Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
The paper introduces token warping, a novel approach to improve MLLM visual reasoning under viewpoint changes, by manipulating image tokens in ViT-based MLLMs instead of raw pixels. They find that backward token warping, which samples source-view tokens based on a target-view grid, is more stable and semantically coherent than forward warping or pixel-based methods. Experiments on the new ViewBench benchmark show that token warping significantly improves MLLM performance in reasoning from nearby viewpoints, surpassing existing baselines.
Token warping lets MLLMs see around corners, fixing their fragile understanding of viewpoint changes where pixel warping fails.
Can warping tokens, rather than pixels, help multimodal large language models (MLLMs) understand how a scene appears from a nearby viewpoint? While MLLMs perform well on visual reasoning, they remain fragile to viewpoint changes, as pixel-wise warping is highly sensitive to small depth errors and often introduces geometric distortions. Drawing on theories of mental imagery that posit part-level structural representations as the basis for human perspective transformation, we examine whether image tokens in ViT-based MLLMs serve as an effective substrate for viewpoint changes. We compare forward and backward warping, finding that backward token warping, which defines a dense grid on the target view and retrieves a corresponding source-view token for each grid point, achieves greater stability and better preserves semantic coherence under viewpoint shifts. Experiments on our proposed ViewBench benchmark demonstrate that token-level warping enables MLLMs to reason reliably from nearby viewpoints, consistently outperforming all baselines including pixel-wise warping approaches, spatially fine-tuned MLLMs, and a generative warping method.