Useful in cases audio samples DC offset is not ~0.0, where
other detectors will fail to detect silence.

According to Dave Airlie:

> <airlied> but I think ignoring it should be fine, I can't see any
> other way to get the imaeg extents correct for other usage
> <Lynne> what width/height should be used for the images?
> the final presentable dimensions, or the coded dimensions?
> <airlied> if you don't want noise I think the presentable dims
> <airlied> the driver should round up the allocations internally,
> but if you are going to sample from the images then w/h have to be
> the bounds of the image you want
> <airlied> since otherwise there's no way to stop the sampler from
> going outside the edges

Apparently, the alignment values are informative, rather than mandatory,
but the spec's wording makes it sound as if they're mandatory.

Signed-off-by: James Almer <jamrial@gmail.com>

Fixes warnings on x86-32 and Windows.
Signed-off-by: James Almer <jamrial@gmail.com>

This prevents multiple NULL accesses - if yuv_sampler exists, then
everything required for it to be destroyed also exists.

Previously floats where scaled up to 32bit int, but floats do not
have 32bits in their mantisse so a quarter of the bits where nonsense.

It seems no fate test is affected by this change, which is interresting
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Also reduce neeeded temporary storage by half
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Old code used about 7 times as many cpu cycles as new
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Old code needed about 6 times as long as new with defaults in afftfilt
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

The new function is much more precise
For default beta it is slightly slower, but its speed is already at the
worst case in that comparison
while the replaced function becomes much slower for larger beta
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

0th order modified bessel function of the first kind are used in multiple
places, lets avoid having 3+ different implementations
I picked this one as its accurate and quite fast, it can be replaced if
a better one is found
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Fixes: signed integer overflow: -9223372036854775808 - 2082844800 cannot be represented in type 'long'
Fixes: 58384/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6428383700713472

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpegSigned-off-by: Michael Niedermayer <michael@niedermayer.cc>

This commit documents most of the Vulkan filters. Some of this was
copy-pasted from equivalent OpenCL filters.

When seeking through MBAFF-coded H264, this can happen. Decoding calls end_frame
without calling start_frame. We are unable to decode this, as no frame
state has been set.

Happens for both VAAPI and Vulkan. Could be an issue elsewhere, hence
the individual commit.

There are circumstances where the flag isn't set but the skip mode
frames are. So don't use the inferred bit which has other inputs
when deciding to pass the skip mode frames to the device.

This fixes some decoding bugs on intel av1

Thanks to Dave Airlie for figuring out a lot of the parameters.

Thanks to Dave Airlie for figuring out a lot of the parameters.

For compatibility with vf_libplacebo

For two reasons:
1. We now create a vulkan 1.3 device
2. libplacebo master currently requires a vulkan 1.3 device