How do I authenticate?

The embed widget uses a site key passed as a query param and locked to your allowed hosts. The REST API uses a secret API key sent as a Bearer token. Create both in the developer dashboard.

What's the difference between a site key and an API key?

A site key is public and safe to expose in the browser for the iframe widget — it's restricted to your allow-listed domains. An API key is secret and used for server-to-server REST calls. Never expose an API key client-side.

Yes — official open-source SDKs for Node.js, Python, PHP, Ruby, and Go wrap the API and handle the submit-then-poll flow for you. You can also use a plain iframe embed or call the REST API directly. See the docs for install and quickstart.

What categories are supported?

Clothing plus accessories like eyewear. You pass a category (and optional subcategory) and we route the request to the right try-on model.

How are try-ons processed?

Try-ons run as asynchronous jobs — you start a job and poll for its status, which keeps your requests fast and your integration resilient.

How is API usage priced?

Per try-on, with volume tiers. See the docs for current rates, or talk to sales for higher-volume plans.

For developers

Embed try-on in minutes

Photoreal virtual try-on, ready to drop into your product. Paste an iframe with a site key or call the REST API — no model training, no infra to run.

Get a site key Read the docs

Embed widget · REST API · site-key host control

After: A photoreal virtual try-on result generated through the TryItOn API

Before: A photoreal virtual try-on result generated through the TryItOn API

BeforeAfter

Two lines of integration, one fitting room

Generate a key in the developer dashboard, then drop in an official SDK — Node.js, Python, PHP, Ruby, or Go — or call the API directly.

// npm install tryiton
import { TryItOn } from "tryiton";

const client = new TryItOn({ apiKey: process.env.TRYITON_API_KEY });

const jobId = await client.tryOnClothes({
  modelImage: "https://example.com/model.jpg",
  garmentImage: "https://yourstore.com/product.jpg",
  category: "clothing",
  subcategory: "tops",
});

const [resultUrl] = await client.waitForResult(jobId);

Go to dashboard Read the docs

What the API gives you

The same engine behind our consumer app, exposed as a clean, documented API.

Clothing & accessories

Pass a category and route any garment, eyewear, or accessory through the right try-on model automatically.

Async jobs

Kick off a try-on, poll for status, and pull the result — typically ready in ~5–10 seconds. Built to fit into background queues and webhooks.

Site-key host control

Scope each key to an allow-list of domains so the widget only runs where you ship it.

Before / after output

Get back clean, photoreal result images ready to drop into your own compare UI.

Real bodies, real fits

Everything you need to ship

Official SDKs, keys, docs, and a live widget to copy from.

Official SDKs

Open-source clients for Node.js, Python, PHP, Ruby, and Go — install, authenticate, run a try-on.

Browse the SDKs

API documentation

Endpoints, parameters, and auth — everything to integrate try-on into your stack.

Open the docs

Developer dashboard

Create API keys and site keys, set allowed hosts, and watch your usage.

Open the dashboard

Designing a try-on integration that feels native

The widget and API snippets above get a result on the screen. Shipping something production-grade is about the shape around them — how a request flows, how you handle an asynchronous result, what images you feed in, and how the wait reads to a user.

A typical request flow

Most integrations follow the same arc, whether you embed or call the API directly.

Resolve the inputs: Take the shopper's photo and the product's garment image URL for the page they're on.
Kick off the job: Embed the widget, or POST to the API and get back a job you can track.
Render the result: Show the finished image in your own before/after view once the job completes.

Handling asynchronous results

Because try-ons run as jobs, plan for the gap between request and result rather than blocking on it. A render is typically ready in ~5–10 seconds, but treat that as a guide, not a guarantee — poll for completion instead of assuming a fixed delay.

Poll with backoff: Check job status on an interval that widens over time instead of hammering the endpoint.
Persist by job id: Key results to the job so a refresh or a returning session can pick the look back up.
Cache the output: Store finished images so a repeat view of the same garment doesn't pay for a second render.

Input requirements that affect quality

Garbage in shows up in the render — a few input rules keep results sharp.

Reachable garment images: Pass a stable, publicly fetchable product image URL, not a transient or auth-gated one.
Clean product shots: A clear image of the item on a plain background routes and renders best.
Right category: Send the category that matches the item so the request hits the correct try-on model.

Designing the wait

A try-on takes a moment, so treat the loading state as part of the feature: show progress rather than a frozen button, keep the original product image in view as the “before,” and reveal the result with a compare slider so the shopper can flip between the two. The widget already does this — match it if you build your own UI.

Frequently asked questions

Trying something new can raise a few questions — here are the ones we hear most.

The embed widget uses a site key passed as a query param and locked to your allowed hosts. The REST API uses a secret API key sent as a Bearer token. Create both in the developer dashboard.
A site key is public and safe to expose in the browser for the iframe widget — it's restricted to your allow-listed domains. An API key is secret and used for server-to-server REST calls. Never expose an API key client-side.
Yes — official open-source SDKs for Node.js, Python, PHP, Ruby, and Go wrap the API and handle the submit-then-poll flow for you. You can also use a plain iframe embed or call the REST API directly. See the docs for install and quickstart.
Clothing plus accessories like eyewear. You pass a category (and optional subcategory) and we route the request to the right try-on model.
Try-ons run as asynchronous jobs — you start a job and poll for its status, which keeps your requests fast and your integration resilient.
Per try-on, with volume tiers. See the docs for current rates, or talk to sales for higher-volume plans.

Build with TryItOn

Generate a key, paste the snippet, and ship a fitting room today.

Get a site key Read the docs

No model training, no GPUs to run — just an API call.