Commit Graph

421 Commits

Author SHA1 Message Date
Paul Wilkins
befb0393c5 Fix crash when --tune=ssim is selected.
Crash fix only. No functional change or testing.

Change-Id: I0c6d114d024c29fc11ae61666f5938f11b01dd6a
2013-03-28 12:48:30 +00:00
John Koleszar
28d9202ed4 Merge "Cleaning up rate control code." into experimental 2013-03-27 12:29:00 -07:00
Ronald S. Bultje
7c70145914 Merge "Add col/row-based coefficient scanning patterns for 1D 8x8/16x16 ADSTs." into experimental 2013-03-26 19:17:08 -07:00
Ronald S. Bultje
3c77ab4c0f Merge "Redo banding for all transforms." into experimental 2013-03-26 19:16:44 -07:00
Ronald S. Bultje
c6efbbcfe4 Merge "Use above/left (instead of previous in scan-order) as token context." into experimental 2013-03-26 19:16:24 -07:00
Deb Mukherjee
23144d2345 Implicit weighted prediction experiment
Adds an experiment to use a weighted prediction of two INTER
predictors, where the weight is one of (1/4, 3/4), (3/8, 5/8),
(1/2, 1/2), (5/8, 3/8) or (3/4, 1/4), and is chosen implicitly
based on consistency of the predictors to the already
reconstructed pixels to the top and left of the current macroblock
or superblock.

Currently the weighting is not applied to SPLITMV modes, which
default to the usual (1/2, 1/2) weighting. However the code is in
place controlled by a macro. The same weighting is used for Y and
UV components, where the weight is derived from analyzing the Y
component only.

Results (over compound inter-intra experiment)
derf: +0.18%
yt: +0.34%
hd: +0.49%
stdhd: +0.23%

The experiment suggests bigger benefit for explicitly signaled weights.

Change-Id: I5438539ff4485c5752874cd1eb078ff14bf5235a
2013-03-26 16:58:56 -07:00
Ronald S. Bultje
d9094d8fd3 Add col/row-based coefficient scanning patterns for 1D 8x8/16x16 ADSTs.
These are mostly just for experimental purposes. I saw small gains (in
the 0.1% range) when playing with this on derf.

Change-Id: Ib21eed477bbb46bddcd73b21c5c708a5b46abedc
2013-03-26 16:46:13 -07:00
Ronald S. Bultje
3120dbddb1 Redo banding for all transforms.
Now that the first AC coefficient in both directions use the same DC
as their context, there no longer is a purpose in letting both have
their own band. Merging these two bands allows us to split bands for
some of the very high-frequency AC bands.

In addition, I'm redoing the banding for the 1D-ADST col/row scans. I
don't think the old banding made any sense at all (it merged the last
coefficient of the first row/col in the same band as the first two of
the second row/col), which was clearly an oversight from the band being
applied in scan-order (rather than in their actual position). Now,
coefficients at the same position will be in the same band, regardless
what scan order is used. I think this makes most sense for the purpose
of banding, which is basically "predict energy for this coefficient
depending on the energy of context coefficients" (i.e. pt).

After full re-training, together with previous patch, derf gains about
1.2-1.3%, and hd/stdhd gain about 0.9-1.0%.

Change-Id: I7a0cc12ba724e88b278034113cb4adaaebf87e0c
2013-03-26 16:46:13 -07:00
Ronald S. Bultje
790fb13215 Use above/left (instead of previous in scan-order) as token context.
Pearson correlation for above or left is significantly higher than for
previous-in-scan-order (absolute values depend on position in scan, but
in general, we gain about 0.1-0.2 by using either above or left; using
both basically just makes this even better). For eob branch skipping,
we continue to use the previous token in scan order.

This helps about 0.9% on derf after re-training on a limited data set.
Full re-training and results on larger-resolution clips are pending.

Note that this commit breaks trellis, so we can probably get further
gains out of it by fixing trellis at some later point.

Change-Id: Iead68e296fc3a105cca746b5e3da9555d6010cfe
2013-03-26 16:46:09 -07:00
Dmitry Kovalev
77c664ade3 Cleaning up rate control code.
Lower case variable names, declaration and initialization on the same line,
removing redundant casts to double.

