MultiModal (2D/3D) Diffusion Models
Work on 2D/3D diffusion model framework that offers high fidelity, controllability, modularity, (re)-usability (adapting existing foundational models) and applicability (data augmentation).
NeurIPS 2024
We present a novel Factor Graph Diffusion Model (FG-DM) framework for modeling the joint distribution of images and conditioning variables (such as segmentation, depth, normal and sketch masks) for improved prompt consistency, high fidelity, and controllable image synthesis.