nvzhuang: 1,407 AI Clothing References Plus a Virtual Try-On Backend

People building AI image workflows keep running into the same gap: a reliable library of garment references, and a way to see how “fitted” a virtual try-on should look. nvzhuang (paipiange/nvzhuang, ★1,862, MIT — created October 29, 2026 and past a thousand stars within three days) answers the first: 1,407 model-free AI-generated garment images in 2:3, sorted into eight categories (spring/summer/autumn/winter, lingerie, party, swimwear, sexy), searchable in both Chinese and English, served as a pure static frontend at paipiange.github.io/nvzhuang. The repo also carries an optional try-on backend (GoFrame): upload a selfie, pick a garment, and it calls GPT Image, Doubao Seedream or Gemini to composite a face-preserving outfit swap — that tier needs your own backend and API keys.

Core features

• 1,407-garment gallery: eight categories (spring 215, summer 254, autumn 244, winter 253, lingerie 199, party 48, swimwear 54, sexy 140), all model-free, downloadable at full resolution.
• Bilingual search: Chinese terms (白, 蕾丝, 连衣裙) and English ones (white, lace, dress) both work; garment attributes localize automatically in Chinese mode.
• Optional AI try-on: upload a portrait plus a gallery garment (or your own, top and bottom accepted separately) and the GoFrame backend composites the swap through a generation model.
• A companion pose library: the same author’s ai-pose-gallery-zh contributes 1,439 multi-angle pose references for pose-locked generation.
• Commercial-friendly: MIT for both code and images, with a compliance audit report linked from the README.

Typical use cases

• E-commerce and fashion creators drafting visuals for 1,400 outfits before any photoshoot.
• Workflow builders combining garment references, pose references and the try-on backend into a virtual studio pipeline.
• Frontend developers self-hosting a gallery: fork the static site, data included.

Unlike general-purpose image editors, this one is vertical to the clothing domain: gallery, search and try-on ship as one pipeline. If your focus is retouching generated images rather than swapping outfits, our earlier coverage of image-editing tools fits better; and for agent-driven review flows, the approach in ux-ui-agent-skills — objective gates codified as skills — transfers well.

Quick start

Browse paipiange.github.io/nvzhuang directly — no deployment, no keys. To enable try-on, clone the repo and configure the backend/ GoFrame service with a model API key (steps in the README), then upload a portrait and pick a garment.

Summary

A solid starting shelf for e-commerce, fashion and AI-image creators who need garment references at volume, and the try-on backend is a ready-made starting point for a self-hosted fitting room. Caveats: the garments are AI-generated, so details may not match real-world tailoring; try-on quality depends on the model you configure, and its API quota is on you; the “sexy” category (140 images) deserves your own content review before use. MIT, Go backend, ★1,862 within a week of creation as of November 2, 2026.