This glossary gives plain-English definitions of the essential 3D and augmented reality terms for e-commerce: file formats (GLB, USDZ), AR viewers (AR Quick Look, Scene Viewer, WebXR), creation techniques (photogrammetry, AI), rendering concepts (PBR, textures, polygons) and sharing methods (QR code, iframe, API).
GLB
GLB is the binary version of the glTF format: a single file that bundles a 3D model's geometry, materials, textures and animations.
It is the reference format for 3D on the web and on Android (Scene Viewer, model-viewer). Its extension is .glb and its MIME type is model/gltf-binary.
In Visuality: GLB is the platform's core format: every object is stored as a GLB, whether it was uploaded as is, converted from a USD/USDZ file or generated by AI.
Learn more: GLB or USDZ: which format should you choose? · Prepare your 3D model
glTF
glTF (GL Transmission Format) is an open, royalty-free format from the Khronos Group, designed for the efficient transmission and loading of 3D scenes and models.
Version 2.0, released in 2017, introduced PBR materials; it is also standardised as ISO/IEC 12113:2022. A glTF asset comes either as a .gltf file (JSON plus separate binary and image files) or as a self-contained .glb.
In Visuality: Visuality imports the binary .glb variant, which is easier to handle than a .gltf with separate files.
Learn more: GLB or USDZ: which format should you choose?
USDZ
USDZ is an uncompressed zip archive containing a USD scene and its textures; it is the 3D format read by AR Quick Look on iPhone and iPad.
Introduced by Apple and Pixar in 2018, it fits in a single, easy-to-share file. It opens in augmented reality straight from Safari, Messages or Mail.
In Visuality: you can upload a .usdz file (it is converted to GLB on upload), and the platform automatically generates on its servers the USDZ file served to iPhones and iPads, at the right scale.
Learn more: GLB or USDZ: which format should you choose?
USD (OpenUSD)
USD (Universal Scene Description) is an open-source format and ecosystem created by Pixar to describe, compose and exchange complex 3D scenes.
It comes in text (.usda), binary (.usdc) or packaged (.usdz) form. Since 2023, the Alliance for OpenUSD, founded by Pixar, Adobe, Apple, Autodesk and NVIDIA, has been steering its development.
In Visuality: .usd, .usda, .usdc and .usdz files are accepted on upload and automatically converted to GLB.
Learn more: GLB or USDZ: which format should you choose?
OBJ
OBJ is an old and very widespread text format that describes a 3D model's geometry; its materials are described in a separate .mtl file.
It supports neither animations nor modern PBR materials, and its textures remain separate files. For the web and augmented reality, it is usually converted to GLB.
Learn more: GLB or USDZ: which format should you choose?
FBX
FBX is a proprietary Autodesk format, widely used in video games and animation to exchange models, rigs and animations between 3D applications.
Feature-rich but closed, it cannot be read by web browsers: it is exported to GLB, for example with Blender, before being published on the web or in augmented reality.
Learn more: GLB or USDZ: which format should you choose?
Augmented reality viewers and platforms
AR Quick Look
AR Quick Look is the augmented reality viewer built into iOS, iPadOS and visionOS: it displays a USDZ or Reality file in 3D and AR, with no app to install.
It opens from Safari, Messages, Mail or Notes and runs on ARKit. On a website, a simple link to a USDZ file or a viewer such as model-viewer is enough to launch it.
In Visuality: the augmented reality button opens AR Quick Look on iPhone and iPad, with a USDZ file generated automatically from your GLB.
Learn more: GLB or USDZ: which format should you choose?
Scene Viewer
Scene Viewer is Google's 3D and AR viewer for Android: it displays a glTF/GLB model in augmented reality from a browser or an app.
It runs on ARCore (Google Play Services for AR). Google recommends at most 100,000 triangles, textures no larger than 2048 × 2048 pixels and a file of around 10 MB.
In Visuality: on Android, the viewer launches augmented reality through WebXR in Chrome or, failing that, through Scene Viewer.
Learn more: WebAR vs mobile app: how to choose?
WebAR
WebAR is augmented reality launched from a web browser, via a link or a QR code, with no app to download.
Depending on the device, it relies on WebXR, Scene Viewer or AR Quick Look. Its main strength is that it is frictionless; a native app is still preferable for offline use or very advanced features.
In Visuality: Visuality is a WebAR solution: your customers open augmented reality from a link, a QR code or a page on your website.
Learn more: WebAR vs mobile app: how to choose?
WebXR
The WebXR Device API is a W3C specification that lets web pages access headsets and smartphones to deliver virtual and augmented reality experiences.
Its augmented reality mode is used by Chrome on ARCore-compatible Android devices. On iPhone and iPad, web-based augmented reality goes through AR Quick Look.
