SPATIALALIGN: Aligning Dynamic Spatial Relationships in Video Generation

TL;DR

SPATIALALIGN is a framework that improves text-to-video models by better capturing dynamic spatial relationships specified in prompts, using a novel geometry-based metric and fine-tuning techniques.

Contribution

The paper introduces SPATIALALIGN, a self-improvement method with a new DSR-SCORE metric and a fine-tuning approach to enhance spatial relationship accuracy in T2V models.

Findings

Significant improvement in spatial relationship accuracy

Introduction of DSR-SCORE for quantitative evaluation

Effective fine-tuning method for DSR alignment

Abstract

Most text-to-video (T2V) generators prioritize aesthetic quality, but often ignoring the spatial constraints in the generated videos. In this work, we present SPATIALALIGN, a self-improvement framework that enhances T2V models capabilities to depict Dynamic Spatial Relationships (DSR) specified in text prompts. We present a zeroth-order regularized Direct Preference Optimization (DPO) to fine-tune T2V models towards better alignment with DSR. Specifically, we design DSR-SCORE, a geometry-based metric that quantitatively measures the alignment between generated videos and the specified DSRs in prompts, which is a step forward from prior works that rely on VLM for evaluation. We also conduct a dataset of text-video pairs with diverse DSRs to facilitate the study. Extensive experiments demonstrate that our fine-tuned model significantly out performs the baseline in spatial relationships. The code will be released in Link. Project page: https://fengming001ntu.github.io/SpatialAlign/