I’m glad most of the users are really enjoying Precise. I’ve been raving about it for weeks, and I’m glad the reaction is widely shared.
This might be a bit of an opinion, but so far Precise is vastly superior to its predecessors including Mini, Sharp, HQ etc.
On a sidenote, I know it’s probably an industry secret, but I do wonder what kind of technological breakthrough the Topaz team were able to make, that makes Precise 2.5 so much better than its previous versions? As someone who had used both Precise 1 & 2, it’s crazy how much better 2.5 really is compared to them.
Same problem here right after install. I rebooted and Sharp and HQ both stared to work. I have the same problem with the other three. I also had the problem with Fast 1, bug report is here:
The main issue I see with this model at this time is that it struggles with small text and numbers. Sometimes it will get it right, sometimes, as the background behind the text shifts it gives a garbled inaccurate recreation. It’s hit or miss depending on the frame.
Had a popup that smartappcontrol blocked neuroserver from executing a file something with a pyd ending . I can’t provide the details of the message as microsoft decided that after you click on a notification you will no longer need it and deletes it.
Edit: after restarting topaz it started processing.
Fully agree with you. I try SLHQ on SLM output, 1x and the sharpening is very good. Downside is SLHQ took exactly same speed of SLM which is very slow (0.4fps on my 5090)
Didn’t mean to @ you Kevin. Not sure how to remove that.
A few other things I noticed when testing yesterday.
It takes a couple minutes for the model to load. This is when the RAM gets fully eaten up and almost stalls the computer. Loading the model takes longer than other models. Maybe twice as long as starlight mini.
Going from 2x to 3x does not exponentially increase processing time. In fact it only increases processing time by ~9.5%.
Here is some data. Anyone else seeing the same thing?
