stereo & 16bit is untested due to lack of samples
Signed-off-by: Martin Storsjö <martin@martin.st>
(cherry picked from commit 5166376f24)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
init_get_bits() takes a number of bits and not a number of bytes as
its size argument.
Signed-off-by: Alex Converse <alex.converse@gmail.com>
(cherry picked from commit 46b004959b)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
In some places 0/mb_height were used in place of start_mb_y/end_mb_y.
Fixes SA00049, SA00058, SA10091, SA10097, SA10131, SA20021, SA30030
Improves PSNR in SA00054, SA00059, SA00060, SA10096, SA10098, SA20022,
SA30031, SA30032, SA40012, SA40013
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 1cf82cab08)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Previously, we would leave the left edge uninitialized, which led to
CBP prediction errors on slice edges, e.g. in SA10098.vc1.
(cherry picked from commit d4b9974465)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Correct computation of implicit weight tables when referencing pictures
that are marked for long reference.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
(cherry picked from commit 87cf70eb23)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
The current check on MMCO parameters prohibits a "max long term frame index
plus 1" of 16 (frame idx of 15) for the "set max long term frame index" MMCO.
Fix this off-by-one error to allow the full range of legal values.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
(cherry picked from commit 29a09eae9a)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Audio found is not triggered on a CCE because a CCE alone has no output.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
(cherry picked from commit d8425ed4af)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
This removes out-of-edge motion compensation artifacts (easily spotted green
blocks in avplay, gray blocks in transcoding), for example here:
http://samples.libav.org/samples/real/tv_watching_t1.rm
Signed-off-by: Diego Biurrun <diego@biurrun.de>
(cherry picked from commit 331971116d)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Custom PAR num/denum are in 1-256 range.
Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
Signed-off-by: Diego Biurrun <diego@biurrun.de>
(cherry picked from commit 0e86965514)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
bits_per_raw_sample is used in video as well, where sample_fmt is not used.
(cherry picked from commit d271d5b215)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
On some platforms EINVAL could be positive, ensure we return negative values.
(cherry picked from commit e5985185d2)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Some versions of the GNU assembler do not handle 64-bit
immediate operands containing arithmetic. Writing the
value out in full works correctly.
Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit fce1e43410)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Set the frame size when decoding DTS audio.
This has the side effect of fixing the computation of timestamps for DTS-HD in compute_pkt_fields. Since frame_size is
not currently set, the duration of a frame is being guessed based on the streams bitrate. But for DTS-HD, the bitrate
currently used is the rate of the DTS core which is much different than the whole DTS-HD stream and leads to a wildly
inaccurate frame duration estimate.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 49c7006c7e)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
filter_mb_fast assumed that qscale_table was padded like many of the other tables.
(cherry picked from commit 5029a40633)
Signed-off-by: Anton Khirnov <anton@khirnov.net>
The previously suggested replacement - av_get_bits_per_sample_fmt() -
was also deprecated.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit ccfa626db8)
Before this, almost all module groups have been used for grouping functions
and fields in structures semantically. This causes them to not appear
properly in the file documentation and needlessly clutters up the "Modules"
index.
Additionally, this commit streamlines some spelling and appearances.
(cherry picked from commit 21a19b7912)
2tap qpel isn't implemented yet for high bit depth, so it just breaks decoding.
(cherry picked from commit 9a0dda8b3a)
Signed-off-by: Reinhard Tartler <siretart@tauware.de>
