Flash-DMD: Towards High-Fidelity Few-Step Image Generation with Efficient Distillation and Joint Reinforcement Learning
Published in IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
Flash-DMD combines efficient distribution-matching distillation with reinforcement learning to produce high-fidelity few-step image generators at a fraction of the conventional training cost.
Recommended citation: Guanjie Chen, Shirui Huang, Kai Liu, Jianchen Zhu, Xiaoye Qu, Peng Chen, Yu Cheng, and Yifu Sun. Flash-DMD: Towards High-Fidelity Few-Step Image Generation with Efficient Distillation and Joint Reinforcement Learning. CVPR, 2026.
Download Paper