I guess that means people can go and run the reference code and start comparing the results for real now.
Hopefully an adaptation will get added to FFMPEG and Handbrake so we can have a play with it too.
now
Full-length encodes will take days to finish, and we are not talking single digits.
An almost, but not really, practical encoder will probably take at least a year of development before it’s ready.
And anyway, do end-users still care about codecs (beyond decoding complexity/hardware support)?
Typical end users do not. Companies do because it will save them money.
Enthusiasts will care because it could save them storage space for equivalent quality, though if the cost of encoding is so high then just in terms of energy costs you may save money just going for a cheaper codec and upgrading storage with the saved money.
Enthusiasts will care because it could save them storage space for equivalent quality
That’s the thing, even if you ignore that such scenarios involve lossy-to-lossy re-encodes (bad), and even when you ignore the general lack of psychoacoustic tuning in new encoders, the advertised so called objective “20%-30% improvement” is not universal, and only applies to bit-starved resolution-maxed encodes.
Your file is 1080p or 720p? you won’t get that improvement, even in not-fit-for-purpose “objective” measures.
You want to encode at a higher bitrate than YouTube to actually get good quality? you won’t get that improvement either.
So if you embark on such a futile journey, you could be wasting a lot of computing power for no, or even negative, gain.
Yeah as a result of this post I decided to look into AV1 grain synthesis again. That’s the only way I can see that AV1 can meaningfully “improve” over h265 right now for noisy content like movies, which is what enthusiasts care about.
Grain synthesis is where you analyze noise on the original file, denoise the video stream and reapply artificial noise of a similar style as the original at decode time.
It relies on a lot of assumptions:
- The denoise doesn’t kill your quality
- The grain synthesis looks convincingly like real film grain.
- decoding devices will support rendering the grain
From a short experiment, I found that VLC was able to render the synthetic grain, but MPV(.net) did not out of the box. I had to play around with gpu rendering modes to get it to work.
As for transcoding, it’s unclear what happens to the synthetic grain, whether it is burnt in or simply ignored. At least one or two people have reported it will just be ignored and you’ll get a weirdly smooth movie.