Research map / Image generation / Autoregressive and latent generation
LiveFigure: Generating Editable Scientific Illustration with VLM Agents
Chenyang Shao, Jiahe Liu, Fengli Xu, Yong Li
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 the Experience module (ablation) | LiveFigure · Ablation | Executability (%): 40.0% -> 83.5% | Improved |
| add the Experience module (ablation) | LiveFigure · Ablation | VLM score: 6.52 -> 7.28 | Improved |
| add the Skill library (ablation) | LiveFigure · Ablation | VLM score: 5.47 -> 7.28 | Improved |
| add the Skill library (ablation) | LiveFigure · Ablation | Executability (%): 77.3% -> 83.5% | Improved |
| add targeted visual refinement (Stage III) (ablation) | LiveFigure · Ablation | VLM score: 6.78 -> 7.28 | Improved |
| add visual prior planning (ablation) | LiveFigure · Ablation | VLM score: 6.81 -> 7.28 | Improved |
| add visual prior planning (ablation) | LiveFigure · Ablation | Executability (%): 82.3% -> 83.5% | No clear effect |
| add Stage 1 visual prior to vanilla LLM (ablation) | Vanilla LLM · Stage-wise | Debug turns: 3.40 -> 3.50 | Hurt |
| add Stage 2 skills and experience (ablation) | Vanilla LLM + Stage 1 · Stage-wise | Executability (%): 32.5% -> 83.5% | Improved |
| add richer method description to input (V1 to V2) | LiveFigure · Scientific figure test set | VLM average score: 7.30 -> 7.89 | Improved |
| remove Skill library (ablation) | LiveFigure · Edit distance | Median edit distance: 8 -> 31 steps | Hurt |
| remove Experience module (ablation) | LiveFigure · Edit distance | Median edit distance: 8 -> 20 steps | Hurt |
Closest papers by meaning
- DeTikZify: Synthesizing Graphics Programs for Scientific Figures and Sketches with TikZ (NeurIPS 2024 · similarity 0.84)
- AutomaTikZ: Text-Guided Synthesis of Scientific Vector Graphics with TikZ (ICLR 2024 · similarity 0.84)
- Factuality Matters: When Image Generation and Editing Meet Structured Visuals (ICLR 2026 · similarity 0.84)
- Paper2Poster: Towards Multimodal Poster Automation from Scientific Papers (NeurIPS 2025 · similarity 0.83)
- P2P: Automated Paper-to-Poster Generation and Fine-Grained Benchmark (ICLR 2026 · similarity 0.82)
- ChartE$^{3}$: A Comprehensive Benchmark for End-to-End Chart Editing (ICML 2026 · similarity 0.82)
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.