Learn more: WebAR vs mobile app: how to choose?
ARKit
ARKit is Apple's augmented reality framework for iOS and iPadOS, launched in 2017: it handles motion tracking, surface detection and light estimation.
It powers AR Quick Look and native AR apps. On devices with a LiDAR Scanner, surface detection is faster and more accurate.
Learn more: What is augmented reality?
ARCore
ARCore is Google's augmented reality platform, at version 1.0 since 2018: it provides motion tracking, environmental understanding and light estimation.
On Android, it is distributed through Google Play Services for AR and only works on a list of certified devices. It powers Scene Viewer and WebXR in Chrome.
Learn more: What is augmented reality?
model-viewer
model-viewer is an open-source web component maintained by Google that displays an interactive glTF/GLB model on a web page and launches augmented reality on mobile.
A single HTML tag handles rotation, zoom, animations and the choice of AR mode: WebXR, Scene Viewer or AR Quick Look.
In Visuality: Visuality's web viewer is built on model-viewer.
Learn more: 3D viewer for websites
3D viewer
A 3D viewer is a module that displays an interactive 3D model on a web page: users can rotate it, zoom in and launch augmented reality.
It replaces or complements the photos on a product page. Its file size and loading speed matter as much as the quality of the model.
In Visuality: every object has a viewer you can share by public link, QR code or iframe, with unlimited views.
Learn more: 3D viewer for websites
3D model creation
Photogrammetry
Photogrammetry rebuilds a textured 3D model from dozens of photos of an object taken from every angle.
Mobile apps such as RealityScan or Polycam automate the processing. Results are very accurate for matte, textured objects, less so for shiny, transparent or plain surfaces.
In Visuality: a guide explains how to capture with RealityScan, whose ZIP export can be uploaded straight into your object library.
Learn more: Create a 3D model without 3D modeling
LiDAR scanning
LiDAR scanning measures the distance to surfaces with a laser sensor, found on recent iPhone and iPad Pro models, to quickly capture the geometry of an object or a room.
It excels with large volumes and interiors; for small, detailed objects, photogrammetry often produces sharper textures. Many apps combine both techniques.
Learn more: How to create a 3D model for augmented reality
AI 3D generation (image-to-3D)
AI 3D generation creates a textured 3D model from one or a few photos, using an artificial intelligence model that infers the parts that are not visible.
It is fast and requires no technical skills, but it has to invent the sides that were not photographed: accuracy depends on the quality of the photos and the shape of the object.
In Visuality: generation is started from the library with 1 photo (simple mode) or 4 views (detailed mode), for 1 credit; every plan includes a monthly credit allowance.
Learn more: AI and 3D: transform a photo into a 3D model · Generate a 3D model with AI
Gaussian splatting
Gaussian splatting represents a scene with millions of small, coloured, semi-transparent 3D blobs instead of polygons, for photorealistic real-time rendering.
Popularised in 2023 by researchers at Inria, the technique renders reflections and complex materials very well, but it produces heavy files and is poorly supported by mesh-based AR viewers (GLB, USDZ).
Learn more: How to create a 3D model for augmented reality
Retopology
Retopology means rebuilding a clean, lightweight mesh on top of a dense model (scan, sculpt, AI output), with well-organised polygons.
It makes UV unwrapping, animation and real-time display easier. The fine details of the original model are then baked into a normal map.
Learn more: How to create a 3D model for augmented reality
Decimation
Decimation automatically reduces the polygon count of a mesh by merging vertices, making a model lighter while preserving its silhouette.
Faster than retopology, it can damage sharp edges and UVs if it is too aggressive. It is the standard way to make a scan or an AI model with millions of triangles viewable on mobile.
In Visuality: when the USDZ file for iPhone is generated, very heavy models are automatically decimated to about 200,000 triangles.
Learn more: Prepare your 3D model
3D rendering and optimisation
Mesh
The mesh is the surface of a 3D model, made of vertices connected into polygons, most often triangles.
Its density determines both the accuracy of the shapes and the size of the file.
Learn more: How to create a 3D model for augmented reality
Polygon and triangle
A polygon is a flat face of the mesh; for real-time display, every face is split into triangles, the usual unit for measuring a model's complexity.
The triangle count directly affects display smoothness and the memory used by the smartphone.
In Visuality: we recommend staying under 100,000 polygons per object, especially if several objects are displayed together.
Learn more: Prepare your 3D model
Texture
A texture is an image applied to the surface of a 3D model to give it its colour, apparent relief or material properties.
Resolution weighs heavily on memory: a 4096 × 4096 pixel texture takes up four times as much memory as a 2048 × 2048 pixel one.
In Visuality: textures in the USDZ file generated for iPhone and iPad are capped at 2048 pixels.
