CoDeFont improves one-shot font style transfer across diverse scripts
CoDeFont: One-Shot Font Generation via Multi-Scale Factor-Specific Representation Learning
Computer Vision and Pattern Recognition
Summary
Creating a new font style from just one example letter is very hard, especially for complex alphabets. The authors propose CoDeFont, which breaks down style and shape features from different spatial levels in the example to better understand the font’s look. This lets the system recreate different letters in the same style more accurately, even for complicated scripts like Bengali. Their tests show CoDeFont works better than previous methods for generating fonts from a single example.
What this means in practice
- •For graphic designers: Generate new font characters from a single example letter to quickly create cohesive typefaces.
- •For typography software developers: Integrate multi-scale style transfer modules to support automatic font design from minimal input samples.
Authors
Rejoy Chakraborty, Prasun Roy, Saumik Bhattacharya, Umapada Pal
Abstract
One-shot Font Generation (OFG) aims to synthesize a target glyph by transferring the stylistic characteristics of a single reference glyph while preserving the structure of the target character. However, accurately capturing font style from a single reference image remains challenging due to the limited stylistic information available in a single glyph, and this challenge is further amplified for complex scripts. We propose CoDeFont, a script-agnostic OFG framework that addresses this challenge through multi-level style-content factorization and progressive style-content reconciliation. CoDeFont independently aggregates complementary style and structural information across multiple spatial levels to obtain factor-specific representations, while a contrastive objective explicitly encourages separation between style and content embeddings. During reconstruction, the disentangled representations are progressively injected through multi-level style conditioning, enabling stylistic information to be incorporated at different decoding resolutions while preserving the underlying glyph structure. We evaluate CoDeFont on structurally distinct Bengali and Latin scripts across seen and unseen font styles. Extensive quantitative and qualitative analyses demonstrate consistent improvements over existing state-of-the-art methods, highlighting the effectiveness of the proposed framework for OFG across diverse writing systems.