platform/upstream/libvpx.git
13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Wed, 24 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Tue, 23 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Tue, 23 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Tue, 23 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoRecalibration of bits per MB tables
Paul Wilkins [Mon, 22 Nov 2010 13:17:35 +0000 (13:17 +0000)]
Recalibration of bits per MB tables

The baseline bits per MB prediction tables have been
re calibrated based on the assumption that bits per mb
is inversely proportional to the quantizer level.

Change-Id: Ibd355c7acac4b8053dda1baf1032fe35f11da7f7

13 years agoMerge "Added extra two pass stats gathering."
Paul Wilkins [Mon, 22 Nov 2010 12:11:20 +0000 (04:11 -0800)]
Merge "Added extra two pass stats gathering."

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Sun, 21 Nov 2010 05:05:06 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Sun, 21 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoAdded extra two pass stats gathering.
Paul Wilkins [Fri, 12 Nov 2010 11:35:30 +0000 (11:35 +0000)]
Added extra two pass stats gathering.

Added code to record spend so far against planed budget.

Change-Id: I5a3335346fa1771b2b1219df9f6127f9993d2594

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Fri, 19 Nov 2010 05:05:03 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoremove low pass filtering from two 4x4 intra prediction
Yaowu Xu [Mon, 1 Nov 2010 21:04:01 +0000 (14:04 -0700)]
remove low pass filtering from two 4x4 intra prediction

In the process of developing new intra prediction modes, tests have
shown removal of the low pass filtering from B_HE_PRED and B_VE_PRED
has an overall minor positive impact in both PSNR and SSIM metric.
Overall difference is about 0.1%. The change shall also have a small
positive impact on speed. Intuitively, this change should also reduce
some of the tendency of "flattening"

Change-Id: I3c43b0daca833c6eff77d00f19c811f9ef9368a3

13 years agochanged MAX_PSNR to 100
Yaowu Xu [Thu, 18 Nov 2010 17:10:30 +0000 (09:10 -0800)]
changed MAX_PSNR to 100

Changing the MAX_PSNR to 100 to allow testing of further experiments
on extending quantizer range to near lossless. With an effective
quantizer of 1, encoder achieves ~68DB, which is consistent with
fdct/idct round trip error.

Change-Id: I7b6d0e94a8936968ef42e82e63ebb13999c36832

13 years agoextends the range of tokens
Yaowu Xu [Tue, 16 Nov 2010 21:58:21 +0000 (13:58 -0800)]
extends the range of tokens

Extending the value range of tokens allows further experiments on
extending quantizer range. Encoder and decoder were verified to
produce matching reconstructed buffers by tests with forced
quantized value of 1.

Change-Id: I12faf92832867870b6f71ddeafbf643f1040086d

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Thu, 18 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Thu, 18 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Thu, 18 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoremove warning
Pascal Massimino [Thu, 18 Nov 2010 00:50:02 +0000 (16:50 -0800)]
remove warning

was having: "vp8/encoder/onyx_if.c:5365: warning: comparison of unsigned expression >= 0 is always true"

13 years agoMerge "Removed unnecessary checks."
Scott LaVarnway [Wed, 17 Nov 2010 19:28:22 +0000 (11:28 -0800)]
Merge "Removed unnecessary checks."

13 years agoRemoved unnecessary checks.
Scott LaVarnway [Wed, 17 Nov 2010 18:15:13 +0000 (13:15 -0500)]
Removed unnecessary checks.

macro_block_yrd and vp8_rdcost_mby are not called for SPLITMV.

