Cross-domain Compositing with Pretrained Diffusion Models
This provides a versatile tool for image editing and data augmentation, though it is incremental as it builds on existing diffusion models.
The paper tackles the problem of cross-domain compositing tasks like image blending and CG2Real translation by using off-the-shelf diffusion models with a localized, iterative refinement scheme, resulting in higher quality and realistic outputs without needing annotations or training.
Diffusion models have enabled high-quality, conditional image editing capabilities. We propose to expand their arsenal, and demonstrate that off-the-shelf diffusion models can be used for a wide range of cross-domain compositing tasks. Among numerous others, these include image blending, object immersion, texture-replacement and even CG2Real translation or stylization. We employ a localized, iterative refinement scheme which infuses the injected objects with contextual information derived from the background scene, and enables control over the degree and types of changes the object may undergo. We conduct a range of qualitative and quantitative comparisons to prior work, and exhibit that our method produces higher quality and realistic results without requiring any annotations or training. Finally, we demonstrate how our method may be used for data augmentation of downstream tasks.