avcodec: Remove libdcadec, we already have it merged internally

avcodec/h264_slice: Check PPS more extensively when its not copied

Fixes Ticket5371
Fixes null pointer dereference

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

avformat/segment: set format options for all segments

Fixes ticket #5318.

Reviewed-by: Stefano Sabatini <stefasab@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>

fate: Remove vsync drop from some h264 tests

Note some tests need vsync drop to produce exact timestamps, these seem not to
need it. quite likely many more dont need it either, ive not checked beyond finding
one that needs it and the ones which have it removed

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

fate: Modify a random h264 test to also test the -framerate option

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

avcodec/audiotoolboxenc: Fix compile error on OS X 10.8.

Fixes error "libavcodec/audiotoolboxenc.c:294:50: error: use of undeclared
identifier 'kAudioCodecPropertyPacketSizeLimitForVBR'."

That was added to 10.9:
https://developer.apple.com/library/mac/releasenotes/General/APIDiffsMacOSX10_9/AudioUnit.html

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

lavc/flicvideo: Implement padding in COPY chunks.

Reviewed-by: Reimar

avformat/rtpdec: Remove stray debug av_log()

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

avcodec/exr: add support for B44 and B44A compression

avcodec/jpeg2000dec: add slice threading support

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

avcodec/jpeg2000dec: account two last bytes from end of bytestream as EOC marker

This silences missing EOC marker when decoding j2k files with Psot set to 0.

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

lavfi: Add coreimage filter for GPU based image filtering on OSX.

lavu/dict: Add new flag to allow multiple equal keys.

sws/aarch64/yuv2rgb: save a few mul and add

27ms to 26ms with UHD 2160 input.

avcodec/libopenjpegenc: fix mixed declarations and code

Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/libopenjpegdec: fix mixed declarations and code

Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/libutvideodec: copy frame so it has reference counters when refcounted_frames is set

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

avutil/parseutils: mark args as static const

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

Added more tests to libavutil/parseutils.c

    - Added tests for av_find_info_tag().
    - Added test for av_get_known_color_name()

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

avformat/rtpdec_jpeg: fix low contrast image on low quality setting

Original mail and my own followup on ffmpeg-user earlier today:

I have a device sending out a MJPEG/RTP stream on a low quality setting.
Decoding and displaying the video with libavformat results in a washed
out, low contrast, greyish image. Playing the same stream with VLC results
in proper color representation.

Screenshots for comparison:

  http://zevv.nl/div/libav/shot-ffplay.jpg
  http://zevv.nl/div/libav/shot-vlc.jpg

A pcap capture of a few seconds of video and SDP file for playing the
stream are available at

  http://zevv.nl/div/libav/mjpeg.pcap
  http://zevv.nl/div/libav/mjpeg.sdp

I believe the problem might be in the calculation of the quantization
tables in the function create_default_qtables(), the attached patch
solves the issue for me.

The problem is that the argument 'q' is of the type uint8_t. According to the
JPEG standard, if 1 <= q <= 50, the scale factor 'S' should be 5000 / Q.
Because the create_default_qtables() reuses the variable 'q' to store the
result of this calculation, for small values of q < 19, q wil subsequently
overflow and give wrong results in the calculated quantization tables. The
patch below uses a new variable 'S' (same name as in RFC2435) with the proper
range to store the result of the division.

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

vc2enc: optimize and simplify quantization

Everything except ORing the sign is now done in the LUT.

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

avformat/mpegtsenc: Fix used service

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

avformat/mpegtsenc: Keep track of the program for each service

Simplifies code

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

Refactor libavutil/parseutils.c

All tests were in the main method which produces a long main. Now, each test
is in his own method.

I think this produces a more clear code and follows more with the main
priority of FFmpeg "simplicity and small code size"

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

fate: Add test that also decodes packed_bframes.avi

No new reference samples are needed for this as the file already exists
for testing the bitstream filter

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

vc2enc: simplify calc_slice_sizes()

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

vc2enc: correctly zero out the slice size cache

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

vc2enc: macro out slice size rounding

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

lavc/psymodel: check for av_malloc failure

No idea why in commit 01ecb7172b684f1c4b3e748f95c5a9a494ca36ec the
checks were removed; this can lead to NULL pointer dereferences. This
effectively reverts that portion of the commit.

Reviewed-by: Benoit Fouet <benoit.fouet@free.fr>
Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanag@gmail.com>

avcodec/dvaudiodec: now that we got samples, fix 12bit case

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

lavc: add AudioToolbox encoders

Fixes trac #4828

lavc: add AudioToolbox decoders

Part of trac #4828

all: move ff_exp10, ff_exp10f, ff_fast_powf to lavu/ffmath.h

The idea is to use ffmath.h for internal implementations of math functions.
Currently, it is used for variants of libm functions, but is by no means
limited to such things.

Note that this is not exported; use lavu/mathematics for such purposes.

