Read rate enforcement delayed till first decoded frame is obtained, to
speed up init of output streams.
Thanks to Linjie Fu <linjie.justin.fu@gmail.com> for the initial patch.
if input start time is not 0 -t is inaccurate doing stream copy,
will record extra duration according to input start time.
it should base on following cases:
input video start time from 60s, duration is 300s,
1. stream copy:
ffmpeg -ss 40 -t 60 -i in.mp4 -c copy -y out.mp4
open_input_file() will seek to 100 and set ts_offset to -100,
process_input() will offset pkt->pts with ts_offset to make it 0,
so when do_streamcopy() with -t, exits when ist->pts >= recording_time.
2. stream copy with -copyts:
ffmpeg -ss 40 -t 60 -copyts -i in.mp4 -c copy -y out.mp4
open_input_file() will seek to 100 and set ts_offset to 0,
process_input() will keep raw pkt->pts as ts_offset is 0,
so when do_streamcopy() with -t, exits when
ist->pts >= (recording_time+f->start_time+f->ctx->start_time).
3. stream copy with -copyts -start_at_zero:
ffmpeg -ss 40 -t 60 -copyts -start_at_zero -i in.mp4 -c copy -y out.mp4
open_input_file() will seek to 120 and set ts_offset to -60 as start_to_zero option,
process_input() will offset pkt->pts with input file start time,
so when do_streamcopy() with -t, exits when ist->pts >= (recording_time+f->start_time).
0 60 40 60 360
|_______|_____|_______|_______________________|
start -ss -t
This fixes ticket #9141.
Signed-off-by: Shiwang.Xie <shiwang.xie666@outlook.com>
xmllint (silently) replaces the ' with " when fixing and validating the output
of ffprobe in fate-ffprobe_xsd.
Reviewed-by: Tobias Rapp <t.rapp@noa-archive.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Besides being unnecessary it is also safer: If the error for an
unrecognized option were triggered (which seems to be impossible right
now), it might be that the stream whose codecpar is accessed is NULL.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Otherwise the rate emulation logic in `transcode_step` never gets
hit, and the unavailability flag never gets reset, leading to an
eternal loop with some rate emulation use cases.
This change was missed during the rework of ffmpeg.c, in which
encoder initialization was moved further down the time line in
commit 67be1ce0c6 . Previously,
as the encoder initialization had happened earlier, this state was
not possible (flow getting as far as hitting the rate emulation logic,
yet not having the encoder initialized yet).
Fixes#9160
Use AVSTREAM_EVENT_FLAG_NEW_PACKETS instead, which should provide the
same information in this case.
Finishes removing all uses of this field as started by 87f0c8280c.
Signed-off-by: James Almer <jamrial@gmail.com>
This also allows to exclusively use pointers to const AVCodec in
fftools/ffmpeg_opt.c.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Deprecated in c29038f304.
The resample filter based upon this library has been removed as well.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: James Almer <jamrial@gmail.com>
The "packets_and_frames" element has been added to ffprobe.xsd in
0c9f0da0f7 but apparently removing the
check in ffprobe.c has been forgotten.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
The MJPEG encoder supports some pixel format/color range combinations
only when strictness is set to unofficial or less. Before commit
059fc2d9da said encoder's pix_fmts array
only included the pixel formats supported with default strictness.
When strictness was <= unofficial, fftools/ffmpeg_filter.c used
an extended list of pixel formats instead of the encoder's including
the pixel formats only supported when strictness <= unofficial.
Said commit turned the logic around: The encoder's pix_fmts array now
included all pixel formats and fftools/ffmpeg_filter.c instead used
a small list of all pixel formats supported when strictness is >
unofficial and the encoder's pixel formats instead. In particular,
the codec's pix_fmt is not used when strictness is normal.
This works for the mjpeg encoder; yet it did not work for other
(hardware-based) mjpeg encoders, because the check for whether one is
using the MJPEG encoder is wrong: It just checks the codec id.
So if one used strict unofficial with a hardware-accelerated MJPEG
encoder before commit 059fc2d9da, the unofficial (non-hardware)
pixel formats of the MJPEG encoder would be used; since said commit
the codec's pixel formats are overridden at ordinary strictness
by the ordinary MJPEG pixel formats. This leads to format conversion
errors lateron which were reported in #9186.
