The original cleanup of frame_id_alloc_mask was done in free_frame_priv
which may be called after the start_frame for the next few frames,
causing the frame_id slots to be exhausted. Instead, decide the in-use
frame_id slots by checking the frame_id present in the DPB as we need it.
We always use two slots, even when only one L0 reference is supported
by the driver. However we still need to set the correct value for the
ref_order_hint of the second slot.
Fixes regression from bf9f921ef7 ("avcodec/hw_base_encode: restrict size of next_prev")
vaapi_get_image_format() will be called in vaapi_map_frame().
So it's a double check, drop it to avoid redundancy.
Signed-off-by: nyanmisaka <nst799610810@gmail.com>
There's a race condition for YUV420P when mapping from pix_fmt
to VA fourcc, both I420 and YV12 could be found by pix_fmt.
Currently, vaapi_get_image_format() iterates over the query results
of pix_fmt and returns the first matching result in the order
declared in the driver. This may result in an incorrect image_format.
Now use fourcc to find the image_format.
Fixes:
ffmpeg -hwaccel vaapi -hwaccel_output_format vaapi -i INPUT \
-vf scale_vaapi=format=yuv420p,hwmap,format=yuv420p \
-vframes 1 -f rawvideo -y yuv420p.yuv
Signed-off-by: nyanmisaka <nst799610810@gmail.com>
Fixes: index 50 out of bounds for type 'INTFLOAT [40][2]'
Fixes: 401661737/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_LATM_fuzzer-4866055713652736
Someone knowing AAC well should review this, there is likely a nicer fix
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Instead of undconditionally using the first input. This covers the case of
one layer fully obscuring another layer, in which case that should become
the new "base" layer.
This prevents leaking stale metadata from previous frames, for example if
an overlay temporarily obscures this input and then un-obscures it again. It
is worth pointing out that this does change the semantics subtly, because of
the smoothing period on detected HDR metadata, but I argue that the new
behavior is an improvement, as it will avoid leaking past metadata that is
definitely no longer relevant after an image is unobscured.
Sometimes, one input fully obscures another. In this case, we can skip
actually rendering any input below the obscuring one.
The reason I don't simply start the main render loop at `idx_start` will
become apparent in the following commit.
We can't use pl_frame_is_cropped() on this dummy frame, but we need to
determine the reference frame before we can map the real output, so to
resolve this conflict, we just reimplement the crop detection logic using
the output link dimensions.
It is possible for pl_queue_update() to return PL_QUEUE_OK, but to generate
an empty frame mix. This happens if the first frame of that input is in the
future.
In this case, we should skip an input as not active, similar to inputs that
have already reached EOF.
Instead of copying over the entire target and changing a few fields,
set the entire struct to a whitelist of safe properties that we want to
persist on the intermediate texture.
In particular, this avoids leaking irrelevant state related to the
acquire/release callbacks, e.g., which can otherwise cause deadlocks
when the same vulkan frame is attempted to be acquired twice.
This creates a new codec id for mxf vbi_vanc_smpte_436M streams.
This makes it easier to use from other [de]muxers and bitstream filters.
It's just the data in Table 7 (starts on page 13) of:
https://pub.smpte.org/latest/st436/s436m-2006.pdf
Signed-off-by: Jacob Lifshay <programmerjake@gmail.com>
They were not sorted before, sort using: LC_ALL=C sort --ignore-case
since that's what file-contents-sorter (added in next commit) supports.
Signed-off-by: Jacob Lifshay <programmerjake@gmail.com>
This commit adds a ProRes RAW hardware implementation written in Vulkan.
Both version 0 and version 1 streams are supported.
The implementation is highly parallelized, with 512 invocations dispatched
per every tile, with generally 4k tiles on a 5.8k stream.
Thanks to unlord for the 8-point iDCT.
Benchmark for a generic 5.8k RAW HQ file:
6900XT: 63fps
7900XTX: 84fps
6000 Ada: 120fps
Intel: 9fps
