Trigger word is flm
The model was trained on 3000 frames from various documentaries filmed in the 1960s to try and capture diversity in people, objects and places.
Description
FAQ
Comments (4)
Pretty cool! Thanks! But was it trained on AI upscaled images? Some of the examples seem to exhibit some of the pitfalls of denoised/upscaled images (blurry and splotchy)
The films were all 1080p. But I had to crop the black bars off the sides and crop the top to remove the large watermark that was in every frame. I don't think they were upscaled. Not from what I could tell watching the videos. I first had 480p videos too and left the watermarks in, but the results were much worse.
@humblemikey Awesome, I actually tested it and my own model exhibits it as well, even without the lora so all is good! The aesthetic is very much there so thanks again! :D
will there be any 1.5 version
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.