Associate Professor, Yonsei University
1 paper at NeurIPS 2025
A rectified flow-based generative model that improves generation quality and straightens ODE solution trajectories by incorporating real images via slerp into the reflow training process.
