Win 11 doesn’t work the same as Win 10. It gives the option to “Pin to start”, but not the option to create a desktop shortcut. Thanks for the suggestion though. I will see if I can find a way to create a desktop shortcut instead. Perhaps I can do it from the actual file in the Programs folder.
Huh?
Sure this is with StartAllBack, but normal Windows 11 just hides it under ‘More Options’ or something like that.
So not what I was thinking. I also have found a few models that are mich faster with the old version, but that’s super odd that it would have the same speed with a 3060ti and a 4080. Almost like it’s not even using the GPU.
Yes,I check it in afterburner, when i convert, the usage of 4080 is lower than 50%…
1 Like
If exporting images from ‘trim’ it would be convenient, to after, be able to export Wav on the same trim so one dont need to import the full video and sync the video to the Tiffs to get access to audio (in sync).
Holy cow just trying the new 3.0.12 coming from 2.6.2 I don’t like it at all.
What is add noise? It seems to do nothing even at max setting.
Grain was changed from luma based to chroma based, which is good, but they removed the size setting!?! This was by far the most useful setting! Grain is terrible now, I get much more natural results with 2.6.2. This 3.0.12 is basically useless for upscaling past 1080p, the grain is too small!
And now full screen preview requires an extra step via the dropdown menu or playing with an external player. both unnecessary changes. Sure playing externally is nice but not needed. And now it won’t play fluidly internally or preview live while processing. wth man.
And within 10 minutes of messing around with it, I got two hard unrecoverable green screen crashes which required a total system reboot. with 6900xt, 3900x.
Idk man I’m not sure about the new version, they were on the right track but this seems like 2 steps forward, 10 steps back. Especially the GUI. Even more simplified yet somehow harder to use. Where did all the different models go? It was easier to just select your model and roll, now everything is hidden behind some dropdown menu with a few selections that you have to check every dropdown menu to find out what’s even included. SMH.
3 Likes
Yes i can ! Actually 4080 doesnt suck at all but the lazy Topaz developers do …
This software is ZERO optimized for the RTX cards tensor cores - especially the last generations .
I am afraid this is the truth my friend and this is very sad … but the worst thing is that it seems they dont care at all to do something about this … Thats why $1500 GPU get the same speed as $300 GPU even its about 5 times more powerful …
So you have same problem with me? That was sad 
,
I buy 4080 for tvai but i got nothing…
1 Like
I get my GTX 1080 that doesnt have a single Tensor core to be faster then my RTX 3060 which is absurd but its a sadly fact … 
And already have heard a lot of people do complain that their 4090 are slower then 3080 … As i said - lazy Topaz developers …
1 Like
You should have seen the new Beta 3.1.0.4.b 
It’s blistering fast. RTX 3080 Ti here, 4K upscale using Proteus. Full-length movie, ETA 5:23 hours (as opposed to the usual 17-20 hours). 95% CPU saturation + near full GPU utilization! 7.3 FPS! (from 1080p → 4K).
6 Likes
So, I might not have read every single post in here, but this statement is false. Up until yesterday, people have complained that they didn’t get the increase they expected from a 40 series upgrade, but never said it was worse or the same.
2 Likes
I get a lockup problem on a Mac M1 usually 1 out of 15 renders. I don’t know why. Plenty of SSD space, 16GB RAM.
It is almost like it lost connection to a server but without an error message.
It happens at any percentage of the rendering, randomly.
I tried now to delete the models before reinstalling them completely using the predefined folder in the preferences from the models. Emptied that completely. Only the authorization file is left.
But now it doesn’t download the models.
And though the message is appropriate - because there are no models at this time - why doesn’t it try to download then?
Or: if it wouldn’t just start processing like this what do I need to do to make it download them again?
They’ve been telling other people with model issues to do, what you just did, to fix it. Restart everything, if you haven’t.
Maybe you’ll get lucky and the update later on today will fix it for you.
Are you being serious? Because that’s huge news!
Very serious. 
That’s wonderful news. Any indication when that will be publically be released?
They said this was the last beta, before a 3.1 release; but don’t shoot me if I misunderstood.
This program isn’t using tensor cores at all? I don’t remember that being on the roadmap but I hope they fix that! As someone else said, why aren’t high performing GPUs being utilized fully? I’m hoping that the goal is to make the program work for the wider audience and then to focus on the higher end cards to really get that speed boost.
1 Like
