Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2509.24299 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916976040869888 |
|---|---|
| author | Chen, Hanqi Zhao, Zhongyin Chen, Ye Liang, Zhujin Ni, Bingbing |
| author_facet | Chen, Hanqi Zhao, Zhongyin Chen, Ye Liang, Zhujin Ni, Bingbing |
| contents | Scalable Vector Graphics (SVG) is a code-based representation for 2D visuals. Leveraging recent advances in large language models (LLMs), we study text-to-SVG generation and address two persistent gaps: weak generalization and poor adherence to input instructions. We present SVGThinker, a reasoning-driven framework that aligns the production of SVG code with the visualization process and supports the full set of SVG primitives. Our pipeline first renders each primitive in sequence and uses a multimodal model to annotate the image and code; we then build stepwise updates that mirror the incremental addition of primitives. On this data, we train an LLM with supervised fine-tuning that exposes its chain-of-thought as intermediate reasoning, improving robustness and reducing errors and hallucinations. Experiments against state-of-the-art baselines show that SVGThinker produces more stable, editable, and higher-quality SVGs while preserving the structural advantages of vector graphics. Unlike image-based methods, our outputs enable precise and hierarchical editing, opening new directions for design, content creation, and automated graphics generation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_24299 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | SVGThinker: Instruction-Aligned and Reasoning-Driven Text-to-SVG Generation Chen, Hanqi Zhao, Zhongyin Chen, Ye Liang, Zhujin Ni, Bingbing Computer Vision and Pattern Recognition Scalable Vector Graphics (SVG) is a code-based representation for 2D visuals. Leveraging recent advances in large language models (LLMs), we study text-to-SVG generation and address two persistent gaps: weak generalization and poor adherence to input instructions. We present SVGThinker, a reasoning-driven framework that aligns the production of SVG code with the visualization process and supports the full set of SVG primitives. Our pipeline first renders each primitive in sequence and uses a multimodal model to annotate the image and code; we then build stepwise updates that mirror the incremental addition of primitives. On this data, we train an LLM with supervised fine-tuning that exposes its chain-of-thought as intermediate reasoning, improving robustness and reducing errors and hallucinations. Experiments against state-of-the-art baselines show that SVGThinker produces more stable, editable, and higher-quality SVGs while preserving the structural advantages of vector graphics. Unlike image-based methods, our outputs enable precise and hierarchical editing, opening new directions for design, content creation, and automated graphics generation. |
| title | SVGThinker: Instruction-Aligned and Reasoning-Driven Text-to-SVG Generation |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.24299 |