 2D-DCT ssse3
2014-05-23 16:02:09 -07:00
Jingning Han
59c3f446fe Merge "Inverse 16x16 2D-DCT SSSE3 implementation" 2014-05-23 16:01:22 -07:00
Jingning Han
74a854fcab Re-enable unit test on neon implementation of 16x16 idct
The previous change only tunes forward transform. It doesn't affect
the neon implementation of the inverse transform. Hence turn the
unit test on.

Change-Id: I4f0f43783b98814d1eee53182209f9669d538140
2014-05-23 15:11:16 -07:00
Jingning Han
e4a758bbd4 Turn on unit test for inverse 16x16 2D-DCT ssse3
Change-Id: I02250d1364ca2e637585a8db020cba9fd4a95664
2014-05-23 15:11:05 -07:00
Jingning Han
48b0891370 Inverse 16x16 2D-DCT SSSE3 implementation
This commit enables the SSSE3 implementation of full inverse 16x16
2D-DCT. The unit runtime goes down from 1642 cycles to 1519 cycles,
about 7% speed-up.

Change-Id: I14d2fdf9da1fb4ed1e5db7ce24f77a1bfc8ea90d
2014-05-23 15:09:35 -07:00
Yaowu Xu
7e673cb641 Merge "Use extreme values for input in convovle tests" 2014-05-23 14:45:30 -07:00