ZazimFind
API Reference

Step-by-step guides for integrating ZazimFind's AR engine. Call our REST API directly or use one of our SDKs to build high-fidelity 3D experiences.

Authentication & Quickstart

Sign up at console.zazimfind.com to obtain your API key. Then make your first request — upload at least three images and receive a 3D model in seconds.

1. Submit an image

curl -X POST https://api.zazimfind.com/v1/render3d \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "image=@product.jpg" \
  -F "output_format=glb"

2. Check processing status

curl https://api.zazimfind.com/v1/status/req_abc123 \
  -H "Authorization: Bearer YOUR_API_KEY"

3. Completed response

{
  "status": "completed",
  "model_url": "https://cdn.zazimfind.com/models/abc123.glb",
  "thumbnail_url": "https://cdn.zazimfind.com/thumbs/abc123.png",
  "metrics": {
    "chamfer_distance": 0.32,
    "psnr": 28.4
  }
}

Code Samples

Production-ready samples for the most popular environments. All samples assume your API key is stored in an environment variable.

Python

import zazhree_client

client = zazhree_client.ZazhreeClient(api_key="YOUR_API_KEY")

with open("product.jpg", "rb") as f:
    result = client.render_3d(image=f, output_format="glb")

print(result.model_url)  # https://cdn.zazimfind.com/models/...

JavaScript (Node.js)

import { ZazhreeClient } from '@zazimfind/api';
import { createReadStream } from 'fs';

const client = new ZazhreeClient({ apiKey: process.env.ZAZIMFIND_KEY });

const result = await client.render3d({
  image: createReadStream('product.jpg'),
  outputFormat: 'glb',
  webhookUrl: 'https://your.domain/webhooks/zazimfind',
});

console.log(result.modelUrl);

React Component

import { ZazhreeViewer } from '@zazimfind/web-sdk';

function ProductPage() {
  return (
    <ZazhreeViewer
      modelUrl="https://cdn.zazimfind.com/models/abc123.glb"
      arMode="webxr"
      enableRotation={true}
    />
  );
}

FAQs

Endpoints

Base URL: https://api.zazimfind.com/v1 — all requests require a Bearer token in the Authorization header.

Webhooks

Configure a webhook_url in your render request. ZazimFind POSTs a HMAC-SHA256 signed payload to your endpoint on completion. Always verify the X-Zazimfind-Signature header before processing.

{
  "event": "render.completed",
  "request_id": "req_abc123",
  "model_url": "https://cdn.zazimfind.com/models/abc123.glb",
  "thumbnail_url": "https://cdn.zazimfind.com/thumbs/abc123.png",
  "metadata": { "your_custom_id": "value" }
}

Rate Limits

Web SDK

Install via npm for programmatic control over upload, model generation, and rendering. TypeScript definitions included; supports React, Vue, and Angular wrappers.

npm install @zazimfind/web-sdk

Mobile SDKs

Unity & XR

Import ZazimFindUnity.unitypackage from the Developer Portal. Use the ZazhreeModelImporter component to download and display glTF models at runtime with AR Foundation plane detection and XR Interaction Toolkit support.

using ZazimFind.Unity;

public class ProductLoader : MonoBehaviour {
    async void Start() {
        var model = await ZazhreeClient.Render3dAsync(imageBytes);
        var go = ZazhreeModelImporter.Import(model.url);
        go.transform.position = new Vector3(0,0,2);
    }
}

Model Card — Zazhree v1.2

We publish model cards for every core algorithm as part of our commitment to transparent, responsible AI. Each card includes performance benchmarks, known limitations, and ethical considerations.

Data Retention & Privacy

Images are retained only for the duration of processing (max 24 hours) unless the user explicitly opts in to the training dataset via a separate consent toggle.

• Opted-in images are anonymised (EXIF stripped, random ID applied) and stored for up to 1 year to improve model accuracy.
• Users can withdraw consent at any time; data is removed from active training within 30 days.
• ZazimFind never sells or rents data to third parties.
• All processing occurs on AWS London (eu-west-2) — UK sovereign cloud.

Accessibility & Localisation

This documentation is WCAG 2.1 AA compliant — keyboard-navigable with visible focus indicators, descriptive alt text on all images, syntax-highlighted code blocks with screen-reader announcements, and support for high-contrast and reduced-motion preferences.

Planned locales (Q2 2027): en-GB, fr, de, zh. Localised content includes API error messages, date formats, and currency examples. User locale is auto-detected via Accept-Language; a manual switcher is available in the footer.