Research map / Image generation / Text-to-image diffusion
CLIPDraw: Exploring Text-to-Drawing Synthesis through Language-Image Encoders
Kevin Frans, Lisa Soros, Olaf Witkowski
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 |
|---|---|---|---|
| Represent drawings with Bézier strokes rather than optimizing pixels | CLIPDraw · Text-to-drawing synthesis comparison | human recognizability | Improved |
| Use a fixed pre-trained generator's latent space | BigGAN Optimization · Out-of-distribution prompts | prompt coverage | Hurt |
| Remove image augmentation (ablation) | CLIPDraw · Text-to-drawing synthesis comparison | human recognizability | Hurt |
| Increase stroke count (ablation) | CLIPDraw · The Eiffel Tower | drawing detail | Improved |
| Use a low stroke count (ablation) | CLIPDraw · Text-to-drawing synthesis | simplicity of representation | Improved |
| Add negative prompt “Words and text” (ablation) | CLIPDraw · Hashtag | text in drawing | Improved |
| Add negative prompts “Purple” and “Red” (ablation) | CLIPDraw · A torii gate | color match to intended negative prompt | Improved |
| Add negative prompt “many sailboats” (ablation) | CLIPDraw · A realistic painting of a sailboat | number of sailboats depicted | Improved |
| Use negative prompts as fine-grained controls (ablation) | CLIPDraw · Text-to-drawing synthesis | reliability of intended adjustment | Mixed |
Closest papers by meaning
- DialogDraw: Image Generation and Editing System Based on Multi-Turn Dialogue (AAAI 2025 · similarity 0.84)
- CLIPDrag: Combining Text-based and Drag-based Instructions for Image Editing (ICLR 2025 · similarity 0.83)
- SketchAgent: Language-Driven Sequential Sketch Generation (CVPR 2025 · similarity 0.83)
- Gradient-based Visual Explanation for Transformer-based CLIP (ICML 2024 · similarity 0.82)
- Text to Sketch Generation with Multi-Styles (NeurIPS 2025 · similarity 0.82)
- Open Vocabulary Semantic Scene Sketch Understanding (CVPR 2024 · 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.
