Johann Koenig [Tue, 20 Mar 2018 17:06:05 +0000 (17:06 +0000)]
Merge "build: remove stale .git files"
Johann Koenig [Tue, 20 Mar 2018 17:05:17 +0000 (17:05 +0000)]
Merge "reland "use intrinsics for 'emms'""
Johann [Mon, 19 Mar 2018 22:57:29 +0000 (15:57 -0700)]
build: remove stale .git files
These were used for an older style of Visual Studio configurations.
Change-Id: I51f07b30ad51c4da0c5caf1ede36cdb69b2d2b19
Johann [Wed, 14 Mar 2018 23:57:57 +0000 (16:57 -0700)]
reland "use intrinsics for 'emms'"
Only target 32bit builds. Visual Studio does not define _mm_empty for
64bit configurations.
Rename emms.asm and remove from 32 bit builds to avoid empty file
warnings.
Don't check register state on 64bit builds.
BUG=webm:1500
This reverts commit
60beb781c140b61c1957abd2a6717d2e9a831933.
Change-Id: I5ac4cf6c67249ff24f7da19792144de20527bfce
Jerome Jiang [Mon, 19 Mar 2018 19:00:55 +0000 (19:00 +0000)]
Merge "VP8: Fix out of range index for mvcost."
James Zern [Sat, 17 Mar 2018 00:47:56 +0000 (17:47 -0700)]
CopyFrameTest: reduce max size for 32-bit targets
avoids potential OOM when allocating 3 buffers for 16383x16383; 3840 is
used as a replacement
this test was missed in:
215bddf32 vpx_scale_test: reduce max size for 32-bit targets
Change-Id: I515adf5999c6ef1724394ccd62d677134bd35e6d
Jerome Jiang [Wed, 21 Feb 2018 23:28:54 +0000 (15:28 -0800)]
VP8: Fix out of range index for mvcost.
Clamp index between 0 and MVvals.
Bit exact for speed -8, -6 and -4 on RTC set.
BUG=b/
72510002
Change-Id: I61bdb02a0924e157b3c1980f74fbbfe5ce51bc44
James Zern [Sat, 17 Mar 2018 00:47:56 +0000 (17:47 -0700)]
vpx_scale_test: reduce max size for 32-bit targets
avoids potential OOM when allocating 3 buffers for 16383x16383; 3840 is
used as a replacement
Change-Id: I92116ab69b10db6820fc651d3626bd9699700208
Marco Paniconi [Wed, 14 Mar 2018 17:03:31 +0000 (10:03 -0700)]
vp9-svc: Frame dropper for SVC.
If a given spatial layer decides to drop, due to the
buffer/overshoot conditions for that layer, then drop
that current spatial layer and all spatial layers above.
In the current implementation the svc frame counter
(and hence the pattern for the non-flexible SVC case)
are updated on frame drops.
Also add last spatial layer encoded to the pkt.
This is useful for RTC applications that enable
frame dropping for SVC.
Update to the SVC datarate tests:
enabled frame dropper on all SVC datarate tests, and
made a fix to properly set the temporal_layer_id, which
works now even on frame drops.
Change-Id: If828c193f3cb6b1839803fd52fe9fbbda5b5a039
James Zern [Fri, 16 Mar 2018 03:52:02 +0000 (03:52 +0000)]
Merge "libs.mk,vcxproj generation: split srcs in invocation"
James Zern [Fri, 16 Mar 2018 02:11:52 +0000 (02:11 +0000)]
Merge "Revert "vp9_loopfilter.c: zero lfl_uv""
James Zern [Thu, 15 Mar 2018 21:43:29 +0000 (21:43 +0000)]
Revert "vp9_loopfilter.c: zero lfl_uv"
This reverts commit
13d0955b250bcb7eac99034e7b1677d3d026b569.
Reason for revert:
this should be investigated further to ensure the memset is really
necessary outside of the static analysis pass.
Original change's description:
> vp9_loopfilter.c: zero lfl_uv
>
> The initialization depends on cm and mi_row which static
> analysis does not approve of.
