Image-Based CLIP-Guided Essence Transfer
This addresses the need for semantic image editing in computer vision, offering a novel approach for tasks like domain adaptation, but it is incremental as it builds on existing StyleGAN and CLIP models.
The paper tackles the problem of essence transfer, which involves editing a source image to incorporate high-level semantic attributes from a target image, distinguishing it from style transfer. The result is a method that achieves identity preservation and feature transfer without facial recognition, showing superiority over existing techniques in experiments.
We make the distinction between (i) style transfer, in which a source image is manipulated to match the textures and colors of a target image, and (ii) essence transfer, in which one edits the source image to include high-level semantic attributes from the target. Crucially, the semantic attributes that constitute the essence of an image may differ from image to image. Our blending operator combines the powerful StyleGAN generator and the semantic encoder of CLIP in a novel way that is simultaneously additive in both latent spaces, resulting in a mechanism that guarantees both identity preservation and high-level feature transfer without relying on a facial recognition network. We present two variants of our method. The first is based on optimization, while the second fine-tunes an existing inversion encoder to perform essence extraction. Through extensive experiments, we demonstrate the superiority of our methods for essence transfer over existing methods for style transfer, domain adaptation, and text-based semantic editing. Our code is available at https://github.com/hila-chefer/TargetCLIP.