I took an SD video and fed it to Topaz AI v5 using Rhea and upsampled it. The processing took 13 hours to run on a MacBook Pro Max M1 64GB of RAM. The results are really disappointing.
I put the original in a Premiere sequence and masked it so the right side is the Topaz output and the left side is the original SD video. The Topaz output is clearer barely but it is hard to tell the difference.
Is this just as good as it is or am I missing something? I used stock settings for Rhea.
Odd that I can’t put YouTube or Google Drive links…
You can upload zip files directly on these forums if they’re less than 50MB.
I have yet to get any good results from Rhea. Mostly Proteus Manual settings does the best on SD videos for me. And I would never push it up to 4K, just FHD.
If you want to share a sample of the original input clip, I can run some tests and see if I have any recommendations on which settings to use. SD to 4k may be too big of a jump for good results, depending on the quality of the original file.
You can securely submit your file(s) to our Dropbox using the link below. Please be sure to send me a note to let me know you sent something. Submit File to Dropbox
Topaz Video AI doesn’t support going beyond 4x the input resolution. If you choose an output resolution that is more than 4x the input, TVAI will enhance the file to 4x and then use lanczos for the rest. If your input is 360p, the file will be enhanced to 1440p, and then enlarged to the final resolution with lanczos.
