Towards Unified Semantic and Controllable Image Fusion: A Diffusion Transformer Approach
Published in IEEE Transactions on Pattern Analysis and Machine Intelligence, 2026
DiTFuse studies how diffusion transformers can unify semantic image fusion and controllable multimodal fusion in a single framework. Chengjie Jiang is a co-first author.
Recommended citation: Jiayang Li*, Chengjie Jiang*, Junjun Jiang, Pengwei Liang, Jiayi Ma, and Liqiang Nie. Towards Unified Semantic and Controllable Image Fusion: A Diffusion Transformer Approach. IEEE TPAMI, 2026.
Download Paper