avcodec/h264_slice: remove duplicate unconditional picture_structure setting code

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/cfhd: Fix errors with valid files.

avcodec/cfhd: Check the number of tag/value pairs

Fixes potentially long loop
Fixes: CID1351382/11
Reviewed-by: Kieran Kunhya <kieran@kunhya.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/avf_ahistogram: assert that variables are initialized by switch()

Silences: CID1351397

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/avf_showfreqs: assert that variables are initialized by switch()

Silences: CID1351396

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_zoompan: Remove dead store

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/avf_avectorscope: assert that variables are initialized by switch()

Silences: CID1351390
Silences: CID1351391

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vaf_spectrumsynth: assert that variables are initialized by switch()

Silences: CID1351387
Silences: CID1351388
Silences: CID1351389

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc/filters: fix metadata example

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/f_metadata: rename "string" into "same_str"

Rename function option value "string" into "same_str". Remove obsolete
"length" option.

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

avfilter/f_metadata: whitespace clean-up

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

avfilter/f_metadata: add starts_with string function

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

avfilter/f_metadata: add support for file output

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

avfilter/f_metadata: avoid float rounding problems

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

avfilter/vf_nnedi: Simplify away some trivial mem*()

Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

fate: Update fate-sub2video for 5486d7fa91f7233e6d9aa9ba8de7fe9287402637

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

MAINTAINERS: add my name to the parts I'm maintaining

Not updated since the AAC Encoder.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

x86/vf_blend: Add SSE2 optimization for screen

10x faster than C.

Reviewed-by: Paul B Mahol <onemda@gmail.com>

x86/vf_blend: Move multiplying to a macro

Reviewed-by: Paul B Mahol <onemda@gmail.com>

avfilter/dualinput: use pts provided by framesync

This fixes stalled output pts when main stream ends.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avformat/hlsenc: Fix filename and options

Regression since bc9a5965c815cf7fd998d8ce14a18b8e861dd9ce

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

vc2enc: use project-standard inclusion guards

This was first reported on the mailing list in an earlier revision of this
encoder but was forgotten from the final commit.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

avformat: add vc2 as an allowed rawenc Dirac extension

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

avcodec: add a native SMPTE VC-2 HQ encoder

This commit adds a new encoder capable of creating BBC/SMPTE Dirac/VC-2 HQ
profile files.

Dirac is a wavelet based codec created by the BBC a little more than 10
years ago. Since then, wavelets have mostly gone out of style as they
did not provide adequate encoding gains at lower bitrates. Dirac was a
fully featured video codec equipped with perceptual masking, support for
most popular pixel formats, interlacing, overlapped-block motion
compensation, and other features. It found new life after being stripped
of various features and standardized as the VC-2 codec by the SMPTE with
an extra profile, the HQ profile that this encoder supports, added.

The HQ profile was based off of the Low-Delay profile previously
existing in Dirac. The profile forbids DC prediction and arithmetic
coding to focus on high performance and low delay at higher bitrates.
The standard bitrates for this profile vary but generally 1:4
compression is expected (~525 Mbps vs the 2200 Mbps for uncompressed
1080p50). The codec only supports I-frames, hence the high bitrates.

The structure of this encoder is simple: do a DWT transform on the
entire image, split it into multiple slices (specified by the user) and
encode them in parallel. All of the slices are of the same size, making
rate control and threading very trivial. Although only in C, this encoder
is capable of 30 frames per second on an 4 core 8 threads Ivy Bridge.
A lookup table is used to encode most of the coefficients.

No code was used from the GSoC encoder from 2007 except for the 2
transform functions in diracenc_transforms.c. All other code was written
from scratch.

This encoder outperforms any other encoders in quality, usability and in
features. Other existing implementations do not support 4 level
transforms or 64x64 blocks (slices), which greatly increase compression.

As previously said, the codec is meant for broadcasting, hence support
for non-broadcasting image widths, heights, bit depths, aspect ratios,
etc. are limited by the "level". Although this codec supports a few
chroma subsamplings (420, 422, 444), signalling those is generally
outside the specifications of the level used (3) and the reference
decoder will outright refuse to read any image with such a flag
signalled (it only supports 1920x1080 yuv422p10). However, most
implementations will happily read files with alternate dimensions,
framerates and formats signalled.

Therefore, in order to encode files other than 1080p50 yuv422p10le, you
need to provide an "-strict -2" argument to the command line. The FFmpeg
decoder will happily read any files made with non-standard parameters,
dimensions and subsamplings, and so will other implementations. IMO this
should be "-strict -1", but I'll leave that up for discussion.

There are still plenty of stuff to implement, for instance 5 more
wavelet transforms are still in the specs and supported by the decoder.

