Lesson 8.1Lesson 8.1 · VR & Immersive Design Review
VR Headsets & How VR Works
Two eyes, a tracked head and ninety frames a second - the plumbing that turns a rendered scene into a place your body believes it is standing in
A monitor shows you a picture of a room. A headset puts you in the room - because it draws a different image for each of your eyes and repaints both the moment you move your head.
The magic of VR is not resolution or graphics. It is a trick played on two systems your brain trusts absolutely: your two eyes see the world from slightly different positions, and your inner ear expects the view to move exactly when your head does. Satisfy both, precisely and fast enough, and your body simply concludes it is standing somewhere it is not.
That is the whole game. A VR headset renders one image per eye to give you depth, and it tracks your head in space so the view answers your motion instantly. Get the timing right - roughly ninety fresh frames every second - and presence arrives. Get it slightly wrong and you feel queasy. This lesson is about that plumbing, because understanding it is what lets you set up VR well in the next lesson and keep it comfortable in the one after.
Depth = two eye images. Presence = head tracking. Comfort = 90fps. The headset shows and tracks; your GPU renders.
Stereoscopy: two eyes, two images, and the depth that follows
Look at a coffee cup and close one eye, then the other. The cup jumps sideways, because each eye sees it from a slightly different position - your eyes sit about 6.3 centimetres apart (the interpupillary distance, or IPD). Your brain fuses those two slightly-different views into a single image and reads the difference between them as depth. This is stereoscopy, and it is the first thing a VR headset does that a monitor cannot.
Inside the headset there are two small displays (or two halves of one panel), one in front of each eye. Unreal renders the scene twice every frame - once from a virtual left eye and once from a virtual right eye, offset by your IPD - and sends each image to the correct display. Because the two renders differ exactly the way your real eyes would differ, your brain fuses them and the building gains true, felt depth: a column is genuinely in front of the wall behind it, a ceiling is genuinely up there. This is why VR reveals scale and proportion that a flat walkthrough only hints at.
Between each display and your eye sits a lens that focuses the close-up panel so it appears comfortably far away, and stretches the image across your vision. That lens distorts the picture, so the engine pre-distorts each eye image in the opposite way (barrel distortion to cancel the lens pincushion) - a step OpenXR and the headset runtime handle for you. The practical upshot for an architect: because the engine is drawing the scene twice, VR costs you roughly double the rendering work of a flat view. That single fact drives almost every performance decision in Module 9. Keep it in mind - it is the price of standing inside the room.
One render per eye, offset by ~6.3 cm. Two images -> your brain reads depth. Cost: ~2x the pixels.
Head tracking and 6DoF: the view moves because you moved
Depth alone is not presence. The second, and more powerful, trick is head tracking: the headset knows where your head is and which way it is pointing, and it feeds that pose to Unreal every frame so the virtual camera matches your real head. Turn to look at a window and the window is there. Crouch to check a worktop height and your eyeline drops. Your vestibular system - the balance sensors in your inner ear - expects the world to shift precisely as you move, and when the picture obeys, your body accepts the illusion completely.
Good VR tracks six degrees of freedom (6DoF). Three are rotation - pitch (nod), yaw (shake), roll (tilt) - and three are translation - moving your head left/right, up/down, and forward/back. A cheap phone-in-a-holder or an old headset offers only 3DoF: it knows where you are looking but not that you leaned in, so you cannot physically step around a column. 6DoF is what makes an architectural review real - you lean over a balustrade, duck under a beam, walk up to a joint and put your face next to it. Every headset in this course is 6DoF.
Modern standalone headsets achieve this with inside-out tracking: cameras on the headset itself watch the room and compute your position, so no external sensors are needed (this is how a Meta Quest works out of the box). Older or high-end tethered systems like the original Valve Index use outside-in tracking with base stations mounted in the room - more setup, historically a touch more precise. Either way, the headset also tracks your two hand controllers in 6DoF, which become your pointing, grabbing and teleporting tools in the design review to come.
3DoF = look around only. 6DoF = look AND step, lean, duck. Architecture needs 6DoF.