The solution to this is to check AVCodec.name instead of its id.
Fixes ticket #9186.
Tested-by: Eoff, Ullysses A <ullysses.a.eoff@intel.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently said list contains only the pixel formats that are always
supported irrespective of the range and the value of
strict_std_compliance. This makes the MJPEG encoder an outlier as all
other codecs put all potentially supported pixel formats into said list
and error out if the chosen pixel format is unsupported. This commit
brings it therefore in line with the other encoders.
The behaviour of fftools/ffmpeg_filter.c has been preserved. A more
informed decision would be possible if colour range were available
at this point, but it isn't.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Only one character is actually rewritten.
Fixes truncation warnings, such as
warning: ‘strncpy’ output truncated before terminating nul copying 3 bytes from a string of the same length [-Wstringop-truncation]
in gcc 10.2.0
This cap is currently used to mark multithreading-capable codecs that
wrap external libraries with their own multithreading code. The name is
highly confusing for our API users, since libavcodec ALWAYS handles
thread_count=0 (see commit message in previous commit). Therefore rename
the cap and update its documentation to make its meaning clear.
The old name is kept deprecated until next+1 major bump.
If the window is resized it was possible that xpos pointed outside the
visualization texture. By rearranging the overflow check we make sure this (and
a crash) does not happen.
We also don't have to use xleft for start position, as that is 0 anyways, and
if we ever want to take into account xleft then the texture should be
positioned accordingly when rendering.
Signed-off-by: Marton Balint <cus@passwd.hu>
The obstacle to do so was in filter_codec_opts: It uses searches
the AVCodec for options via the AV_OPT_SEARCH_FAKE_OBJ method, which
requires using a void * that points to a pointer to a const AVClass.
When using const AVCodec *, one can not simply use a pointer that points
to the AVCodec's pointer to its AVClass, as said pointer is const, too.
This is fixed by using a temporary pointer to the AVClass.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
It only affects the old and deprecated avcodec_decode_(video2|audio4)
API which is no longer used here.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
avcodec_find_best_pix_fmt_of_2 has been moved to libavutil in
617e866e25.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The FF_API macros are private and must not be used by external callers.
As the fields in question are to be removed without replacement, just
drop them.
The fields are:
AVPacket.convergence_duration
AVCodecContext.time_base
AVCodecContext.timecode_frame_start
AV_PIX_FMT_FLAG_PSEUDOPAL pixel descriptor flag
The metadata company_name, product_name, product_version from input
file will be deleted to avoid overwriting information
Please to test with below commands:
./ffmpeg -i ../fate-suite/mxf/Sony-00001.mxf -c:v copy -c:a copy out.mxf
and
./ffmpeg -i ../fate-suite/mxf/Sony-00001.mxf -c:v copy -c:a copy \
-metadata company_name="xxx" \
-metadata product_name="xxx" \
-metadata product_version="xxx" \
out.mxf
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
The st->codec values are updated based on the lowres factor by
avformat_find_stream_info() when it runs an instance of the decoder internally,
and the same thing happens in ffmpeg.c when we open ist->dec_ctx with
avcodec_open2(), so these assignments are redundant.
Signed-off-by: James Almer <jamrial@gmail.com>
As per signal() help (man 2 signal) the semantics of using signal may
vary across platforms. It is suggested to use sigaction() instead.
Reviewed-by: Zane van Iperen <zane@zanevaniperen.com>
Signed-off-by: Andriy Gelman <andriy.gelman@gmail.com>
The first stats is printed after the initial stats_period has elapsed. With a large period,
it may appear that ffmpeg has frozen at startup.
The initial stats is now printed after the first transcode_step.
At present, progress stats are updated at a hardcoded interval of
half a second. For long processes, this can lead to bloated
logs and progress reports.
Users can now set a custom period using option -stats_period
Default is kept at 0.5 seconds.
They add considerable complexity to frame-threading implementation,
which includes an unavoidably leaking error path, while the advantages
of this option to the users are highly dubious.
It should be always possible and desirable for the callers to make their
get_buffer2() implementation thread-safe, so deprecate this option.
We now have the possibility of getting AVFrames here, and we should
not touch the muxer's codecpar after writing the header.