The encoder can be lossless, given a high enough bitrate.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

options_table: update maximum bitrate limit

The type of the option has been changed but the limit was apparently forgotten.
Some video codes can handle bitrates of over ~2.2 Gbps (like VC-2).

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

avformat/segment: Fix header_filename handling

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

sws/output: fix ordered dither threshold for mono output

This makes sure that white stays white and black black

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Merge commit '9f61abc8111c7c43f49ca012e957a108b9cc7610'

This also deprecates our old duplicated callbacks.

* commit '9f61abc8111c7c43f49ca012e957a108b9cc7610':
  lavf: allow custom IO for all files

Merged-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

ffmpeg_filter: Factor get_compliance_unofficial_pix_fmts() out

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg_filter: Add missing pixel formats to strict_std_compliance case for ljpeg

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg_filter: Add YUV 444 to strict_std_compliance case for mjpeg

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/f_metadata: fix few logic errors

Signed-off-by: Paul B Mahol <onemda@gmail.com>

tests: Add test for proper header guard

Reviewed-by: Clément Bœsch <u@pkh.me>

avfilter: add metadata filters

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/vf_nnedi: Fix memleak

Fixes CID1351359

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

nut: Add PAL8 support

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/dirac: Fix memleak of dsh on error

Fixes CID1351360

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavf/mov: Extend extracting XMP in mov files using UUID Box

The UUID is based on http://www.adobe.com/devnet/xmp.html

The patch is made according to XMP SPECIFICATION PART 3 - STORAGE IN
FILES See Table 8

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

configure: Enable GCC vectorization on ≥4.9 on x86

4.9 was released precisely nine years after the first GCC version with
autovectorizer (4.0) and six years after the first GCC version with
`-ftree-vectorize` default to enabled on `-O3` (4.3). We've given GCC
enough time to fix those bugs.

FATE passes here on a x86-64 machine with both GCC 4.9.2 and 5.3.1.

Some optimization hotspots benefit greatly from this change, especially
those without handwritten assembly. For instance, the main function in
vf_phase is now 1.6x faster (1.2x overall) on my machine.

avfilter/af_agate: fix memleak of out frame

Fixes CID1351358

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

vf_blend: Add SSE2 optimization for multiply

5 times faster than C, 3 times overall.

ffmpeg_opt: Fix memleaks in "manually set programs" loop

Fixes CID1351356, CID1351357

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_swaprect: add timeline support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avfilter/af_anequalizer: Fix memleak of args

Fixes CID1351355

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vaf_spectrumsynth: Move "break" up

Fixes CID1351347

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/af_aformat: remove deprecated syntax from options description

Signed-off-by: Paul B Mahol <onemda@gmail.com>

x86/cpu: set avxslow cpuflag on btver2 CPUs

They are also slow when using 256 bit wide registers

Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/vf_blend: add multiply128 mode

Signed-off-by: Paul B Mahol <onemda@gmail.com>

diractab: Fix header guard name

huffyuvencdsp: Undefine "i" macro after each use

libvpxenc: Allow setting tune parameter

diracdec: Pass DWTPlane to dwt init

diracdec: Split DWTPlane struct from Plane

dirac_dwt: Rename init2 to init

The functions are all private.

dirac_dwt: Don't pass information in context as arguments

lavfi/fieldmatch: fix fields copy when plane height is odd

Fixes Ticket 5187

avfilter: add swaprect filter

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/utils: Add AV_PIX_FMT_GBRAP16?E to avcodec_align_dimensions2()

Found-by: kierank
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/mp3enc: Assert that the header we assembled is valid

Silences: CID1351343

The header is calculated by the code above the changed hunk, it is
thus asserted that the header is always correct.

Reviewed-by: "Ronald S. Bultje" <rsbultje@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter: add streamselect and astreamselect filter

Signed-off-by: Clément Bœsch <u@pkh.me>
Signed-off-by: Paul B Mahol <onemda@gmail.com>

build: use intermediate lcov coverage file

Otherwise the 'lcov -q --remove' run fails with the following error:
lcov: ERROR: cannot write to coverage.info!

Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

build: fix lcov with src link

When out-of-tree builds now use a relative path, the '-b' option of lcov
is not needed, so just pass the current directory to it in this case.

Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

configure: fall back to using full path if src is a directory

Or when building in-tree.
Also don't try to remove src on distclean in these cases.

Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

cmdutils: realign for some additional filters with very long name

Signed-off-by: Paul B Mahol <onemda@gmail.com>

x86/dcadec: add ff_lfe_fir0_float_{sse,sse2,avx,fma3}

Up to ~4 times faster on x86_64, ~8 times on x86_32 if compiling using x87 fp math.