Why VR insists on about 90 frames a second
On a monitor, 30fps is watchable and 60fps is smooth. VR is far less forgiving, and the reason is physical rather than aesthetic. When you turn your head, the time between your motion and the matching image reaching your eyes - motion-to-photon latency - must be tiny, ideally under about 20 milliseconds. If the view lags your head even slightly, your inner ear and your eyes disagree, and that sensory conflict is exactly what triggers VR sickness: the cold sweat and nausea that ends a client review early.
To keep latency low and motion smooth, VR headsets run at high, fixed refresh rates - commonly 72, 90 or 120Hz. Ninety hertz has become the comfort benchmark, and it means the engine has about 11 milliseconds to draw both eye images - half the time a 60fps desktop frame gets, for double the pixels. This is the brutal arithmetic of VR performance: you have made the rendering job twice as big and given yourself less than half the time. Miss the budget and frames drop, latency spikes, and comfort collapses.
Headset runtimes defend against the occasional missed frame with reprojection (Meta calls its versions Asynchronous Timewarp and Spacewarp): if a fresh frame is not ready, the runtime re-shifts the last frame to match your latest head pose so the world still tracks your motion. It is a genuine safety net, but a crutch, not a cure - lean on it constantly and you see warping and judder. The honest rule for architectural VR: treat 90fps as a hard floor, not a target, and build the scene so you rarely need reprojection at all. Module 9 (hardware-and-vr-performance) is where you learn to actually hit it; here, just internalise why the number is non-negotiable.
60fps desktop = 16ms per frame. 90fps VR = 11ms for TWO eyes. Half the time, double the work.
The headset landscape: Quest, Index, Vive, Varjo - and OpenXR above them all
You do not need the most expensive headset to do excellent architectural VR, but it helps to know the field. Headsets split first into standalone and tethered. A standalone headset - the Meta Quest family is the dominant example - has its own processor and battery and runs untethered, which makes it superb for taking to a client meeting or a site. Its onboard chip is modest, though, so a heavy Unreal scene usually runs by streaming from a PC over a cable (Quest Link) or Wi-Fi (Air Link), using the PC's graphics card while the Quest acts as the display and tracker. A tethered headset - the Valve Index, HTC Vive family, or the high-end Varjo Aero and XR-4 - stays plugged into a powerful workstation and simply shows what that machine renders.
The differences that matter for archviz are resolution (sharper text and finer detail on the wall), field of view (how much of your vision is filled - the Index and Varjo are wider than most), refresh rate (Index reaches 120/144Hz), and comfort and lens quality. Varjo sits at the top for photoreal fidelity and near retinal-resolution clarity, at a professional price; the Index remains a benchmark for tracking and comfort; the Quest wins overwhelmingly on cost, convenience and the fact that a client can be handed one with no wires.
What frees you from betting on any one device is OpenXR - a royalty-free open standard, backed by the Khronos Group, that sits between your Unreal project and whatever headset is plugged in. You build your VR scene against OpenXR once, and it runs on Quest, Index, Vive, Varjo and future hardware without engine-specific rewrites. This is why the next lesson enables the OpenXR plugin rather than any vendor plugin: you are targeting the standard, not the gadget. For a practice, that means the VR experience you build this year survives the hardware refresh next year.
Stereoscopy
Rendering a separate image per eye, offset by your IPD, to create felt depth
The core of VR depth perception; it also means the engine draws the scene twice, roughly doubling the rendering cost.
6DoF (six degrees of freedom)
Tracking head rotation AND translation - look, step, lean, duck
What makes architectural review real; 3DoF headsets only track where you look, not that you moved. All course hardware is 6DoF.
OpenXR
A royalty-free open standard between the engine and any headset
Build once against OpenXR and run on Quest, Index, Vive and Varjo without vendor lock-in. The plugin the next lesson enables.
Refresh rate (72/90/120Hz)
How many times per second the headset repaints, fixed by the device
~90Hz is the comfort floor; the engine gets ~11ms to draw both eyes. Miss it and motion-to-photon latency spikes and sickness follows.
