Topaz Video AI Early Access v3.0.0-5

Hello Everyone,

We have another update for Topaz Video AI 3.0 with lots of fixes and improvements. Please provide your feedback.

Windows | Mac
Headers, Dylibs, & Libs for custom FFmpeg builds

Changes from v3.0.0-4

  • Multi-select for inputs (full screen view only), previews, and exports
    • Use shift, control/command click for selection. Right click selects all
  • Filters may be applied to batches of videos
  • Stop and re-start processing
  • Order of processing may be changed
  • Failed or errored previews & exports may be reprocessed
  • Faster queue operations for input/preview/exports
  • Playhead position is now kept for each video
  • Gaia and Theia models work on CPU & iGPU as expected

Known Issues

  • Preview mismatch for certain VOB files
  • Some videos can still have some seeking issues in Preview/Export
  • Color space / HDR issues
  • Keyboard navigation missing
  • Intel codec not working on some machines
  • Some quality mismatch between 2.6.4 and 3.0
  • Output bitrate is not reset to Preferences bitrate when importing a new file
  • Different view mode is still missing
  • Failure can happen when resizing to a big resolution like 4k or 8k without a warning if machine runs out of memory
  • Speed can be slower than 2.6.4 with some models on some machines. We’ll be working on speed optimization in the next weeks
  • Proteus with Auto + Stabilization may cause artifacts
  • Preview with image sequence doesn’t work or plays at 25fps
  • Aspect ratio crop missing

Problems and Solutions:

  • Make sure Max. Memory Usage in the Preferences is set to a high number
  • If the Gaia model fails on Mac, try re-installing the app using the DMG file
  • If export fails, try using Transcode/None/Auto in Audio Settings in Preferences

Proteus Auto Caveats
The adjustments will not be visible and the sliders will show values of “0” when in auto mode. However, each frame is being analyzed for optimal settings while processing. If you keep Auto enabled and modify the sliders, the settings you choose will be value-adjustment percentages. For example, if Auto estimates the Sharpen slider with a value of 50, and you manually select -50, then the actual value applied will be 25.

CLI Usage
CLI mode has returned for Topaz Video AI 3. Detailed usage instructions are available here, on our support website .

Frequently Asked Questions can be found here .

Please upload videos that are not working or having any type of conflicts to this DropBox Folder .



Progress bar colour also needs to be changed to something other than the same colour as the selected export job!


  • some of the issue on Exit the software while processing was still going on has been fixed. the “close video” on the export page, now offer a " are you sure", when an export is processed…

but one error still remain lol… we can still close one video in the source list, even if the same video is processed on the export page ! so if you have videos exporting, that during this time you import more video, and launch some previews, and by mistake hit the close video button… well… everything is lost !
please, one more warning message to write !!!

  • lost of tiny stuff have been fixed in this release, but are not written in the change’s would be nice to write them too !

  • two things about the batch export page !
    1/ one suggestion : selecting a video select it, even several videos ! very quick way to do it, love it ! but to unselect one, we need to do control + left click ! it would be much faster i think to be able to unselect ONE video but just click on it again ! (the control + click can stay of course).

2/ double clicking on one of the video, selected or not, provide an exit on the preview page and reset the whole batch selection. For someone who would import 100 videos to do batch or more, unselect several ones, and by mistake double clicking on it, would have to redo it totally. maybe this feature is not the best… and the double clicking must be removed to avoid such mistakes.


Had to run it a few times just to be sure but i’ll be damned…it’s finally clearing the temp folder on exit.

Nvmd, now it’s not…idk what’s going on. Also why is fullscreen still not full screen…you can still grab the edges.

This version forces the models and json files back to the old “ProgramData” location regardless of settings in preferences. (Windows version)

Because it’s using a poor, non-standard Window manager.

Does the Reduce Jittery Motions option finally work on Stabilization when processing longer videos?

Just wondering, will there be a live preview based on the current frame? Because setting parameters is done blindly, I need to constantly preview to see what I am setting. For me, it would be enough to show the current frame, but I’d like to see the effect of the change (like in the case of say Gigapixel AI or Photo AI).
So will there be a live preview of the current frame?


@suraj @ida.topazlabs why is the encoding always so slow on export? when are you really going to increase the encoding speed? thank you

1 Like

I have no idea what’s going wrong. The program is still useless. Factor 10x slower and the filter quality is still bad. It’s a mystery to me why people go online with such an early version. That’s totally frightening!? Disappointed greetings


I stay on the old version for me, 2.6.4, the version 3 is totally a regression of VEAI which is sad. :cry:


I’m only waiting for stabilization to be fixed and then I go back to 2.64 as Artemis & Moire is all I need. Only model that doesn’t introduce painted details.

1 Like

I’m agree with you for that :wink:

The fact that now 2 files or set of images are now being created at the same time (one for the preview) is a big part of that slowdown.
We absolutely need a “disable preview” option when exporting.


VEAI does not take advatage of resources available
AMD Ryzen 7 5800x @4GHz 32GB RAM + NVIDIA RTX 3060 @1897 MHz 12GB memory

Load is only:
CPU ffmpeg.exe 23%, VEAI 8%; RAM 13,6/32GB
GPU 51%; Memory 37%


After testing this version I feel a bit confused. No matter what processing mode I choose, the improvement is minor and barely visible, even if I ramp up several additional options. edit: it looked better in my second attempt. :eyes:

I also wonder what an application is good for that matches the demands of other third party tools when it is lacking of basic functions, stability and usability?! Would I really care about if the Porsche is compatible with Ford if it is still not able to stable drive a straigt line?!

  • processing a second instance fails on my 8 GB RTX even when I lower the RAM usage in the preferences and there is a noticeable color shift in the output still. “stop processing” ended the processing with no option to continue. a pause option was not yet found instead.

Welcome to the club of frustrated users of alpha/beta/whatsoever prerelease versions of VEAI something

You are right. I originally thought VE AI gets a new GUI by keeping the old functionality. Instead of testing the new GUI with the old functionality I spend most of the time with struggeling around with the new GUI and missing of the old functionality. :worried:


Looks like I’m the only testing this, and it still don’t work.

“Picture size 4294967288x4294967292 is invalid
Error reinitializing filters!
Failed to inject frame into filter network: In
2022-09-21-22-31-3-Main.tzlog (2.8 MB)
valid argument
Error while processing the decoded data for stream #0:0”

One more try with the Reduce Jittery Motions and another crash after 7%. C: drive gets filled, even if the temp folder has been set to drive D:. Maybe it’s the virtual memory that is getting filled. Although I have a 32GB of RAM, and I think that should be enough when processing 29.97fps 720x544 file with the same output size and frames.

2022-09-22-07-14-49-Main.tzlog (1.7 MB)