Hello. I’m wondering if anyone has found particular settings in Topaz Video AI for sharpening that work particularly well. Iris seems to be a good model, and one of the renders came out well. But the software interface is such that I can’t see what settings were used for that render.
I’ve done a lot of trial and error with settings trying to get back to that render result, but no luck so far. The footage is quite out of focus.
Any thoughts are greatly appreciated.
I played with it recently for standards conversion and scaling up.
I can not tell you what works best but I think the results get tagged with a suffix of letters telling you the model used 
I actually ran out of time to test more thoroughly. Seems I could have spent a long time noodling but the machine I was using was slow and the clips long so I got something that wasn’t terrible and moved on 
I found Dione Robust the best model, but for some reason this model is only avaible on the OFX/AE version…dont know why. But t is soo clean, so sharpen, so good. Dont need to upscale, just add details on the base resolution.
Look at this example;
7450921-uhd_4096_2160_25fps_iris3.zip (5.9 MB)
I know there is a iris3 suffix but it is a mistake, the model used is Dione Robust.
But if the footage is so out of focus i’m not quite sure this will work in this case. The only solution i can think of is doing some ai training using nuke’s copycat or the FlameSimpleML, an example here.
Hey that looks quite good. Would the OFX version work in Flame? We don’t even use AE.
The suffix of the model in the render name doesn’t say enough — there are so many other sliders. I cannot reproduce the one quite good result I got, no matter all the combinations of variables I’ve tried…
Only Davinci Resolve and After Effects, only Windows and Mac. The example above was just the model selected and Auto, no manual tweaking at all. Could give a hand if you need to.