I know that there's a reference to vid model and we can use video with it, so I wonder if we can do a nudify workflow from this? like make the video of a dancing girl get nudified?
it works just by default without any special workflow just by prompting: "naked female dancing/garment falls off revealing her...". Better results are achieved if you give it a reference image of a naked body so it basically does a swap of what you want.
there's no separate audio prompt, you can describe the audio in the normal prompt and/or connect a "Load audio" node if you want a reference. (same voice/effects and so on)
I can't believe that...they honestly have ZERO checks server side, I literally typed a RANDOM email that doesn't event exist and I was able to register...damn vibecoders
User on reddit has turned it into a character reference sheet generator, really works well if you know how to prompt it. I however don't, lol. This will be a major boon until they release the H3 image model. Really struggling here trying to get the prompting right in this use case.
Are you able to nudify a person using the exact reference video? Like just taking the clothes pff but it is the exact same video. Like a tiktok video? Or is there any workflow other than h3 that is capable in doing this?