Unlike other tf.*.conv2d layers, tf.nn.conv2d does not create many
nodes (within a scope) in the graph, it just acts like other layers.
tf.nn.conv2d only creates one node in the graph, and no internal
nodes such as 'kernel' are created.
The format of native model file is also changed, a flag named
has_bias is added, so change the version number.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
Allows the creation of the sdtp atom while remuxing MP4 to MP4. This
atom is required by Apple devices (iPhone, Apple TV) in order to accept
2160p medias.
The flac parser uses a fifo to buffer its data. Consequently, when
searching for sync codes of flac packets, one needs to take care of
the possibility of wraparound. This is done by using an optimized start
code search that works on each of the continuous buffers separately and
by explicitly checking whether the last pre-wrap byte and the first
post-wrap byte constitute a valid sync code.
Moreover, the last MAX_FRAME_HEADER_SIZE - 1 bytes ought not to be searched
for (the start of) a sync code because a header that might be found in this
region might not be completely available. These bytes ought to be searched
lateron when more data is available or when flushing.
Unfortunately there was an off-by-one error in the calculation of the
length to search of the post-wrap buffer: It was too large, because the
calculation was based on the amount of bytes available in the fifo from
the last pre-wrap byte onwards. This meant that a header might be
parsed twice (once prematurely and once regularly when more data is
available); it could also mean that an invalid header will be treated as
valid (namely if the length of said invalid header is
MAX_FRAME_HEADER_SIZE and the invalid byte that will be treated as the
last byte of this potential header happens to be the right CRC-8).
Should a header be parsed twice, the second instance will be the best child
of the first instance; the first instance's score will be
FLAC_HEADER_BASE_SCORE - FLAC_HEADER_CHANGED_PENALTY ( = 3) higher than
the second instance's score. So the frame belonging to the first
instance will be output and it will be done as a zero length frame (the
difference of the header's offset and the child's offset). This has
serious consequences when flushing, as returning a zero length buffer
signals to the caller that no more data will be output; consequently the
last frames not yet output will be dropped.
Furthermore, a "sample/frame number mismatch in adjacent frames" warning
got output when returning the zero-length frame belonging to the first
header, because the child's sample/frame number of course didn't match
the expected sample frame/number given its parent.
filter/hdcd-mix.flac from the FATE-suite was affected by this (the last
frame was omitted) which is the reason why several FATE-tests needed to
be updated.
Fixes ticket #5937.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
A threshold of 1 is sufficient for simple_dump_cut.webm, 10 is used
just to be sure the next truncated file doesnt cause the same issue
Obvious alternative fixes are to simply accept that the file is broken or to
write some advanced error concealment or to
simply accept that the decoder wont stop at the end of input.
Fixes: Ticket 8069 (artifacts not the differing md5 which was there before 1afd246960)
Fixes: simple_dump_cut.webm
Fixes: regression of 1afd246960
fate-vp5 changes because the last frame is truncated and now handled
differently.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Because the lavf_container is sometimes called with only 2 arguments,
fate tests produce bash errors like this:
tests/fate-run.sh: 299: test: =: unexpected operator
This commit fixes this.
Reviewed-by: Limin Wang <lance.lmwang@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Right now, the concat filter does not set the frame_rate value on any of
the out links. As a result, the default ffmpeg behaviour kicks in - to
copy the framerate from the first input to the outputs.
If a later input is higher framerate, this results in dropped frames; if
a later input is lower framerate it might cause judder.
This patch checks if all of the video inputs have the same framerate, and
if not it sets the out link to use '1/0' as the frame rate, the value
meaning "unknown/vfr".
A test is added to verify the VFR behaviour. The existing test for CFR
behaviour passes unchanged.
the info can be saved in dnn operand object without regenerating again and again,
and it is also needed for layer split/merge, and for memory reuse.
to make things step by step, this patch just focuses on c code,
the change within python script will be added later.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
This makes the code bitexact between platforms.
Intermediate timestamps between frames are preserved.
