MaskFlow: Discrete Flows For Flexible and Efficient Long Video Generation

TL;DR

MaskFlow introduces a novel framework combining discrete representations and flow-matching for efficient, high-quality long video generation, capable of producing videos ten times longer than training sequences with flexible sampling modes.

Contribution

The paper presents MaskFlow, a unified approach that enables efficient long video generation using frame-level masking and flow-matching, extending sequence length significantly beyond training data.

Findings

Achieves competitive FVD scores on FFS and DMLab datasets.

Supports both autoregressive and full-sequence generation modes.

Demonstrates training-free application to timestep-dependent and independent models.

Abstract

Generating long, high-quality videos remains a challenge due to the complex interplay of spatial and temporal dynamics and hardware limitations. In this work, we introduce MaskFlow, a unified video generation framework that combines discrete representations with flow-matching to enable efficient generation of high-quality long videos. By leveraging a frame-level masking strategy during training, MaskFlow conditions on previously generated unmasked frames to generate videos with lengths ten times beyond that of the training sequences. MaskFlow does so very efficiently by enabling the use of fast Masked Generative Model (MGM)-style sampling and can be deployed in both fully autoregressive as well as full-sequence generation modes. We validate the quality of our method on the FaceForensics (FFS) and Deepmind Lab (DMLab) datasets and report Frechet Video Distance (FVD) competitive with state-of-the-art approaches. We also provide a detailed analysis on the sampling efficiency of our method and demonstrate that MaskFlow can be applied to both timestep-dependent and timestep-independent models in a training-free manner.