Topaz Video AI 5.2.1

Ask them - do they have a licensed product or obtained the other way?

1 Like

It continues to be impossible to do more than one preview of a video, because the app tries to combine the second preview with the first. IOW, two 10 second previews 30 mins apart on the timeline become a single 30 min 20 sec preview. You have to delete your first completed preview before you can do another one.

6 Likes

Officially purchased license, of course.

2 Likes

is there a way to use the new rhea model with a more stable version like v5.0.4 ?

I never saw or could use the new Rhea Model, cause a 5 seconds preview would cause 30 minutes render time in my case even I have a high performance system with high GPU performance.

There is something wrong in 5.2, cause also the other models have now very high rendering times.
In the older version it tooks me for 5 seconds only some seconds for rendering th preview in all models.

1 Like

think i found what caused it, it was the integrated GPU disabled (i always have it disabled).
when its re-enabled in Device manager, it stopped switching to AUTO after 1 preview.
idk why it mattered but it did.

This sounds logic.
I wait for the next working version, cause I don’t would adjust things in my windows system only for Topaz Video AI

The Mac can’t use Nvidia GPU’s, only AMD, so no CUDA. The incredible thing is, one GPU has 16GB, the second only 4GB, but somehow, one adds to the other and together they perform better than one would expect.

Yeah. Sounds like a good task for AI to solve. How hard would it be to make a scene detection AI?

That’s why I said CUDA like. Whatever they’re using, AMD was fastest in VEAI 2.6.

I’ve been getting good results with “Iris” and the release of the model “Rhea” sounded very promising, so I wanted to try it on a 4K video.
The first thing I noticed is that it has a huge impact on performance.
Where I was getting 5 fps using an Nvidia 4070 Super with an i7 CPU the performance fell back to 0.6 fps.

The whole reason for buying a 4070 Super was that low framerate I was getting.

The Rhea model was able to take out more grain and noise and the end quality seemed to be better than Iris, but I’m not totally convinced. The face was a bit too jelly to my liking.
I would need more mileage with “Rhea” to evaluate it, but that’s not going to happen with these speeds, so I’m afraid I need to stick with “Iris”.

1 Like

I have the same issue. It went from around 8fps to 2-3fps using the same settings.

1 Like

Really pissed off by this update…

This new license tier offers […] commercial usage rights […]

Does it mean that at the moment, output cannot be used commercially?! Or from this new license takes away the possibility to use the actual output commercially?

multi-GPU support for teams

It’s more likely that teams have for each member their own GPU than a system with more than one shared. This just sounds to me like a way to justify something that is more of a commercial bully move than a feature to be proud of. If you want to change the terms, do it. But do not make excuses.
‘We need to divert so much money into multi-GPU development, that we decided to make it a PRO feature, to dedicate more funds to make it work better and more reliably’ this would have been much, much more ‘acceptable’ announcement.

I’m a new customer of yours, this doesn’t make me trust your announcements very much, as others are already saying that you previously announced you would not take multi-GPU away.

Sorry,
had to voice my concerns.

This said, Rhea looks great!

5 Likes

I am merely comparing the models Rhea and Iris using the same Topaz Video 5.2.0
It seems to be 10x slower.

This would mean days of rendering instead of 6 hours.

1 Like

I wonder why it’s happening on my system that has no iGPU (Ryzen 2700X).

The software exhibits strange behavior by changing AI processor settings to “Auto” even when we manually set it to “GPU”. When in “Auto” mode, the software uses the CPU for processing. Please try to change it to “GPU” after each render. With every update, this software becomes more problematic.

5 Likes

When I close a video file in Video AI it is still told to be in use and can’t be moved or deleted. This issue was reported in the beta period.

Did you remove it from the list of sources? I find it will stay ‘in-use’ until you remove it from the sources list in VAI.

ah I did not remove it from the list. I will try that trick. edit: that worked. :slight_smile:

1 Like

And Auto doesn’t even work correctly. It’s supposed to select the fastest available processing device. So even on that setting, it should still be selecting your GPU. Ideally, if you have more than one GPU it should be selecting whichever one of them is the fastest.

3 Likes

Same problem for me… please fix it