Change-Id: I2224d3c8725df526d48426447482768d543752f1

13 years agoMerge "FIXED bug in when CONFIG_SEGMENTATION NOT DEFINED" into experimental
Suman Sunkara [Wed, 17 Nov 2010 17:07:24 +0000 (12:07 -0500)]
Merge "FIXED bug in when CONFIG_SEGMENTATION NOT DEFINED" into experimental

13 years agoFIXED bug in when CONFIG_SEGMENTATION NOT DEFINED
Jim Bankoski [Wed, 17 Nov 2010 16:30:24 +0000 (11:30 -0500)]
FIXED bug in when CONFIG_SEGMENTATION NOT DEFINED

13 years agoReplaced recode loop test with a function call
Paul Wilkins [Wed, 17 Nov 2010 15:12:04 +0000 (15:12 +0000)]
Replaced recode loop test with a function call

Replaced existing code to decide if a frame recode is required
with a function call. This is to simplify addition of extra clauses
that may be needed for the planned constrained quality mode.

Also fixed a bug where by alt ref not considered in the test.

Change-Id: I3d40bb21abe3e19f8456761e6849deb171738b60

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Wed, 17 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Wed, 17 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Wed, 17 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoMerge "Comments for alt ref flags."
Fritz Koenig [Wed, 17 Nov 2010 00:11:39 +0000 (16:11 -0800)]
Merge "Comments for alt ref flags."

13 years agoComments for alt ref flags.
Fritz Koenig [Tue, 16 Nov 2010 23:16:24 +0000 (15:16 -0800)]
Comments for alt ref flags.

Clarify what the alt ref flags do when encoding.

Change-Id: I71f78e0f42edae633fb91840f29dfbe64362c44c

13 years agoMerge "correct errors in token alphabet descriptions"
Yaowu Xu [Tue, 16 Nov 2010 22:06:44 +0000 (14:06 -0800)]
Merge "correct errors in token alphabet descriptions"

13 years agoMerge branch 'experimental' of ssh://on2-git.corp.google.com:29418/libvpx into test
Suman Sunkara [Tue, 16 Nov 2010 21:31:59 +0000 (16:31 -0500)]
Merge branch 'experimental' of ssh://on2-git.corp.google.com:29418/libvpx into test

Conflicts:
configure

Change-Id: Id874dc46b13e8b5da4179fc3b48e354ec313a2cd

13 years agoMerge branch 'experimental' of ssh://on2-git.corp.google.com:29418/libvpx into test
Suman Sunkara [Tue, 16 Nov 2010 20:09:26 +0000 (15:09 -0500)]
Merge branch 'experimental' of ssh://on2-git.corp.google.com:29418/libvpx into test

Conflicts:
vp8/common/blockd.h
vp8/decoder/decodemv.c
vp8/decoder/decodframe.c
vp8/decoder/demode.c
vp8/decoder/onyxd_if.c
vp8/decoder/onyxd_int.h
vp8/encoder/encodeframe.c

Change-Id: Ic379f4dffaded9796dc19d56be304d3f8527c61f

13 years agoAdd x86_64-darwin10-gcc target.
tomfinegan [Tue, 16 Nov 2010 19:52:05 +0000 (14:52 -0500)]
Add x86_64-darwin10-gcc target.

Adds native build configuration for Snow Leopard.  Useful when
users configure without arguments on OSX 10.6.

Change-Id: I0bd63912a25bbfb9d4c8d58a781d0f390792429c

13 years agochanges to start experimenting with color segmentation prediction modes.
Jim Bankoski [Tue, 16 Nov 2010 19:38:40 +0000 (14:38 -0500)]
changes to start experimenting with color segmentation prediction modes.

13 years agocorrect errors in token alphabet descriptions
Yaowu Xu [Tue, 16 Nov 2010 18:51:43 +0000 (10:51 -0800)]
correct errors in token alphabet descriptions

There were a few errors in the comment section that describe VP8 token
alphabet table.

Change-Id: Ie6728a0e08bc3798893221b60408d5b201064bdc

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Tue, 16 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Tue, 16 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoRemove stack shadowing for x86-x64 for SAD functions.
Fritz Koenig [Thu, 11 Nov 2010 18:15:58 +0000 (10:15 -0800)]
Remove stack shadowing for x86-x64 for SAD functions.

x86-64 passes arguments in registers.  There is no need to push
them to the stack before using them.