The timebase is simplified.
Rounding differs from doubles in cases where timestamps/durations
are "funny"
Suggested-by: jb
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This reverts commit a9dacdeea6.
This patch effectively made the decoder output vfr content out of samples
where cfr is expected.
Addresses ticket #7880.
Signed-off-by: James Almer <jamrial@gmail.com>
The packet counting based approach caused excessive sdt/pat/pmt for VBR, so
let's use a timestamp based approach instead similar to how we emit PCRs.
SDT/PAT/PMT period should be consistent for both VBR and CBR from now on.
Also change the type of sdt_period and pat_period to AV_OPT_TYPE_DURATION so no
floating point math is necessary.
Fixes ticket #3714.
Signed-off-by: Marton Balint <cus@passwd.hu>
background:
DNN (deep neural network) is a sub module of libavfilter, and FATE/dnn
is unit test for the DNN module, one unit test for one dnn layer.
The unit tests are not based on the APIs exported by libavfilter,
they just directly call into the functions within DNN submodule.
There is an issue when run the following command:
build$ ../ffmpeg/configure --disable-static --enable-shared
make
make fate-dnn-layer-pad
And part of error message:
tests/dnn/dnn-layer-pad-test.o: In function `test_with_mode_symmetric':
/work/media/ffmpeg/build/src/tests/dnn/dnn-layer-pad-test.c:73: undefined reference to `dnn_execute_layer_pad'
The root cause is that function dnn_execute_layer_pad is a LOCAL symbol
in libavfilter.so, and so the linker could not find it when build dnn-layer-pad-test.
To check it, just run: readelf -s libavfilter/libavfilter.so | grep dnn
So, add dependency in fate/dnn Makefile with ffmpeg static libraries.
This is the same method used in fate/checkasm
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
This fixes make fate issue for frame thread scale in my local testing
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
At the moment scene change detection score uses all planes to detect scene
changes. In this regard this is similar how the frozen frames detection works.
However, in classic encoding scene change detection typically only uses the Y
plane.
We might get more resonable scores for scene change if we also use only
the Y plane for calculating the score if the pixel format is YUV. Although
this will require additional work once packed YUV formats are added,
because for the moment the generic scene sad score calculation has no way
to ignore some components in a packed format.
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
cuda_runtime.h as well as dynlink_loader.h used nonstandard inclusion
guards with an AV_ prefix, although these files are not in an libav*/
path. So change the inclusion guards and adapt the ref file of the
source fate test accordingly.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
why change .4 to .25, it's for:
one scenecut(pkt_pts=20040) isn't detected by 0.4 threshold
why not change to 0.3 instead of 0.25:
it will miss the scenecut(pkt_pts=20040) after applying the next
patch which enables yuvj420
for fate testing, it's better to catch all scenecut scenes.
Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
The tests previously rounded the timestamps. Its better in a fate test to preserve
the data from the demuxer and decoder.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Commit cd48318035 added support for NV24 and NV42, including several
fate tests for these formats, but did not include the reference files
for the tests filter-pixdesc-nv24 and filter-pixdesc-nv42. As a result,
these two tests were broken.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The implementation is pretty straight-forward. Most of the existing
NV12 codepaths work regardless of subsampling and are re-used as is.
Where necessary I wrote the slightly different NV24 versions.
Finally, the one thing that confused me for a long time was the
asm specific x86 path that did an explicit exclusion check for NV12.
I replaced that with a semi-planar check and also updated the
equivalent PPC code, which Lauri kindly checked.
These are the 4:4:4 variants of the semi-planar NV12/NV21 formats.
These formats are not used much, so we've never had a reason to add
them until now. VDPAU recently added support HEVC 4:4:4 content
and when you use the OpenGL interop, the returned surfaces are in
NV24 format, so we need the pixel format for media players, even
if there's no direct use within ffmpeg.
Separately, there are apparently webcams that use NV24, but I've
never seen one.
