[benwaggoner]

Yes, this should absolutely work. The grain synthesis model is really sidecar metadata and a post-decode renderer. It's nothing encoding or AV1 specific.