>
> Clears a static analysis warning:
> warning: The right operand of '+' is a garbage value
> const loop_filter_thresh *lfi = lfthr + *lfl;
>
> Change-Id: I8c863ced2b1e9a7e10103b7281098f20941a6ca2
TBR=johannkoenig@google.com,marpan@google.com,builds@webmproject.org,jianj@google.com
Change-Id: Icadb6438fbcddba747622f06f2eadebdb333edf6
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Johann Koenig [Thu, 15 Mar 2018 21:16:43 +0000 (21:16 +0000)]
Merge changes I5501d0d6,I8c863ced,I19895d06,Ifa39353d,I09bd209b
* changes:
vp9_resize.c: assert vp9_highbd_resize_plane conditions
vp9_loopfilter.c: zero lfl_uv
vp8 rdopt.c: zero rd.[rate_uv|distortion_uv]
vp8 mfqe: zero map[]
temporal svc: zero layer_target_bitrate
Johann Koenig [Thu, 15 Mar 2018 19:26:16 +0000 (19:26 +0000)]
Merge "remove spatial svc experiment"
Linfeng Zhang [Thu, 15 Mar 2018 19:21:36 +0000 (19:21 +0000)]
Merge "Add vp9_highbd_iht16x16_256_add_neon()"
James Zern [Tue, 27 Jun 2017 01:55:09 +0000 (18:55 -0700)]
libs.mk,vcxproj generation: split srcs in invocation
this avoids truncation under mingw which would result in link failures
BUG=webm:1434
Change-Id: I6eb45d94f02966532b3cdf02860a5bf2e5d3efef
Linfeng Zhang [Thu, 15 Mar 2018 18:05:08 +0000 (18:05 +0000)]
Merge changes I9e0bf2c7,I695b4090
* changes:
Fix a bug in vp9_highbd_iht8x8_64_add_neon
Fix a bug in vp9_highbd_iht4x4_16_add_neon()
Marco Paniconi [Thu, 15 Mar 2018 00:00:27 +0000 (17:00 -0700)]
vp9-svc: Bugfix to dyanmic enabling/disabling of layers.
Fix a bug when middle and top spatial layer are skip encoded
(disabled) and then re-enabled again, during the sequence.
Issue is that pending_frame_count in the packing may
be incremented on middle layer, even though that layer is skipped
(not encoded and hence zero size). Fix is to add size check.
Modified existing unitest to reproduce the issue.
Change-Id: I86d806a112d468e06b04fbf7c46ae07db9e0ad93
Johann [Wed, 14 Mar 2018 21:45:23 +0000 (14:45 -0700)]
vp9_resize.c: assert vp9_highbd_resize_plane conditions
Clears static analysis warnings similar to the low bitdepth version:
commit
c4367b9b51782969d3b37ead6694d19ef9bfb58c
Author: James Zern <jzern@google.com>
Date: Wed Mar 18 14:34:30 2015 -0700
vp9_resize_plane: quiet some static analysis warnings
Change-Id: I5501d0d6ad7c7720d746d53ec07078cb9051d0d7
Johann [Tue, 13 Mar 2018 00:23:41 +0000 (17:23 -0700)]
remove spatial svc experiment
Change-Id: Ifda11caaf992d10f2d93d6cd1d07b79b6047be05
Johann Koenig [Wed, 14 Mar 2018 22:00:08 +0000 (22:00 +0000)]
Merge "spatial svc: set window_size to 15"
Johann [Wed, 14 Mar 2018 21:22:08 +0000 (14:22 -0700)]
vp9_loopfilter.c: zero lfl_uv
The initialization depends on cm and mi_row which static
analysis does not approve of.
Clears a static analysis warning:
warning: The right operand of '+' is a garbage value
const loop_filter_thresh *lfi = lfthr + *lfl;
Change-Id: I8c863ced2b1e9a7e10103b7281098f20941a6ca2
Johann [Wed, 14 Mar 2018 20:25:05 +0000 (13:25 -0700)]
vp8 rdopt.c: zero rd.[rate_uv|distortion_uv]
These values are not consistently set before calling update_best_mode.
In vp9_rdopt.c they are individual values instead of a struct and are
zero'd at declaration.
Clears a static analysis warning:
warning: The right operand of '-' is a garbage value
RDCOST(x->rdmult, x->rddiv, (rd->rate2 - rd->rate_uv - other_cost),
warning: The right operand of '-' is a garbage value
(rd->distortion2 - rd->distortion_uv));
Change-Id: I19895d062e7c0ac67937126ebc5dcb0afd3a2931
Johann [Wed, 14 Mar 2018 20:10:08 +0000 (13:10 -0700)]
vp8 mfqe: zero map[]
The loop appears to set map[i] with the intention of running
the 'j' loop up to that point. However, without zero'ing map[]
first the behavior is unpredictable.
Fixes a static analysis warning:
warning: Branch condition evaluates to a garbage value
for (j = 0; j < 4 && map[j]; ++j) {
Change-Id: Ifa39353d8aa5cc47b467a7d3d8cdd3b5319fd997
Johann [Wed, 14 Mar 2018 19:52:49 +0000 (12:52 -0700)]
temporal svc: zero layer_target_bitrate
These values are set in main() from user input. Ensure
they are cleared out first.
