Image to Image

Modify existing visuals using text guidance. Replace latent noise with VAE encoding for targeted transformations.