Results of FATE tests change as the MXF and Matroska muxers actually
write down the field/frame coding type of a stream in their
respective headers. Before this change, these values in codecpar
would only be set after the muxer was initialized. Now, the
information is also available for encoder and muxer initialization.
Additionally, reap the first rewards by being able to set the
color related encoding values based on the passed AVFrame.
The only tests that seem to have changed their results with this
change seem to be the MXF tests. There, the muxer writes the
limited/full range flag to the output container if the encoder
is not set to "unspecified".
- For video, this means a single initialization point in do_video_out.
- For audio we unfortunately need to do it in two places just
before the buffer sink is utilized (if av_buffersink_get_samples
would still work according to its specification after a call to
avfilter_graph_request_oldest was made, we could at least remove
the one in transcode_step).
Other adjustments to make things work:
- As the AVFrame PTS adjustment to encoder time base needs the encoder
to be initialized, so it is now moved to do_{video,audio}_out,
right after the encoder has been initialized. Due to this,
the additional parameter in do_video_out is removed as it is no
longer necessary.
This way the old max queue size limit based behavior for streams
where each individual packet is large is kept, while for smaller
streams more packets can be buffered (current default is at 50
megabytes per stream).
For some explanation, by default ffmpeg copies packets from before
the appointed seek point/start time and puts them into the local
muxing queue. Before, it getting utilized was much less likely
since as soon as the filter chain was initialized, the encoder
(and thus output stream) was also initialized.
Now, since we will be pushing the encoder initialization to when the
first AVFrame is decoded and filtered - which only happens after
the exact seek point is hit as packets are ignored until then -
this queue will be seeing much more usage.
In more layman's terms, this attempts to fix cases such as where:
- seek point ends up being 5 seconds before requested time.
- audio is set to copy, and thus immediately begins filling the
muxing queue.
- video is being encoded, and thus all received packets are skipped
until the requested time is hit.
The user has no business modifying the underlying AVCodec.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The AVFilterInOuts normally get freed in init_output_filter() when
the corresponding streams get created; yet if an error happens before
one reaches said point, they leak. Therefore this commit makes
ffmpeg_cleanup free them, too.
Fixes ticket #8267.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Threaded input can increase smoothness of e.g. x11grab significantly. Before
this patch, in order to activate threaded input the user had to specify a
"dummy" additional input, with this change it is no longer required.
Signed-off-by: Marton Balint <cus@passwd.hu>
This useful, because by ffprobe's very nature, you use it to probe
a file and find out what it is. Requiring every format private option
to be known to the demuxer forces one to run ffprobe twice, if one
wants to use ffprobe in a generic way.
For example, say one wants to probe all user-uploaded files, while
also ignoring edit lists for any MP4s that are uploaded. Currently,
you'd have to run ffprobe twice: once to identify the format, and
once again to actually probe the metadata you want. After this
patch, you could set -ignore_editlist 1 on every call and only
probe once.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Currently, ffmpeg inserts scale filter by default in the filter graph
to force the whole decoded stream to scale into the same size with the
first frame. It's not quite make sense in resolution changing cases if
user wants the rawvideo without any scale.
Using autoscale/noautoscale as an output option to indicate whether auto
inserting the scale filter in the filter graph:
-noautoscale or -autoscale 0:
disable the default auto scale filter inserting.
ffmpeg -y -i input.mp4 out1.yuv -noautoscale out2.yuv -autoscale 0 out3.yuv
Update docs.
Suggested-by: Mark Thompson <sw@jkqxz.net>
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
The previous code here did not handle passing a frames context when
ffmpeg itself did not know about the device it came from (for example,
because it was created by device derivation inside a filter graph), which
would break encoders requiring that input. Fix that by checking for HW
frames and device context methods independently, and prefer to use a
frames context method if possible. At the same time, revert the encoding
additions to the device matching function because the additional
complexity was not relevant to decoding.
Also fixes#8637, which is the same case but with the device creation
hidden in the ad-hoc libmfx setup code.
This can support encoders which want frames and/or device contexts. For
the device case, it currently picks the first initialised device of the
desired type to give to the encoder - a new option would be needed if it
were necessary to choose between multiple devices of the same type.
