LaWa: Using Latent Space for In-Generation Image Watermarking
This addresses the need for imperceptible watermarking to trace AI-generated images, offering a domain-specific solution for generative models.
The paper tackles the problem of malicious use of AI-generated images by proposing LaWa, an in-generation watermarking method for Latent Diffusion Models that modifies the latent space to embed watermarks, achieving high robustness against image transformations and outperforming prior works in perceptual quality, robustness, and computational complexity.
With generative models producing high quality images that are indistinguishable from real ones, there is growing concern regarding the malicious usage of AI-generated images. Imperceptible image watermarking is one viable solution towards such concerns. Prior watermarking methods map the image to a latent space for adding the watermark. Moreover, Latent Diffusion Models (LDM) generate the image in the latent space of a pre-trained autoencoder. We argue that this latent space can be used to integrate watermarking into the generation process. To this end, we present LaWa, an in-generation image watermarking method designed for LDMs. By using coarse-to-fine watermark embedding modules, LaWa modifies the latent space of pre-trained autoencoders and achieves high robustness against a wide range of image transformations while preserving perceptual quality of the image. We show that LaWa can also be used as a general image watermarking method. Through extensive experiments, we demonstrate that LaWa outperforms previous works in perceptual quality, robustness against attacks, and computational complexity, while having very low false positive rate. Code is available here.