AV1@Scale: Film Grain Synthesis, The Awakening

It's a bit frustrating that the footage is first shot, then denoised in post, then renoised in post, then denoised in encoding and then renoised at decoding.

Only if you are aware of it, which 99.9% of people consuming video content are not. It’s simply an unimportant implementation detail (from a viewer’s perspective who doesn’t really care about bitrate-as-cost).

You worry too much, it's all fake. What you think is the "footage" will be many layers from various sources composited together.

Fake lights, fake shadows, fake sky, ...