Topaz Video AI v3.0.6

Hello everyone!

Another release of Topaz Video AI is now available.

Released December 6th, 2022

Windows | Mac (.dmg) | Mac (.pkg)

Changes from v3.0.5:

  • Autoplay disabled in preview
  • Default FPS for image sequences can be changed
  • Fixed H.265 encoders on NVIDIA GTX cards
  • Version number in Get Info on Mac is now correct
  • Known Issues

    • Some users may experience reduced performance compared to 2.6.4
    • Preview experience needs to be improved
    • Attempting to export without any filters selected may result in an error

    File Submission Dropbox: Submit Files


    Even more? How bad can it get?


    Any progress on this…


    Still only one audio track (first) is being copied from source file.

    This is ridiculous.

    I had hopes that this is something very minor that can be resolved quickly, but just lost the 30d refund period and there is no human-friendly way to reach the support really.


    Thank you for all your hard work. Can you explain the audio settings please.
    Only auto works for me

    1 Like

    The drop in file don’t work as well yet on version 3. It worked on version 2

    Stand corrected this new version fixed it thank you :grinning:

    It works on my Laptop but not on my main pc :angry:

    1 Like

    So we’re still doing release versions that are lower numbered than the previous beta? :\

    1 Like

    you’re a beta tester, so you have access to the thread. one of the developper, Suraj, just replied to this question and some other. I don’t know if i’m allowed to quote him here but it would be good that someone from the team does it here then.

    My point is, that comment about adding a Pause option was near 12 months ago. Beta does not have a pause button. I purchased this software on release of 3.0 and it includes 12 months update rights. I feel they will have to extend this for everyone as the current versions are still raw beta versions and I cannot process video as required. Slow, crashing, failing to start exports with unknown errors. Beta software is not Production ready release for purchase. Feeling like guinea pigs…


    you asked a question on a specific thing. I just told that someone replied about it, that’s all. I’m not the one to tell the other things.

    Great job. I have a question. I read the Support thing, but the definition of “Max Processes” is kinda vague.

    Should I set it to 2? Which is the default option, or more if I have a 4 core CPU?

    What will it exactly change in terms of stability / performance? So far I have only tested it with a HD 630 IGP and no difference was found, but it´s a low power GPU anyway. Thank you!

    Just in case, I render videos one by one, so no batch rendering is used in my case.


    With 3.0.6, the entire preview is black now, with no apparent way to get it back. :frowning: See below.

    I have an i9 12900k, 64G RAM, and a RTX 3080 Ti (Windows 11).

    The release notes say something like ‘Autoplay disabled in preview’, except I see no way to enable it again (or whether it’s even related). At least 3.0.5 showed something in Preview


    GPU high load is not always the best performance solution.

    I checked that with a progressive 1920x1080 50fps input enhanced 2x to 3840x2160 50fps

    1. Proteus Relative to Auto:
      ~37% GPU; 30%CPU; 2.0fps
    2. Gaia HQ:
      ~90%GPU; 14% CPU; 1.6fps

    Output quality at 200% in both cases no remarkable difference

    Is there ever going to be a ‘Turbo’ option for those of us that don’t need or want to do batches of videos?

    Basically, a mode that will allow the software to use as much CPU/GPU and RAM as the hardware will allow to do do a single video file as quick as possible?

    I really like my software to fully use the expensive hardware I purchased to "make things faster! in the first place!

    It’s frustrating to work on a 10 SD minute video and see only 20% of resources being used and 30 minutes on the clock to tick down.


    When MPEG2-TS is processed with ddv3

    • Error message from AI engine: model failed…

    I’m having a hard time creating videos since October.

    Request for beta tester sent

    Thank you johnnystar from the previous thread

    As you can see from this screenshot (Windows Taskmanager + CPUID Hardware Monitor) a PC with 8 cores (=16 processors) does load individual processors up to 100% while the average in total is below 30%. It all depends on how many parallel processers are requested by one vector/tensor task to get finished before the next can start.

    Yes but can I have an option to use as much GPU and CPU as possible on one file?

    That’s all I (and probably a lot of casual users) want.

    Why do I have to wait an hour at 30% GPU/CPU usage when I could wait like 15 minutes (whatever) at 90% usage?

    1 Like