Up until now, the length field of most level 1 elements has been written
using eight bytes, although it is known in advance how much space the
content of said elements will take up so that it would be possible to
determine the minimal amount of bytes for the length field. This
commit changes this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Given that in both the seekable as well as the non-seekable mode dynamic
buffers are used to write level 1 elements and that now no seeks are
used in the seekable case any more, the two modes can be combined; as a
consequence, the non-seekable mode automatically inherits the ability to
write CRC-32 elements.
There are no differences in case the output is seekable; when it is not
and writing CRC-32 elements is disabled, there can still be minor
differences because before this commit, the EBML ID and length field
were counted towards the cluster size limit; now they no longer are.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now the EBML Header length field has been written with eight
bytes, although the EBML Header is always so small that only one byte
is needed for it. This patch saves seven bytes for every Matroska/Webm
file.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
The spec in https://xiph.org/vorbis/doc/v-comment.html states that
the metadata keys are case-insensitive, so don't change the case
and update the fate test case.
Fix#7784
Reviewed-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
The transcode() helper function will already prepend the TARGET_PATH to
the sample path, if its a relative path. This avoids an issue on
Windows, where the relative path check could fail.
write_tmcd allows tmcd track to be created with any mode but in
mov_write_header, index for first tmcd track is only set for modes
MP4 or MOV, causing a crash if tmcd creation is attempted with other
modes.
* commit 'f8df5e2f31a5ba7b30a0e1caaaf5a03c753b3f9b':
tests: Add a convenience function for video-only lavf tests
Merged-by: James Almer <jamrial@gmail.com>
* commit 'a70eac7a9b193e8434b5bed90bd72aa4cb688363':
tests: Convert image2pipe tests to non-legacy test scripts
Merged-by: James Almer <jamrial@gmail.com>
When a JACOsub subtitle has two timestamps, they represent its start and
end times (http://unicorn.us.com/jacosub/jscripts.html#l_times); the
duration is the difference between the two, not the sum of the two.
The subtitle end times in the FATE test for this were wrong as a result;
fix them too. (This test is based on JACOsub's demo.txt, and the end
time computed for the last line using @ now matches what the comments
there say it should be.)
Also tested in practice using MPV, a LaserDisc, and some authentic 1993
JACOsub files.
Signed-off-by: Adam Sampson <ats@offog.org>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If we fill with black then the generated palette will have one color more
than what the user requested. This also resulted in unwanted black specks in
the output of paletteuse, especially when generating small palettes.
Use av_ts2str() for AVFrame.pkt_dts/pts to avoid print the
pkt_dts/pts as negative number like:
"0, 3616613, -9223372036854775808, 1001, 3110400, 0x75e37a65"
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
The VP3/4/5/6 reference decoders all use three IDCT versions: one for the
DC-only case, another for blocks with more than 10 coefficients, and an
optimised one for blocks with up to 10 AC coefficents. VP6 relies on the
sparse 10 coefficient version, and without it, IDCT drift occurs.
Fixes: https://trac.ffmpeg.org/ticket/1282
Signed-off-by: Peter Ross <pross@xvid.org>
init add three test examples:
1. check no endlist at the end
2. check endlist at the end
3. check hls_list_size 0 full list
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
Change the some options location in avcodec_options to make code more
readable. And update the fate test with this change.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Now "-c copy" works.
Update FATE files.
Demuxer only split file into packets, no data is trimmed.
Encoder & muxer currently expect completely another format
where muxer writes stuff like disposal method which should
be really encoder job.
With this patch muxer only modifies delay between two packets.
Codec copy need to have same behavior between demuxer and
muxer to work correctly.
Fixes#6640.
The header guards were unnecessarily non-standard and the c file
inclusion trick means the files dont't have standard licence
headers.
Based on a patch by: Martin Vignali <martin.vignali@gmail.com>
This is needed because of 32bit float formats (which are difficult to
store in 16bits)
This also fixes undefined behavior found by fate
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
ISMV lacks any sort of edit list support, as well as tfxd is
effectively the PTS of the fragment for most intents and purposes.