The data of an attachment file is put into an AVCodecParameter's
extradata. The corresponding size field has type int, yet there was no
check for the size to fit into an int. As a consequence, it was possible
to create extradata with negative size (by using a big enough max_alloc).
Other errors were also possible: If SIZE_MAX < INT64_MAX (e.g. on 32bit
systems) then the file size might be truncated before the allocation;
and avio_read() takes an int, too, so one would not have read as much
as one desired.
Furthermore, the extradata is now padded as is required.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When QSV is enabled in FFmpeg, the command "ffmpeg -hwaccels" shows a
duplicate entry in acceleration methods for QSV:
Hardware acceleration methods:
vaapi
qsv
drm
opencl
qsv
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
PRId64 and PRIu64 already expand to the complete specifier; adding
another 'd' at the end is wrong and just leads to warnings that say
that only an option like '-frames:v 2d' will be used, although said
option won't be accepted at all ('Expected int64 for frames:v but found
2d').
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Each time the sub2video structure is initialized, the sub2video
subpicture is initialized together with the first received heartbeat.
The heartbeat's PTS is utilized as the subpicture start time.
Additionally, add some documentation on the stages.
It's a duplicate of the properly implemented nvdec libavcodec hwaccel
Reviewed-by: Timo Rothenpieler <timo@rothenpieler.org>
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes#8094.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -9223372036854775808 - 9223372036854775807 cannot be represented in type 'long'
Fixes: Ticket8142
Found-by: Suhwan
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For audio packets with dts != AV_NOPTS_VALUEs the dts was converted
twice to the muxer's timebase during streamcopy, once as a normal
packet and once specifically as an audio packet. This has been changed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
fix crash when used the command like:
- ffmpeg -h protocol
- ffmpeg -h protocol=
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
After 06ec9c4746 we check for these
functions in configure (which will succeed in cygwin), but cmdutils.c
only includes windows.h if _WIN32 is defined (which it isn't in cygwin).
Retain the old intent from before 06ec9c4746,
that these functions only would be used when _WIN32 is defined, while
only using them if configure has agreed that they do exist.
Signed-off-by: Martin Storsjö <martin@martin.st>
Before this fix, ffmpeg -h full | grep map get the command dump
like:
-map [-]input_file_id[:stream_specifier][,sync_file_id[:stream_s set input stream mapping
^
|
truncated
after this fix, we can get full option dump.
Found-by: vacingfang <vacingfang@tencent.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
CPB side_data is copied when stream-copying (see init_output_stream_streamcopy()),
but it shall not be copied when the stream is decoded.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In these cases, we must pass the full path of the file to ffprobe
(as the current working dir on the remote system, e.g. when invoked
with "ssh remote ffprobe ..." isn't the wanted one).
The input filename passed to ffprobe is also included in the output,
which is part of the reference test data. Add a new option to
ffprobe to allow overriding what path is printed, to keep the
original relative path in the tests.
An alternative approach could be an option to allow requesting omitting
the file name from the dumped data, and updating the test references
accordingly.
Signed-off-by: Martin Storsjö <martin@martin.st>
These functions aren't available when building for the restricted
UWP/WinRT/WinStore API subsets.
Normally when building in this mode, one is probably only building
the libraries, but being able to build ffmpeg.exe still is useful
(and a ffmpeg.exe targeting these API subsets still can be run
e.g. in wine, for testing).
Signed-off-by: Martin Storsjö <martin@martin.st>
Freeing this was forgotten in ad899522.
Fixes#8315 and #8316.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
add error handle if av_asprintf return null.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
The "type" entry was hardcoded with an trailing comma, even if it was
the only entry in the section.
Fixes ticket #8228.
Signed-off-by: James Almer <jamrial@gmail.com>
do_streamcopy() has a packet that gets zero-initialized first, then gets
initialized via av_init_packet() after which some of its fields are
oerwritten again with the actually desired values (unless it's EOF): The
side data is copied into the packet with av_copy_packet_side_data() and
if the source packet is refcounted, the packet will get a new reference
to the source packet's data. Furthermore, the flags are copied and the
timestamp related fields are overwritten with new values.
