Dual Diffusion Implicit Bridges for Image-to-Image Translation
This method addresses data separation and privacy issues in image translation for domains like synthetic and high-resolution datasets, though it is incremental as it builds on existing diffusion models.
The paper tackles the problem of image-to-image translation without joint training on source and target domains, which hinders data privacy and adaptability, by introducing Dual Diffusion Implicit Bridges (DDIBs) that use independently trained diffusion models and achieve translation via ODE-based steps with cycle consistency up to discretization errors.
Common image-to-image translation methods rely on joint training over data from both source and target domains. The training process requires concurrent access to both datasets, which hinders data separation and privacy protection; and existing models cannot be easily adapted for translation of new domain pairs. We present Dual Diffusion Implicit Bridges (DDIBs), an image translation method based on diffusion models, that circumvents training on domain pairs. Image translation with DDIBs relies on two diffusion models trained independently on each domain, and is a two-step process: DDIBs first obtain latent encodings for source images with the source diffusion model, and then decode such encodings using the target model to construct target images. Both steps are defined via ordinary differential equations (ODEs), thus the process is cycle consistent only up to discretization errors of the ODE solvers. Theoretically, we interpret DDIBs as concatenation of source to latent, and latent to target Schrodinger Bridges, a form of entropy-regularized optimal transport, to explain the efficacy of the method. Experimentally, we apply DDIBs on synthetic and high-resolution image datasets, to demonstrate their utility in a wide variety of translation tasks and their inherent optimal transport properties.