把视觉经验做成可以检索、复制、比较和复用的系统。
We build visual reference libraries, prompt systems, character workflows, and interfaces for repeatable AI creation.
Live Lab · Catalog · Compare · Source · Contribute
One-line version: HosieryLab is a structured hosiery visual library where an image, its metadata, and its reusable prompt layers stay connected.
HosieryLab turns a vague request like “black tights” into a controlled visual reference:
- Denier and opacity — 5D, 15D, 20D, 40D, 60D, 100D and beyond
- Color and surface — black, white, ivory, neutrals, matte, satin, gloss
- Construction — pantyhose, knee-highs, thigh-highs, welt, toe, heel, knit, pattern
- Prompt layers — full-body context plus a focused hosiery specification
- Website-first access — images, metadata and prompts are connected through the live catalog for people and tools to reuse
![]() 5D · PURE WHITE · KNEE-HIGH |
![]() 15D · JET BLACK · PANTYHOSE |
![]() 60D · JET BLACK · PANTYHOSE |
![]() 100D · JET BLACK · PANTYHOSE |
![]() 15D · IVORY · PANTYHOSE |
![]() 50D · IVORY · PANTYHOSE |
| Project | What it is | Best entry |
|---|---|---|
| HosieryLab | A searchable visual and prompt library for hosiery | Open the live catalog |
| Character Library | A separate home for consistent AI characters | Browse the repository |
| Aurora Profile | The public index for the lab and its experiments | Read the source |
Find a specimen → inspect the image → copy one prompt layer → render → compare one variable
The system is deliberately split into small decisions. Change the Denier, opacity, color, finish, or construction without losing the rest of the scene.
- Creators: use the catalog as a visual reference before generating.
- Prompt engineers: copy the full-body prompt or the single hosiery layer.
- MCP users: use the website as the source for connected image, metadata, and prompt records.
- Contributors: report missing constructions, confusing labels, broken links, or useful workflows in Issues.
- Searchable hosiery catalog with structured filters
- Separate full-body and hosiery prompt layers
- English and Chinese website paths
- Image and metadata delivery through the live site
- Public contributor guide and issue templates
- More reusable MCP examples for image and prompt workflows







