I need this too. My PC is in my bedroom, and the noise is quite considerable. I guess I could split my video in chunks, render them separately, and then merge them? Anyone ever tried that?
Since this is a feature request for “Video Enhance AI” (2.x), do we need a separate one for “Video AI” (3.x)? Is there even a chance to get it for 2.x? I’m currently using both since they have different problems/features.
Honestly I find it bewildering that this hasn’t been a feature from the very start. It seems to me that this would be heavily requested considering how long a single video can render. In my case it’s more than 2 days for a 2 houre video.
Bro. Something’s wrong. I have an i9 12900K, 64G RAM also, and a RTX 3080 Ti. Upscaling a full-length 1080p movie to 4K, on VEAI 3.0.4, takes ca. 23 hours. That is still a long time. But 10+ f-ing days on your system is insane. That can’t be right.
He uses a Quadro M6000 , not an A6000 or something - so 10 days+ is quite imaginable. The M6000 is a very old workstation card, even a GeForce 1660 Super is faster, it’s only 5% faster than the 980 TI, so you can imagine the relative performance. It has a nice memory size, but its processing power is more than just outdated…
For situations like yours, I recently found a very useful little progam, called Process Lasso. With it, you can extremely easily suspend processes, such as TVAI (why, the process you actually suspend will be ffmpeg). And resume afterwards.
On my i9 12900K, I now also use Process Lasso [Pro] to ‘bind’ ffmpeg’s core affinity (for every time ffmpeg is running) to cores 0-15 (all P-Cores), and TVAI itself to all E-Cores, as the latter basically does little else than call ffmpeg. Same for command line x265 processes now: I bound x265 to all P-Cores only (needed for speed), so I no longer have to park all E-Cores, which are now free to use for the rest of my machine.
I use Video AI on Mac, and I’m currently upscaling a 4k video to 8k to be able to zoom in a bit in processing later. The Macbook Pro with M1Max still needs 1-2 days for this due to the length of the video clip (concert recording).
I also need a pause method for this. I can’t have Video AI continue to draw 90 watts when I’m on battery power and need the laptop to work on something else. I even needed to stop the processing at one point at about 75% progress, hoping I could resume, but I only could start it again from scratch, which overwrote the already processed 75% output file, so I couldn’t even decide to use that and skip the rest.
I looked at them, but they usually work on M1/M2 by pinning CPU-intensive tasks to the efficiency cores. Topaz Video AI however leaves most of the CPU capacity unused, and runs almost exclusively on the GPU and ML cores, which AFAIK none of these tools can manage. That’s what I’m seeing on performance monitoring tools.
There is already a (roundabout) way to suspend and resume TVAI processing on a Mac (may work on Windows too?) without any additional software. Instead of using the “Export” button within TVAI use the menu items Process | Show Export Command then “Copy”. Then use Process | Open Command Prompt. Then paste the copied command into the Terminal window and Enter. That should start the export process. Now you can use CTRL-Z to suspend the ffmpeg processing. To resume, type fg (and Enter). In order to completely stop the ffmpeg process while it’s running, use CTRL-C.
It’s ridiculous that this feature has not been added yet. So many new things, some of them useless (like the UI rework) yet this remains untackled.
Anyway I will not be renewing my licence and stay on the fallback version until this has been released.
And if that is of any importance to you I suggest you do the same.