Workshop — map the trick your headset is playing
Before you touch Unreal, build a felt understanding of stereoscopy, 6DoF and the frame budget. This is a reasoning-and-observation exercise; a headset helps but is not required.
A browser to research headset specs; optionally any 6DoF VR headset to feel the tracking. No Unreal yet - that begins next lesson.
Goal: understand how a headset produces depth, tracking and comfort Inputs: any browser; a headset if you have one (Quest, Index or borrowed) Time: ~30 minutes
- 1Do the one-eye test on a real object: close each eye in turn and watch it shift. That shift is the parallax your headset recreates by rendering one image per eye. Write down, in your own words, how two offset images become depth.
- 2If you have any 6DoF headset, put it on in a safe cleared space and physically step around a virtual object, then lean in close and duck under something. Note every motion a 3DoF headset could NOT let you do - those are exactly the moves an architectural review needs.
- 3Look up the refresh rate and per-eye resolution of three headsets - a Meta Quest, a Valve Index and a Varjo. Make a small table. Which would you take to an on-site client pitch, and which to a high-fidelity studio review, and why?
- 4Compute the budget yourself: at 90fps, how many milliseconds does the engine get per frame, and remember it must draw BOTH eyes in that time. Compare it to a 60fps desktop frame. Write one sentence on why VR performance is so much harder.
- 5Explain motion-to-photon latency to a friend who has never used VR, using only the inner-ear-versus-eyes idea. If they understand why lag makes people queasy, you have the concept.
You’ll walk away with
A one-page brief that (a) explains stereoscopy and 6DoF in your own words, (b) tabulates three headsets on resolution, refresh rate and tethered/standalone, and (c) states the 90fps frame budget and why it is halved by two eyes.
Three altitudes on the same idea
Read the band that fits you — or all three.
Knowing the plumbing changes what you buy and promise. When a developer asks for a VR walkthrough, you now know the real question is not the headset brand but whether your workstation can push two eye-images at 90fps, and whether a standalone Quest streaming from your PC will do for an on-site pitch. Understanding 6DoF, stereoscopy and the frame budget lets you scope VR honestly instead of over-promising a queasy demo.
Stereoscopy is exactly why VR sells interiors that a moodboard cannot. Real depth and true scale let a client feel a ceiling height, a corridor width or the reach to a high shelf with their body. Because the headset renders your material and light for each eye at full scale, finishes read honestly - and a wireless Quest you can hand across a table makes that experience effortless to share in a meeting.
This is the conceptual bedrock recruiters assume you have. Being able to explain motion-to-photon latency, why VR needs ~90fps, and the difference between 3DoF and 6DoF marks you as someone who understands the medium, not just someone who put on a headset once. Learn it now on paper and the hands-on setup in the next lesson - and the performance work in Module 9 - will make immediate sense.
“A better VR headset means better graphics - the headset does the rendering, so buy the most powerful one.”
Do it yourself
No headset needed - reason it through.
- 1Why does a VR headset render the scene twice per frame, and what does that cost you?
- 2In one sentence, what is the difference between 3DoF and 6DoF, and why does architecture need 6DoF?
- 3Roughly how many milliseconds does the engine get to draw both eyes at 90fps?
- 4What is motion-to-photon latency, and why does high latency make people sick?
- 5Why does enabling the OpenXR plugin, rather than a vendor plugin, protect your work against a hardware change?
The one line to carry out
Peer-reviewed journals & authoritative standards
- 01Virtual reality — Wikipedia, 2026.
- 02Head-mounted display — Wikipedia, 2026.
- 03Stereoscopy — Wikipedia, 2026.
- 04Six degrees of freedom — Wikipedia, 2026.
- 05OpenXR — Wikipedia, 2026.
Now that you understand what a headset is doing, we open Unreal and make it happen: enabling the OpenXR plugin, starting from the VR template, meeting the VR Pawn, and getting your building running on a headset.
The author
Amogh N P
Architect, interior designer, and creative polymath. Studio Matrx began in his notebooks — his vision of design made honest, useful, and open to everyone. Its Academy is written and taught in his memory, and free, forever.
More about Amogh →