Change-Id: I7ea3905bed827aa6faac11a78401b85e448b57f9
2013-03-26 11:25:58 -07:00
John Koleszar
441e2eab1b Add an in-loop deringing experiment
Adds a per-frame, strength adjustable, in loop deringing filter. Uses
the existing vp9_post_proc_down_and_across 5 tap thresholded blur
code, with a brute force search for the threshold.

Results almost strictly positive on the YT HD set, either having no
effect or helping PSNR in the range of 1-3% (overall average 0.8%).
Results more mixed for the CIF set, (-0.5 min, 1.4 max, 0.1 avg).
This has an almost strictly negative impact to SSIM, so examining a
different filter or a more balanced search heuristic is in order.

Other test set results pending.

Change-Id: I5ca6ee8fe292dfa3f2eab7f65332423fa1710b58
2013-03-26 08:23:24 -07:00
Deb Mukherjee
49dcc71493 Merge "Modeling default coef probs with distribution" into experimental 2013-03-26 07:13:13 -07:00
Deb Mukherjee
fd18d5dffe Modeling default coef probs with distribution
Replaces the default tables for single coefficient magnitudes with
those obtained from an appropriate distribution. The EOB node
is left unchanged. The model is represeted as a 256-size codebook
where the index corresponds to the probability of the Zero or the
One node. Two variations are implemented corresponding to whether
the Zero node or the One-node is used as the peg. The main advantage
is that the default prob tables will become considerably smaller and
manageable. Besides there is substantially less risk of over-fitting
for a training set.

Various distributions are tried and the one that gives the best
results is the family of Generalized Gaussian distributions with
shape parameter 0.75. The results are within about 0.2% of fully
trained tables for the Zero peg variant, and within 0.1% of the
One peg variant.

The forward updates are optionally (controlled by a macro)
model-based, i.e. restricted to only convey probabilities from the
codebook. Backward updates can also be optionally (controlled by
another macro) model-based, but is turned off by default. Currently
model-based forward updates work about the same as unconstrained
updates, but there is a drop in performance with backward-updates
being model based.

The model based approach also allows the probabilities for the key
frames to be adjusted from the defaults based on the base_qindex of
the frame. Currently the adjustment function is a placeholder that
adjusts the prob of EOB and Zero node from the nominal one at higher
quality (lower qindex) or lower quality (higher qindex) ends of the
range. The rest of the probabilities are then derived based on the
model from the adjusted prob of zero.

Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-25 23:43:38 -07:00
Paul Wilkins
99a4939ec3 Merge "Disable zero bin mode boost." into experimental 2013-03-22 10:59:43 -07:00
Paul Wilkins
bfe7666142 Merge "Minor code clean up" into experimental 2013-03-22 10:53:12 -07:00
Paul Wilkins
ec080fa9de Disable zero bin mode boost.
As things stand the zero bin mode boost is hurting somewhat.
In part this seems to be because the boost applied as is
interferes with the rd mode selection loop.

Average gains (derf 0.072, yt 0.243, ythd 0.179 std-hd 0.212%)

Change-Id: Icaecea3908d9a7352370e49b8fa822f2c2c49dc1
2013-03-22 17:43:43 +00:00
Paul Wilkins
815734e5fb Minor code clean up
Change-Id: Ifa864e0acb253b238b03cdeed0fe5d6ee30a45d8
2013-03-22 17:42:45 +00:00
Dmitry Kovalev
56f3a2c663 Code cleanup: lower case variable names.
Renaming Width to width, Height to height and Version to version in
several structs and function signatures.

Change-Id: I084c3f7e747cb2ce3345aff27a3dff9b13a87543
2013-03-20 16:41:30 -07:00
Paul Wilkins
d8ffee4526 Changes to rd error_per_bit calculation.
Specifically changes to retain more precision
especially at low Q through to the point of use.

