Video Latent Flow Matching: Optimal Polynomial Projections for Video Interpolation and Extrapolation

TL;DR

This paper introduces Video Latent Flow Matching (VLFM), a novel method that uses polynomial projections in latent space for efficient video interpolation and extrapolation, leveraging pre-trained image models and the HiPPO framework.

Contribution

The paper proposes a new approach combining polynomial projections and the HiPPO framework for improved video modeling and frame rate flexibility.

Findings

Effective video interpolation and extrapolation demonstrated on multiple datasets.

Theoretical guarantees of bounded approximation error and robustness.

Compatibility with pre-trained image generation models.

Abstract

This paper considers an efficient video modeling process called Video Latent Flow Matching (VLFM). Unlike prior works, which randomly sampled latent patches for video generation, our method relies on current strong pre-trained image generation models, modeling a certain caption-guided flow of latent patches that can be decoded to time-dependent video frames. We first speculate multiple images of a video are differentiable with respect to time in some latent space. Based on this conjecture, we introduce the HiPPO framework to approximate the optimal projection for polynomials to generate the probability path. Our approach gains the theoretical benefits of the bounded universal approximation error and timescale robustness. Moreover, VLFM processes the interpolation and extrapolation abilities for video generation with arbitrary frame rates. We conduct experiments on several text-to-video datasets to showcase the effectiveness of our method.