Clears a static analysis warning:
warning: The right operand of '*' is a garbage value
1000.0 * rc->layer_target_bitrate[0] / rc->layer_framerate[0];
Change-Id: I09bd209be5aff31b87597a24d37a9673fa99381b
Johann Koenig [Wed, 14 Mar 2018 19:06:57 +0000 (19:06 +0000)]
Revert "use intrinsics for 'emms'"
This reverts commit
118a57045bf5b49ab7c2f7f930543b9217fd422e.
Reason for revert: Fails on Visual Studio builds:
vpxmdd.lib(vpx_ports_emms_mmx.obj) : error LNK2019: unresolved
external symbol _m_empty referenced in function
vpx_clear_system_state
Original change's description:
> use intrinsics for 'emms'
>
> BUG=webm:1500
>
> Change-Id: I3235d8c2abc01dd3a35e14c5cbcfe20283ff8fb2
Change-Id: Ia9c40bc103c57cced83353249c55218eaf2f0b0c
Marco Paniconi [Wed, 14 Mar 2018 20:13:34 +0000 (20:13 +0000)]
Merge "vp9-svc: Fix to update layer counters when layer is skipped."
Johann [Wed, 14 Mar 2018 01:31:26 +0000 (18:31 -0700)]
spatial svc: set window_size to 15
Static analysis does not recognize that output_rc_stat guards
the usage of window_size. Clears this warning:
The right operand of '>' is a garbage value
if (frame_cnt > (unsigned int)rc.window_size) {
set_rate_control_stats sets window_size to 15. Zeroing it
just introduces another static analysis warning.
Change-Id: Ieee7b81a385f986e42189101cfa39279e519b368
Marco Paniconi [Wed, 14 Mar 2018 18:49:25 +0000 (11:49 -0700)]
vp9-svc: Fix to update layer counters when layer is skipped.
Update layer counters when layer is skipped,
for any spatial layer.
Change-Id: Ie37c4a16ccafdef3390b651dec473beb5d926896
Johann [Wed, 14 Mar 2018 01:22:40 +0000 (18:22 -0700)]
spatial svc: zero sizes
This should be taken care of by parse_superframe_index but
the static analysis is not recognizing it because it depends
on 'marker' which is read from the bitstream.
Clears a static analysis warning:
The right operand of '*' is a garbage value
rc.layer_encoding_bitrate[layer] += 8.0 * sizes[sl];
Change-Id: I8ee48a98f907bc7b46869fd27a351f33e2e7de71
Johann [Wed, 14 Mar 2018 00:37:28 +0000 (17:37 -0700)]
spatial svc: remove vpx_svc_get_message
Print error messages as they are encountered. This was the default
behavior.
Removes a static analysis warning regarding the use of strncat:
Null pointer argument in call to string length function
As this is the only use of strncat in the library, remove it and the
associated public function.
Change-Id: Id55305c5a4d65f11da88c3a2203ff824200f526f
Linfeng Zhang [Wed, 28 Feb 2018 01:06:01 +0000 (17:06 -0800)]
Add vp9_highbd_iht16x16_256_add_neon()
BUG=webm:1403
Change-Id: I2293c11666786be276909d48ee78dacb40a89e25
Linfeng Zhang [Tue, 6 Mar 2018 00:25:08 +0000 (16:25 -0800)]
Fix a bug in vp9_highbd_iht8x8_64_add_neon
This bug was introduced in
29b6a30c.
BUG=webm:1403
Change-Id: I9e0bf2c7a01d8ff1c714c12236f7985b772b0540
Johann [Wed, 14 Mar 2018 00:11:48 +0000 (17:11 -0700)]
spatial svc: remove unused locals
Clears static analysis warning:
Value stored to 'tl' is never read
Change-Id: If047a74f508288c63d5b83ed0f3ad34f791f9312
Johann Koenig [Tue, 13 Mar 2018 23:51:08 +0000 (23:51 +0000)]
Merge "use intrinsics for 'emms'"
Linfeng Zhang [Tue, 13 Mar 2018 23:10:00 +0000 (16:10 -0700)]
Fix a bug in vp9_highbd_iht4x4_16_add_neon()
This bug was introduced in
36363304.
BUG=webm:1403
Change-Id: I695b409047e41ab7e0460981524310d78753751a
Johann [Tue, 13 Mar 2018 21:37:39 +0000 (14:37 -0700)]
spatial svc: rescope sl
sl was passed to set_frame_flags_bypass_mode, triggering
an uninitialized variable warning. Inside the function it
is only used as a local variable.