This commit replaces this by using av_packet_ref() to both initialize
the packet as well as populate its fields with the right values (unless
it's EOF again in which case the packet will still be initialized). The
differences to the current approach are as follows:
a) There is no call to a deprecated function (av_copy_packet_side_data())
any more.
b) Several fields that weren't copied before are now copied from the source
packet to the new packet (e.g. pos). Some of them (the timestamp related
fields) may be immediately overwritten again and some don't seem to be
used at all (e.g. pos), but in return using av_packet_ref() allows to forgo
the initializations.
c) There was no check for whether copying side data fails or not. This
has been changed: Now the program is exited in this case.
Using av_packet_ref() does not lead to unnecessary copying of data,
because the source packets are already always refcounted (they originate
from av_read_frame()).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The option tables of the various fftools (in particular ffprobe) are
arrays of OptionDef; said type contains a union of a pointer to void and
a function pointer of type int (*)(void *, const char *, const char *)
as well as a size_t. Some entries (namely the common entry for writing a
report as well as several more of ffprobe's entries) used the pointer to
void to store a pointer to functions of type int (*)(const char *) or
type int (*)(const char *, const char *); nevertheless, when the functions
are actually called in write_option (in cmdutils.c), it is done via a
pointer of the first type.
There are two things wrong here:
1. Pointer to void can be converted to any pointer to incomplete or
object type and back; but they are nevertheless not completely generic
pointers: There is no provision in the C standard that guarantees their
convertibility with function pointers. C90 lacks a generic function
pointer, C99 made every function pointer a generic function pointer and
still disallows the convertibility with void *.
2. The signature of the called function differs from the signature
of the pointed-to type. This is undefined behaviour in C99 (given that
C90 lacks a way to convert function pointers at all, it doesn't say
anything about such a situation). It only works because none of the
functions this patch is about make any use of their parameters at all.
Therefore this commit changes the type of the relevant functions
to match the type used for the call and uses the union's function
pointer to store it. This is legal even in C90.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Respect default disposition when select audio/video
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
The implementation will use some default in this case. The empty string
is not a meaningful device for any existing hardware type, and indeed
OpenCL treats it identically to no device already to work around the lack
of this setting on the command line.
This patch improves the logs when the message "cur_dts is invalid" appears.
If helps to identify which stream generates the trouble,
and the status of the stream.
A lot of users suffers with the message, and the origin varies.
The improved message can help to discover the cause.
Signed-off-by: Andreas Hakon <andreas.hakon@protonmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Right now, the code check for no filter description, but if we use a
filter_complex, the code will use the AVFrame.duration which could be
wrong in case of using fps filter.
How to reproduce the problem:
ffmpeg -f lavfi -i testsrc=duration=1 -vf fps=fps=50 -vsync 1 -f null -
output 50 frames
ffmpeg -f lavfi -i testsrc=duration=1 -filter_complex fps=fps=50 -vsync 1 -f null -
output 51 frames
With this commit, the same command will always output 50 frames.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This way re-initializations properly update end_pts, enabling
sub2video_heartbeat to call sub2video_update as expected to re-init
the sub2video AVFrame's contents and to feed a frame into the filter
chain.
This then fixes memory usage ballooning due to framesync waiting
for secondary input in case of no actual subtitle samples being present
for a while in source after a re-init occurs.
Fully discarded streams can't be selected for output or mapped or filtered.
Previously, a few packets from such streams, probably buffered for
stream probing, would get smuggled into output files.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Previously if only -y was used then nothing happened, if only -x then zero
window height was set which is undefined in SDL and caused a black window.
From now on if only one dimension is set we calculate the other based on
size/aspect of the shown content.
Signed-off-by: Marton Balint <cus@passwd.hu>
PTS is in microseconds, so correct field name is out_time_us.
Old field out_time_ms kept for now - will be removed after a suitable transition
period.
Fixes#7345
Currently extra_hw_frames can't be applied to qsv since it
doesn't call function avcodec_get_hw_frames_parameters().
Give an option to fix ticket #7261 though it is not a perfect soultion
(allocate the minimum pool size internally and automatically).
Signed-off-by: Zhong Li <zhong.li@intel.com>
SDL from version 2.0.8 has support for full range YUV and specifying
BT601/BT709 color space for YUV->RGB conversion.
Signed-off-by: Marton Balint <cus@passwd.hu>
