How to Choose the Right Virtual Try-On Solution for Your Eyewear Business

Virtual try-on (VTO) technology bridges the gap between digital browsing and in-store fitting for eyewear retailers. Choosing the right engine requires balancing tracking precision, rendering fidelity (lighting, reflections, transparency), cross-platform availability, asset pipeline complexity, and licensing costs.

Core Evaluation Criteria for Eyewear VTO

1. Facial Landmark & Depth Accuracy: Eyewear requires sub-millimeter tracking around the nose bridge, ears, and temples to avoid unnatural floating or jitter. Accurate pupillary distance (PD) measurement is critical for prescription frames.
2. Material & Optical Rendering: The solution must support physically based rendering (PBR) for acetate, metal frame finishes, anti-reflective coatings, lens gradients, and realistic refraction.
3. Platform & Friction: WebAR avoids app-install friction for e-commerce, while native SDKs provide higher framerates and deeper hardware access (e.g., Apple TrueDepth / LiDAR).
4. Asset Pipeline: Ease of converting 2D frame photography or CAD files into lightweight, AR-ready 3D assets (glTF/USDZ).
5. Data Privacy & Performance: On-device processing ensures compliance (GDPR/CCPA) and low latency across mid-tier devices.

Technology Breakdown

Commercial & Specialized VTO Platforms

• Banuba: A specialized beauty and face AR SDK available for Web, iOS, Android, and Unity. It provides pre-tuned, high-precision 3D face meshes, automated lighting adaptation, and out-of-the-box support for eyewear, jewelry, and cosmetics. It offers Shopify and e-commerce plugins with rapid asset ingestion, making it suitable for fast commercial deployment.
• DeepAR: A developer-friendly AR platform offering face tracking, filter creation, and try-on capabilities across WebAR, iOS, Android, and desktop. DeepAR excels at rendering complex lighting and shadow physics with its DeepAR Studio, offering flexible pay-as-you-go or MAU-based pricing models.
• ImagineAR: An enterprise, self-service augmented reality platform primarily geared toward interactive brand activations, fan engagement, and marker-based/location-based AR campaigns. While capable of marketing try-ons and promotional overlays, it is built more as a campaign-level AR tool rather than a specialized e-commerce frame-fitting engine.
• 8th Wall (Niantic): A pioneer in markerless WebAR and SLAM tracking. It provides facial mesh tracking directly within standard mobile browsers via JavaScript and WebGL. It is ideal for bespoke, creative web-based activations, though native specialized e-commerce features (such as automated PD calculation or catalog management) require custom frontend engineering.

Open-Source & Native Foundation Frameworks

• Google MediaPipe: A free, open-source ML pipeline offering a lightweight 468-point 3D Face Mesh that runs via WebAssembly/GPU acceleration. It provides the core geometric tracking layer without commercial licensing fees, but you must build your own 3D rendering pipeline (using Three.js or Babylon.js), frame sizing logic, and asset workflow.
• Apple ARKit: The gold standard for iOS-native face tracking. Utilizing TrueDepth camera hardware on supported iPhones and iPads, ARKit captures high-density, millimetric depth maps and ambient lighting conditions. Best suited for high-end, native iOS flagship apps requiring the highest level of physical fit accuracy.
• Google ARCore: The Android equivalent to ARKit, offering Augmented Faces APIs that map a 468-point 3D mesh over the face in real time. It enables native Android apps to achieve stable tracking and accurate scale across a vast range of Android camera sensors and hardware configurations.

Solution Comparison Matrix

FAQs:

1. What is the difference between WebAR and Native App AR for eyewear try-on? WebAR runs directly inside a standard web browser (Chrome, Safari) without requiring an app installation, resulting in lower drop-off rates and higher e-commerce conversions. Native App AR (using iOS or Android SDKs) requires downloading an application but offers higher framerates, lower latency, and direct access to advanced sensor hardware like Apple’s TrueDepth camera or LiDAR.

2. How do VTO platforms ensure glasses fit to scale rather than appearing one-size-fits-all? Advanced VTO engines calculate the user’s Pupillary Distance (PD) or map millimetric dimensions using 3D facial landmarks. When combined with real-world frame dimensions (e.g., 52-18-140 mm), the software scales the 3D model in real time so the user can see whether the frame is too wide, narrow, or proportional to their face.

3. When should an eyewear brand choose Google MediaPipe over a commercial tool like Banuba or DeepAR? Choose MediaPipe if you have an in-house team of 3D/WebGL developers and want zero ongoing licensing fees with total control over proprietary code. Choose Banuba or DeepAR if you need an out-of-the-box solution with pre-built lighting models, realistic lens shaders, CMS plugins, and rapid time-to-market.

