Search papers, labs, and topics across Lattice.
This paper introduces KeyFrame-Compass, a comprehensive benchmark designed to evaluate keyframe-conditioned video generation across various application domains and settings. The study reveals that existing models struggle to balance faithful keyframe execution with overall video quality, particularly as the density of keyframe constraints increases. Key findings indicate that most open-source models inadequately interpret storyboard-grid inputs as temporally ordered sequences, highlighting significant limitations in current video generation systems.
Current video generation models face a critical trade-off between faithfully executing keyframes and producing natural-looking videos, with performance degrading under increased keyframe density.
Video generation increasingly relies on keyframe-based workflows, where creators specify a sequence of reference images to guide generation. Although recent models support multi-keyframe conditioning, it remains unclear whether they can faithfully reproduce the prescribed keyframes while maintaining overall video quality. We present KeyFrame-Compass, the first comprehensive benchmark for evaluating keyframe-conditioned video generation. The benchmark contains 386 carefully curated samples spanning three application domains, two video structures, two prompt granularities, two conditioning formats, and four keyframe densities, enabling controlled analysis under diverse generation settings. We further introduce an automated evaluation framework that jointly measures keyframe execution and overall video quality. Specifically, we decompose keyframe execution into six complementary metrics covering presence, fidelity, temporal ordering, localization, persistence, and uniqueness, while assessing overall video quality through evidence-grounded MLLM judgments augmented with specialized perception models. Experiments on nine representative video generation systems reveal several fundamental limitations. Current models exhibit a clear trade-off between faithful keyframe execution and natural video synthesis. Their performance further degrades as keyframe constraints become denser and most open-source models also fail to interpret storyboard-grid inputs as temporally ordered keyframe sequences.
