sleepy-zone/ai-ref-library-cn

This project is a Chinese-language browsing and search front end for three reference image sets published by illustrator あぎ / ギア: face references (1,499 images), faceless grey-mannequin poses (288 poses, 1,439 images) and costume-only references (1,407 sets). It is aimed at people generating AI images who want to find a face, pose or outfit reference quickly and copy a ready prompt. The interface itself contains only thumbnails; full-resolution originals are hosted elsewhere.

What changes in the workflow

The repository translates the three galleries' categories and tags into Chinese, adds bilingual search, and puts all three libraries behind one entry point with a one-click copy of the image prompt README. The upstream galleries remain separate sites face library, pose library, costume library; this project's contribution is the combined, Chinese-searchable index rather than new source material.

How it runs and where images come from

Open index.html directly from disk; the README states it works under file:// without a server. If the browser blocks local scripts, a static server in the directory is the fallback. Thumbnails are embedded in data/*.js, so the gallery does not show broken images in any environment README.

The "download original" button first tries a locally configured directory, but only when that directory is set in config.js, the file exists, and the page was opened via file://. Otherwise it redirects to the author's source site, which the README says is slow from mainland China and recommends downloading originals instead. Bulk originals are offered through Baidu Pan and Quark Pan links README. Putting the downloaded originals in the parent directory, matching config.js, enables fast offline use.

Licensing is the main constraint

The three libraries have different terms. The face library is CC0 and images may be redistributed. The pose and costume libraries allow free commercial use, modification and reference, but redistribution of the image files themselves, including modified versions, is prohibited README. The repository holds only thumbnails for search and explicitly asks users not to repackage originals. All three sets are AI-generated and involve no real people; the costume library's author checked brand similarity but does not guarantee non-infringement, so commercial users must judge for themselves. The repository declares no license of its own and is provided as-is without warranty.

Fit

It suits people who work with these three specific galleries and want Chinese tags and search. It is not a general reference library, it does not supply originals, and the pose/costume license makes it unsuitable as a redistribution source.