Comment on fMRI-to-image with contrastive learning and diffusion priorsparentComments−RC_ITR3yIf you can retrieve an image using a latent vector, it’s trivial to reconstruct it (decently well) with a diffusion model.
Comments
If you can retrieve an image using a latent vector, it’s trivial to reconstruct it (decently well) with a diffusion model.