This fixes 15acc84f10cefd98b2f8dbd2eac2cc92c5a3f851 where ebx
was not getting preserved on x86.

Change-Id: I1214b5f818a0201f75ab6ad7d5c6f448e09b16c2

13 years agoMerge "Bad cost tables used in ARNR filtering."
Paul Wilkins [Mon, 15 Nov 2010 17:55:35 +0000 (09:55 -0800)]
Merge "Bad cost tables used in ARNR filtering."

13 years agoBad cost tables used in ARNR filtering.
Paul Wilkins [Mon, 15 Nov 2010 17:47:12 +0000 (17:47 +0000)]
Bad cost tables used in ARNR filtering.

The use of incorrect mv costing tables in the ARNR sub-pel
filtering code led to corruption of the altref buffer in some cases,
particularly at low data rates.

The average gain from this fix is about 0.3% but there are a few
extreme cases where nasty and visible artifacts manifested and
for these few data points the improvement is > 10%.

PGW and AWG

Change-Id: I95cc02b196a433e71d0d2bd2b933fe68ed31e796

13 years agoMerge "make rdmult adaptive for intra in quantizer RDO"
Yaowu Xu [Mon, 15 Nov 2010 17:22:45 +0000 (09:22 -0800)]
Merge "make rdmult adaptive for intra in quantizer RDO"

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Sun, 14 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Sat, 13 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Sat, 13 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Sat, 13 Nov 2010 05:05:04 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoFixed bug first cluster timecode of webm file is wrong.
Frank Galligan [Thu, 4 Nov 2010 03:33:00 +0000 (23:33 -0400)]
Fixed bug first cluster timecode of webm file is wrong.

When the first pts equaled 0 ivfenc was incorrectly increasing the
pts by 1. I changed the pts and last pts to be signed. I also set
the default value of last pts to -1.

Change-Id: I30bcec5af9b16d93fa9e3abbea7764b133e9cd73

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Fri, 12 Nov 2010 05:05:03 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Fri, 12 Nov 2010 05:05:03 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Fri, 12 Nov 2010 05:05:03 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agomake rdmult adaptive for intra in quantizer RDO
Yaowu Xu [Thu, 11 Nov 2010 05:16:17 +0000 (21:16 -0800)]
make rdmult adaptive for intra in quantizer RDO

This intends to correct the tendency that VP8 aggressively favors rate
on intra coded frames. Experiments tested different numbers in [0, 1]
and found 9/16 overall provided about 2-4% gains for all-intra coded
clips based on vpx-ssim metric. The impact on regular encoded clips
is much smaller but positive overall. Overall impact on psnr is also
positive even though very small.

Change-Id: If808553aaaa87fdd44691f9787820ac9856d9f8a

13 years agoquantizer: fix assertion in fast quantizer path
John Koleszar [Thu, 11 Nov 2010 17:41:07 +0000 (12:41 -0500)]
quantizer: fix assertion in fast quantizer path

The fast quantizer assembly code has not been updated to match the new
exact quantizer, which was made the default in commit 6adbe09.
Specifically, they are not aware of the potential for the coefficient
to be scaled, which results in the quantized result exceeding the range
of the DCT. This patch restores the previous behavior of using the
non-shifted coefficients when in the fast quantizer code path, but
unfortunately requires rebuilding the tables when switching between the
two.

Change-Id: I0a33f5b3850335011a06906f49fafed54dda9546

13 years agoUse of temporal context for encoding delta updates.
Suman Sunkara [Thu, 23 Sep 2010 22:25:33 +0000 (18:25 -0400)]
Use of temporal context for encoding delta updates.

- Used three probability approach for temporal context as follows:
P0 - probability of no change if both above and left not changed
P1 - probability of no change if one of above and left has changed
P2 - probability of no change if both above and left have changed

In addition, a 1 bit/frame has been used to decide whether to use temporal context or to encode directly.  The cost of using both the schemes is calculated ahead and the temporal_update flag is set if the cost of using temporal context is lower than encoding the segment ids directly.

