Hi everyone, sorry to bug people for help right away, but I’m kinda stuck and I’ve been digging around to no success.
So, I’ve followed to basic tutorial to get to the training step with a slight exception (rather than using a source video, I downloaded a celebrity photoset from this site, and put the images in the aligned folder). I extracted my images from the dst video, and both image sets show up in the view aligned results portion. When I try to run the quick96 training, however, it says that “png is not a dfl file” for each dst aligned image. I’ve tried having the dst extract save them as jpgs instead, as well as converted the dst video to an mp4 rather than an MOV, but nothing has worked. I am using an AMD Radeon card, if that makes a difference (which I hope it doesn’t).
Has anyone seen this issue before, and have any idea what I should be doing about it? Thank you very much, very excited to be working with this program, and hopefully learning more and more as I get going.
The aligned images should be jpg files. Did you extract the aligned images after extracting the frames? Are you using a really old version of DeepFaceLab? The dst aligned files should be square jpg images, just like the source faceset you downloaded.