kaonashi-tyc/zi2zi

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2745 stars · 475 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

zi2zi is a TensorFlow implementation of a conditional GAN (based on pix2pix) for learning and transferring Chinese and East Asian character typefaces. It generates new font styles from example fonts using category embeddings, category and constant losses, and optional label shuffling.

Use cases

• generate a new Chinese calligraphy font from a few sample characters
• convert text rendered in one font style to another style
• create brush writing or cursive script typefaces automatically
• generate Korean hangul typefaces from Chinese font training
• interpolate between font styles for animation effects

When to choose

• you need to synthesize East Asian typefaces in new styles from limited samples
• you want a research baseline or reference implementation of conditional GAN font style transfer
• you are working with TensorFlow 1.x and have CUDA GPUs available

When to avoid

• you need a maintained library or support for modern TensorFlow versions
• you want to generate Latin or non-East-Asian scripts without adaptation
• you lack GPU resources or need production-ready font generation

Facets

library · maturity abandoned

machine-learning deep-learning image-processing machine-learning deep-learning graphics python gan pix2pix style-transfer chinese-characters calligraphy font-generation tensorflow natural-language-processing gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="kaonashi-tyc/zi2zi")