Research map / Image generation / Editing and style transfer
DG-Font: Deformable Generative Networks for Unsupervised Font Generation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 two rather than one FDSC module (ablation) | DG-Font · Chinese font generation | FID: 47.99 -> 46.15 | Helped |
| Add a third FDSC module (ablation) | DG-Font · Chinese font generation | FID: 46.15 -> 47.12 | Hurt |
| Replace both FDSC modules with ordinary U-Net skip connections (ablation) | DG-Font · Chinese font generation | LPIPS: 0.1006 -> 0.1114 | Hurt |
| Add content-encoder deformable convolution layers (ablation) | DG-Font baseline · 187 handwritten fonts | L1 loss | Helped |
| Add an FDSC module (ablation) | DG-Font with content-encoder deformable convolutions · 187 handwritten fonts | character structure completeness | Helped |
| Add offset normalization (ablation) | DG-Font with FDSC · 187 handwritten fonts | similarity to target style | Helped |
| Add the second FDSC module (ablation) | DG-Font with one FDSC module · Chinese font generation | generation performance: 0.0563 -> 0.0562 L1 loss; 0.2005 -> 0.1994 RMSE; 0.7549 -> 0.7580 SSIM; 47.99 -> 46.15 FID | Helped |
| Set lambda_cnt from 0 to 0.1 (ablation) | DG-Font · 400 seen fonts | FID: 52.91 -> 46.15 | Helped |
| Increase lambda_cnt from 0.1 to 1 (ablation) | DG-Font · 400 seen fonts | L1 loss: 0.0562 -> 0.0591 | Hurt |
| Set lambda_img from 0 to 0.1 (ablation) | DG-Font · 400 seen fonts | FID: 59.29 -> 46.15 | Helped |
| Increase lambda_img from 0.1 to 1 (ablation) | DG-Font · 400 seen fonts | L1 loss: 0.0562 -> 0.0627 | Hurt |
| Set lambda_offset from 0 to 0.5 (ablation) | DG-Font · 400 seen fonts | FID: 52.60 -> 46.15 | Helped |
Closest papers by meaning
- Generate Like Experts: Multi-Stage Font Generation by Incorporating Font Transfer Process into Diffusion Models (CVPR 2024 · similarity 0.87)
- FontDiffuser: One-Shot Font Generation via Denoising Diffusion with Multi-Scale Content Aggregation and Style Contrastive Learning (AAAI 2024 · similarity 0.86)
- VecGlypher: Unified Vector Glyph Generation with Language Models (CVPR 2026 · similarity 0.85)
- VQ-FONT: Few-Shot Font Generation with Structure-Aware Enhancement and Quantization (AAAI 2024 · similarity 0.84)
- DreamText: High Fidelity Scene Text Synthesis (CVPR 2025 · similarity 0.84)
- VecFusion: Vector Font Generation with Diffusion (CVPR 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.