Change-Id: If743626e9e10fd41d135e3b4ad6196dc4dc90172
Johann [Mon, 5 Mar 2018 22:27:04 +0000 (14:27 -0800)]
use intrinsics for 'emms'
BUG=webm:1500
Change-Id: I3235d8c2abc01dd3a35e14c5cbcfe20283ff8fb2
Johann [Mon, 12 Mar 2018 21:54:48 +0000 (14:54 -0700)]
autodetect macOS High Sierra
Add darwin17 target
Change-Id: I349a2f6a0396c59269f567a03ae813e3e59ccefa
Johann Koenig [Mon, 12 Mar 2018 20:56:30 +0000 (20:56 +0000)]
Merge "vp8 temporal_filter: ignore return value"
Johann [Tue, 31 Jan 2017 23:45:12 +0000 (15:45 -0800)]
vp8 temporal_filter: ignore return value
Clears up static clang analysis warning regarding a dead store.
Change-Id: I6a90e6fd5f2775d933c46c7553811635bd2def21
Marco Paniconi [Mon, 12 Mar 2018 19:18:35 +0000 (12:18 -0700)]
vp9-svc: Update layer frame counters when layer is skipped.
When an enhancement spatial layer is skipped, we should check
for updating the layer frame counters.
Change-Id: Ib79d0955c62fb465f59ef2f9ac45240ae2614d7b
Marco Paniconi [Mon, 12 Mar 2018 05:09:14 +0000 (22:09 -0700)]
vp9-SVC: Fix to choose_partition when LAST ref is NULL.
This causes assert to trigger in choose_partitioning().
This can happen in some cases when enhancement layers
are enabled midway during the stream.
Change-Id: I69c3c8b4b1e3f1c7d8d7294d633ca5ddca148e8b
James Zern [Thu, 8 Mar 2018 21:13:09 +0000 (21:13 +0000)]
Merge "vpx_scale_test: add w/h output to alloc failure"
Paul Wilkins [Thu, 8 Mar 2018 13:34:15 +0000 (13:34 +0000)]
Merge "Change to KF frame boost calculation."
James Zern [Thu, 8 Mar 2018 07:07:51 +0000 (23:07 -0800)]
vpx_scale_test: add w/h output to alloc failure
Change-Id: Ib5df91d9fcd7fe973a2f7d8e73a204259beddc07
James Zern [Thu, 8 Mar 2018 03:41:27 +0000 (03:41 +0000)]
Merge changes If96fd6f1,I84b27337
* changes:
Fix a bug in vp9_iht16x16_256_add_neon()
Fix a bug in vp9_iht8x8_64_add_neon()
Johann [Tue, 6 Mar 2018 02:51:10 +0000 (18:51 -0800)]
add worst-case frame size cap
The largest frame is currently in choose_partitioning:
warning: stack frame size of 44156 bytes in function 'choose_partitioning'
but adding HBD amplifies other things:
warning: stack frame size of 51480 bytes in function 'dec_build_inter_predictors'
Add some padding for sanitizer and variances between compilers.
BUG=webm:1498
Change-Id: I0d94d4f94d25dafafca9d7484881c2ce5f8de371
Linfeng Zhang [Tue, 6 Mar 2018 00:14:35 +0000 (16:14 -0800)]
Fix a bug in vp9_iht16x16_256_add_neon()
This bug was introduced in
88c23864.
BUG=webm:1403
Change-Id: If96fd6f102be6b9bda866e55e574257287746f4a
Linfeng Zhang [Mon, 5 Mar 2018 23:16:49 +0000 (15:16 -0800)]
Fix a bug in vp9_iht8x8_64_add_neon()
This bug was introduced in
b14b616d.
BUG=webm:1403
Change-Id: I84b2733734982e52b66548850d61758c772b5494
Johann [Sun, 4 Mar 2018 20:31:04 +0000 (12:31 -0800)]
move vp8 encodeopt to block_error_sse2
The file contains sse2 implementations related to various block error
functions. Update the .mk file to include it only when sse2 is
requested.
BUG=webm:1500
Change-Id: I67b766faed425fd7a96db8541b13c69670b65fec
James Zern [Sat, 3 Mar 2018 20:47:24 +0000 (12:47 -0800)]
disable vp9_highbd_iht{4x4_16,8x8_64}_add_neon
these causes test vector failures
BUG=webm:1403
Change-Id: I08218f0bf26651eb367ece4feec6d704e0189bd8
James Zern [Sat, 3 Mar 2018 20:47:24 +0000 (12:47 -0800)]
disable vp9_iht8x8_64_add_neon
this causes test vector failures
BUG=webm:1403
Change-Id: I7d37a05fbf4641ea352c947053aa4eaeb7f5c318
James Zern [Sat, 3 Mar 2018 20:47:24 +0000 (12:47 -0800)]
disable vp9_iht16x16_256_add_neon
this causes test vector failures
BUG=webm:1403
Change-Id: Ifdb5b270c5cc70be5689e4fbda2ada3724cc65c3
Marco Paniconi [Fri, 2 Mar 2018 18:47:04 +0000 (10:47 -0800)]
vp9-svc: Disable partition_reuse unless 2x2 scale.
