RISE-Video: Can Video Generators Decode Implicit World Rules?

TL;DR

RISE-Video introduces a new benchmark for evaluating video generators' ability to understand and reason over implicit world rules, highlighting current models' limitations in complex, constraint-driven scenarios.

Contribution

This paper presents RISE-Video, a novel reasoning-oriented benchmark with a multi-dimensional evaluation protocol and automated assessment pipeline for Text-Image-to-Video models.

Findings

Models show significant deficiencies in reasoning over implicit constraints.

The benchmark reveals gaps in temporal consistency and physical rationality.

Automated evaluation correlates well with human judgment.

Abstract

While generative video models have achieved remarkable visual fidelity, their capacity to internalize and reason over implicit world rules remains a critical yet under-explored frontier. To bridge this gap, we present RISE-Video, a pioneering reasoning-oriented benchmark for Text-Image-to-Video (TI2V) synthesis that shifts the evaluative focus from surface-level aesthetics to deep cognitive reasoning. RISE-Video comprises 467 meticulously human-annotated samples spanning eight rigorous categories, providing a structured testbed for probing model intelligence across diverse dimensions, ranging from commonsense and spatial dynamics to specialized subject domains. Our framework introduces a multi-dimensional evaluation protocol consisting of four metrics: \textit{Reasoning Alignment}, \textit{Temporal Consistency}, \textit{Physical Rationality}, and \textit{Visual Quality}. To further support scalable evaluation, we propose an automated pipeline leveraging Large Multimodal Models (LMMs) to emulate human-centric assessment. Extensive experiments on 11 state-of-the-art TI2V models reveal pervasive deficiencies in simulating complex scenarios under implicit constraints, offering critical insights for the advancement of future world-simulating generative models.