What are the best settings to upscale Ai-videos? Pika platform produce 1024 x 576 low quality compressed videos. What’s the best approach? Which model is best and settings for fix low quality, blurry, heavy compressed videos?
My settings at the moment:
Upscale to 4k
Motion Deblur
Proteus - Fine - Tune/Enhance
Recover Original Detail = 20
Encoder ProRes 422 Std
With these settings, 3 seconds video takes about 20 minutes rendering, with 361 MB file size, which is a lot of time and size. I have NVIDIA GTX 1080.
After that I downscale it using Davinci Resolve to 1080p to get better quality, then I bring it back to Topaz to enhance using Upscale to HD resolution to get better quality. Encoder H265 : Main10
Tbh, I still learning what’s the best method to do it, because Ai videos have great potentials if we could upscale them, do you suggested better workflow?
Yes, At least 2k. I need it to look crisp and clean, professionally digested. I go for the 4k, hence down scaling a 4k to 1080 would result in a better results.
I can’t tell if I have better settings because I dunno how your result looks like. I tried Iris first and then Iris with a 2nd step Proteus. Here are the results.
Wow! Great results I see … Thanks for the efforts. I have questions to ask:
Which settings were for 1w1_chr2_iris2_prob3? both have the same name, different is the chr2. this one which is 15.5 is better result I see.
What Setup/GPU do you use, and how long it took you to render?
I followed your exact settings and parameters, still I don’t get the same results … It’s obvious from the text on the lower right side. I did a comparison between yours and mine … there’s a difference, what do you think?
Even though I didn’t go for the extreme 4K … there’s a huge difference between our results … Can you share with me tips I might missed something because I followed your settings exactly.
