Research map / Image generation / Text-to-image diffusion
WebRPG: Automatic Web Rendering Parameters Generation for Visual Presentation
Zirui Shao, Feiyu Gao, Hangdi Xing, Zepeng Zhu, Zhi Yu, Jiajun Bu, Qi Zheng, Cong Yao
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 VAE to compress rendering parameters instead of flat one-dimensional input (ablation) | WebRPG-AR · WebRPG test set | Overall FID: 0.9702 -> 0.1281 | Improved |
| add hierarchical (DOM tree) embedding instead of 1D positional embedding (ablation) | WebRPG-AR · WebRPG test set | Overall FID: 0.3003 -> 0.1281 | Improved |
| add semantic embedding from MarkupLM (ablation) | WebRPG-AR · WebRPG test set | Overall FID: 0.1797 -> 0.1281 | Improved |
| add character count embedding (ablation) | WebRPG-AR · WebRPG test set | Overall FID: 0.1575 -> 0.1281 | Improved |
| add Z_mask (ablation) | WebRPG-AR · WebRPG test set | Overall FID: 0.1487 -> 0.1281 | No clear effect |
| use autoregressive decoding instead of diffusion model | WebRPG baselines · WebRPG test set | Overall FID: 62.021 -> 0.1281 | Improved |
| use trained WebRPG-AR instead of few-shot GPT-4 | rendering parameter generation · WebRPG test set (10% subset) | Overall FID: 4.2141 -> 0.1324 | Improved |
| prompt LLMs (StarCoder2, CodeLlama) to generate rendering parameters rather than leaving CSS absent | open-source code LLMs · WebRPG test set (10% subset) | Overall FID: plain HTML 8.5342 vs StarCoder2-7b 11.899 and CodeLlama-13b 9.2826 | Hurt |
Closest papers by meaning
- P2P: Automated Paper-to-Poster Generation and Fine-Grained Benchmark (ICLR 2026 · similarity 0.83)
- Web2Code: A Large-scale Webpage-to-Code Dataset and Evaluation Framework for Multimodal LLMs (NeurIPS 2024 · similarity 0.82)
- AnyDoc: Enhancing Document Generation via Large-Scale HTML/CSS Data Synthesis and Height-Aware Reinforcement Optimization (CVPR 2026 · similarity 0.82)
- LayoutNUWA: Revealing the Hidden Layout Expertise of Large Language Models (ICLR 2024 · similarity 0.81)
- AutoPresent: Designing Structured Visuals from Scratch (CVPR 2025 · similarity 0.81)
- DesignDiffusion: High-Quality Text-to-Design Image Generation with Diffusion Models (CVPR 2025 · similarity 0.81)
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.
