GitHub radar
1439 photo references of human poses for drawing, 3D and AI art
This is a fully mirrored and translated reference library of human poses — 288 poses, 5 camera angles each. Instead of describing a pose in words, you grab a ready photo and use it as a reference.
A gallery database of 1439 pose cards (2878 photos) with filters by pose type and camera angle, opens right in your browser.
When it helps
Useful if you draw, build 3D models, shoot photos, or generate AI images and video with characters and want a stable pose. Skip it if character poses don't matter to you or you want to avoid explicit content.
Pros
- No GPU needed — runs in a plain browser
- A photo beats a text description for stable results
- Filter by pose type and camera angle, no Chinese needed
- Download the full-size photo with a single click
Cons
- Some poses are explicitly sexual — a niche tool
- Photos stay copyrighted, not cleared for commercial use
How to set it up — step by step
- 1Open your AI agent (Claude Code, Codex) in the project folder where you'll generate the image.
- 2Send the agent the text from the block below.
- 3Describe in words the pose your character needs.
- 4Check the photo the agent found and ask it to use it as a pose reference for generation.
- 5Generate the image or video with that reference and compare it to a text description.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Open the gallery demo at https://ai-pose-gallery.anhao.app/ (repo: https://github.com/anhao/ai-pose-gallery-zh), find a pose that fits what I describe, save the reference photo, and tell me how to feed it as a pose reference when generating an image or video.
Runs on a regular laptop — just a browser, no graphics card needed.
Open on GitHub▌ More finds