Learn more: Prepare your 3D model
UV (UV unwrapping)
UV unwrapping flattens the surface of a 3D model to define how each pixel of a texture maps onto it, like a sewing pattern.
Poorly unwrapped UVs cause stretched textures or visible seams. Models from scans or AI often have highly fragmented UVs.
Learn more: How to create a 3D model for augmented reality
PBR (physically based rendering)
PBR (Physically Based Rendering) simulates how light really behaves, based on material parameters: base colour, roughness and metalness.
glTF 2.0 uses the metallic-roughness PBR model, which ensures consistent rendering from one viewer to another and under the room's lighting in augmented reality.
Learn more: How to create a 3D model for augmented reality
Normal map
A normal map is a texture that simulates fine relief (seams, grain, engravings) by changing how light hits the surface, without adding polygons.
It makes a lightweight model look highly detailed: this is the technique used to keep a scan's detail after decimation or retopology.
Learn more: How to create a 3D model for augmented reality
LOD (level of detail)
LOD (Level of Detail) means providing several increasingly simplified versions of a model, displayed depending on distance or device performance.
Widely used in video games, it is rarely needed for a single product viewed up close, but useful for scenes containing many objects.
Learn more: Prepare your 3D model
Draco compression
Draco is an open-source library from Google that compresses 3D mesh geometry; glTF supports it through the KHR_draco_mesh_compression extension.
It greatly reduces the size of the geometry, but not of the textures, and adds decompression time when loading. model-viewer can read Draco-compressed GLB files.
Learn more: GLB or USDZ: which format should you choose?
HDR lighting and IBL
IBL (Image-Based Lighting) lights a 3D model using a panoramic HDR image of the environment, producing realistic reflections and shading.
Web viewers use a default environment or a custom HDR image. In augmented reality, some platforms estimate the room's real lighting and apply it to the model.
In Visuality: an option lets you adapt the object's lighting to the real environment in augmented reality.
Learn more: Background image customization
The augmented reality experience
Augmented reality (AR)
Augmented reality overlays digital content, such as a 3D object, in real time onto the view of the real world captured by a smartphone's or headset's camera.
Unlike virtual reality, the real environment stays visible: you can see a piece of furniture at its actual size in your living room before buying it.
Learn more: What is augmented reality?
Plane detection
Plane detection finds flat horizontal surfaces (floor, table) and vertical ones (wall) in the camera image so that virtual objects can be placed on them.
It is the step that comes before placement in augmented reality. On plain, dark or very reflective surfaces, detection is slower, so a well-lit room works best.
Learn more: Floor and wall placement in augmented reality
Floor or wall anchoring
Anchoring attaches a virtual object to a detected surface, the floor or a wall, so that it stays put while the user moves around it.
Wall placement suits frames, mirrors, wall lights or shelves; floor placement suits furniture and machinery.
In Visuality: each object can be set to sit on the floor or hang on the wall.
Learn more: Floor and wall placement in augmented reality
1:1 scale (true-to-size)
1:1 scale displays the virtual object at its actual dimensions, which is essential for judging how much space a product takes up in augmented reality.
It requires the 3D model to be exported in the right units: in glTF, the unit of length is the metre.
In Visuality: you can adjust the object's size and stop customers from resizing it in augmented reality.
Learn more: Manage the size of a 3D object
Occlusion
Occlusion hides the parts of a virtual object that are behind a real element, such as a person or a piece of furniture, so that it looks truly part of the scene.
It relies on depth estimation (cameras, LiDAR or AI); availability varies by device and viewer.
Learn more: What is augmented reality?
Sharing and integration
AR QR code
An AR QR code opens a 3D viewer or an augmented reality experience directly when it is scanned with a smartphone camera.
It connects a physical medium (catalogue, packaging, trade show stand) to the product's 3D model.
In Visuality: a QR code is available for every shared object.
Learn more: QR code and augmented reality
Iframe (embedding)
An iframe is an HTML tag that embeds an external page, such as a 3D viewer, inside a page of your website.
It is the simplest way to integrate: a snippet to copy and paste, compatible with most CMSs and online stores.
In Visuality: each object provides an iframe snippet ready to paste into a product page.
Learn more: 3D viewer for websites
REST API
A REST API lets two pieces of software exchange data using standard HTTP requests, for example to create or update 3D objects from another system.
It is used to automate repetitive tasks, such as syncing a product catalogue with a 3D platform.
In Visuality: the REST API is used with an API key available in your dashboard.
Learn more: Integrations and API
3DModel structured data
3DModel is a schema.org type that describes a 3D model in a page's code, including its file URL and format, for search engines.
Google documents it as a recommended property (subjectOf) of merchant listing structured data, pointing to a .gltf or .glb file; 3D display in Google mainly goes through Merchant Center.
Learn more: Google 3D: your products in 3D on Google