Presentation

· Contributors · Organizations · Search Program · My Schedule · My Recommendations · Happening Now

Canvas-to-Image: Compositional Image Generation with Multimodal Controls

DescriptionCanvas-to-Image is a unified framework enhancing diffusion models' compositional and multimodal control through a single canvas interface. By encoding subject references, bounding boxes, and poses into one composite canvas, it enables integrated visual-spatial reasoning. This approach streamlines heterogeneous control signals, allowing models to generate high-fidelity images with precise structural alignment.

Authors

Yusuf Dalva

Virginia Tech University

University of California Merced