tyxsspa/AnyText

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

4874 stars · 304 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

AnyText is the official implementation of a diffusion-based model for multilingual visual text generation and editing in images, accepted as an ICLR 2024 Spotlight paper. It includes inference and training code, benchmark datasets (AnyText-benchmark, AnyWord-3M), and online demos on ModelScope and HuggingFace.

Use cases

• generate images containing accurate text in multiple languages
• edit or replace text rendered inside existing images
• add stylized text to AI-generated artwork
• train a custom text-rendering diffusion model
• evaluate text rendering accuracy in generated images
• merge AnyText weights with community SD1.5 models or LoRAs

When to choose

• you need legible, correctly spelled text inside generated images, especially for Chinese or other non-Latin scripts
• you want a research-grade, open-source text-in-image generation model with training code and datasets
• you want to integrate text rendering into Stable Diffusion 1.5 pipelines

When to avoid

• you only need simple image captioning or OCR without generation
• you need production text layout typesetting rather than diffusion-based generation
• you cannot run GPU inference with at least 8GB memory

Facets

library · maturity active

image-processing machine-learning deep-learning stable-diffusion artificial-intelligence image-processing computer-vision python cross-platform text-generation text-editing diffusion-models multilingual aigc iclr-2024 research-code natural-language-processing gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="tyxsspa/AnyText")