Hi, just wondering if the diffusers repo already provides a way to invert images (or I guess latents in the case of latent diffusion models) into noise that denoise via DDIM back into the original inputs as described in the DDIM paper, sections 4.3/5.4?
2 Likes