Definition: A passthrough camera is a forward-facing camera in a closed XR headset. It films the real environment and shows it as live video on the displays in front of the eyes. This method is called video see-through. Software adds virtual content to the camera image. This turns a VR headset into a device for mixed reality.

In B2B mechanical engineering, passthrough cameras are used for: mixed reality training next to the real system, layout planning in the factory hall, trade show presentations of virtual machines in the real booth, service instructions, and safe work with a headset, because the environment stays visible.

Advantage: One device for VR and MR. Users see colleagues, obstacles and real machines, while virtual content appears on top.

English: Passthrough Camera, Video See-Through (VST)
German: Passthrough-Kamera, Durchsichtkamera, Video-Durchsicht

How It Works

From the camera image to the display

  1. Capture: Two color cameras at approximately eye level film the environment.
  2. Determine depth: A depth sensor or software estimates distances, so that virtual objects appear correctly in front of or behind real objects (occlusion).
  3. Correct the perspective: The cameras are not exactly at the position of the eyes. Therefore the system recalculates the image for the eye position (reprojection). Otherwise distances look wrong.
  4. Overlay: The render engine places 3D content precisely over the video.
  5. Display: The combined image appears on the displays, new in each frame.

Difference from optical see-through

Headsets such as the HoloLens use optical see-through: users look through transparent lenses, and the content is projected onto them. The user sees reality without delay, but virtual content often looks pale. Video see-through, by contrast, shows everything on displays. This allows opaque, high-contrast content and a seamless switch to full VR, but it brings latency and limited image quality.

Devices with Passthrough (as of 2026)

Device Passthrough Classification
Meta Quest 3 Two 4 MP color cameras, depth sensor Widely used standalone headset
Meta Quest 3S Color passthrough without a dedicated depth sensor Lower-cost variant
Apple Vision Pro Cameras with R1 chip, 12 ms photon-to-photon latency according to Apple visionOS, with M5 since 2025
Samsung Galaxy XR Two high-resolution passthrough cameras First device with Android XR (2025)
Varjo XR-4 Two 20 MP cameras Enterprise headset for high fidelity

On Quest 3 and 3S, developers can use the Passthrough Camera API to apply their own image analysis to the camera image, for example for object detection. The user must give explicit permission first.

B2B Applications

Training next to the real machine

An operator stands in front of a real filling system. The headset shows work steps, highlighted controls and virtual parts that the operator will install later. Dangerous conditions, for example a running unit, are simulated, while the real system stays safely switched off.

Safety in the factory hall

In full VR, users see neither forklifts nor cables on the floor. Passthrough makes the environment visible. Many headsets switch to the camera image automatically when the user comes near an obstacle. But headset use in active production areas still needs a risk assessment and marked-off areas.

Layout planning and design review

A virtual prototype is placed at a 1:1 scale at its future location in the factory hall. Planners check the distances to columns, escape routes and adjacent systems before the foundation is poured.

Sales and trade shows

At the booth, a large virtual system stands next to a real exhibit. Visitors still see the sales representative and can talk to them normally. This lowers the barrier compared to closed VR.

Service

Technicians can go through maintenance steps on the real machine with on-screen instructions. For longer on-site assignments, tablet AR and optical headsets are often the more practical choice because of weight and battery life.

Limitations: Latency and Image Quality

  • Latency: Milliseconds pass between the camera image and the display. The longer the delay, the more likely discomfort and inaccurate grasping become during head movements. Manufacturers give their own values, and independent measurements are rarely comparable.
  • Resolution and sharpness: Small text on type plates or displays is often hard to read in the camera image. Test this in advance for tasks where users read measured values.
  • Light: Dark areas of the factory hall cause image noise, and strong backlight washes out the image.
  • Distortion at close range: Hands and objects directly in front of the face can look slightly distorted, because reprojection reaches its limits there.
  • Data privacy: Cameras film colleagues and operational areas. You must clarify whether images leave the device (data privacy, data security).

Passthrough in the Platform Mix

A passthrough headset is not the right tool for every task. A good approach is one shared 3D model that runs on different channels, depending on the situation:

Channel Strength Typical use
Web 3D in the browser No device, only a link Quotation phase, first impression
Tablet/smartphone AR Quickly available, familiar On-site service, placement at the customer site
Kiosk terminal Robust, no instruction necessary Booth, showroom
Passthrough headset Hands-free, 1:1 scale in the room Training, layout review
Full VR Full control over the scene Hazard simulation, remote review

Practical tip: Before a rollout in production, run a pilot in the real factory hall, with typical lighting, protective equipment and real routes. Only there will you see whether image quality, tracking and wearing comfort are sufficient for a full shift.

Frequently Asked Questions (FAQ)

Is passthrough the same as augmented reality?

Not quite. Passthrough is a technical method to make reality visible. Augmented reality describes the result: digital content in the real environment. AR also works on smartphones, tablets or optical headsets, without any passthrough headset.

Can I safely operate a machine through the camera image?

For training and orientation, yes. For safety-critical operations, do not rely on the camera image alone. Latency, a limited field of view and possible system crashes are reasons to practice such steps without a headset or on switched-off systems.

Do I need a depth sensor?

It improves how real objects hide virtual objects (occlusion) and how the device captures the room. Devices without a dedicated sensor estimate depth with software. For many training sessions, this is sufficient.

Can an application access the camera images?

The platforms handle this differently. On some devices, only the system processes the images. On others, apps get access after permission. Your company should define which apps get camera access, ideally centrally through MDM.

Is a high-end headset worth it?

When fine details of the real environment are important, for example when you compare components or read displays, enterprise devices give much better image quality. For layout planning and presentations, a standalone headset is often sufficient. In addition to the purchase price, the total cost counts: accessories, management, hygiene, replacement devices, and whether the headset needs a powerful PC. A comparison in your own use case tells you more than any data sheet.

Industry Standards and Technologies

XR technologies: - Mixed reality – combination of real and virtual - Augmented reality – digital content in reality - Virtual reality – fully virtual environment - Extended reality – the XR spectrum - Spatial computing – computing in space - Optical see-through – optical transparency

Devices and platforms: - visionOS – Apple's XR operating system - HoloLens – optical MR headset - Sensors – cameras and depth sensors - LiDAR – depth measurement with a laser

Tracking and display: - SLAM – spatial mapping and positioning - Motion tracking – tracking of movement - Real-time rendering – display in real time - Frame rate – frames per second