AI summaryⓘ
The authors developed POSTERHARNESS, a system to check if AI models create scientific posters correctly, focusing on things like readable labels and avoiding fake scientific images. Their method separates the design of the poster layout from inserting actual scientific figures, using placeholders that get replaced later. This makes it easier to measure if the poster follows rules, like matching aspect ratios and using real figures only. They tested it on papers from physics and AI, finding it helps prevent fake images but also introduces challenges with placeholder design and text quality. They compare their system with another method, noting different pros and cons without claiming one is better.
text-rich image modelsscientific poster designvisual summary designplaceholdersfigure provenanceaspect ratiotemplate criticfailure taxonomyVLM (Vision-Language Model)Paper2Poster
Authors
Tianyi Yang, Dawei Fu, Youpeng Wu, Zixun Kou, Linrui Chen, Ruobing Jiang, Zijian Wang, Qiang Li
Abstract
Text-rich image models can now design poster-scale layouts, but we lack ways to measure whether they honor scientific communication contracts: legible labels, prescribed aspect ratios, and -- above all -- abstaining from fabricated scientific figures. We present POSTERHARNESS, an auditable harness reframing poster generation as measurable instruction-following tasks, with a pilot benchmark and failure taxonomy. POSTERHARNESS uses a placeholder-first contract to separate two jobs models otherwise conflate. The model performs visual-summary design: typography, reading path, color, and background -- but never draws data-bearing figures. Every figure region must be an empty labeled placeholder; a deterministic compositor inserts real source-paper figures at detected coordinates. This makes properties measurable: placeholder count and ID accuracy, blankness, aspect-ratio compliance, abstention from synthesized graphics, public-text hygiene, and source-figure provenance -- with failures logged as explicit rejections, not hidden in plausible-looking output. We instantiate the harness on 12 papers (6 HEP, 6 AI/ML-adjacent) and report three findings. (i) A counterfactual probe shows the placeholder contract drives VLM-counted synthesized figures from 34 to 0 across three papers. (ii) A failure taxonomy identifies blocking contracts: placeholder geometry, placeholder QA, template critic, and public text. (iii) Comparison with Paper2Poster shows a trade-off: PosterHarness yields higher-resolution artifacts, lower white-canvas fraction, and stronger VLM visual preference; the deterministic baseline retains slightly more PosterQuiz-style information and runs faster. We report this as regime characterization, not a superiority claim. All artifacts, prompts, manifests, and audit scripts are released as a reusable evaluation component.