Thus, if b-frames are requested without negative CTS offsets you
end up with N frames' worth of delay (tfxd PTS plus the CTS offset
of the first sample). Negative CTS offsets enable the first sample
to have CTS=DTS, and thus a/v desync due to b-frame reorder delay
is avoided.
Fixes vorbis mp4 audio files, with edit list specified. Since
st->skip_samples is not set in case of vorbis , ffmpeg computes the
start_time as negative.
Signed-off-by: Sasi Inguva <isasi@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Add tests for upmixing and downmixing with audio channel counts that
have a corresponding default layout and also tests where there is no
default layout.
Update the existing "stereo4" test so it actually outputs stereo like
the other stereo tests. Rename the previous "stereo4" test into
"upmix1".
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
Set make variable KEEP to non-zero value to preserve temp files
when a test has passed.
Helpful in diagnosing failed tests when test outfile is some type of
single hash and does not reveal differences in processed output.
da9cc22d5b allowed the MOV muxer to relay a custom stream handler name,
whether populated from the input stream or user-set. However, the entry
key didn't match the key set by the MOV demuxer, so it wasn't
effective. Fixed.
Due to the change, four FATE refs have to be updated. Verified that the
target payload of the tests hasn't changed in terms of CRC.
verify that the stco atom is upgraded to co64 when the addition of moov
size to the offsets results in an overflow
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If start_time is not set, ffmpeg takes the duration from the global
movie instead of the per stream duration.
Signed-off-by: Sasi Inguva <isasi@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Generic C implementation of vf_blend performs reads and writes of 16-bit
elements, which requires the buffers to be aligned to at least 2-byte
boundary.
Also, the change fixes source buffer overrun caused by src_offset being
added to to test handling of misaligned buffers.
Fixes: #7226
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This improves performance and makes qtrle behave more similar to other decoders.
Libavcodec does generally not output known duplicated frames, instead the calling Application
can insert them as it needs.
Fixes: Timeout
Fixes: 6383/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QTRLE_fuzzer-6199846902956032
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This uses any devices it can find on the host system - on a system with no
hardware device support or in builds with no support included it will do
nothing and pass.
This new optional flag makes it easier to deal with mpegts
samples where the PMT is updated and elementary streams move
to different PIDs in the middle of playback.
Previously, new AVStreams were created per PID, and it was up
to the user to figure out which streams had migrated to a new PID
(by iterating over the list of AVProgram and making guesses), and
switch seamlessly to the new AVStream during playback.
Transcoding or remuxing these streams with ffmpeg on the CLI was
also quite painful, and the user would need to extract each set
of PIDs into a separate file and then stitch them back together.
With this new option, the mpegts demuxer will automatically detect
PMT changes and feed data from the new PID to the original AVStream
that was created for the orignal PID. For mpegts samples with
stream_identifier_descriptor available, the unique ID is used to
merge PIDs together. If the stream id is not available, the demuxer
attempts to map PIDs based on their position within the PMT.
