Post Content Post navigation Rebalancing Reference Frame Dominance to Improve Motion in Image-to-Video Models LIFT and PLACE: A Simple, Stable, and Effective Knowledge Distillation Framework for Lightweight Diffusion Models