This approach has given around 20% reduction in cost of bits needed to encode segmentation ids.

Change-Id: I44a5509599eded215ae5be9554314280d3d35405

13 years agoRevert "Remove stack shadowing for x86-64"
Fritz Koenig [Thu, 11 Nov 2010 16:19:37 +0000 (08:19 -0800)]
Revert "Remove stack shadowing for x86-64"

This reverts commit 15acc84f10cefd98b2f8dbd2eac2cc92c5a3f851.

Change-Id: Ia640be8cbc134432914849c1750f62575ea084e6

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Thu, 11 Nov 2010 14:25:10 +0000 (09:25 -0500)]
Merge remote branch 'origin/master' into experimental

Conflicts:
configure

Change-Id: Ifa63e4610657f75cb953aa7ca08f997267612cc0

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Thu, 11 Nov 2010 14:22:46 +0000 (09:22 -0500)]
Merge remote branch 'internal/upstream' into HEAD

Conflicts:
configure

Change-Id: I1c7bae5241f999387cae3f2abf2dfc84fe3f6651

13 years agoMerge "Relax rate control for last few frames"
Paul Wilkins [Thu, 11 Nov 2010 10:39:20 +0000 (02:39 -0800)]
Merge "Relax rate control for last few frames"

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Thu, 11 Nov 2010 05:05:03 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

13 years agoconfigure : Incorrect syntax in configure
Fritz Koenig [Wed, 10 Nov 2010 22:51:49 +0000 (14:51 -0800)]
configure : Incorrect syntax in configure

Check to see if postproc was enabled when enabling the
postproc visualizer was wrong.

Fix for bug introduced in Change Ia74f357d

Change-Id: I4bee9ad2caee3cfe3bac6972047f6af7c54cad4e

13 years agoMerge "Remove stack shadowing for x86-64"
Fritz Koenig [Wed, 10 Nov 2010 22:36:10 +0000 (14:36 -0800)]
Merge "Remove stack shadowing for x86-64"

13 years agoFDCT optimizations.
Fritz Koenig [Thu, 21 Oct 2010 17:53:15 +0000 (10:53 -0700)]
FDCT optimizations.

Fixed up the fdct for mmx and 8x4 sse2 to match them
most recent changes.

Change-Id: Ibee2d6c536fe14dcf75cd6eb1c73f4848a56d719

13 years agopostproc : Re-work posproc calling to allow more flags.
Fritz Koenig [Thu, 4 Nov 2010 23:03:36 +0000 (16:03 -0700)]
postproc : Re-work posproc calling to allow more flags.

Debugging in postproc needs more flags to allow for specific
block types to be turned on or off in the visualizations.

Must be enabled with --enable-postproc-visualizer during
configuration time.

Change-Id: Ia74f357ddc3ad4fb8082afd3a64f62384e4fcb2d

13 years agoRelax rate control for last few frames
Paul Wilkins [Wed, 10 Nov 2010 10:09:45 +0000 (10:09 +0000)]
Relax rate control for last few frames

VBR rate control can become very noisy for the last few frames.
If there are a few bits to spare or a small overshoot then the
target rate and hence quantizer may start to fluctuate wildly.

This patch prevents further adjustment of the active Q limits for
the last few frames.

Patch also removes some redundant variables and makes one small bug fix.

Change-Id: Ic167831bec79acc9f0d7e4698bcc4bb188840c45

13 years agoTuning for the more exact quantizer.
Paul Wilkins [Mon, 8 Nov 2010 15:28:54 +0000 (15:28 +0000)]
Tuning for the more exact quantizer.

Small changes to the default zero bin and rounding tables.
Though the tables are currently the same for the Y1 and Y2 cases
I have left them as separate tables in case we want to tune this later.

