Hidir Yesiltepe
Hidir Yesiltepe is a Ph.D. student in the Department of Computer Science. He is advised by Pinar Yanardag.
His research focuses on controllable and long-form video synthesis with text-to-video diffusion models, especially making pretrained video generators more programmable, efficient, and capable without expensive retraining. He studies how inference-time reparameterizations and architectural changes in diffusion transformers can improve long-horizon generation, motion control, restoration, dynamic view synthesis, and memory-efficient autoregressive video generation.
Publications
Projects
In the News
-
Article Item