For SVC, if any of the layer scale ratios are not
2x2, then disable the partiton_reuse, which assumes
2x2 scaling between layers.
Change-Id: I8b3163de0826052bbb1bfe03554a074c89510558
Marco Paniconi [Thu, 1 Mar 2018 17:45:11 +0000 (09:45 -0800)]
vp9-svc: Fix to downsampling filter phase_shift.
Set phase_shift = 0 if the scale factors are
above 3/4. Removes artifact for scale factors
close to 1.
phase_shift = 8 is to get an averaging filter
(decimated pixel aligns to 8/16, midway between source pixels),
and only makes sense for scale factors multiples of
2 (1/2, 1/4,...).
Removes artifact for high scaling ratios.
Change-Id: Id0a85869d6c6156dda0032c697ded2de78fad6bd
James Zern [Thu, 1 Mar 2018 07:49:39 +0000 (23:49 -0800)]
iadst16x16_256_add_half1d: fix array size
t[] is indexed from 0..11
Change-Id: I7d0021f1795c6608354c8770843ea9dfdea66f97
paulwilkins [Wed, 21 Feb 2018 11:18:09 +0000 (11:18 +0000)]
Change to KF frame boost calculation.
This change is targeted mainly at higher resolutions where typically
the average error per MB is much smaller. hence this patch replaces
a fixed error per MB factor with a tiered value.
It also adds in a fixed offset value that acts as a minimum return score.
Note also minor fix to debug stats output.
The results are overall beneficial (-ve) on our test sets, most notably for
higher definition formats (see below - overall psnr, ssim, psnr hvs)
low res: 0.184 -0.262 -0.166
mid res: 0.094 0.075 0.049
hd res: -0.752 -0.300 -0.800
NF 2K: -0.353 1.095 -0.302
NF 4K: -1.245 -0.578 -1.205
The most notable negative case is pierseaside 2K which appears to be worse by
8-10% (which has a big impact on the overall gain for the NF 2K set). Closer
inspection reveals that the drop does not relate to the key frame boost
per se as in both cases the key frame substantially undershoots its target. Rather
this is a side effect relating to the initial Q range allowed for the key frame and
a poor initial complexity estimate. This will hopefully be improved in a later
patch.
Change-Id: I4773ebe554782f4024c047c3c392c763a3fe843b
Linfeng Zhang [Wed, 28 Feb 2018 18:26:39 +0000 (18:26 +0000)]
Merge "Add vp9_iht16x16_256_add_neon()"
James Zern [Tue, 27 Feb 2018 22:15:05 +0000 (22:15 +0000)]
Merge "datarate_test: correct last_pts_ref_ type"
Linfeng Zhang [Tue, 27 Feb 2018 01:22:36 +0000 (17:22 -0800)]
Add vp9_iht16x16_256_add_neon()
BUG=webm:1403
Change-Id: I1413cc3dfcb62143ba04fe9b0f8d8b010fdf69b6
James Zern [Tue, 27 Feb 2018 07:03:02 +0000 (23:03 -0800)]
datarate_test: correct last_pts_ref_ type
use vpx_codec_pts_t to match last_pts_; this quiets a conversion warning
under visual studio
Change-Id: I3f1c146fc13f2edfb515d76730a9ef063846bf69
Linfeng Zhang [Tue, 27 Feb 2018 01:00:10 +0000 (17:00 -0800)]
Fix a bug in create_s16x4_neon()
This bug exposes when 2nd argument is negative, and the higher 32 bits
would be all 1s.
Change-Id: I189ee8cd3753fde00a34847e7a37cde2caa4ba72
Linfeng Zhang [Sat, 24 Feb 2018 01:24:32 +0000 (01:24 +0000)]
Merge "Clean test/dct_test.cc with testing::Combine"
Linfeng Zhang [Thu, 22 Feb 2018 22:26:05 +0000 (14:26 -0800)]
Clean test/dct_test.cc with testing::Combine
Change-Id: I910fd34e4a06a73568b597ccb194c8395c2e6d08
Linfeng Zhang [Fri, 23 Feb 2018 01:42:59 +0000 (01:42 +0000)]
Merge "Add vp9_highbd_iht8x8_16_add_neon()"
Jerome Jiang [Fri, 23 Feb 2018 00:21:56 +0000 (00:21 +0000)]
Merge "VP9 SVC: Datarate test for dynamic bitrate change."
