Also deduplicate printing json and summary output.
Reviewed-by: Kyle Swanson <k@ylo.ph>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Descriptor buffers were a neat attempt at organizing descriptors.
Simple, robust, reliable.
Unfortunately, driver support never caught on, and neither did validation
layer support.
Now they're being replaced by descriptor heaps, which promises to be
the future. We'll see how it goes.
fixup 80229c1
[scale_vulkan @ 0000028b1c2c1300] scale:31: error: 'texture' : no matching overloaded function found
scale:31: error: 'return' : cannot convert return value to function return type
Signed-off-by: nyanmisaka <nst799610810@gmail.com>
loudnorm provides stats output that's meant to be used for two-pass
normalization. These stats are often interleaved with ffmpeg's stream
descriptions and other output, making them difficult to parse and pass
to the second pass. The new stats_file option enables writing the stats
to a separate file, or to standard output, for simple parsing and other
programmatic usage.
Signed-off-by: Adam Jensen <adam@acj.sh>
Given we now align both dimensions when allocating buffers, don't reinitialize
the pool when dealing with dimension changes that will not affect the existing
pool size.
Signed-off-by: James Almer <jamrial@gmail.com>
Initial checkin of OCIO filter.
Initial checkin of OCIO filter.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Change for the right C++ library, should work on linux too.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Adding inverse when using display/view.
Removed comments.
Removed code that was setting the CICP values. Hopefully this can be done through OCIO at some point.
Config cleanup - need a modified require_cpp to handle namespacing.
Switch to using require_cpp so that namespace can be used.
Adding documentation.
Sadly a bit of linting went in here, but more importantly added a threads option to split the image into horizontal tiles, since OCIO was running rather slow.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Adding context parameters.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Add the OCIO config parameter.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Make the min threads 1 for now, reserve 0 for later if we can automatically pick something.
Also added a few comments.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
This is using ffmpeg-slicing.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Adding OCIO filetransform.
Making sure everything is using av_log rather than std::cerr.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Updating the tests so they would work without additional files.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Adding the file-transform documentation.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Adding copyright/license info.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Removing tests, since this is optional code.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Code cleanup.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Typo.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
I went the wrong way, av_log is expecting \n
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Fix indenting to 4 spaces.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Fixing lint issues and a spelling mistake.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Code formatting cleanup to match conventions.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Whitespace removal.
Signed-off-by: Sam.Richards@taurich.org <Sam.Richards@taurich.org>
Starting with PyTorch 2.6, the hooks API was redesigned so it no longer depends on the device type.
As part of this change, the XPU initialization function was renamed from initXPU() to init().
Add version check to support both old and new LibTorch versions.
Signed-off-by: younengxiao <steven.xiao@amd.com>
When using `dnn_processing` filter with torch backend, FFmpeg hangs indefinitely because no inference is actually performed.
Resolve this problem by add "else" branch for synchronous execution path.
Usage:
ffmpeg -i input.mp4 -vf scale=224:224,format=rgb24,dnn_processing=dnn_backend=torch:model=sr_model_torch.pt:device=cpu output.mp4
Fixes peak detection for samples like in https://github.com/haasn/libplacebo/issues/358
This might introduce flickering on some other samples, but was agreed
upstream that they are better overall.
Signed-off-by: Kacper Michajłow <kasper93@gmail.com>
Return the data in an AVFrame instead. This is what several users
({find,cover}_rect*) want anyway. This also avoids accessing
AVFrame.format (an int) via an enum AVPixelFormat*.
*: This commit actually avoids two frame copies For find_rect:
av_frame_clone() contained an implicit alloc+copy.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Also rename it to ff_pixfmt_is_in(). This is more type-safe;
in particular, it is required to support -fshort-enum.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
A constant named 8 has value 8 is redundant. These constants cannot
prevent user from setting other vaues. Just describe valid values
in the help message.
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
In config_output() for direction=DU/RL, the position
is initialized to s->sono_size, which equals h or w
when bar=0. That position is later used as an
in-bounds pixel coordinate without clamping in draw(),
causing writes past the end of the output planes.
Repro:
ffmpeg -f lavfi -i sine=frequency=1000:sample_rate=44100 \
-filter_complex "[0:a]showcwt=s=640x512:bar=0:direction=du[v]" \
-map "[v]" -frames:v 1 -f null -
AddressSanitizer: heap-buffer-overflow ... WRITE of size 1
Initialize and wrap the DU/RL position to sono_size - 1
(or 0 when empty), preventing out-of-bounds row/column
writes when bar=0 while preserving existing slide behavior.
Add chroma_location option so that, in the subsampled-to-subsampled case, the destination's chroma siting can be changed from the source's without having to use other filters. Useful, for example, when converting BT.2020 to BT.709, where the former customarily uses "top left" and the latter "left."
Update documentation.
Closes: #21185
The limiter_buf was being allocated with buf_size (3000ms worth of
samples) instead of limiter_buf_size (210ms worth of samples).
This resulted in allocating approximately 14x more memory than needed
for the limiter buffer. At 192kHz stereo:
- buf_size: 1,152,000 samples = 9.2 MB
- limiter_buf_size: 80,640 samples = 0.6 MB
- Wasted: ~8.2 MB per filter instance
The bug was introduced when the limiter buffer was added but the
wrong size variable was used in av_malloc_array().
This patch adds screen capture support via AMF vsrc_amf:
ffmpeg -y \
-filter_complex "vsrc_amf=framerate=120:capture_mode=keep_framerate" \
-c:v hevc_amf \
-frames:v 300 \
output_grab.mp4
If the HW frames pool is insufficient,
increase extra_hw_frames in filter_complex.
This avoids undefined behavior when -fshort-enums is enabled, where accessing enum arrays through int pointers causes incorrect memory reads.
Signed-off-by: niyinghao <niyinghao@xiaomi.com>
We use INT32_MIN/MAX so as to ensure we dont have to deal with 64bit width or height
on a int is int64 system. int64 width would overflow in a system where we assume the product of 2
values fit in int64
Fixes: #YWH-PGM40646-14
Found-by: An0n99X
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Update execute_model_th to support asynchronous execution by pushing tasks to the pending queue and notifying the worker thread.
Signed-off-by: Raja Rathour <imraja729@gmail.com>
Reviewed-by: Wenbin Chen <wenbin.chen@intel.com>
Reviewed-by: Guo Yejun <yejun.guo@intel.com>