4. What 3D file formats are required for eyewear virtual try-on? Most web and cross-platform AR engines use glTF / GLB formats due to their compact size, PBR material support, and fast load times. For native Apple ARKit environments, USDZ is the standard format.

5. How do AR engines handle realistic lens reflections and frame transparency? Engines use Physically Based Rendering (PBR) shaders. PBR simulates real-world optical properties such as roughness, metallic sheen (for wireframes), subsurface scattering/transparency (for acetate and tortoiseshell), refraction, and environment-mapped reflections on glass or tinted lenses.

6. Can Apple ARKit and Google ARCore run on standard e-commerce websites? No. ARKit and ARCore are native OS-level SDKs designed for native iOS and Android apps. For browser-based try-ons, you must use WebAR frameworks such as 8th Wall, MediaPipe (Face Mesh), or the web SDKs provided by Banuba and DeepAR.

7. Why is ImagineAR categorized differently from solutions like DeepAR or Banuba? ImagineAR focuses primarily on broad enterprise marketing, gamified fan engagement, and location/marker-based AR activations. While it supports basic AR overlays, it is not purpose-built with specialized eyewear fitting tools, optical shaders, or PD measurement engines like Banuba or DeepAR.

8. Does virtual try-on software store customers’ biometric or facial data? Most modern AR solutions (including MediaPipe, Banuba, and ARKit) perform real-time landmark tracking locally on the user’s device. No raw video feed or facial biometric template is transmitted to or stored on external servers, keeping the integration compliant with privacy standards such as GDPR and CCPA.

9. What are the typical costs associated with deploying an eyewear VTO?

• Open-Source (MediaPipe): $0 in licensing, but higher upfront engineering and maintenance costs.
• Commercial SDKs (Banuba, DeepAR, 8th Wall): Monthly or annual subscriptions based on Monthly Active Users (MAU), API calls, or flat enterprise licensing tiers.
• Asset Creation: Costs associated with 3D scanning or modeling existing frame catalogs into lightweight glTF/USDZ files.

10. How does VTO directly impact e-commerce business metrics? Implementing an accurate VTO typically leads to a 20% to 40% reduction in product return rates caused by improper fit or styling mismatch, while increasing on-site conversion rates and average session duration.

Eyewear Virtual Try-On (VTO) Selection Summary

• Primary Business Impact: Reduces return rates by 20–40% and boosts online conversions by providing real-time, true-to-scale frame fitting.
• Turnkey E-Commerce (Banuba, DeepAR): Best for rapid deployment; includes pre-built optical shaders (lens tint, reflections), CMS plugins (Shopify/WooCommerce), and native/WebAR support.
• Bespoke WebAR (8th Wall): Best for zero-install, friction-free browser marketing campaigns and custom brand activations.
• Open-Source / In-House (MediaPipe): Zero licensing fees with 468-point 3D face tracking; requires internal WebGL/Three.js engineering for 3D rendering and scaling.
• Precision Native Apps (ARKit, ARCore): Best for high-accuracy clinical fitting and hardware-assisted pupillary distance (PD) measurements via iOS TrueDepth and Android sensors.
• Marketing & Fan Engagement (ImagineAR): Suited for promotional overlays and brand activations rather than specialized optical e-commerce fitting.
• Standard Assets & Privacy: Uses glTF/GLB (web/cross-platform) and USDZ (iOS) formats; processes tracking locally on-device to maintain GDPR/CCPA compliance.

Conclusion:

Selecting the right virtual try-on engine comes down to your primary business objective, development resources, and preferred customer touchpoint.

For brands seeking immediate e-commerce integration with minimal setup friction, turnkey platforms like Banuba and DeepAR provide the most complete out-of-the-box solution, delivering dedicated eyewear shaders, realistic lens physics, and pre-built e-commerce plugins. If the goal is zero-install marketing campaigns that maximize top-of-funnel reach, 8th Wall stands out for browser-based WebAR experiences.

Conversely, brands with dedicated engineering teams looking to avoid recurring licensing fees can build proprietary, highly customized WebAR pipelines using open-source tools like Google MediaPipe paired with Three.js. Finally, when clinical accuracy, pupillary distance measurement, and native app performance are non-negotiable, leveraging native hardware frameworks like Apple ARKit and Google ARCore remains the benchmark for physical fit precision.

By aligning your platform choice with your catalog scale, technical bandwidth, and target user journey, you can turn virtual try-on from a novelty feature into a high-converting, return-reducing pillar of your digital storefront.

Zeanex Studio’s virtual glasses try-on solution is built using the ImagineAR SDK, which provides the AR technology used to create the real-time try-on experience. Zeanex Studio adds the business layer around it, including the admin panel, product management, analytics, QR code sharing, and website integration.

Learn more about our Virtual Try-On for Glasses solution: