Recent video models increasingly support generation, reference conditioning, and editing within a single model, yet typically treat these as separate operations over fixed inputs. Practical video creation, however, unfolds across multiple shots, requiring a model to generate from text, follow a reference, or edit source footage while maintaining shared history. Researchers from Hugging Face formalize this setting as interactive multi-shot video creation (IMVC) and introduce ContextMaster, a unified model with a role-aware context representation.
An interactive model must retain access to an expanding history without letting the context read cost at each denoising step grow. ContextMaster combines reusable clean context states with fixed-budget sparse context routing and uses ConstraintSink to keep task constraints visible. To address the dual challenges of sparse context access and inference with few denoising steps, the authors propose a two-stage privileged context distillation framework, which transfers full context behavior from a dense teacher through consistency distillation and then refines deployment rollouts with distribution matching.
Experiments on the three primitive tasks—generation, reference conditioning, and editing—demonstrate improved task fulfillment and consistency across shots over specialized baselines. User studies further validate flexibly composed workflows, while the model reaches 16 FPS on a single GPU, enabling real-time interactive creation.