Jingning Han [Wed, 9 Sep 2015 19:29:57 +0000 (12:29 -0700)]
Fix ioc warnings related to sub8x8 reference frame
Access scaled reference frame in the sub8x8 rate-distortion
optimization loop only when the current test mode is an inter mode.
This prevents an ioc warning triggered by sending intra_frame index
to fetch scaled reference frame.
Marco [Wed, 9 Sep 2015 00:05:31 +0000 (17:05 -0700)]
SVC sample encoder: Change some settings/update.
Change settings for 1 pass CBR.
And only use SET_SVC control(s) if there is at least 1 layer (spatial or temporal).
This allows sample encoder to also work for 1 layer case.
Jingning Han [Sun, 6 Sep 2015 01:28:16 +0000 (18:28 -0700)]
Fix the sub8x8 block inter prediction with scaled reference frame
Sync the encoder's buffer offset calculation for sub8x8 block motion
compensated prediction with scaled reference frame to match the
decoder's behavior. This resolves an enc/dec mismatch issue when
sub8x8 inter mode with scaled is turned on.
Scott LaVarnway [Wed, 2 Sep 2015 20:15:52 +0000 (13:15 -0700)]
VPX: subpixel_8t_ssse3 asm using x86inc
This is based on the original patch optimized for 32bit
platforms by Tamar/Ilya and now uses the x86inc style asm.
The assembly was also modified to support 64bit platforms.
Make update_map/temporal_update fields implicit for keyframes.
These frame types cannot make bitstream parsing depend on previous
frames, so the hypothetical combinations of e.g. keyframe=1 and
update_map=0 or keyframe=1 and temporal_update=1 are non-sensical.
Therefore, make it impossible to code such combinations in the vp10
bitstream header.
Marco [Thu, 3 Sep 2015 19:55:51 +0000 (12:55 -0700)]
Fix to dynamic resize mode under change_config().
If the encoder dynamic resize is triggered and change config()
is then called, it will reset the current (resized) codec width/height
back to the the config (unresized) width/height (which will then
prevent the resizing action from occurring in encoder_loop).
Avoid this by checking for a change in the config width/height
before resetting the cm->width/height.
Unify coding order of MC filters between blocks and frame header.
In VP9, the order for frame header was: [0] smooth, [1] regular, [2]
sharp, [3] bilinear. Per-block, the order was [0] regular, [1] smooth
and [2] sharp. For VP10, swap smooth/regular in the frame header so
that the block ordering and frame header ordering are interchangeable.
* changes:
test: limit the valid image size on OS/2
configure: add -Zhigh-mem to LDFLAGS on OS/2
configure: disable PIC on OS/2
Makefile: add $(STACKREALIGN) to CFLAGS for vp9_reconintra.c
x86inc.asm: fix NASM compilation
* changes:
Only build multithreaded functions on mt builds.
Don't build calc_psnr for high bit depth.
Enable missing dual lpf test
Remove unused VP10 functions.
Mark VP10 functions as 'INLINE'
Remove unused functions from test files
Only build append_negative_gtest_filter when it is used.
Add INLINE decoration to static test functions
James Zern [Sat, 29 Aug 2015 03:25:29 +0000 (03:25 +0000)]
Merge changes from topic 'vp9-enc-static-fns'
* changes:
vp9_mcomp: make search functions private
vp9_mbgraph: use vp9_full_pixel_search(HEX)
vp9_temporal_filter: use vp9_full_pixel_search(HEX)
vp9_firstpass: make vp9_init_subsampling private
vp9_encoder: make vp9_alloc_compressor_data private
Johann Koenig [Fri, 28 Aug 2015 23:18:18 +0000 (16:18 -0700)]
Check CONFIG_VP<N> instead of encoder and decoder
If either the encoder or the decoder is enabled, CONFIG_VP<N> will be
set. This simplifies the conditional and passes the chromium update
script when CONFIG_ values are passed in with 'yes' and 'no' values.
This was failing because it was checking against empty strings but
they are set to 'no'
Marco [Mon, 24 Aug 2015 20:00:50 +0000 (13:00 -0700)]
Fix to golden update in cyclic_refresh.
For one pass CBR: only check for updating refresh_golden
if ext_refresh_frame_flags_pending is not set (i.e., == 0).
And move the resetting of ext_refresh_frame_flags_pending = 0
down to after the encode_loop (and account for dropped frames).
This is to prevent changing refresh_golden flga when the user
supplies the reference/update flags.
Johann Koenig [Thu, 27 Aug 2015 20:58:53 +0000 (20:58 +0000)]
Merge changes I53b5bdc5,Ib81168a7,Ie0113945
* changes:
Only build ssse3 filter functions on 64 bit
Clean up unused function warnings in vp8 encoder
Clean up unused function warnings in vp8 onyx_if.c
Yaowu Xu [Tue, 25 Aug 2015 02:09:03 +0000 (19:09 -0700)]
Limit new motion vector size
This commit adds clamp of new vectors similar to the logic in RD loop.
Such clamp is not necessary from the perspective of VP8 bitstream, but
is added to improve ChromeCast mirroring's robustness.
Marco [Mon, 24 Aug 2015 21:59:54 +0000 (14:59 -0700)]
vp8: modifcatiion to skin map computation.
For each block in pickinter: use average of four middle
pixels (instead of single pixel) to set skin map.
This can help a little in reducing false skin detection in
some cases.
hui su [Mon, 24 Aug 2015 18:40:04 +0000 (11:40 -0700)]
Avoid setting rate multiplier as 0
In high bitdepth setting, the rate multipier may be set as 0. In
lossless mode, the RD cost would always be 0, resulting in bad
partition and prediction mode choices.
projects / libvpx / log