There is now some adjustment of the zbin based on the prediction mode.
Previously this was restricted to an adjustment for gf/arf 0,0 MV.

The exact quantizer now marginal outperforms and is the default.

The overall average gain is about 0.5%

Change-Id: I5e4353f3d5326dde4e86823684b236a1e9ea7f47

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Wed, 10 Nov 2010 05:05:06 +0000 (00:05 -0500)]
Merge remote branch 'origin/master' into experimental

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Wed, 10 Nov 2010 05:05:05 +0000 (00:05 -0500)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge "improve average framerate calculation"
John Koleszar [Tue, 9 Nov 2010 16:52:16 +0000 (08:52 -0800)]
Merge "improve average framerate calculation"

13 years agoMerge remote branch 'internal/upstream-experimental' into HEAD
John Koleszar [Mon, 8 Nov 2010 13:34:46 +0000 (08:34 -0500)]
Merge remote branch 'internal/upstream-experimental' into HEAD

Conflicts:
configure

Change-Id: I952d3a50ddd243990d37c5f931bd91a4244d6284

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Sat, 6 Nov 2010 04:05:08 +0000 (00:05 -0400)]
Merge remote branch 'origin/master' into experimental

13 years agoMerge commit 'fix integer promotion bug in partition size check'
John Koleszar [Fri, 5 Nov 2010 20:49:17 +0000 (16:49 -0400)]
Merge commit 'fix integer promotion bug in partition size check'

Change-Id: I4081917b46013fa8f4218cade8bd12cb2d013aee

13 years agofix integer promotion bug in partition size check
John Koleszar [Thu, 4 Nov 2010 20:59:26 +0000 (16:59 -0400)]
fix integer promotion bug in partition size check

The check '(user_data_end - partition < partition_size)' must be
evaluated as a signed comparison, but because partition_size was
unsigned, the LHS was promoted to unsigned, causing an incorrect
result on 32-bit. Instead, check the upper and lower bounds of
the segment separately.

Change-Id: I6266aba7fd7de084268712a3d2a81424ead7aa06

13 years agoMerge remote branch 'origin/master' into experimental
John Koleszar [Fri, 5 Nov 2010 16:30:33 +0000 (12:30 -0400)]
Merge remote branch 'origin/master' into experimental

Conflicts:
configure
ivfenc.c
vp8/common/alloccommon.c
vp8/common/onyxc_int.h
vp8/vp8_cx_iface.c

13 years agoimprove average framerate calculation
John Koleszar [Thu, 4 Nov 2010 19:05:45 +0000 (15:05 -0400)]
improve average framerate calculation

Change Ice204e86 identified a problem with bitrate undershoot due to
low precision in the timestamps passed to the library. This patch
takes a different approach by calculating the duration of this frame
and passing it to the library, rather than using a fixed duration
and letting the library average it out with higher precision
timestamps. This part of the fix only applies to vpxenc.

This patch also attempts to fix the problem for generic applications
that may have made the same mistake vpxenc did. Instead of
calculating this frame's duration by the difference of this frame's
and the last frame's start time, we use the end times instead. This
allows the framerate calculation to scavenge "unclaimed" time from
the last frame. For instance:

  start |  end  | calculated duration
  ======+=======+====================
    0ms    33ms   33ms
   33ms    66ms   33ms
   66ms    99ms   33ms
  100ms   133ms   34ms

Change-Id: I92be4b3518e0bd530e97f90e69e75330a4c413fc

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Fri, 5 Nov 2010 04:05:05 +0000 (00:05 -0400)]
Merge remote branch 'internal/upstream' into HEAD

13 years agoMerge "vpxdec: report parse errors from webm_guess_framerate()"
John Koleszar [Fri, 5 Nov 2010 02:18:53 +0000 (19:18 -0700)]
Merge "vpxdec: report parse errors from webm_guess_framerate()"

13 years agoMerge remote branch 'internal/upstream' into HEAD
John Koleszar [Fri, 5 Nov 2010 01:50:37 +0000 (21:50 -0400)]
Merge remote branch 'internal/upstream' into HEAD

Conflicts:
vp8/common/alloccommon.c
vp8/common/onyxc_int.h
vp8/vp8_cx_iface.c
vpxenc.c

13 years agovpxdec: report parse errors from webm_guess_framerate()
John Koleszar [Thu, 4 Nov 2010 18:54:51 +0000 (14:54 -0400)]
vpxdec: report parse errors from webm_guess_framerate()

If this function fails silently, the nestegg context is destroyed and
future nestegg calls will segfault.

