Yang, J. (CSE) – Towards Controllable and Compositional Generative Vision
Diffusion-based text-to-image models can generate impressive images, but they largely treat an image as a single, flat output, which makes precise editing of individual elements difficult. This proposal studies layered […]