Research map / Vision-language and multimodal / CLIP and zero-shot
FontCopilot: Towards Generalist Multimodal Large Language Models for Holistic Chinese Font Engineering
Yu Liu, Yang Liu, Ying Gao, Yang Ding, Cunrui Wang
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 FontCopilot SVG-code MLLM instead of instruction-tuned Qwen3-VL-8B | glyph vectorization · Glyph vectorization | IoU: 0.865 -> 0.925 | Improved |
| use FontCopilot instead of non-learning tracing tools for raster fidelity | glyph vectorization · Glyph vectorization | IoU / CD: IoU 0.982 (AIT) vs 0.925; CD 0.0041 vs 0.0355 | Hurt |
| use FontCopilot for node compression instead of Qwen3-VL-8B | vector optimization · Vector optimization | Reduction Ratio (%): 71.88 -> 76.52 | Improved |
| use FontCopilot for defect detection | defect detection · Defect detection | IoU: 0.875 -> 0.938 | Improved |
| use FontCopilot for geometric correction | geometric correction · Geometric correction | IoU: 0.825 -> 0.908 | Improved |
| replace manual refinement with FontCopilot outputs | font production · Human evaluation | fix time (s): 184.1 -> 24.5 | Improved |
| raise number of few-shot reference glyphs K beyond 3 (ablation) | FontCopilot · Reference count sensitivity K 0-8 | geometric error | No clear effect |
| raise visual injection layer count L beyond 3 (ablation) | FontCopilot · Injection layers L 0-8 | geometric error | Hurt |
| use graphics-based methods vs FontCopilot for raster-to-vector accuracy | Potrace/AIT · Human evaluation V | rating (1-5): 4.15 (AIT) vs 4.12 (FontCopilot) | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- StarVector-8B — StarVector: Generating Scalable Vector Graphics Code from Images and Text (AAAI 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- VQ-FONT: Few-Shot Font Generation with Structure-Aware Enhancement and Quantization (AAAI 2024 · similarity 0.89)
- DeepCalliFont: Few-Shot Chinese Calligraphy Font Synthesis by Integrating Dual-Modality Generative Models (AAAI 2024 · similarity 0.88)
- VecGlypher: Unified Vector Glyph Generation with Language Models (CVPR 2026 · similarity 0.87)
- Decoupling Layout from Glyph in Online Chinese Handwriting Generation (ICLR 2025 · similarity 0.86)
- Font-Agent: Enhancing Font Understanding with Large Language Models (CVPR 2025 · similarity 0.86)
- IF-Font: Ideographic Description Sequence-Following Font Generation (NeurIPS 2024 · similarity 0.84)
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.