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/dcadsp: rename lfe_fir_float functions

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

dirac_dwt: Make x86 files/functions names consistent

diracdsp: Make x86 files/functions names consistent

lavc/dirac_dwt: fix building without asm

clang needs HAVE_MMX to be first in order to avoid an undefined
reference error.

avcodec/h264: Fix segfault in 4:2:2 chroma deblock with 32-bit msvc

Using rNm and x86inc's stack allocation with a negative value at the same
time isn't supported, and caused the original stack pointer to be clobbered
when using a compiler that doesn't support stack alignment.

lavf/asfenc: add support for setting packet size

This can provide a manual workaround for ticket #4230.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Marton Balint <cus@passwd.hu>

lavf/asfenc: add support for storing creation time

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Marton Balint <cus@passwd.hu>

lavf/asfenc: add support for storing languages

Checked compatiblity with VLC, Windows Media Player 12 and Windows Media ASF
Viewer 9 series.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Marton Balint <cus@passwd.hu>

lavf/asfenc: check the number of streams in header

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Marton Balint <cus@passwd.hu>

lavf/asfenc: add AVClass to context

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Marton Balint <cus@passwd.hu>

avfilter/split: support any channel count for asplit filter

Signed-off-by: Paul B Mahol <onemda@gmail.com>

avcodec/h264: mmxext 4:2:2 chroma deblock/loop filter

2.6 times faster (366 vs. 142 cycles)

lavc/rawdec: Use 16-byte line alignment for B1W0 and B0W1 video in nut

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/dcadsp: replace intptr_t with ptrdiff_t

Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

x11grab: fixed next frame capture time calculation

The next frame time could slip, causing the frame rate to drop until
 frames were dropped. Now will capture at the next correct moment instead.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/mov: dont print frma warning when format is the same

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/af_apulsator: assert that pathes leaving uninitialized variables do not occur

Fixes: CID1341581, CID1341582
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

fate: fix fate-aac-pns-encode failures on non-sse x86_32 targets

Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/vf_stack: free input pads name on uninit

Signed-off-by: Paul B Mahol <onemda@gmail.com>

lavf/isom: Add more code points for H264.

Defined by the MP4 registration authority.
Fixes ticket #5217.

avfilter/af_afftfilt: Set last_expr to the default

This fixes a null pointer dereferece
Fixes: CID1351346
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

dca_parser: ensure frame duration matches the actual sample rate

The parser only parses the core, and thus has a duration relative
to the core sample rate only, not the actual stream sample rate.

FATE references changed due to now correct timestamps.

x86/emms: empty the mmx state unconditionally on supported targets

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/libutvideoenc: Add NULL termintor to options

Fixes segfaults

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

swscale/swscale-test: Check av_image_fill_linesizes() for failure

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avutil/aes_crt: free AVAESCTR struct properly

Reviewed-by: Andreas Cadhalpun <andreas.cadhalpun@googlemail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

mjpegenc: Remove duplicate initializer

This was causing c99conv to crash.

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

libutvideo: Unbreak

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

diradec: split tables away to a separate diractab file

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

diracdec: move the MAX_DWT_LEVELS macro to dirac.h

Used by the VC-2 encoder.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

Merge commit '68395f8c99393c281a08139d20a7a04398b2fd04'

* commit '68395f8c99393c281a08139d20a7a04398b2fd04':
  qsvenc: fix a typo

Merged-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

Merge commit 'cc602061ee860b041013397e27a036b85cd87b09'

* commit 'cc602061ee860b041013397e27a036b85cd87b09':
  x86inc: Add debug symbols indicating sizes of compiled functions

Merged-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

Merge commit '002c47798da0c43a053822c8041144798d49ed84'

This is a NOOP.

* commit '002c47798da0c43a053822c8041144798d49ed84':
  x86inc: Avoid creating unnecessary local labels
  x86inc: Simplify AUTO_REP_RET
  x86inc: Use more consistent indentation
  x86inc: Preserve arguments when allocating stack space
  x86inc: Improve FMA instruction handling
  x86inc: Be more verbose in assertion failures
  x86inc: Make cpuflag() and notcpuflag() return 0 or 1

Merged-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

Merge commit 'd336bfcf69fee159e9dba5e5e486ddb1aba61aab'

* commit 'd336bfcf69fee159e9dba5e5e486ddb1aba61aab':
  pixdesc: fix and extend doxy for av_pix_fmt_get_chroma_sub_sample()

Merged-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

Merge commit 'c0c4d7a0a556ec66e3068d36a883e84d1efb0690'

* commit 'c0c4d7a0a556ec66e3068d36a883e84d1efb0690':
  configure: Correctly add openssl cflags and libs

Merged-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>