Change-Id: Ie6a0ea284ab9ddfa97b1843ef8030a953937c8cd

13 years agoMerge "postproc : Update visualizations."
Fritz Koenig [Thu, 4 Nov 2010 18:28:18 +0000 (11:28 -0700)]
Merge "postproc : Update visualizations."

13 years agopostproc : Update visualizations.
Fritz Koenig [Thu, 4 Nov 2010 17:35:02 +0000 (10:35 -0700)]
postproc : Update visualizations.

Change color reference frame to blend the macro block edge.
This helps with layering of visualizations.

Add block coloring for intra prediction modes.

Change-Id: Icefe0e189e26719cd6937cebd6727efac0b4d278

13 years agoIncrease the resolution of default timebase
Yaowu Xu [Wed, 3 Nov 2010 19:56:31 +0000 (12:56 -0700)]
Increase the resolution of default timebase

The old value 1000 was too low, which caused the effective duration and
frame rate calculation to have an 1% error for typical 30 frame/second
inputs. Symptom of the issue has been that most 2 pass encodings were
undershooting target bit rate by 1% or so for 30 fps input.

Change-Id: Ice204e86f844ceb9ce973456f2b995cc095283cf

13 years agovpxenc: require width and height for raw streams
John Koleszar [Wed, 3 Nov 2010 17:58:40 +0000 (13:58 -0400)]
vpxenc: require width and height for raw streams

Defaulting to 320x240 for raw streams is arbitrary and error-prone.
Instead, require that the width and height be set manually if they
can't be parsed from the input file.

Change-Id: Ic61979857e372eed0779c2677247e894f9fd6160

13 years agoMerge "fix pipe support on windows"
John Koleszar [Wed, 3 Nov 2010 00:01:54 +0000 (17:01 -0700)]
Merge "fix pipe support on windows"

13 years agopostproc : Fix display of motion vectors.
Fritz Koenig [Tue, 2 Nov 2010 20:29:13 +0000 (13:29 -0700)]
postproc : Fix display of motion vectors.

Split motion vectors were all being treated as 4x4
blocks.  Now correctly handle 16x8, 8x16, 8x8, 4x4
blocks.

Change-Id: Icf345c5e69b5e374e12456877ed7c41213ad88cc

13 years agoMerge "SSSE3 version of fast quantizer"
Scott LaVarnway [Tue, 2 Nov 2010 13:27:29 +0000 (06:27 -0700)]
Merge "SSSE3 version of fast quantizer"

13 years agofix pipe support on windows
John Koleszar [Tue, 2 Nov 2010 13:11:57 +0000 (09:11 -0400)]
fix pipe support on windows

STDIO streams are opened in text mode by default on Windows. This patch
changes the stdin/stdout streams to be in binary mode if they are being
used for I/O from the vpxenc or vpxdec tools.

Fixes issue #216. Thanks to mw AT hesotech.de for the fix.

Change-Id: I34525b3ce2a4a031d5a48d36df4667589372225b

13 years agoMerge "postproc : Added SPLITMV visualization, fix line constrain."
Fritz Koenig [Mon, 1 Nov 2010 21:41:41 +0000 (14:41 -0700)]
Merge "postproc : Added SPLITMV visualization, fix line constrain."

