Technical introduction
FullOnMobile runs cloud GPU pipelines for human segmentation, keypoint detection, garment warping & texture rendering, plus multimodal outfit recommendation inference. Clients interact via REST APIs or an embeddable mobile H5 SDK.
Base URL (illustrative):
Secure authentication
All computing calls require an API key. Send it in the Authorization header. Rotate keys from the merchant dashboard. Rate limits and anomaly detection reduce malicious batch calling of GPU endpoints.
Try-on API
POST /tryon
Submit a portrait reference and garment asset IDs. Returns a job id; poll or use webhooks for the rendered result URL. Portrait objects support auto-expire deletion.
- Supports tops, bottoms, dresses, shoes, and accessories as garment types expand.
- Async jobs recommended for high-resolution outputs.
Outfit recommendation API
POST /outfit/recommend
Provide body/style signals and an anchor SKU. Returns coordinated sets: tops + bottoms + shoes + jewelry.
H5 SDK integration
Load the SDK script, initialize with your public key and theme, then mount on a product detail container. The SDK opens the try-on sheet, handles camera/upload permissions, and emits commerce events.
For mini-programs, wrap the same flow in a web-view bridge. Offline AI mirrors can deep-link via QR payloads that hydrate outfit state on mobile.
Privacy & anti-abuse
- Clear portrait data usage statement required in merchant UX.
- Uploaded images support timed deletion; avoid long-term face storage by default.
- API keys, per-key quotas, and Cloudflare edge acceleration for global access.
- Webhook signatures verify result callbacks.