SPATIALALIGN: Aligning Dynamic Spatial Relationships in Video Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Fengming, Cham, Tat-Jen, Zheng, Chuanxia
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915820973588480
author Liu, Fengming
Cham, Tat-Jen
Zheng, Chuanxia
author_facet Liu, Fengming
Cham, Tat-Jen
Zheng, Chuanxia
contents 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/
format Preprint
id arxiv_https___arxiv_org_abs_2602_22745
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SPATIALALIGN: Aligning Dynamic Spatial Relationships in Video Generation
Liu, Fengming
Cham, Tat-Jen
Zheng, Chuanxia
Computer Vision and Pattern Recognition
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/
title SPATIALALIGN: Aligning Dynamic Spatial Relationships in Video Generation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.22745