Jerome Jiang [Thu, 22 Feb 2018 00:51:12 +0000 (16:51 -0800)]
VP9 SVC: Datarate test for dynamic bitrate change.
Change-Id: Ie1cd990dcb19a4cc18de4a2e487791f399c4b3cb
Kyle Siefring [Wed, 21 Feb 2018 23:12:07 +0000 (23:12 +0000)]
Merge "Fold adds in 16->32-bit converts in SSE2/AVX2 fDCT"
Linfeng Zhang [Wed, 21 Feb 2018 01:21:31 +0000 (17:21 -0800)]
Add vp9_highbd_iht8x8_16_add_neon()
BUG=webm:1403
Change-Id: I11efb652f1aee371c71eee2d29e33793e4736832
Johann [Tue, 20 Feb 2018 22:49:02 +0000 (14:49 -0800)]
remove deprecated 'register' keyword
Will be removed in C++17:
http://en.cppreference.com/w/cpp/language/storage_duration
Change-Id: Iadce5e2b974c707799fa939f3ff1c420fb79a871
Jerome Jiang [Mon, 12 Feb 2018 23:26:07 +0000 (23:26 +0000)]
Merge "VP9 ROI test clean up regarding bool type flag."
Jerome Jiang [Mon, 12 Feb 2018 19:01:29 +0000 (11:01 -0800)]
vp9_cx_iface: Remove else when returning from the other branch.
Change-Id: I2fc15ec25cc5587cafc6621176d0a6d7c376fc7c
Jerome Jiang [Mon, 12 Feb 2018 18:36:14 +0000 (10:36 -0800)]
VP9 ROI test clean up regarding bool type flag.
Clean up code to make use_roi_ flag a bool.
Change-Id: I5b606ca19f8543840259d1cc79fe3301a2a70d30
Kyle Siefring [Mon, 5 Feb 2018 05:15:29 +0000 (00:15 -0500)]
Fold adds in 16->32-bit converts in SSE2/AVX2 fDCT
Changes in the function size in bytes (in lieu of performance metrics)
Before After Diff
vpx_fdct32x32_avx2 29564 -> 28334 -1230
vpx_fdct32x32_sse2 38053 -> 36309 -1744
Change-Id: Ie0b3e6ed7c3f2e9ea45f9d6a1ce1e27d068cee6b
Jerome Jiang [Sat, 10 Feb 2018 16:39:43 +0000 (08:39 -0800)]
VP9 ROI: reset use_roi_ in datarate test.
Change-Id: I51765ce6c3c8e8646852c4da47b12a0198892c52
Jerome Jiang [Fri, 9 Feb 2018 22:47:00 +0000 (14:47 -0800)]
VP9 ROI: Fix errors in example encoder.
Fix some errors in the vpx_temporal_svc_encoder.
Change-Id: Id93f449364dcf72c826ca931df3c8c3d3b80100f
Jerome Jiang [Fri, 9 Feb 2018 19:01:52 +0000 (19:01 +0000)]
Merge "Reland "Add ROI support for VP9.""
Jerome Jiang [Mon, 29 Jan 2018 18:14:40 +0000 (10:14 -0800)]
Reland "Add ROI support for VP9."
Extended ROI struct suitable for VP9.
ROI input from user is passed into internal struct and applied on every frame
(except key frame).
Enabled usage of all 4 VP9 segment features (delta_qp, delta_lf, skip,
ref_frame) via the ROI map input.
Made changes to nonrd_pickmode for the ref_frame feature.
Only works for realtime speed >= 5.
AQ_MODE needs to be turned off for ROI to take effect.
Change example in the sample encoder: vpx_temporal_svc_encoder.c to be suitable
for VP9.
Add datarate test.
Bump up ABI version.
BUG=webm:1470
Change-Id: I663b8c89862328646f4cc6119752b66efc5dc9ac
Jerome Jiang [Fri, 9 Feb 2018 18:54:55 +0000 (18:54 +0000)]
Merge "Revert "Add ROI support for VP9.""
Jerome Jiang [Fri, 9 Feb 2018 18:53:54 +0000 (18:53 +0000)]
Revert "Add ROI support for VP9."
This reverts commit
4e5b4b58483e1f38e37acd49b809d725b4f66c26.
Reason for revert: Commit message inaccurate.
