GlamAR's SDKs and APIs are designed to deliver high-quality AR and AI product experiences across platforms. To ensure consistent performance, please follow the requirements below across device, browser, and network layers.
| Platform | Supported Environments | Performance Notes |
|---|
| Web SDK | Chrome (desktop/mobile), Safari (iOS 11+), Firefox | Hardware-dependent performance |
| Android SDK | Android 10+ (API 29+) | Optimal performance on Android 11+ |
| iOS SDK | iOS 13+ | Best on iPhone XR/11+ |
| React Native | Android & iOS (follows native SDK specs) | Bridge overhead applies |
| Flutter | Android & iOS | Dart VM performance considerations |
| Smart Mirror | Android-powered only (see detailed specs below) | Requires mid-range+ hardware |
| Tablets | Android tablets fully supported; iPads Web only | Landscape orientation recommended |
❗ In-app browsers (Instagram, Facebook WebView, etc.) are not supported.
Browser Compatibility (Web SDK Only)
| Browser | Android | iOS | Desktop | Expected FPS Range |
|---|
| Chrome | 66+ | N/A | 66+ | 8-28 FPS (hardware-dependent) |
| Safari | N/A | 11+ | 11.1+ | 6-24 FPS (WebKit limitations) |
| Firefox | 60+ | N/A | 60+ | 8-25 FPS (WebGL performance) |
| Edge | N/A | N/A | 88+ | 8-28 FPS (Chromium-based) |
📌 Performance Notes:
- Chrome on iOS uses Safari's WebKit engine — performance mirrors Safari
- Entry-level devices: 6-12 FPS across all browsers
- Mid-range devices: 14-25 FPS with Chrome/Edge performing best
- Flagship devices: 24-28+ FPS with optimal WebGL support
Device Compatibility
| Hardware Tier | CPU Examples | Expected AR VTO Performance | Recommended Use Cases |
|---|
| Entry-Level | Snapdragon 4xx, Helio P22/P35 | 10-15 FPS | Basic try-on experiences |
| Mid-Range | Snapdragon 6xx/7xx, Dimensity 700-900 | 20-28 FPS | Full-featured AR VTO |
| Flagship | Snapdragon 8 Gen+, Apple A14+ | 30+ FPS | Premium AR experiences |
Tablet Support
| OS | Support Level | Performance Tier | Notes |
|---|
| Android | ✅ Full Support | Mid-Range to Flagship | Works with native SDK or Chrome |
| iPadOS | ⚠️ Limited Support | Mid-Range | Safari camera APIs, Web SDK only |
| Device Model | Hardware Tier | Expected FPS | Details |
|---|
| iPad Pro (2021+) | Flagship | 26-30 FPS | 12MP front camera + Apple Neural Engine |
| Samsung Galaxy Tab S8/S9 | Flagship | 28-30 FPS | Excellent selfie cam + strong processors |
| Lenovo P12 Pro / Yoga Tab | Mid-Range | 18-24 FPS | Mid-range Android options with decent optics |
Optimization Tips: Ensure adequate lighting and stable device positioning for best detection performance.
Smart Mirror Support
| Component | Recommendation |
|---|
| Camera | Logitech Brio 4K / Intel RealSense D415 |
| CPU | Intel i5/i7 10th Gen+, or macOS M1/M2 chip |
| RAM | Minimum 8 GB |
| GPU | Not required unless running on-device ML |
| Display | ≥15", Full HD, touch-enabled preferred |
🛠️ Our Android SDK can run on most kiosk builds with mid-range+ hardware specifications.
Camera Requirements
Specs
| Use Case | Minimum Spec | Notes |
|---|
| Smart Mirrors | 720p+, fixed, high-lumen front camera | Must support low-latency capture |
| Web SDK (General) | 480p+, browser-permissioned camera | Avoid over/underexposed lighting |
Permissions
| Platform | Permissions |
|---|
| Web SDK | Browser camera access |
| Android SDK | android.permission.CAMERA |
| iOS SDK | NSCameraUsageDescription in Info.plist |
| React Native | Permissions via react-native-permissions |
| Flutter | Permissions via permission_handler plugin |
While GlamAR does not require motion-tracking or depth-sensing cameras for most use cases, they can significantly enhance accuracy in professional or in-store environments — particularly for PD measurement, mirror-based experiences, or high-precision facial tracking.
If you're deploying GlamAR in a smart mirror, full-length in-store setup, or for exact sizing use cases, we recommend installing a motion-tracking or depth camera to improve detection stability and accuracy.
