Topaz Video 1.0.4 - Patch

I think double speed notifications comes from fault resolution comparison, you must always check output file what it really did, because shown output resolution while rendering could be displayed wrong. However, I certainly can’t detect such a speed boost it’s lame als always :smirking_face:

I think they are faking it, to be honest.

1 Like

Maybe, because they announced it in 2024 and the beta version hasn’t even been released yet.

4 Likes

Personally, after updating, I didn’t notice any difference in speed. And I’m still using version 7.1.5 of Video AI. I’m getting 5.48-5.50 seconds per frame in the best-case scenario, and sometimes peaks at 5.60 seconds with all performance settings maxed out and nothing else running in the background. I was getting this speed before the driver update.

1 Like

But why is Adobe paying their customers to provide high quality videos for model training? I hardly thinks it’s faking it, when they have invested millions into the project. Adobe openly admits they are struggling with technological challenges, and won’t release a half-baked product, like Topaz Labs did 5 years ago. Essentially, I paid Topaz Labs to be a beta tester, but on the other hand, Adobe pays their customers for helping to move the project forward.

We’ll see this happen in the coming years; it’s impossible to say for sure right now, because Adobe hasn’t said a word about this project, which will be two years in the making. They’re either keeping it a secret, or they might have postponed it due to technical issues or because they’ve been busy with more important projects. Or they couldn’t handle it and abandoned the project.

The project hasn’t been abandoned, because Adobe research engineers just published an article in June, describing the progress they are making and also the roadblocks they face. Unlike Topaz Labs, Adobe has vast finical resources to continue the project, which means they don’t need to release software that isn’t ready, just to raise revenue. This is how Topaz Labs did it, and I paid good money, just to be a beta tester, because what I saw in end results was nothing like their marketing hype described in 2020.

Topaz already released its latest update in v7. There have been no updates since then, no new models, no optimizations. They’re currently writing code for Cloud solutions because it’s more profitable. Let’s wait for Adobe’s app, whenever it’s released. This is a pressing need for me right now; I’m restoring old videos 24/7. I need a faster and higher-quality model. SLM and SLS aren’t working anyway, so I’m using SeedVR2.5.

Adobe engineers aren’t smart enough to figure out how to add av1 support into premiere. There’s zero chance they figure out video upscaling.

4 Likes

Unfortunately, I agree. Topaz was a wonderful program. What I see now is a constant marketing tool for NEXT. That seems to be the main point of their programs.

2 Likes

They do have a new beta with add final grain filters and a new high quality source denoise filter.
Hopefully they don’t release it until it’s actually ready.

Topaz will lose ground if they continue like this. The stated roadmap is to improve Project Starlight, but releasing new variants that are then left in development in order to release yet another new model doesn’t really correspond to the roadmap. They also wanted to find out how the Stable Diffusion model can be influenced to give us parameters, but we don’t see any of this. It’s not too late yet, and I hope something will come along.

We’ll see who laughs last.

Any updates with CUDA Tile? https://x.com/nvidianewsroom/status/1996976316453679381
This is new CUDA technology can give +20-30 and +50% perfomance as people saying

3 Likes

Yes please!

2 Likes

Oh well, there goes 6 months of updates while they learn how to reprogram everything again.

2 Likes

Unless they were warned about this a year ago, I think 6 months is too soon of a guess for this to get implemented into most existing programs.

ChatGPT could do it in 5 days.

Sounds like a plan. It’s not like they can let us down more than they already have at this point—if it happens to introduce issues.