Fixes regressions with stream copy and output timebase/fps being twice as fine as needed
Makes the timebase and ticks per frame handled identical which should make the
code easier to understand and work with. It does not solve the problem without
st->codec access
Suggested-by: Hendrik Leppkes Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
lavf/matroskaenc: use mkv_check_tag_name consistently
Previously, we used a different list of checks when deciding whether to
write a set of tags at all than we did when deciding whether to write an
individual tag in the set. This resulted in sometimes writing an empty
tag master and seekhead. Now we use mkv_check_tag_name everywhere, so
if a dictionary is entirely composed of tags we skip, we don't write a
tag master at all.
This affected the test file, since "language" was on one list but not
the other, so we were writing an empty tag master there. The test hash
is updated to reflect that change.
Philip Langdale [Sun, 4 Sep 2016 16:47:29 +0000 (09:47 -0700)]
cuvid: Implement flush to support seeking in media players
Right now, if we attempt to use cuvid in a media player and then
try to seek, the decoder will happily pass out whatever frames were
already in flight before the seek.
There is both the output queue in our code and some number of frames
within the cuvid decoder that need to be accounted for.
cuvid doesn't support flush, so our only choice is to do a brute-force
re-creation of the decoder, which also implies re-creating the parser,
but this is fine.
The only subtlty is that there is sanity check code in decoder
initialisation that wants to make sure the HWContextFrame hasn't already
been initialised. This is a fair check to do at the beginning but not
after a flush, so it has to be made conditional.
Signed-off-by: Philip Langdale <philipl@overt.org> Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Philip Langdale [Sun, 4 Sep 2016 16:47:28 +0000 (09:47 -0700)]
cuvid: Add hwaccels and decoders for remaining supported formats
cuvid/nvdecode also supports mpeg1, mpeg2, h.263/mpeg4-asp and mjpeg.
It should, in theory, also support wmv3 via the vc1 support, given
that vdpau supports this. However, it failed to play wmv3 samples
which vdpau played correctly, so I'm not sure what to make of it.
Signed-off-by: Philip Langdale <philipl@overt.org> Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
avfilter/af_atempo: Make independent of the channel layout
Reviewed-by: Paul B Mahol <onemda@gmail.com> Reviewed-by: Pavel Koshevoy <pkoshevoy@gmail.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avformat/hlsenc: fix incomplete write of subtitle HLS playlist
This patch is a fix for the bug with ticket number 5812
(https://trac.ffmpeg.org/ticket/5812).
Basically, the bug is that the last segment and the "#EXT-X-ENDLIST"
tag are not being written to the subtitle playlist file when
converting a file with subtitles to the HLS format. The reason for
this seem to be that the filename for the subtitle file is being freed
too early. This solution is simply to free that memory after it's last
use in the "hls_window" function.
Signed-off-by: Franklin Phillips <franklinphillips9p8@inbox.lv> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
vf_colorspace: Allow overriding input color properties
The filter needs input frames with color properties filled out by
the decoder. Since this is not always possible, add input options to
the filter so that user may override color space, color primaries,
transfer characteristics, and color range, as well as a generic option
to set all properties at once.
For some reason, when compiling with gcc-asan and a recent enough gcc
version(seen on 5.3+ so far), linking dlopen works without -ldl, but
dlsym fails with:
undefined reference to symbol 'dlsym@@GLIBC_2.2.5'
So this patchs checks for both dlopen and dlsym to work for determining
if -ldl is needed.
Michael Graczyk [Thu, 1 Sep 2016 18:17:30 +0000 (11:17 -0700)]
libavcodec/opus: Add channel mapping 2 to extradata parser
This allows libavcodec/opus to demux ambisonics in an ogg/opus container.
Channel mapping family 2 is being added in this standards track IETF draft:
tools.ietf.org/html/draft-ietf-codec-ambisonics-00
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Mark Thompson [Wed, 31 Aug 2016 19:48:45 +0000 (20:48 +0100)]
lavu/timestamp: Avoid C++-unfriendly code in user header
Including this header in a C++11 program (inside extern "C") will
throw an error because it looks like a user-defined literal. Add a
space between the two tokens to avoid the problem.
Erkki Seppälä [Wed, 31 Aug 2016 11:35:44 +0000 (14:35 +0300)]
avformat/movenc: added ability to use original stream ids as track ids instead of regenerating them
Sometimes it's useful to be able to define the exact track numbers in
the generated track, instead of always beginning at track id 1. Using
the option use_stream_ids_as_track_ids now copies the use stream ids
to track ids. Dynamically generated tracks (ie. tmcd) have their track
numbers defined as continuing from the highest numbered stream id.
Signed-off-by: Erkki Seppälä <erkki.seppala.ext@nokia.com> Signed-off-by: OZOPlayer <OZOPL@nokia.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Steven Liu [Tue, 30 Aug 2016 09:26:54 +0000 (17:26 +0800)]
avformat/hlsenc: add warning for append_list and hls_init_time option
When use append_list mode, the hls_init_time set nouse,
Because the append_list only support append at the old m3u8 end
cannot set init segments durations at the middle of the list.
That's invalid. and show a warning message for user.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Simon Hailes [Tue, 30 Aug 2016 08:40:14 +0000 (09:40 +0100)]
avformat/crypto: add seeking support on read
cyrpto allows reading of data which has been aes-128-cbc encrypted given a key and an iv. But it did not handle filetypes which require seeking... e.g. it failed on an encrypted .mp4 file.
prior to this modification, the transcode would fail.
Note also: crypto previously maked both reads and writes as streamed, which caused the whole file
to be read before the transcode started. Now, for read only, if the underlying layer is not marked as streamed,
then crypto is not. This should enable efficient reading of encrypted containers which require seeking.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
projects / ffmpeg / log