Research map / Image generation / Text-to-image diffusion
NeuralSVG: An Implicit Representation for Text-to-Vector Generation
Sagi Polaczek, Yuval Alaluf, Elad Richardson, Yael Vinker, Daniel Cohen-Or
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 NeuralSVG instead of VectorFusion at 16 shapes | text-to-SVG · 128-prompt CLIP evaluation | R-Precision: 48.03 -> 67.18 | Improved |
| use NeuralSVG instead of SVGDreamer at 16 shapes | text-to-SVG · 128-prompt CLIP evaluation | R-Precision: 43.30 -> 67.18 | Improved |
| use NeuralSVG instead of VectorFusion at 16 shapes | text-to-SVG · 128-prompt CLIP evaluation | Text-Image Sim.: 23.47 -> 26.94 | Improved |
| use NeuralSVG instead of SVGDreamer at 16 shapes | text-to-SVG · 128-prompt CLIP evaluation | Text-Image Sim.: 20.89 -> 26.94 | Improved |
| baselines use many more shapes (64 or 256) than NeuralSVG's 16 | VectorFusion, SVGDreamer · 128-prompt CLIP evaluation | R-Precision: 67.18 vs 83.46 and 85.03 | Hurt |
| use 64 or 256 shapes in baselines vs NeuralSVG 16 | VectorFusion, SVGDreamer · 128-prompt CLIP evaluation | Text-Image Sim.: 26.33, 26.58 vs 26.94 | No clear effect |
| add the two-branch MLP (separate control points and colors) instead of a single MLP (ablation) | NeuralSVG · text prompts (qualitative) | visual quality | Improved |
| add the neural representation instead of directly optimizing shapes (ablation) | NeuralSVG · text prompts (qualitative) | visual quality | Improved |
| add dropout over shapes (ablation) | NeuralSVG · text prompts (qualitative) | semantic meaning of early shapes | Improved |
Closest papers by meaning
- Chat2SVG: Vector Graphics Generation with Large Language Models and Image Diffusion Models (CVPR 2025 · similarity 0.90)
- SVGDreamer: Text Guided SVG Generation with Diffusion Model (CVPR 2024 · similarity 0.89)
- Empowering LLMs to Understand and Generate Complex Vector Graphics (CVPR 2025 · similarity 0.87)
- InternSVG: Towards Unified SVG Tasks with Multimodal Large Language Models (ICLR 2026 · similarity 0.87)
- SVGBuilder: Component-Based Colored SVG Generation with Text-Guided Autoregressive Transformers (AAAI 2025 · similarity 0.86)
- StarVector: Generating Scalable Vector Graphics Code from Images and Text (AAAI 2025 · similarity 0.86)
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.
