Anime Image2Prompt
Upload an anime image or illustration to generate structured Danbooru tags and a copy-ready prompt. The model identifies visual features, characters, and source franchises separately.
64 1024
Uses fewer image tokens to reduce peak VRAM, with a small possible loss of detail.
The model loads on startup. A cold start may take several minutes.
This model is designed for anime images and illustrations. Outputs may be incomplete or contain sensitive-content tags, so please review them. View the model on Hugging Face
Uploaded images are used only for the current inference request. This app does
not intentionally persist images or generated tags.