With this change, I am able to playback and transcode/remux these
two samples which previously caused issues:
https://tmm1.s3.amazonaws.com/pmt-version-change.tshttps://kuroko.fushizen.eu/videos/pid_switch_sample.ts
I also have another longer sample in which the PMT changes
repeatedly and ES streams move to different pids three times
during playback:
https://tmm1.s3.amazonaws.com/multiple-pmt-change.ts
Demuxing this sample with the new option shows several new log
messages as the PMT changes are handled:
[mpegts] detected PMT change (program=1, version=3/6, pcr_pid=0xf98/0xfb7)
[mpegts] re-using existing video stream 0 (pid=0xf98) for new pid=0xfb7
[mpegts] re-using existing audio stream 1 (pid=0xf99) for new pid=0xfb8
[mpegts] re-using existing audio stream 2 (pid=0xf9a) for new pid=0xfb9
[mpegts] detected PMT change (program=1, version=6/3, pcr_pid=0xfb7/0xf98)
[mpegts] detected PMT change (program=1, version=3/4, pcr_pid=0xf98/0xf9b)
[mpegts] re-using existing video stream 0 (pid=0xf98) for new pid=0xf9b
[mpegts] re-using existing audio stream 1 (pid=0xf99) for new pid=0xf9c
[mpegts] re-using existing audio stream 2 (pid=0xf9a) for new pid=0xf9d
[mpegts] detected PMT change (program=1, version=4/5, pcr_pid=0xf9b/0xfa9)
[mpegts] re-using existing video stream 0 (pid=0xf98) for new pid=0xfa9
[mpegts] re-using existing audio stream 1 (pid=0xf99) for new pid=0xfaa
[mpegts] re-using existing audio stream 2 (pid=0xf9a) for new pid=0xfab
[mpegts] detected PMT change (program=1, version=5/6, pcr_pid=0xfa9/0xfb7)
Signed-off-by: Aman Gupta <aman@tmm1.net>
Generates color bar test patterns based on EBU PAL recommendations.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
Adds tests for the hue angle and brightness filter parameters.
Renames the existing saturation parameter test for consistency.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
The artificial sample file sei-1.h264 contains five frames (IDR P B I B)
and the following SEI message types:
* Buffering period
* Picture timing
* Pan-scan rectangle (display as 4:3)
* User data registered, containing A/53 closed captions (captions match
frame content, including reordering)
* Recovery point (at the I frame)
* Display orientation (identity transformation)
* Mastering display (with arbitrary contents)
* Undefined SEI type 1234 (containing ascending bytes)
fix the warning: "function declaration isn’t a prototype", in C
int foo() and int foo(void) are different functions. int foo()
accepts an arbitrary number of arguments, while int foo(void) accepts 0
arguments.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Uses the same mechanism as other codecs - conformance test files are
passed through the metadata filter (which, with no options, reads the
input and writes it back) and the output verified to match the input.
The specs says that the the first color component in the color array is
not alpha, but simply 0.
Fixes 0 alpha of fate-suite/cvid/catfight-cvid-pal8-partial.mov
Signed-off-by: Marton Balint <cus@passwd.hu>
The track's media duration from the mdhd atom takes precedence
over both the stts and elst atom for calculating and setting
the track's total duraion.
Technically, we shouldn't be using the stts atom at all for
calculating stream durations.
This fixes incorrect stream and final packet durations on files
with edit lists that are longer than the media duration.
The FATE changes are expected, and output is more correct (the
AAC frame is not 1028 samples).
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Add previously omitted overlap smooting and loop filtering for
frame/field-interlace pictures. For progressive pictures switch to the
re-implemented versions of overlap smooting and loop filtering.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The existing implementation did out-of-bounds reference pixel replication for
progressive reference frames. In interlaced reference frames both the even and
odd line on the horizontal edges need to be replicated.
Fixes#3262.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
- Parse schm atom to get different encryption schemes.
- Allow senc atom to appear in track fragments.
- Allow 16-byte IVs.
- Allow constant IVs (specified in tenc).
- Allow only tenc to specify encryption (i.e. no senc/saiz/saio).
- Use sample descriptor to detect clear fragments.
This doesn't support:
- Different sample descriptor holding different encryption info.
- Only first sample descriptor can be encrypted.
- Encrypted sample groups (i.e. seig).
- Non-'cenc' encryption scheme when using -decryption_key.
Signed-off-by: Jacob Trimble <modmaker@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Some ADTS streams can have multiple ID3 tags between frames. This
change parses all of them, rather than just the first one.
Signed-off-by: Mattias Amnefelt <mattiasa@avm.se>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avdevice_register_all() is still required to register devices into
lavf (this is required due to lavd being somewhat of a hack).
Signed-off-by: Josh de Kock <josh@itanimul.li>
On modern x86 systems its around 2x faster. For systems without
FPUs it'll be slower, but our policy is to prefer floating point
implementations and to let users decide what's best (or just not
compile them on systems without FPUs).
