Open atonderski opened 1 month ago
Hi Adam, How exactly would you like to condition on the image? In the paper, I show results for unclip-like adapters which invert a clip image embedding. As this is a form of img2img, maybe that's already sufficient for you?
Nick
Hi,
Thanks for the very interesting work! I'm interested in using this approach for img2img conditioning. Have you done any experiments in that direction? Do you foresee any issues or caveats that one should consider?
Cheers, Adam