Problem
Research questions and friction points this paper is trying to address.
scientific poster generation
editable output
design diversity
print-ready
multi-agent system
PosterMELD: Multi-Agent Paper-to-Poster Generation for Controllable Design Diversity with Editable Print-Ready Outputs
🤖 AI Summary
Existing methods for scientific poster generation struggle to simultaneously achieve editability, design diversity, and print-ready quality, often masking request-level failures. This work proposes a template-guided multi-agent generation framework that employs capacity-aware slot guidance for content composition, integrates a deterministic gating mechanism with a vision-language model (VLM) review process for failure detection and repair, and outputs editable PPTX files alongside high-resolution PNG posters. The approach introduces explicit design controls to produce diverse variants and establishes, for the first time, the Print-Ready Rate (PRR) and CHE scoring metrics to evaluate generation quality. Evaluated on 621 papers, the method achieves a PRR of 81.3%—5.2× and 3.4× higher than PosterGen and P2P, respectively—and attains the highest CHE score at a cost of only $0.38 per request.
📝 Abstract
Scientific poster construction compresses a long multimodal paper into a readable, editable canvas. Existing systems hide request-level failures by scoring only completed outputs; direct image generation is not element-editable, while coding-agent workflows are costly. PosterMELD is a template-conditioned multi-agent pipeline: capacity-aware slots guide writing before rendering, and deterministic gates plus vision-language model (VLM) review route failures to bounded repair. Each accepted request exports editable PowerPoint (PPTX) and Portable Network Graphics (PNG) artifacts; explicit design controls yield same-paper variants. Across 621 papers, Print-Ready Rate (PRR) counts requests passing geometric, readability, asset-integrity, and obvious-factual-error checks, with native editability reported separately. A frozen VLM assigns conditional Craftsmanship-Harmony-Expressiveness (CHE) scores to print-ready outputs. PosterMELD attains 81.3% PRR, 3.4 times P2P's rate and 5.2 times PosterGen's, and the highest conditional CHE among generated methods with multiple print-ready outputs. Native editability and explicit design controls are retained at a mean cost of USD 0.38 per request, 3.5% of Codex+Skill's. Code and resources are available at https://github.com/Shannon4Science/PosterMELD.
Innovation
Methods, ideas, or system contributions that make the work stand out.
multi-agent generation
template-conditioned design
editable poster output
vision-language model (VLM) review
print-ready rate (PRR)
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