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Set relevant filter parameters such that the result can easily be
checked with a waveform editor.
In particular, it makes it clear the silence_start is not accurate in
the current code.
test extract color and alpha
with the three main kind of hap frame :
- no snappy compression
- snappy compression and one chunk
- snappy compression and several chunks (16 here)
like the bsf filter need to be used with vtag and encoder edition
also test the information of the target mov for color and alpha
This adds a way for an API user to transfer QP data and metadata without
having to keep the reference to AVFrame, and without having to
explicitly care about QP APIs. It might also provide a way to finally
remove the deprecated QP related fields. In the end, the QP table should
be handled in a very similar way to e.g. AV_FRAME_DATA_MOTION_VECTORS.
There are two side data types, because I didn't care about having to
repack the QP data so the table and the metadata are in a single
AVBufferRef. Otherwise it would have either required a copy on decoding
(extra slowdown for something as obscure as the QP data), or would have
required making intrusive changes to the codecs which support export of
this data.
The new side data types are added under deprecation guards, because I
don't intend to change the status of the QP export as being deprecated
(as it was before this patch too).
enable dump bit stream filter and update opt fate test ref.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Thanks for the discussion. Here's the next version, now with /25 and removed
ff_log2().
The blocksize of the PCM decoder is hard-coded. This creates
unnecessary delay when reading low-rate (<100Hz) streams. This creates
issues when multiplexing multiple streams, since other inputs are only
opened/read after a low-rate input block was completely read.
This patch decreases the blocksize for low-rate inputs, so
approximately a block is read every 40ms. This decreases the startup
delay when multiplexing inputs with different rates.
Signed-off-by: Philipp M. Scholl <pscholl@bawue.de>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes seek for files with empty edits and files with negative ctts
(dts_shift > 0). Added fate samples and tests.
Signed-off-by: Sasi Inguva <isasi@isasi.mtv.corp.google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
To make the best use of existing code, I generalised the wrapper
that currently does yuv420p10 to p010 to support any mixture of
input and output sizes between 10 and 16 bits. This had the side
effect of yielding a working code path for all yuv420p1x formats
to p01x.
External headers are no longer welcome in the ffmpeg codebase because they
increase the maintenance burden. However, in the NVidia case the vanilla
headers need some modifications to be usable in ffmpeg therefore we still
provide them, but in a separate repository.
The external headers can be found at
https://git.videolan.org/?p=ffmpeg/nv-codec-headers.git
Fate-source is updated because of the deleted files, and dynlink_loader.h
license headers were updated with the standard FFmpeg headers.
Signed-off-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
If configure fails before config.fate is generated, the report file misses
some values and gets discarded by the FATE server. In these cases, print
those values as "failed" along with the failing configure command line.
This is needed by later hwaccel code to tell which encoding process was
used for a particular frame, because hardware decoders may only support a
subset of possible methods.
These tests cover specific rounding behaviour, to ensure that I don't
introduce any regressions with the rewritten "activate" callback based
fps filter.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In 16x8 motion compensation, for lower 16x8 region, the input to mpeg_motion() for motion_y was "motion_y + 16", which causes wrong rounding. For 4:2:0, chroma scaling for y is dividing by two and rounding toward zero. When motion_y < 0 and motion_y + 16 > 0, the rounding direction of "motion_y" and "motion_y + 16" is different and rounding "motion_y + 16" would be incorrect.
We should input "motion_y" as is to round correctly. I add "is_16x8" flag to do that.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For B field pictures, the spec says,
> The prediction shall be made from the field of the same parity as the field being predicted.
I did it.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is done mainly in preparation for the SIMD patches.
- for the 8-bit input, decrease the blend factor precision to 7-bit.
- for the 16-bit input, increase the blend factor precision to 15-bit.
- make sure the blend functions are not called with 0 or maximum blending
factors, because we don't want the signed factor integers to overflow.
