ViewFusion: Structured Spatial Thinking Chains for Multi-View Reasoning

TL;DR

ViewFusion introduces a two-stage framework for multi-view spatial reasoning that explicitly models cross-view relations, significantly improving accuracy on view-dependent tasks by separating spatial pre-alignment from question answering.

Contribution

The paper proposes ViewFusion, a novel two-stage approach that explicitly separates spatial pre-alignment from reasoning, enhancing multi-view reasoning capabilities in vision-language models.

Findings

Improves MMSI-Bench accuracy by 5.3% over baseline models.

Achieves larger gains on tasks requiring cross-view alignment.

Stabilizes reasoning behavior through synthetic supervision and reinforcement learning.

Abstract

Multi-view spatial reasoning remains difficult for current vision-language models. Even when multiple viewpoints are available, models often underutilize cross-view relations and instead rely on single-image shortcuts, leading to fragile performance on viewpoint transformation and occlusion-sensitive cases. We present ViewFusion, a two-stage framework that explicitly separates cross-view spatial pre-alignment from question answering. In the first stage, the model performs deliberate spatial pre-thinking to infer viewpoint relations and spatial transformations across views, forming an intermediate workspace that goes beyond a simple re-description. In the second stage, the model conducts question-driven reasoning conditioned on this workspace to produce the final prediction. We train ViewFusion with synthetic reasoning supervision followed by reinforcement learning using GRPO, which improves answer correctness while stabilizing the intended two-stage generation behavior. On MMSI-Bench, ViewFusion improves accuracy by 5.3\% over Qwen3-VL-4B-Instruct, with the largest gains on examples that require genuine cross-view alignment.