3D Model Specifications
| Field | Recommended Value |
|---|
| Format | .glb, .gltf (preferred), .fbx, .usdz |
| Max Size | 10 MB |
| Texture Resolution | 2048 x 2048 (2K) |
| Hosting | PixelBin-hosted or imported via secure URL |
Network Requirement
| Metric | Minimum | Recommended |
|---|
| Downlink | 2 Mbps | 3–5 Mbps |
| Latency | < 300 ms | < 150 ms |
| CDN Support | Yes | ✅ (PixelBin CDN) |
- Models are fetched via PixelBin and cached per session
- Works under 4G+, Wi-Fi, and 5G conditions
Our AR VTO SDK utilizes advanced face detection and 478-point 3D facial landmark tracking models that require significant computational resources. Performance is categorized by hardware capabilities:
Entry-Level Hardware
| Component | Specification | Performance Expectation |
|---|
| CPU | ARM Cortex-A53/A55, Snapdragon 4xx series, MediaTek Helio P22/P35 | 10-15 FPS AR VTO |
| GPU | Adreno 505/506, PowerVR GE8300/8320, Mali-G52 MP1/MP2 | Basic AR experience |
| RAM | 3-4 GB | Adequate for standard features |
| Android | Android 10-12 (API 29-31) | Optimizations planned for future release |
| iOS | iPhone 7/8, iPad (6th gen) | Limited AR performance |
Mid-Range Hardware
| Component | Specification | Performance Expectation |
|---|
| CPU | ARM Cortex-A76/A78, Snapdragon 6xx/7xx, MediaTek Dimensity 700-900 | 20-28 FPS AR VTO |
| GPU | Adreno 618/640, Mali-G76 MP4/MP12, PowerVR GM9446 | Smooth AR experience |
| RAM | 6-8 GB | Optimal for all features |
| Android | Android 11-14 (API 30-34) | Full feature support |
| iOS | iPhone XR/11/12, iPad Air (3rd gen+) | Excellent performance |
Flagship Hardware
| Component | Specification | Performance Expectation |
|---|
| CPU | ARM Cortex-X1/X2, Snapdragon 8 Gen 1+, Apple A14 Bionic+ | 30+ FPS AR VTO |
| GPU | Adreno 730+, Mali-G710/G715, Apple GPU (A14+) | Premium AR experience |
| RAM | 8+ GB | Maximum performance |
| Android | Android 14+ (API 34+) | Latest optimizations |
| iOS | iPhone 13+, iPad Pro (M1/M2) | Peak performance |
Face Detection & Landmark Tracking
| Hardware Tier | Face Detection FPS | Landmark Tracking FPS | Overall AR VTO FPS |
|---|
| Entry-Level | 15-20 | 12-15 | 10-15 |
| Mid-Range | 25-30 | 20-25 | 20-28 |
| Flagship | 30+ | 30+ | 30+ |
| Platform | Entry-Level | Mid-Range | Flagship | Notes |
|---|
| Web SDK (Chrome) | 8-12 FPS | 18-25 FPS | 28+ FPS | WebGL performance dependent |
| Web SDK (Safari) | 6-10 FPS | 14-20 FPS | 24+ FPS | WebKit limitations apply |
| Android Native | 12-15 FPS | 22-28 FPS | 30+ FPS | Optimal performance |
| iOS Native | 10-14 FPS | 20-26 FPS | 30+ FPS | Metal rendering advantage |
| React Native | 10-13 FPS | 18-24 FPS | 26-30 FPS | Bridge overhead |
| Flutter | 11-14 FPS | 19-25 FPS | 28-30 FPS | Dart VM performance |
Model Processing Benchmarks
Face Detection Model
| Hardware Tier | Input Resolution | Processing Time | Memory Usage |
|---|
| Entry-Level | 192x192 | 80-120ms | 150-200 MB |
| Mid-Range | 192x192 | 40-60ms | 200-250 MB |
| Flagship | 192x192 | 20-35ms | 250-300 MB |
3D Landmark Model
| Hardware Tier | Input Resolution | Processing Time | Memory Usage |
|---|
| Entry-Level | 256x256 | 100-150ms | 200-300 MB |
| Mid-Range | 256x256 | 50-80ms | 300-400 MB |
| Flagship | 256x256 | 25-45ms | 400-500 MB |
| Scenario | Entry-Level | Mid-Range | Flagship | Notes |
|---|
| First-time Load | 4-8 seconds | 2-4 seconds | 1-2 seconds | Includes fetch + compile |
| Cached Model Load | 1.5-2.5s | 800ms-1.2s | 400-700ms | Browser/app caching |
| Optimized GLB | 3-6 seconds | 1.5-3s | 800ms-1.8s | Compressed models |
Entry-Level Hardware Optimizations
🚧 Coming Soon: The following SDK-level optimizations are planned for future releases to enhance performance on entry-level devices:
- Automatic frame skipping (process every 2nd frame) - Planned
- Dynamic camera resolution (reduce to 480p on low-end devices) - Planned
- Progressive model loading (load models in stages) - Planned
- Simplified AR overlays (reduced complexity rendering) - Planned
- Performance mode detection (auto-detect device capabilities) - Planned
⚠️ Current Limitation: These optimizations require SDK-level implementation and cannot be achieved through application-layer modifications. Entry-level device performance will be significantly improved once these features are released.
Mid-Range Hardware Optimizations
- Standard 720p camera resolution
- Full feature set enabled
- Adaptive quality based on thermal throttling
- Preload frequently used models
Flagship Hardware Optimizations
- Maximum 1080p camera resolution
- Enhanced AR effects and animations
- Multi-model parallel processing
- Advanced rendering techniques
Memory Requirements by Hardware Tier
| Hardware Tier | Minimum RAM | Recommended RAM | Peak Memory Usage |
|---|
| Entry-Level | 3 GB | 4 GB | 600-800 MB |
| Mid-Range | 4 GB | 6 GB | 800-1.2 GB |
| Flagship | 6 GB | 8+ GB | 1.2-1.8 GB |
| Connection Type | Model Download | Real-time Performance | Recommended Tier |
|---|
| 3G/Slow 4G | 15-30 seconds | Stuttering possible | Entry-Level+ |
| 4G/LTE | 3-8 seconds | Smooth operation | Mid-Range+ |
| 5G/WiFi | 1-3 seconds | Optimal performance | All tiers |
✅ Built-in SDK Optimizations:
Our SDK automatically includes the following performance enhancements:
- Intelligent Model Caching: Models are cached after first download to eliminate repeated network requests
- CDN-Optimized Delivery: PixelBin CDN reduces initial load times with global edge distribution
- Progressive Loading: Models load incrementally to improve perceived performance and reduce wait times
- Thermal Management: Automatic performance scaling prevents device overheating and maintains stability
⚠️ Note: Entry-level hardware optimizations are currently in development at the SDK level. These performance improvements cannot be implemented at the application layer and require core SDK enhancements.