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanag@gmail.com>

Added a selftest to libavutil/display.c

    - Check if av_display_rotation_get() gets the correct degrees
    - Check if av_display_rotation_set() sets the correct matrix
    - Check if av_display_matrix_flip() changes correct the matrix

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

vc2enc: add non-experimental support for all video formats from spec

Until now, for formats which were in the spec but not in the encoder's
list of supported formats required the -strict -1 flag. This enables
support for all video formats which are specified, all the way from
QSIF525 to 8K.

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

swresample/swresample: Remove "less than" comparissions of enums

Found-by: wm4
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavf/img2dec: Skip COM when auto-detecting jpeg.

It could theoretically contain invalid data that gets ignored by decoders.

fate: add audiomatch

Testset provided by Justin Greer <justin@zencoder.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

configure: Remove (b)zlib and iconv dependencies for videoltoolbox encoder.

rtpdec: support for VC-2 HQ RTP payload format (draft v1)

ffplay: remove redundant silence buffer

Signed-off-by: Lukasz Marek <lukasz.m.luki2@gmail.com>

avformat/utils: scan a bit farther for a keyframe in mpeg/mpegts (7 sec instead of 5, we already scan 90sec in some cases by default)

Fixes Ticket5305

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

vc2enc: simplify count_hq_slice() and caching

The count_hq_slice() function is always used with a SliceArgs struct

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

vc2enc: fix segfault

Fixes trac bug #5353

Uninitialized memory for the initial quantization index

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

lavf/matroskaenc.c: add early support for colour elements

Adding early support for a subset of the proposed colour elements
according to the latest version of spec:
https://mailarchive.ietf.org/arch/search/?email_list=cellar&gbt=1&index=hIKLhMdgTMTEwUTeA4ct38h0tmE

Like matroskadec, I've left out elements for pix_fmt related things
 as there still seems to be some discussion around these.

The new elements are exposed under strict experimental mode.

Signed-off-by: Neil Birkbeck <neil.birkbeck@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_vectorscope: always flip output vertically

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

avformat/mov: read start_pad from edit list start time if codec is aac

Related to ticket #2324, #2325.

Stream duration still need to be fixed...

Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/utils: increase detected start_time with skip_samples

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

tests/gapless: add gapless aac tests

Signed-off-by: Marton Balint <cus@passwd.hu>

lavf/dv: use c->sys->frame_size in dv_frame_offset()

dv_frame_offset() is static and called only from dv_read_seek(), where
c->sys->frame_size is already used.

This simplifies the incoming codecpar merge where
avctx->{coded_width,coded_height,time_base} are not accessible anymore.

lavf/dv: reindent after previous commit

lavf/dv: do not check for c->sys

dv_extract_video_info() is a static function called only from
avpriv_dv_produce_packet(), where c->sys is made sure to be set.

ffplay: Remove "&& 0" from already disabled debug code

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

avfilter/vf_waveform: set color range for output frames

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

avformat/utils: Do not wait for more than 1 frame on attachments

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

avformat/wtvdec: Set AVFMTCTX_NOHEADER

Needed for noStreams.wtv unless something else forces continued parsing (like looking for more than 1
frame in attachments)

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

avfilter/vf_waveform: add graticule to aflat filter

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

avfilter/vf_waveform: add subsampling input support for remaining filters

Remove achroma filter, as same output can be done with lowpass filter
and multiple components with overlay display.

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

avformat/file: Add crypto to default whitelist

Fixes Ticket5287

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

avcodec/mjpegenc_common: Store approximate aspect if exact cannot be stored

Fixes Ticket5244

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

avfilter/vf_detelecine: Remove redundant declaration

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

lavc/hevc: Allow arbitrary garbage in bytestream as long as at least one NAL unit is found.

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

Fix start_frame handling in detelecine filter

Signed-off-by: Benjamin Steffes <benjaminst123@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Fix detelecine filter for patterns containing 1

Signed-off-by: Benjamin Steffes <benjaminst123@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavf/srtdec: do not be strict wrt timing digit lengths

Fixes a sample with 3-length digits for the seconds reported by wm4.

lavc/aacenc_utils: replace powf(x,y) by expf(logf(x), y)

This is ~2x faster for y not an integer on Haswell+GCC, and should
generally be faster due to the fact that anyway powf essentially does
this under the hood. Made an inline function in lavu/internal.h for this
purpose.

Note that there are some accuracy differences, that should generally be
negligible. In particular, FATE still passes on this platform.

Results in ~ 7% speedup in aac encoding with -march=native, Haswell+GCC.
before:
ffmpeg -i sin.flac -acodec aac -y sin_new.aac  6.05s user 0.06s system 104% cpu 5.821 total

after:
ffmpeg -i sin.flac -acodec aac -y sin_new.aac  5.67s user 0.03s system 105% cpu 5.416 total

This is also faster than an alternative approach that pulls in powf, gets rid of
the crufty NaN checks and other special cases, exploits knowledge about the intervals, etc.
This of course does not exclude smarter approaches; just suggests that
there would need to be significant work on this front of lower utility than
searches for hotspots elsewhere.