Change-Id: Ief5f010f2ca4daaabef49520e7edb46c35daf397
2013-03-18 23:07:51 +00:00
Ronald S. Bultje
a5b54d73e4 Merge "Fix ENTROPY_STATS code in vp9_tokenize.c." into experimental 2013-03-18 15:58:33 -07:00
Ronald S. Bultje
b99dce6881 Fix ENTROPY_STATS code in vp9_tokenize.c.
Change-Id: I9b4cb1e2ce6c6a99cffd473ff2fa7579bd318fcd
2013-03-18 15:39:04 -07:00
Yunqing Wang
6344c84c82 Optimize 8x8 idct function
Wrote sse2 functions of vp9_short_idct8x8 and vp9_short_idct10_8x8.
Compared to c version, the sse2 version is 2X faster. The decoder
test didn't show noticeable gain since 8x8 idct doesn't take much
of decoding time (less than 1% in my test).

Change-Id: I56313e18cd481700b3b52c4eda5ca204ca6365f3
2013-03-18 15:34:14 -07:00
John Koleszar
93529bd7c1 Merge "Replace scaling byte with explicit display size" into experimental 2013-03-18 13:02:07 -07:00
John Koleszar
8a3f55f2d4 Replace scaling byte with explicit display size
If the intended display size is different than the size the frame is
coded at, then send that size explicitly in the bitstream. Adds a new
bit to the frame header to indicate whether the extra size fields
are present.

Change-Id: I525c66f22d207efaf1e5f903c6a2a91b80245854
2013-03-18 12:02:20 -07:00
Paul Wilkins
ef179bce61 Merge "Adapt ARNR filter length and strength." into experimental 2013-03-18 12:00:39 -07:00
John Koleszar
c5b317057b Merge "Fix pulsing issue with scaling" into experimental 2013-03-18 11:57:36 -07:00
John Koleszar
e5d7542447 Merge "Add VP9_GET_REFERENCE control" into experimental 2013-03-18 11:57:31 -07:00
Paul Wilkins
cdb322dd72 Adapt ARNR filter length and strength.
Adjust the filter length and strength for each
ARF group based on a measure of difficulty (the boost)
and the active q range.

Remove lower limit on RDMULT value.

Average gains on the different sets in range 0.4%-0.9%.
However the ARNR changes give a very big boost on a
few clips.

Eg. Soccer ~5%, in derf set and Cyclist ~ 10% in the std-hd set

Change-Id: I2078d78798e27ad2bcc2b32d703ea37b67412ec4
2013-03-18 16:17:04 +00:00
Yaowu Xu
12ade55719 Merge "removed reference to "LLM" and "x8"" into experimental 2013-03-18 08:51:19 -07:00
John Koleszar
9a56ea7e46 Merge "Remove some unused rate control variables" into experimental 2013-03-18 08:36:23 -07:00
John Koleszar
571fce6546 Merge "Fix use of NaN in firstpass" into experimental 2013-03-18 08:36:18 -07:00
Deb Mukherjee
bf7387f6b7 Merge "Context-pred fix to not use top/left on edges" into experimental 2013-03-16 19:09:25 -07:00
Deb Mukherjee
b1921b2f08 Context-pred fix to not use top/left on edges
This fix resolves some of the mismatch issues being seen
recently. While this is the right thing to do when tiling
is used for this experiment, it is not the underlying cause
of the the mismatches.
Something else is causing writing outside of the allowable
frame area in the encoder leading to this mismatch.

Change-Id: If52c6f67555aa18ab8762865384e323b47237277
2013-03-16 09:26:52 -07:00
John Koleszar
b8ac9f2f2c Remove some unused rate control variables
These variables are unused, and are subject to overflowing, causing
assertions when built with -ftrapv.

Change-Id: Ia00a3201af309906c05bcd4b23a643925ed6ea86
2013-03-15 17:53:45 -07:00
John Koleszar
db5f2cb57b Fix use of NaN in firstpass
If the second reference is better than the first in the long term,
it was possible to try to take the fractional exponent of a
negative number, giving an undefined result.

Change-Id: I1dd08286747ceae960eb03bb5d98a383cc9d253b
2013-03-15 17:53:38 -07:00
Christian Duvivier
4418b790a7 Faster vp9_short_fdct16x16.
Scalar path is about 1.5x faster (3.1% overall encoder speedup).
SSE2 path is about 7.2x faster (7.8% overall encoder speedup).

