ReFigBench: Benchmarking Scientific Figure Reconstruction as Editable PowerPoint Artifacts

2026-09-16

AI Summary

研究通过ReFigBench框架评估科学图表转为可编辑PPT的效果，采用直接代码生成和专门PPTX工作流两种方法，揭示了保真度与可编辑性之间的挑战。

Abstract

Multimodal coding agents are expected to turn visual inputs into usable artifacts, and they act through a harness, the layer of tools, context management, and execution environment around the model. Existing evaluations often isolate short tool calls, API traces, or screenshot resemblance, and a low score under these proxies cannot say whether the model saw poorly, planned poorly, or was failed by its harness. We study scientific overview figure reconstruction, an agent task in which a source image must become an editable PowerPoint slide that preserves text, topology, layout, and native document structure. We introduce ReFigBench, a benchmark and evaluation framework built on 1,000 real overview figures retrieved from arXiv papers with full provenance. Coding agents from four model families reconstruct every figure under two workflows, direct code generation and a specialized PPTX workflow, and the strongest model runs inside two commercial harnesses, yielding ten configurations. Evaluation combines deterministic artifact checks, repeated automated scoring by judges from two model families, and blinded human comparisons. Perception remains a bottleneck that iterative rendering only partly repays. Whether workflow effort converts into quality depends on the model together with its harness, since the same model gains from the specialized workflow inside one harness and loses inside the other, and the harness shifts scores even under an identical direct prompt. The specialized workflow erases native connectors in every configuration, human judges still prefer its renderings in most matchups, and even the strongest agent falls short of the rubric ceiling. These results expose the tension between fidelity and editability as the central challenge for practical multimodal document agents.

Problem

• Scientific Figure Reconstruction
• Editable PowerPoint
• Multimodal Document Agents
• Fidelity and Editability

Innovation

• Benchmarking
• Scientific Figure Reconstruction
• Editable PowerPoint
• Harness Interaction
• Workflow Comparison