Fate test changes are due to different rounding.
Signed-off-by: Marton Balint <cus@passwd.hu>
<jamrial> durandal_1707: 8088b5d69c broke the acrossfade test
<@durandal_1707> jamrial: there was test?
<jamrial> durandal_1707: fate-filter-acrossfade
<@durandal_1707> what broke?
<jamrial> what used to be one frame is now two
<@durandal_1707> ahh, just update test
Signed-off-by: James Almer <jamrial@gmail.com>
It tests a useless profile which sounds no better than regular aac and which
takes extremely long to encoder something. Also it has been behind experimental
flag for as long as it has been supported.
Should be removed altogether sometime in the future.
The twoloop coder sounds decent at low bitrates, however at higher bitrates
it sounds worse than the fast coder (which used to be the old twoloop coder
before October 2015) and needs quite a lot more CPU.
Change the default to fast. It has been well tested and has had little changes
over the years so its been confirmed to be quite stable.
Also change its description (not valid for more than a year) and the
documentation.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
The framerate filter was quite convoluted with some filter_frame /
request_frame logic bugs. It seemed easier to rewrite the whole filter_frame /
request_frame part and also the frame interpolation ratio calculation part in
one step.
Notable changes:
- The filter now only stores 2 frames instead of 3
- filter_frame outputs all the frames it can to be able to handle consecutive
filter_frame calls which previously caused early drops of buffered frames.
- because of this, request_frame is largely simplified and it only outputs
frames on flush. Previously consecuitve request_frame calls could cause the
filter to think it is in flush mode filling its buffer with the same frames
causing a "ghost" effect on the output.
- PTS discontinuities are handled better
- frames with unknown PTS values are now dropped
Fixes ticket #4870.
Probably fixes ticket #5493.
Signed-off-by: Marton Balint <cus@passwd.hu>
The PERSIST_RPARAM_A_RExt_Sony_1 bitstream has an out-of-range value
and has therefore been superseded.
It is otherwise identical, and decodes the same.
Signed-off-by: James Almer <jamrial@gmail.com>
It was truncated to int later on anyway. Fate test changes are due to rounding
instead of truncation.
Fixes fate test failures on x86-32 (gcc 4.8 (Ubuntu 4.8.5-2ubuntu1~14.04.1))
after 090b740680.
Signed-off-by: Marton Balint <cus@passwd.hu>
- normalize score to [0..100] instead of [0..85]
- change the default score to 8.2 to roughly keep existing behaviour
- take into account bit depth
- do not truncate to integer
Signed-off-by: Marton Balint <cus@passwd.hu>
Check fread return value to fix build warning as "ignoring
return value of ‘fread’"
Signed-off-by: Jun Zhao <jun.zhao@intel.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Resulted in valgrind errors due to uninitialized memory.
Also updates fate and makes it use the tron sample result.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Every bitstream filter behaves as intended now, so there's no need to
wait for the first packet of every stream.
Signed-off-by: James Almer <jamrial@gmail.com>
Also change note to say that we compare against the officially decoded
samples rather than our own, this was changed long ago.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
The current edit unit cannot be reliably determined for the last packet of a
video stream, because we can't query the start offset of the next edit unit
from the index. This caused missing timestamps for the last video packet.
Therefore from now on, we allow setting the PTS even if we are not sure of the
current edit unit if mxf_set_current_edit_unit returned a specific failure, and
the assumed current edit unit is the last.
Fixes last packet timestamp of:
ffprobe -fflags nofillin -show_packets tests/data/lavf/lavf.mxf -select_streams v
Signed-off-by: Marton Balint <cus@passwd.hu>
Writes one set of field framing information for progressive streams and
two sets for interlaced streams. Fixes ticket #6383.
Unfortunately the OpenDML v1.02 document is not very specific on what
value to use for start_line when frame data is not coming from a
capturing device, so this is just using 0/1 depending on the field order
as a best-effort guess.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