13 years agopostproc : Added SPLITMV visualization, fix line constrain.
Fritz Koenig [Sat, 30 Oct 2010 00:09:33 +0000 (17:09 -0700)]
postproc : Added SPLITMV visualization, fix line constrain.

Now draw 16 vectors for SPLITMV mode.

Fixed constrain line to block divide by zero issues.

Blend block was not centering the shaded area correctly.

Change-Id: I1edabd8b4e553aac8d980f7b45c80159e9202434

13 years agoSSSE3 version of fast quantizer
Scott LaVarnway [Mon, 1 Nov 2010 20:24:15 +0000 (16:24 -0400)]
SSSE3 version of fast quantizer

(test clip: tulip)
For good quality mode with speed=1, this gave the encoder
a small (2 - 3%) performance boost.

Change-Id: I8a1d4269465944ac0819986c2f0be4b0a2ee0b35

13 years agoMerge "Finding first label"
Scott LaVarnway [Mon, 1 Nov 2010 15:42:37 +0000 (08:42 -0700)]
Merge "Finding first label"

13 years agocosmetic: remove alt_ref from vpxenc usage message
John Koleszar [Fri, 29 Oct 2010 15:07:31 +0000 (11:07 -0400)]
cosmetic: remove alt_ref from vpxenc usage message

Undo an automatic transform.

Change-Id: Ie730a6a31b4680b34e54b61691d67c4b3ed2f2aa

13 years agoFinding first label
Scott LaVarnway [Wed, 27 Oct 2010 18:38:33 +0000 (14:38 -0400)]
Finding first label

Using tables for the label count and label offset.

Change-Id: Iac3d5b292c37341a881be0af282f5cac3b3e01eb

13 years agoSave XMM registers in asm functions
Yunqing Wang [Thu, 28 Oct 2010 20:59:03 +0000 (16:59 -0400)]
Save XMM registers in asm functions

XMM6/7 are used in these functions, and need to be saved.

Change-Id: I3dfaddaf2a69cd4bf8e8735c7064b17bac5a14e5

13 years agoMerge "Fix full-search SAD function crash in Visual Studio"
Yunqing Wang [Thu, 28 Oct 2010 20:46:35 +0000 (13:46 -0700)]
Merge "Fix full-search SAD function crash in Visual Studio"

13 years agoMerge branch 'aylesbury'
John Koleszar [Thu, 28 Oct 2010 20:01:03 +0000 (16:01 -0400)]
Merge branch 'aylesbury'

13 years agoFix full-search SAD function crash in Visual Studio
Yunqing Wang [Thu, 28 Oct 2010 19:26:58 +0000 (15:26 -0400)]
Fix full-search SAD function crash in Visual Studio

Unlike GCC, Visual Studio compiler doesn't allocate SAD output
array 16-byte aligned, which causes crash in visual studio.

Change-Id: Ia755cf5a807f12929bda8db94032bb3c9d0c2362

13 years agoCHANGELOG: correct date v0.9.5
John Koleszar [Thu, 28 Oct 2010 13:14:14 +0000 (09:14 -0400)]
CHANGELOG: correct date

Change-Id: I146a7f241efad4f0684cf8613c7fa42bd5cf42f3

13 years agoUpdate CHANGELOG for v0.9.5 (Aylesbury) release
John Koleszar [Wed, 27 Oct 2010 20:27:56 +0000 (16:27 -0400)]
Update CHANGELOG for v0.9.5 (Aylesbury) release

Change-Id: Ic9f05dbbe90480d5b172233c87eaf1d4e2f1b48e

13 years agoEliminate more warnings.
Timothy B. Terriberry [Wed, 27 Oct 2010 23:04:02 +0000 (16:04 -0700)]
Eliminate more warnings.

