Student Researcher (Seed Vision – Long-Range Video Generation) – 2026 Start (PhD)

Work on foundational multimodal generative models for visual creation, focusing on long-range video generation. Develop scalable architectures for consistent motion, identity, and layout, and address temporal drift, motion collapse, and high-frequency detail retention. Explore hierarchical/recurrent latent structures and autoregressive or chunked generation strategies. Build evaluation protocols to measure realism, consistency, and semantic continuity for long videos.