HDR Model + GPU Player: Topaz Video AI Alpha 5.4.0.8.a

[Alpheratz]

It mainly depends on how much metadata the player needs.

Some need everything, others only some

If MediaInfo does not recognize it as SMPTE 2086 (minimum HDR standard), it means that one or more ITU-R standard metadata is missing

[dunmer2007]

Quoting myself due to laziness…

@dunmer2007

It is the media player incompatibility with my system that has ganked my TVAI ALPHA+BETA I cannot even open up my old beta install as I installed the new alpha, and the only way I can open the alpha is to first open the player from the alpha’s install directory… so it looks like the player itself is the problem, at least with my system that is…

I’m uninstalling both new alpha/beta and reinstalling the previous versions to see if that works, it really sucks that my system is rejecting the player’s implementation to the program as I cannot even use the previous beta if that player is installed on my system, I’d install the player itself, but it’s coder-specific, and I’m no coder!

I’m a GUI nerd, so I’m not as versed with command-line programs…

[deepzone]

Or have a border edge around the video at least

[jackkhoros]

Is the TensorRT conversion, which was added for RXL Alpha 3 in version 5.3.4.0.a, present in this version?

[meimeiriver]

@jackkhoros

Is the TensorRT conversion (…), present in this version?

I’d like to know this too, please. From the amount of GPU activity (near 100%), I’d say yes. (And the heat; my god, the heat!) But would like to be sure.

[ForSerious]

Classically, Windows Task Manager is not going to show 100% utilization on the GPU if the tensor cores are the main things being used. But it seems to be inconsistent.

[meimeiriver]

@ForSerious

Classically, Windows Task Manager is not going to show 100% utilization on the GPU if the tensor cores are the main things being used.

Are the Tensor Cores used for Rhea (not XL) too? I recently namely discovered anomalies in Rhea XL. Rhea (v1) appears to be using near 100% GPU too, however.

[ForSerious]

Chances are it’s not using Tensor cores if it says the GPU is at 100%.
I say that it’s inconsistent because the behavior seems to change. I know when TVAI first started using tensor cores, it did not show it. But there have been times since then when I open it and look and it does show it at 100%.
Anyway, maybe try GPUz and see what is says.

[meimeiriver]

@ForSerious

Anyway, maybe try GPUz and see what is says.

GPUz says nothing useful about Tensor cores; but I will explore this link:

Using Nsight Compute or Nvprof to Show Mixed Precision Use in Deep Learning...

[ForSerious]

Sorry. I thought by now, they would. Of course they can’t if Nvidia never makes it an option.

[jackkhoros]

Nobody knows?

[Alpheratz]

HDR Model + GPU Player: Topaz Video AI Alpha 5.4.0.8.a

99% stealth let be honest At least it has been quick fixed which contrasts highly with common uses. Not sure that the full black background is a good idea anyway … Grey background was fine → We can’t see black borders anymore which is a regression Should be fixed to my mind [image] [image] [image]

@henry.z
Are you going to do something with wrecked cropping and black background ?

" Please let us know any problems / bugs you run into !"
This is done

[hazardc]

So now literally every version of topaz crashes when i load videos … the alpha, the beta, and the release version.

Seriously what the hell?

What is the point of doing this if you guys just ignore literally all feedback?

I pay for this program and can’t use it because it just crashes out … This is getting ridiculous. I don’t mind issues with the alphas and betas but to push it to a full release that still has this bug is insane and … these are the kinds of reasons people who test your software usually get kickbacks … because they put up with this stuff. This company USED to do this kind of stuff. Now you don’t even reply to the testers that you claim to want feedback from and they getg stuck with a program that you won’t even extend the license period on when you nuke their setup.