Original change's description:
> Add ROI support for VP9.
>
> Extended ROI struct suitable for VP9.
> ROI input from user is passed into internal struct and applied on every frame
> (except key frame).
>
> Enabled usage of all 4 VP9 segment features (delta_qp, delta_lf, skip,
> ref_frame) via the ROI map input.
> Made changes to nonrd_pickmode for the ref_frame feature.
>
> Only works for realtime speed >= 5.
> AQ_MODE needs to be turned off for ROI to take effect.
>
> Change example in the sample encoder: vpx_temporal_svc_encoder.c to be suitable
> for VP9.
> Add datarate test.
>
> Bump up ABI version.
>
> BUG=webm:1470
>
> Change-Id: I7e0cf6890649adb98a5fda2efb6ae1fa511c7fc9
TBR=yaowu@google.com,jzern@google.com,marpan@google.com,builds@webmproject.org,jianj@google.com
Change-Id: I000dbd81e0c67cb8a0dcde4013ee9bf7afb038f0
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webm:1470
Jerome Jiang [Fri, 9 Feb 2018 16:58:55 +0000 (16:58 +0000)]
Merge "Add ROI support for VP9."
paulwilkins [Wed, 31 Jan 2018 16:42:37 +0000 (16:42 +0000)]
Improved coding on slide show content.
This patch adds in detection of slide show key frame groups.
The detection assumes extremely low or 0 motion for all frames
in the key frame group.
If this case is detected the boost level is set to a very high value
and the min Q to a lower value for the key frame itself.
Alt refs and golden frames are disabled to save bits (up to a limiting
maximum interval currently set to 240 frames).
In test samples that I created, this patch gave rise to a substantial
improvement in overall psnr and a drop in data rate. In some cases the
average psnr fell, however, with the boost and minQ values set as they are.
This is to be expected because previously a relatively poor key frame
could be followed by progressively better alt refs. For example a key
frame at q7.5 but subsequent alt refs improving it to lossless. Given that
average psnr tends to be dominated by the best frames, a ramp like this
from q7.5 to lossless may give a better average psnr than, for example,
coding the entire sequence at q2.5. Overall psnr, however, will be much
better in the latter case. The option exists to boost the key frame further
which would insure much better results for all metrics, but at the expense
of smaller bitrate savings. Given that these samples tend to have very
good quality anyway this seems like a bad trade off.
For slides displayed for several seconds, bitrate savings of >= 20% are likely
and much larger gains are possible in some cases.
Change-Id: Ib4b61e153c55d3f2f561153da13fdb56f397a52b
Marco [Mon, 29 Jan 2018 18:14:40 +0000 (10:14 -0800)]
Add ROI support for VP9.
Extended ROI struct suitable for VP9.
ROI input from user is passed into internal struct and applied on every frame
(except key frame).
Enabled usage of all 4 VP9 segment features (delta_qp, delta_lf, skip,
ref_frame) via the ROI map input.
Made changes to nonrd_pickmode for the ref_frame feature.
Only works for realtime speed >= 5.
AQ_MODE needs to be turned off for ROI to take effect.
Change example in the sample encoder: vpx_temporal_svc_encoder.c to be suitable
for VP9.
Add datarate test.
Bump up ABI version.
BUG=webm:1470
Change-Id: I7e0cf6890649adb98a5fda2efb6ae1fa511c7fc9
paulwilkins [Wed, 7 Feb 2018 15:05:20 +0000 (15:05 +0000)]
Adjust MAXRATE_1080P.
This value was originally set in response to requests from the hardware
team before levels were properly defined for VP9.
Even if a level is not specified for an encode, it imposes a maximum
frame size for videos of dimensions <= 1080P. For larger formats the
limit was set at 250 bits per MB.
This patch modifies the limit to be more in line with the requirements
specified for level 4 (max rate for a 4 frame group of 16 Mbits). If a lower
level is specified at encode time and this mandates a smaller maximum frame
size then the level requirement will still take precedence.
Increasing this value allows for some slide shows or very low motion clips
to code a better quality key frame.
Change-Id: Ic08e0e09c8a918077152190c59732b9a1c049787
Paul Wilkins [Thu, 8 Feb 2018 10:57:44 +0000 (10:57 +0000)]
Merge "Fix file input pointer bug in allocate_gf_group_bits()."
Linfeng Zhang [Wed, 7 Feb 2018 02:12:12 +0000 (18:12 -0800)]
Update iadst NEON functions
Use scalar multiply. No impact on clang, but improves gcc compiling.
