On Tue, 09 Mar 2021 13:53:20 -0300 James Almer <jamr...@gmail.com> wrote:
> On 3/9/2021 1:38 PM, Niklas Haas wrote:
> > + if (s->skip_av1_grain)
> > + image.av1_grain = (struct pl_av1_grain_data) {0};
>
> [...]
>
> > + if (!s->skip_av1_grain)
> > + av_frame_remove_side_data(out, AV_FRAME_DATA_FILM_GRAIN_PARAMS);
>
> [...]
>
> > + { "skip_av1_grain", "Disable AV1 grain application",
> > OFFSET(skip_av1_grain), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DYNAMIC },
>
> How does this work? Does pl_upload_avframe() support parsing the
> filmgrain side data? If so, that's pretty cool.
Yes. More specifically, `pl_frame_from_avframe` translates the film
grain side data into `pl_av1_grain_data`, and `pl_render_image` applies
film grain using a compute shader.
> Also, I'd prefer if this was inverted. So make it filmgrain or maybe
> apply_filmgrain, and of course set it to 1 by default. And it doesn't
> need to specify it's for AV1 since the AV_FRAME_DATA_FILM_GRAIN_PARAMS
> side data type could be extended to support the H.274 implementation in
> the future.
This makes sense. Changed. Upon further consideration, I'm also not sure
if this should really be enabled by default. As I understand it, this
side data is only present if the user specifically disabled film grain
application in the decoder, so overriding that by just moving it to
vf_libplacebo seems like a surprising default.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