This eliminates a large set of warnings exposed by the Mozilla build
 system (Use of C++ comments in ISO C90 source, commas at the end of
 enum lists, a couple incomplete initializers, and signed/unsigned
 comparisons).
It also eliminates many (but not all) of the warnings expose by newer
 GCC versions and _FORTIFY_SOURCE (e.g., calling fread and fwrite
 without checking the return values).
There are a few spurious warnings left on my system:

../vp8/encoder/encodemb.c:274:9: warning: 'sz' may be used
 uninitialized in this function
gcc seems to be unable to figure out that the value shortcut doesn't
 change between the two if blocks that test it here.

../vp8/encoder/onyx_if.c:5314:5: warning: comparison of unsigned
 expression >= 0 is always true
../vp8/encoder/onyx_if.c:5319:5: warning: comparison of unsigned
 expression >= 0 is always true
This is true, so far as it goes, but it's comparing against an enum,
 and the C standard does not mandate that enums be unsigned, so the
 checks can't be removed.

Change-Id: Iead6cd561a2afaa3d801fd63f1d8d58953da7426

13 years agopostproc: Tweaks to line drawing and blending.
Fritz Koenig [Wed, 27 Oct 2010 19:50:16 +0000 (12:50 -0700)]
postproc: Tweaks to line drawing and blending.

Turned down the blending level to make colored blocks obscure
the video less.
Not blending the entire block to give distinction to macro
block edges.
Added configuration so that macro block blending function can
be optimized.
Change to constrain line as to when dx and dy are computed.
Now draw two lines to form an arrow.

Change-Id: I986784e6abff65ea3e0d1437dfca7d06d44ede71

13 years agoOutput the PSNR for the entire file.
Frank Galligan [Wed, 27 Oct 2010 15:28:56 +0000 (11:28 -0400)]
Output the PSNR for the entire file.

If --psnr option is enabled vpxenc will output PSNR values for the
entire file. Added a \n before final output to make sure the output
is on its own line. Overall and Avg psnr matches the values written
to opsnr.stt file.

Change-Id: Ibac5fa9baf8d5a626ea0d6ba161b484e6e8427ee

13 years agoEliminate more warnings.
Timothy B. Terriberry [Wed, 27 Oct 2010 23:04:02 +0000 (16:04 -0700)]
Eliminate more warnings.

This eliminates a large set of warnings exposed by the Mozilla build
 system (Use of C++ comments in ISO C90 source, commas at the end of
 enum lists, a couple incomplete initializers, and signed/unsigned
 comparisons).
It also eliminates many (but not all) of the warnings expose by newer
 GCC versions and _FORTIFY_SOURCE (e.g., calling fread and fwrite
 without checking the return values).
There are a few spurious warnings left on my system:

../vp8/encoder/encodemb.c:274:9: warning: 'sz' may be used
 uninitialized in this function
gcc seems to be unable to figure out that the value shortcut doesn't
 change between the two if blocks that test it here.

../vp8/encoder/onyx_if.c:5314:5: warning: comparison of unsigned
 expression >= 0 is always true
../vp8/encoder/onyx_if.c:5319:5: warning: comparison of unsigned
 expression >= 0 is always true
This is true, so far as it goes, but it's comparing against an enum, and the C
 standard does not mandate that enums be unsigned, so the checks can't be
 removed.

Change-Id: Iaf689ae3e3d0ddc5ade00faa474debe73b8d3395

13 years agoMerge "postproc: Tweaks to line drawing and blending."
Fritz Koenig [Wed, 27 Oct 2010 20:20:56 +0000 (13:20 -0700)]
Merge "postproc: Tweaks to line drawing and blending."

13 years agopostproc: Tweaks to line drawing and blending.
Fritz Koenig [Wed, 27 Oct 2010 19:50:16 +0000 (12:50 -0700)]
postproc: Tweaks to line drawing and blending.

Turned down the blending level to make colored blocks obscure
the video less.
Not blending the entire block to give distinction to macro
block edges.
Added configuration so that macro block blending function can
be optimized.
Change to constrain line as to when dx and dy are computed.
Now draw two lines to form an arrow.

Change-Id: Id3ef0fdeeab2949a6664b2c63e2a3e1a89503f6c