Search papers, labs, and topics across Lattice.
The paper introduces OSCBench, a new benchmark to evaluate text-to-video (T2V) models on their ability to generate videos that accurately depict object state changes (OSC) described in the text prompt. OSCBench uses instructional cooking data organized into regular, novel, and compositional scenarios to test both in-distribution performance and generalization. Evaluations of six T2V models reveal that they struggle with accurate and temporally consistent OSC, particularly in novel and compositional settings, highlighting OSC as a key bottleneck.
Current text-to-video models can generate visually appealing videos, but they often fail to accurately depict how actions change the state of objects, like a potato being peeled.
Text-to-video (T2V) generation models have made rapid progress in producing visually high-quality and temporally coherent videos. However, existing benchmarks primarily focus on perceptual quality, text-video alignment, or physical plausibility, leaving a critical aspect of action understanding largely unexplored: object state change (OSC) explicitly specified in the text prompt. OSC refers to the transformation of an object's state induced by an action, such as peeling a potato or slicing a lemon. In this paper, we introduce OSCBench, a benchmark specifically designed to assess OSC performance in T2V models. OSCBench is constructed from instructional cooking data and systematically organizes action-object interactions into regular, novel, and compositional scenarios to probe both in-distribution performance and generalization. We evaluate six representative open-source and proprietary T2V models using both human user study and multimodal large language model (MLLM)-based automatic evaluation. Our results show that, despite strong performance on semantic and scene alignment, current T2V models consistently struggle with accurate and temporally consistent object state changes, especially in novel and compositional settings. These findings position OSC as a key bottleneck in text-to-video generation and establish OSCBench as a diagnostic benchmark for advancing state-aware video generation models.
NUS