BUG=webm:1403
Change-Id: I4922e7e033d9e93282c754754100850e232e1529
Linfeng Zhang [Wed, 7 Feb 2018 18:39:52 +0000 (10:39 -0800)]
Clean vp9_highbd_iht4x4_16_add_neon()
Extract common code.
Change-Id: I422150ada1c6915f0ce39b912149994eb3bb3f12
paulwilkins [Tue, 6 Feb 2018 13:54:25 +0000 (13:54 +0000)]
Fix file input pointer bug in allocate_gf_group_bits().
The stats input pointer, when passed in, already points to the
frame after the golden frame so should not be advanced here.
This fix has a small mostly positive effect on results in our test sets
(tested using corpus vbr settings) and gives a gain of almost 0.5%
in overall psnr (plus slightly smaller gains on other metrics) for the
4K set.
The bug also caused a crash in calculate_group_score() in another
patch which allows coding of slides in a slide show as a single
long KF group without ARFs or GFs.
Change-Id: I57a3a24baf442ce55dbc91fba05e056697c63a6f
Linfeng Zhang [Wed, 31 Jan 2018 00:55:22 +0000 (16:55 -0800)]
Update tx_type switch code in idct
Change-Id: Ia244bfd4b4eb9d703653792bc4f64c6f5358ae19
Linfeng Zhang [Wed, 31 Jan 2018 00:01:41 +0000 (16:01 -0800)]
Add vp9_highbd_iht4x4_16_add_neon()
BUG=webm:1403
Change-Id: Id9833e985fb70958cf4bde38f8e6303ed83c12f9
James Zern [Fri, 2 Feb 2018 18:41:46 +0000 (18:41 +0000)]
Merge "inv_txfm_vsx.c: make code c90 compatible"
Jerome Jiang [Fri, 2 Feb 2018 05:32:55 +0000 (05:32 +0000)]
Merge "Fix issue for 0 target bitrate in multi-res build."
Jerome Jiang [Thu, 25 Jan 2018 19:36:53 +0000 (11:36 -0800)]
Fix issue for 0 target bitrate in multi-res build.
For encoding with --enable-multi-res-encoding, with 1 layer, when the
target bitrate is set 0, under these conditions null pointer
will be de-referenced. Fix is to check
cpi->oxcf.mr_total_resolutions > 1. Also added NULL pointer check.
This issue causes crash for asan build in chromium clusterfuzz.
BUG=805863
Change-Id: I9cd25af631395bc9fede3a12fb68af4021eb15f8
James Zern [Fri, 2 Feb 2018 03:39:58 +0000 (19:39 -0800)]
inv_txfm_vsx.c: make code c90 compatible
move for loop declarations to function scope
Change-Id: I84d92a1a6ca6c5ac30aacb0f55d87ca3aef4c98f
James Zern [Thu, 1 Feb 2018 20:44:48 +0000 (20:44 +0000)]
Merge "vp9_scale_test: parameterize filter type"
Paul Wilkins [Thu, 1 Feb 2018 10:21:38 +0000 (10:21 +0000)]
Merge "Further change to code detecting slide transitions."
James Zern [Thu, 1 Feb 2018 07:38:47 +0000 (23:38 -0800)]
vp9_scale_test: parameterize filter type
this allows the test to be sharded more efficiently and speeds up the
run when built with slower configs, e.g., asan.
Change-Id: If6d863b76871e3934704a1079bbf17f4886932c7
Marco [Wed, 31 Jan 2018 23:15:40 +0000 (15:15 -0800)]
vp9-svc: Add condition on allocation for scaled_temp.
scaled_temp frame is used as an intermediate buffer for
2 stage down-sampling: two stages of 1/2 down-sampling
for a target of 1/4x1/4. This is used in 3 layer SVC
to avoid duplicate frame downsampling (on middle layer).
As this allocation is only needed/used when the
number_spatial_layers > 2, add this condition to avoid
unneeded allocation for 1 and 2 spatial SVC.
Change-Id: If342466644f685c1ea3ca5344b581793e5136c09
Marco [Wed, 31 Jan 2018 21:33:27 +0000 (13:33 -0800)]
vp9-svc: Fix to initialize downsampling filters.
For 3 spatial layers with 1/2 downsampling, the
downsampling filter for the middle layer was not
set for the very first frame, so it was defaulting
to the subsample filter (no averaging/phase = 0).
Its not set due to the two stage scaling that is
done for 1/4 on base layer, during which the intermediate
1/2 result is saved for the middle layer.
Fix for now is to set the default downsampling filter
to Bilinear (averaging/non-zero phase) for all layers on
init (vp9_init_layer_context):.
Change-Id: Ic7407810b34c621e7e7420682508d45478bdffcf