Frame Context Packing and Drift Prevention in Next-Frame-Prediction Video Diffusion Models

Open in new window