Research map / Image generation / Text-to-image diffusion
PosterIQ: A Design Perspective Benchmark for Poster Understanding and Generation
Yuheng Feng, Wen Zhang, Haodong Duan, Xingxing Zou
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use closed commercial MLLMs instead of open-source MLLMs | poster understanding · PosterIQ understanding average | Average score: Gemini-2.5-Pro 0.647 vs Qwen3-VL-8B 0.543 | Improved |
| use MLLMs as automatic poster quality raters | all evaluated MLLMs · Overall rating | cosine similarity to human ratings: best 0.483 (Gemma 3N-E4B-IT); GPT-5 0.347 | Hurt |
| move from simple to hard OCR conditions | MLLMs · Hard OCR vs Simple OCR | word recall: GPT-5 0.965 -> 0.496; Grok-4-fast 0.769 -> 0.044 | Hurt |
| Qwen family shows stable performance under heavy visual interference | Qwen3-VL-8B · Hard OCR | word recall: 0.781 vs GPT-5 0.496 | Improved |
| font matching without font-name priors | open-source MLLMs · Font Matching | normalized score: near 0 (e.g. MiniCPM-V-4.5 -0.001) | Hurt |
| larger Qwen3-VL (8B vs 4B) | Qwen3-VL · Style understanding | score: 0.805 -> 0.610 | Hurt |
| larger Qwen3-VL (8B vs 4B) | Qwen3-VL · Text localization | mean IoU: 0.223 -> 0.450 | Improved |
| closed models for layout generation | Gemini-2.5-Pro · Layout Generation | Area Ratio: 0.569 vs open best 0.394 | Improved |
| compare generators on font generation | image generators · Font generation | richness: best 0.391 vs composition up to 0.866 | Hurt |
| iteratively add VLM feedback to revise prompts for a T2I generator (ablation) | GPT-Image-1 and Qwen-Image with GPT-5 / Qwen3-VL · Qualitative poster generation | poster quality (qualitative) | Improved |
Closest papers by meaning
- PosterO: Structuring Layout Trees to Enable Language Models in Generalized Content-Aware Layout Generation (CVPR 2025 · similarity 0.88)
- DesignDiffusion: High-Quality Text-to-Design Image Generation with Diffusion Models (CVPR 2025 · similarity 0.88)
- POSTA: A Go-to Framework for Customized Artistic Poster Generation (CVPR 2025 · similarity 0.88)
- Can Vision–Language Models Assess Graphic Design Aesthetics? A Benchmark, Evaluation, and Dataset Perspective. (ICLR 2026 · similarity 0.88)
- Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers (NeurIPS 2025 · similarity 0.87)
- P2P: Automated Paper-to-Poster Generation and Fine-Grained Benchmark (ICLR 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
