[FFmpeg-devel] [PATCH] [libdav1d.c]: Add option to output all the spatial layers.

[Ronald S. Bultje]

Hi,

On Thu, Nov 14, 2019 at 2:12 PM Andrey Semashev <andrey.semashev at gmail.com>
wrote:

> I think there needs to be some consistency across different lavc
> decoders. If we consider that lavc should produce one decoded frame per
> one encoded one, even if the encoded one contains multiple layers, then
> that should be true for all decoders.
>

Yes. I think one thing that would help is if we had access to more samples
with an expected behaviour. Right now we may have samples, but if all we do
is check md5 without caring what it means, then it's kind of pointless.


> Also, I think having decoded frames from all layers would also be
> useful, but there should be a way to know which layer they belong to.
> AFAIK, lavc currently doesn't provide that information. This mode of
> operation (producing frames for all layers) should be optional.


I agree.

Ronald