Change-Id: I06da5ad0cdae2488431eabf002b0d898d66d8289
2013-03-15 15:55:31 -07:00
Yaowu Xu
82fe8c9f36 Merge "force lossless coding at very high quality end" into experimental 2013-03-14 19:05:17 -07:00
Yaowu Xu
5d9ba7938e Merge "Remove leftover reference to 2nd order dc/ac quant" into experimental 2013-03-14 19:05:11 -07:00
Yaowu Xu
374a17366e force lossless coding at very high quality end
Change-Id: I75fc4eee10bee9efd419d248827290cce8e6d637
2013-03-14 12:31:27 -07:00
Yaowu Xu
f4d2ad6915 Remove leftover reference to 2nd order dc/ac quant
Change-Id: Ib8dacf1d2797743569771b8f699e40e1aeb085cb
2013-03-14 10:46:15 -07:00
John Koleszar
9b7be88883 Fix pulsing issue with scaling
Updates the YV12_BUFFER_CONFIG structure to be crop-aware. The
exiting width/height parameters are left unchanged, storing the
width and height algined to a 16 byte boundary. The cropped
dimensions are added as new fields.

This fixes a nasty visual pulse when switching between scaled and
unscaled frame dimensions due to a mismatch between the scaling
ratio and the 16-byte aligned sizes.

Change-Id: Id4a3f6aea6b9b9ae38bdfa1b87b7eb2cfcdd57b6
2013-03-13 19:10:10 -07:00
John Koleszar
b3c350a1a9 Add VP9_GET_REFERENCE control
This is like VP8_COPY_REFERENCE, but returns a pointer to the reference
frame rather than a copy of it. This is useful when the application
doesn't know what the size of the reference is, as is the case when
scaling is in effect.

Change-Id: I63667109f65510364d0e397ebe56217140772085
2013-03-13 19:08:06 -07:00
Jingning Han
76c12ab9c9 Support +/-2048 motion vector coding
Enable entropy coding of motion vectors up to +/-2048. Also
extend the motion search range accordingly.

Change-Id: Iac2bb015e8934521cef83a19edbe967d9f097436
2013-03-13 14:08:27 -07:00
Yaowu Xu
005552639b removed reference to "LLM" and "x8"
The commit changed the name of files and function to remove obselete
reference to LLM and x8.

Change-Id: I973b20fc1a55149ed68b5408b3874768e6f88516
2013-03-13 08:35:46 -07:00
Paul Wilkins
a2c6f6e945 Merge "disambiguate superframe index in vp9_stop_encode()" into experimental 2013-03-12 16:00:30 -07:00
John Koleszar
872fc3ded8 disambiguate superframe index in vp9_stop_encode()
If the bool-coded partition naturally ends in a byte that matches the
superframe index marker, it could lead to a parse error. This commit
ensures that if such a marker is seen, it is padded out with an
additional zero byte to disambiguate it.

Change-Id: Id977de05745b6fa9ef08afb71e210a2a3ecca02e
2013-03-12 14:30:23 -07:00
Paul Wilkins
21ba242bfd Merge "Change buffer update rules on ARF overlay." into experimental 2013-03-12 11:07:20 -07:00
Paul Wilkins
49d1425d19 Merge "Changes to maximum gf/arf interval." into experimental 2013-03-12 09:59:43 -07:00
Paul Wilkins
8be3056c45 Change buffer update rules on ARF overlay.
When coding the frame that corresponds to the midpoint frame
defining an ARF, do not update the last reference frame buffer.
Previously this buffer was updated meaning that when coding the next
ARF all the reference buffers were the same (or nearly so).
Turning the update off means that the frame before is still available
as an alternative predictor and for use in compound prediction.

Also fixed inconsistency in test for mismatch (patch from JK).

Net average gains (derf 0.049, yt 0.163, yt-hd 0.207, std-hd 0.286)

Change-Id: Ifee21da21ccbb1648ac2eafe890d3ce60562c7bc
2013-03-12 16:57:39 +00:00
Dmitry Kovalev
ff553ba113 Merge "Code cleanup." into experimental 2013-03-11 17:22:21 -07:00