Research map / Image generation / Text-to-image diffusion
Beyond Patches: Global-aware Autoregressive Model for Multimodal Few-Shot Font Generation
Haonan Cai, Yuxuan Luo, Zhouhui Lian
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 |
|---|---|---|---|
| add language-style adapter with text description to 4 visual references | GAR-Font · UFSC (Large) | FID: 7.7155 (8 refs) -> 7.4915 (M4) | Improved |
| add text description to 2 visual references | GAR-Font · UFSC (Large) | RMSE: 0.2816 -> 0.2811 | No clear effect |
| add text description | GAR-Font · UFSC (Large) | Acc(S): 0.1741 -> 0.1688 | Hurt |
| add ViT blocks after CNN (causal) in tokenizer (ablation) | G-Tok · UFUC (Small) | FID: 10.5239 -> 8.4841 | Improved |
| use causal instead of non-causal ViT (ablation) | G-Tok · UFUC (Small) | FID: 8.7504 -> 8.4841 | Improved |
| use pure ViT-6 instead of CNN tokenizer (ablation) | G-Tok · Reconstruction robustness (Small) | FID: 28.4279 -> 98.4270 | Hurt |
| add pixel-level loss (ablation) | GAR-Font AR generator · UFSC (Small) | FID: 9.6696 -> 7.9484 | Improved |
| use soft decoding instead of hard (ablation) | GAR-Font AR generator · UFSC (Small) | FID: 8.3024 -> 7.9484 | Improved |
| decoupled adapter instead of joint training (ablation) | GAR-Font(M4) · UFSC (Large) | FID: 7.9272 -> 7.4915 | Improved |
| add novel font adaptation (NFA-8) post-refinement | GAR-Font(I8) · UFSC (Large) | RMSE: 0.2772 -> 0.2600 | Improved |
| add GRPO structural enhancement (SE) after NFA | GAR-Font(I8,+NFA-8) · UFSC (Large) | FID: 6.5634 -> 8.9851 (RMSE 0.2600 -> 0.2503) | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- Font-Diffuser — FontDiffuser: One-Shot Font Generation via Denoising Diffusion with Multi-Scale Content Aggregation and Style Contrastive Learning (AAAI 2024) · better on 2 of 2 shared results
- IF-Font — IF-Font: Ideographic Description Sequence-Following Font Generation (NeurIPS 2024) · better on 2 of 2 shared results
- CF-Font — CF-Font: Content Fusion for Few-Shot Font Generation (CVPR 2023) · 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
- IF-Font: Ideographic Description Sequence-Following Font Generation (NeurIPS 2024 · similarity 0.90)
- Generate Like Experts: Multi-Stage Font Generation by Incorporating Font Transfer Process into Diffusion Models (CVPR 2024 · similarity 0.89)
- VQ-FONT: Few-Shot Font Generation with Structure-Aware Enhancement and Quantization (AAAI 2024 · similarity 0.89)
- FontDiffuser: One-Shot Font Generation via Denoising Diffusion with Multi-Scale Content Aggregation and Style Contrastive Learning (AAAI 2024 · similarity 0.88)
- VecGlypher: Unified Vector Glyph Generation with Language Models (CVPR 2026 · similarity 0.88)
- Towards Visual Text Design Transfer Across Languages (NeurIPS 2024 · similarity 0.87)
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.
