So I have the entire Star Trek TNG series on blu-ray. I decided to do a 4K upscale of the entire series with Proteus set to Auto, AV1 as encoder, bitrate set to Auto, audio set to Copy, and mkv as the container.
I was initially processing each episode one at a time, it usually goes at about 7.9-8.4fps, each episode taking roughly 2 hours 22minutes.
Well, I decided to try 2 episodes at a time and I’m getting 5.0-5.2fps per episode process. And so 2 episodes at a time takes about 3 hours 33minutes.
The whole thing was a bit counterintuitive, but it’s actually faster this way. Is the reason for this a limitation on how the program processes things? Like why can’t one episode at a time go 10.4fps?
I’ve often wondered why as well. The Topaz devs are well aware of this and if they wanted to they are quite capable of applying that speed gain to a single video. By splitting the video in 2 and processing both halves in parallel and joining up afterwards. Given that all this can be done by the user in the CLI, there’s absolutely no reason or excuse for Topaz not to do this “behind the scenes”.
One day, perhaps a TVAI user with some experience will build their own app which takes the TVAI export command and alters it to apply this split video parallel processing method. Either for a single GPU or even multiple GPUs. Now there’s a thought…
