Description
Halfway through drawing a character you realize you have no idea what that pose looks like from below, or how this outfit folds, and no single gallery has it all. Open Pose & Costume puts poses, costumes and characters into one bilingual workspace, searchable by name, keyword or ID.
Every pose comes in five views (front, top, low, side, three-quarter) with matching depth maps, so checking an action from another angle is one thumbnail click. The costume library holds 2,769 outfits that you can filter as realistic or anime, and the character library has 1,527 references you can narrow by region and hair length. Pick a pose, a costume and a character and you can feed them straight into character generation.
Three kinds of reference in one place: poses, costumes and characters share one search that handles Chinese and English names, keywords and IDs, with filters for category, collection and style.
Multiple views and depth maps: each custom pose has five views, switched per card without touching the others, and depth maps are there when you want a structural guide for drawing or generating.
Realistic and anime outfits: 1,320 upstream anime outfits plus custom ones, filtered apart from the realistic set, with extended categories behind a switch.
Favorites and sharing: save references to My Favorites, and share links keep the language, filters and chosen references.
One-click generation: choose a pose, a costume and a character reference, connect your own image service and generate a character; cost depends on that service.
Run it locally: the project is open source on GitHub, written in TypeScript, and you can deploy your own copy. Check the licensing notes in the repo before using the assets.
Every pose comes in five views (front, top, low, side, three-quarter) with matching depth maps, so checking an action from another angle is one thumbnail click. The costume library holds 2,769 outfits that you can filter as realistic or anime, and the character library has 1,527 references you can narrow by region and hair length. Pick a pose, a costume and a character and you can feed them straight into character generation.
Features
Three kinds of reference in one place: poses, costumes and characters share one search that handles Chinese and English names, keywords and IDs, with filters for category, collection and style.
Multiple views and depth maps: each custom pose has five views, switched per card without touching the others, and depth maps are there when you want a structural guide for drawing or generating.
Realistic and anime outfits: 1,320 upstream anime outfits plus custom ones, filtered apart from the realistic set, with extended categories behind a switch.
Favorites and sharing: save references to My Favorites, and share links keep the language, filters and chosen references.
One-click generation: choose a pose, a costume and a character reference, connect your own image service and generate a character; cost depends on that service.
Run it locally: the project is open source on GitHub, written in TypeScript, and you can deploy your own copy. Check the licensing notes in the repo before using the assets.
