Research map / Image generation / Text-to-image diffusion
Rethinking Layered Graphic Design Generation with a Top-Down Approach
Jingye Chen, Zhaowen Wang, Nanxuan Zhao, Li Zhang, Difan Liu, Jimei Yang, Qifeng Chen
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 |
|---|---|---|---|
| add questionnaire-based result selection (ablation) | Accordion text removal · Text removal | PSNR: 18.01 -> 21.18 | Improved |
| add questionnaire-based result selection (ablation) | Accordion object removal · Object removal | PSNR: 22.92 -> 24.87 | Improved |
| joint multi-task training of the VLM instead of separate (ablation) | Accordion VLM · Design39K | average score: 72.03 -> 72.85 | No clear effect |
| joint training instead of separate (ablation) | Accordion VLM · Design39K original designs | Text Recognition NED: 72.87 -> 68.51 | Hurt |
| joint training instead of separate (ablation) | Accordion VLM · Design39K original designs | Font Accuracy: 24.51 -> 21.62 | Hurt |
| joint training instead of separate (ablation) | Accordion VLM · Design39K backgrounds without text | Object Detection F1: 83.52 -> 86.94 | Improved |
| joint training instead of separate (ablation) | Accordion VLM · Design39K nonsensical text designs | Object Detection F1: 79.27 -> 83.06 | Improved |
| add OCR prompt to the system (ablation) | Accordion VLM · Design39K original designs | paragraph-level OCR NED: 61.28 -> 68.51 | Improved |
| add OCR prompt to the system (ablation) | Accordion VLM · Design39K | text detection F1: 73.12 -> 78.59 | Improved |
| refine Flux reference images with Accordion | Flux · DesignIntention | content relevance (GPT-4V): 7.0 -> 7.4 | Improved |
| top-down Accordion instead of bottom-up COLE | COLE · DesignIntention | GPT-4V average score: 6.0 -> 6.5 | Improved |
| Accordion instead of COLE for aesthetics | COLE · DesignIntention | LAION aesthetic score: 4.72 -> 4.98 | Improved |
Closest papers by meaning
- DesignDiffusion: High-Quality Text-to-Design Image Generation with Diffusion Models (CVPR 2025 · similarity 0.87)
- Graphic Design with Large Multimodal Model (AAAI 2025 · similarity 0.86)
- LayerCraft: Enhancing Text-to-Image Generation with CoT Reasoning and Layered Object Integration (NeurIPS 2025 · similarity 0.86)
- From Elements to Design: A Layered Approach for Automatic Graphic Design Composition (CVPR 2025 · similarity 0.85)
- Desigen: A Pipeline for Controllable Design Template Generation (CVPR 2024 · similarity 0.85)
- Decomposition of Graphic Design with Unified Multimodal Model (ICML 2025 · similarity 0.85)
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.