Reviewed-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanag@gmail.com>

vc2enc_dwt: remove outdated comment

Support for Haar was added a month or so ago.

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

lavf/avidec: Add blurb regarding the skipping of xxpc entries in the index

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

vc2enc: increase the starting value of the  size scaler

In some cases this caused the slice size rounding to generate invalid
slice sizes and overwrite some slices.

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

avfilter/vf_waveform: add subsampled input support for (a)color filter

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

avcodec/error_resilience: remove unneeded and disabled code

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

avcodec/error_resilience: wait for previous frame to be available

This is possibly redundant but its more correct

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

libwebpenc_animencoder: add missing braces to struct initialization

The first member of the WebPAnimEncoderOptions struct is non scalar

Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/motion_est_template: Fix map cache use in qpel_motion_search()

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

avformat/segment: Fix "occured" typo

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

libwebpenc_animencoder: print library messages in verbose log levels

Reviewed-by: James Zern <jzern@google.com>
Signed-off-by: James Almer <jamrial@gmail.com>

libwebpenc_animencoder: zero initialize the WebPAnimEncoderOptions struct

This zeroes the WebPAnimEncoderOptions.verbose field, silencing library info messages
printed to stderr.

Reviewed-by: James Zern <jzern@google.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avformat/avio: Fix unknown protocol handling

Fixes regression since bb8cc89b2986df6f60831b67cd250da312cce1d0

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

Fix detelecine filter for patterns like 3444 or 33333334.

Signed-off-by: Benjamin Steffes <benjaminst123@gmail.com>
Signed-off-by: Carl Eugen Hoyos <cehoyos@ag.or.at>

avcodec/resample: Remove disabled and faulty code

Fixes Ticket5345

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

configure: check for SEC_I_CONTEXT_EXPIRED before enabling SChannel

Fixes build on mingw32, which lacks this constant.

dxva2_h264: fix size alignment asserts

Convert them to av_assert0 in the process

matroskaenc: set the actual PCM bitdepth in the header

The actual bitdepth can be different to the storage format (ie. sample format).
Fixes the stored bitdepth for 24-bit formats like FLAC.

lavc/dxva2_h264: Fix incorrect assert statement.

Signed-off-by: Matt Oliver <protogonoi@gmail.com>

indeo2: Fix banding artefacts

Rename luma table to delta table and change how it is used.

CC: libav-stable@libav.org
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Signed-off-by: Diego Biurrun <diego@biurrun.de>
(cherry picked from commit f8c34f4b8d62afad3f63cf3d9617d73735bef8c1)

indeo2data: K&R formatting cosmetics

Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Signed-off-by: Diego Biurrun <diego@biurrun.de>
(cherry picked from commit d4066a702407352a0648af882c34ea81a404fa2b)

hls: handle crypto in the protocol checks

Fixes issue 5248

hls: read protocol options through the AVIOContext

This reverts commit 9f9ed79d4cb40e5d9093899f8a79086ff23da844.

The hlsopts member was never set anywhere and always NULL, furthermore
the HLS demuxer needs to retrieve the proper options from the underlying
http protocol (cookies, user-agent, etc), so a dummy context won't help.

Instead, use the AVIOContext directly to access the options.

configure: Use lowercase includes/library names for schannel check.

Fixes cross-build on Linux with mingw-w64.

Reviewed-by: Reimar Döffinger
Reviewed-by: Hendrik Leppkes

avfilter/vf_bwdif: Add yadif base information to copyright header

Signed-off-by: Thomas Mundt <loudmax@yahoo.de>
Signed-off-by: James Almer <jamrial@gmail.com>

lavf/segment: change type of increment_tc to BOOL

lavf/segment: add increment_tc option

For example you can split a file, keeping a continuous timecode between
each segment:
ffmpeg -i src.mov -timecode 10:00:00:00 -vcodec copy -f segment \
-segment_time 2 -reset_timestamps 1 -increment_tc 1 target_%03d.mov

Signed-off-by: Stefano Sabatini <stefasab@gmail.com>

avcodec/mediacodec_sw_buffer: remove redundant article

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

lavc/mediacodec: remove stray empty lines

lavc/mediacodec: fix codec_name leak

avfilter/vf_waveform: add forgotten color and acolor filter to switch case

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

avutil/dict: add warning to docs about invalidating existing entries when adding a new entry

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

avutil/dict: do not realloc entries when deleting a non-existing item

Deleting a non-existing item should not invalidate existing entries returned
with av_dict_get.

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

avcodec/pthread_frame: Remove unused variable

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

avcodec/aactab: do not use floats for constants

This may improve the precision of the fixed point encoder/decoder for some
compilers and architectures.

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