Adrian Grange [Tue, 24 Mar 2015 15:55:35 +0000 (08:55 -0700)]
Fix use of scaling in joint motion search
To enable us to the scale-invariant motion estimation
code during mode selection, each of the reference
buffers is scaled to match the size of the frame
being encoded.
This fix ensures that a unit scaling factor is used in
this case rather than the one calculated assuming that
the reference frame is not scaled.
Adrian Grange [Thu, 19 Mar 2015 21:41:43 +0000 (14:41 -0700)]
Restore first ref frame pointer to the correct value
The joint_motion_search function alternates prediction
between two reference frames. In order to reuse existing
code, a pointer to the appropriate reference frame is
written into xd->plane[0].pre[0], that the motion
estimation code assumes points to the reference frame.
If this first reference frame was scaled then the
pointer was incorrectly being reset to point to the
unscaled reference frame rather than the scaled
version.
Yunqing Wang [Thu, 19 Mar 2015 19:28:24 +0000 (12:28 -0700)]
vp8: fix a bug in the internal PSNR calculation
While CONFIG_INTERNAL_STATS=1, PSNR is calculated while encoding.
The aligned width/height were used mistakenly in the calculation.
This patch fixed it, and used the orignal image width/height.
James Zern [Thu, 19 Mar 2015 19:20:14 +0000 (12:20 -0700)]
put spatial svc behind an ABI check
this removes the CONFIG_* checks from public headers, but means
'--enable-experimental --enable-spatial-svc' builds will fail without a
local change to the ABI in vpx_encoder.h. this should be all right for
testing this experiment.
Jingning Han [Wed, 18 Mar 2015 01:40:40 +0000 (18:40 -0700)]
Speed up non-rd mode decision search
This commit makes the encoder to explicitly calculate the SAD
associated with the LAST_FRAME motion vector and compare it to
that of the GOLDEN_FRAME given by integral projection motion
estimation. It skips the expensive sub-pixel motion search over
GOLDEN_FRAME when the LAST_FRAME can provide fairly good motion
compensated prediction quality.
For dark720p speed -6 single thread goes from
33304 b/f, 40.070 dB, 18156 ms ->
33319 b/f, 40.061 dB, 17611 ms
Marco [Mon, 16 Mar 2015 02:19:29 +0000 (19:19 -0700)]
Update to variance partition.
Use force_split to constrain the partition selection.
This is used because in the top-down approach to variance partition,
a block size may be selected even though one of its subblocks may have
high variance.
In this patch the selection of the 64x64 block size will only
be allowed if the variance of all the 32x32 subblocks are also below the threshold.
Stil testing, but some visual improvement for areas near slow moving boundary
can be seen. Metrics for RTC set increase by about ~0.5%.
Jingning Han [Mon, 16 Mar 2015 19:03:31 +0000 (12:03 -0700)]
Refactor column integral projection computation
Move the scaling factor outside column projection. This avoids
repeated calculation of the same scaling factor. Profiling shows
that the percentage of vp9_int_pro_col_sse2 of overall cycles
goes from 2.29% down to 1.88%.
Yaowu Xu [Mon, 16 Mar 2015 16:42:33 +0000 (09:42 -0700)]
vp9_pick_inter_mode(): minor optimizations
1. remove duplicate initialization to mbmi->interp_filter.
2. move mv clamping into ref_frame loop instead of mode checking loop.
3. move the check if last frame is same as golden frame earlier to
avoid initialization of Golden reference related variables.
James Zern [Sat, 14 Mar 2015 01:52:11 +0000 (18:52 -0700)]
usage.dox: fix encoder/decoder subpage list indent
use \li to separate the list items contained in conditionals. this
avoids the encode page becoming a sub-item of decode; likely a problem
in <1.8.3.
+ fix encoder conditional, spelling error
+ correct encode page name to match decode 'Encoding'
James Zern [Sat, 14 Mar 2015 01:49:03 +0000 (18:49 -0700)]
usage.dox: fix doxygen warnings in 1.8.x
use \li to denote list items with \if.
fixes the following likely visible in <1.8.3:
usage.dox: warning: Invalid list item found
usage.dox: warning: End of list marker found without any preceding list items
Jingning Han [Fri, 13 Mar 2015 23:51:29 +0000 (16:51 -0700)]
Fix choose_partitioning threshold setup for speed -5
The compression performance of speed -5 is on average 12.6% better
than speed -6. At lower bit-rates, the gains are typically 20% or
more. For 2-thread encoding, the speed -5 takes about 1.6x time of
speed -6.
Yaowu Xu [Fri, 13 Mar 2015 01:14:43 +0000 (18:14 -0700)]
Added comment for vp9 only codec controls
Comments are updated to reflect that these controls apply to VP9 only,
thereby, to insure the document produced by doxygent to reflect the
same fact too.
Jingning Han [Thu, 12 Mar 2015 23:18:12 +0000 (16:18 -0700)]
Reduce the number of full block SAD calls
This commit uses a 6-point 1-step refine motion search in the
integral projection based full pixel motion estimation, to replace
the current 9-point search.
It reduces runtime cost of speed -6 on some noisy clips, e.g.,
dark720p single thread
33314 b/f, 40.076 dB, 18231 ms ->
33307 b/f, 40.067 dB, 17768 ms
The compression performance for rtc set remains unchanged.
Yaowu Xu [Wed, 11 Mar 2015 17:16:39 +0000 (10:16 -0700)]
vp9_pick_inter_mode(): Use single loop to evaluate inter modes
This commit changes to use single loop to evaluate all inter modes.
There is no impact on compression quality and speed, but allow future
experiment with the order of modes evaluated.
Yunqing Wang [Wed, 11 Mar 2015 18:31:47 +0000 (11:31 -0700)]
Minorly modify model_rd_for_sb_y function
Added a skip_dc check. If skip_dc = 1, we could eliminate calling
of vp9_model_rd_from_var_lapndz(). This gave slight PSNR & SSIM
gain(<0.1%), and no speed change.
projects / libvpx / log