ExplorXR

What Is Inside a VR Headset? Displays, Lenses, Cameras, Chips

By ExplorXR Editorial · · Updated November 3, 2026 · 10 min read

What is inside a VR headset?

A VR headset is six subsystems in a plastic shell: a display for each eye, a lens in front of each display, a ring of cameras and depth sensors, inertial sensors, a mobile chip with memory and storage, and a battery with speakers and microphones. Everything else, the strap, the facial interface, the heat spreaders, exists to hold those parts in the right place and keep them from overheating.

The parts list is also why two headsets with the same resolution can feel nothing alike. Meta’s Quest 3 puts 2064 by 2208 pixels per eye behind pancake lenses, weighs 515 grams, and runs a Snapdragon XR2 Gen 2 with 8GB of RAM. Apple’s Vision Pro puts 23 million pixels of micro-OLED behind its own optics, adds a second chip for sensor data, and lists 750 to 800 grams with the Light Seal and Dual Knit Band, plus a 353 gram battery in your pocket. Valve’s Steam Frame goes the other way: a 190 gram core, 435 grams fully assembled, and a battery in the rear strap.

Those three designs map the current category, and the components explain the trade-offs at the heart of it.

Displays: one panel per eye

Mainstream headsets now use two panels, one for each eye, rather than the single shared panel of earlier designs. Two panels make stereo separation exact, keep the monocular field of view consistent across interpupillary distances, and give each eye the full pixel grid.

The panel technology splits into two families.

LCD. A backlight shines through color filters and liquid crystal shutters. It is cheap, bright and mature. Quest 3 uses dual LCD panels at 2064 by 2208 per eye and 25 pixels per degree, and the review record is consistent about what that costs: no local dimming, no quantum dot layer, and no true black, only a dark grey. Steam Frame also uses LCD, at 2160 by 2160 per eye.

Micro-OLED. Each pixel emits its own light, so black is black and contrast is limited by the panel rather than by a backlight. Vision Pro uses micro-OLED with 23 million pixels across both eyes, a 7.5 micron pixel pitch, 92 percent DCI-P3 coverage, and refresh rates of 90, 96, 100 and 120Hz. The cost is yield and price, which is why micro-OLED appears at the top of the market first, and why Meta’s micro-OLED headset is a $1,300 device with an off-head compute puck rather than a drop-in Quest 3 replacement.

Resolution per eye is the number on the box. Pixels per degree is the number your eye actually sees: Quest 3’s 25 PPD against Quest 3S’s 20 PPD is a difference you notice in text, even though both are Meta headsets at similar prices.

Optics: pancake lenses and the light they give up

The lens decides a headset’s thickness and its brightness budget.

Older and cheaper headsets use Fresnel lenses, a flat element that bends light with concentric prisms. They are cheap and efficient, but they need more distance between the panel and the eye and they are less forgiving about where your pupil sits. Quest 3S uses Fresnel optics and a 73.9mm thick visor. Quest 3 uses pancake optics and a 62.3mm visor.

Pancake optics fold the light path. The display emits polarized light, a half mirror reflects part of it back, quarter wave plates rotate the polarization, and a reflective polarizer sends the light forward again, so the beam makes three passes before it reaches the eye. The folding packs a long focal length into a short module, which is what lets the panel sit close to the lens and the visor stay slim.

The physics charges for that trick. In a conventional pancake design roughly 25 percent of the light emitted by the display reaches the observer, with the rest lost at the half mirror. That is why pancake headsets need bright panels, and why the display and the optics have to be co-designed: a dim panel behind a pancake lens gives a dim image, and a brighter panel costs power and heat.

The upside is optical tolerance. Meta told UploadVR that Quest 3’s lenses have a claimed eyebox of plus or minus 5mm, meaning the image stays sharp even when the lenses are not perfectly centred on your pupils. Fresnel headsets make you hunt for the sweet spot.

Cameras, depth sensors and the ring around the front

Look at the front of a headset and you are looking at the tracking system.

Quest 3 carries two 4 megapixel color cameras for passthrough, four tracking cameras, and an infrared projector that supplies depth for room meshing and hand tracking. Vision Pro carries twelve cameras: two 6.5 megapixel color cameras, four front tracking cameras, two downward facing cameras for face tracking, and four internal infrared cameras for eye tracking, plus a LiDAR scanner and a depth projector. Steam Frame uses four monochrome outward cameras with an infrared illuminator so tracking survives a dark room, and two internal cameras for eye tracking.

The camera count is not a spec sheet flex. Each job needs a different field of view, exposure and wavelength. Wide angle cameras see the room and the controllers, color cameras feed the passthrough pipeline, infrared cameras inside the lens assembly watch your pupils and are invisible to you, and downward cameras read your cheeks and jaw for face tracking. A headset that skips color cameras, as Steam Frame does, saves cost and power and gives up color mixed reality.

The internal cameras are the hardware behind gaze input and foveated rendering, which how VR eye tracking works covers from the sensor side.

IMUs and the loop that keeps the world still

Cameras are too slow to track a fast head turn on their own. A headset also carries an inertial measurement unit, typically an accelerometer and a gyroscope, reporting rotation and acceleration far more often than the cameras refresh. The headset fuses the two: the IMU carries fast motion, and the cameras correct the drift that the IMU accumulates.

That loop is why inside-out tracking works without base stations, and it is measured in milliseconds. Apple’s R1 chip, which exists only to process sensor data on Vision Pro, is specified at 12 milliseconds of photon-to-photon latency, the time from a photon hitting a sensor to the display updating. Miss that budget and the world swims when you turn your head.

The chip, and sometimes a second chip

A standalone headset runs the same class of silicon as a phone. Quest 3 uses a Snapdragon XR2 Gen 2 on a 4nm process with 8GB of RAM. Steam Frame uses a Snapdragon 8 Gen 3 with 16GB. Vision Pro splits the job: an M5 chip runs the apps, and the R1 handles the camera and sensor stream, which keeps the sensor loop off the main processor.

The split exists because a headset has two very different workloads. Rendering wants a wide GPU and memory bandwidth. Tracking wants deterministic, low latency processing that cannot wait behind a frame. Putting both on one chip means the tracking loop competes with the renderer, and in a device with no fan and a battery on your head that competition shows up as heat and dropped frames.

Heat is also why some designs move compute off the head entirely, into a puck on a cable or a tethered box. It is the same trade the category made with PC VR, at a shorter distance and a smaller box.

Power, weight and where the battery sits

Quest 3 puts the battery in the visor, rates about 2.2 hours of average use, and weighs 515 grams, of which the visor alone is 397 grams. Vision Pro moves the battery into a 353 gram pack on a cable, rated up to 2.5 hours of general use, and still lists 750 to 800 grams on the head. Steam Frame puts a 21.6Wh battery in the rear strap, which brings the total to 435 grams and moves the mass behind your head.

Total weight is the number people quote. Weight distribution is the number they feel. A 435 gram headset with the battery at the back can be more comfortable than a lighter headset with everything in the visor, because the torque on your neck depends on where the mass sits. It is the argument behind rear batteries and top straps, and the reason a Quest 3 versus Steam Frame comparison is about more than a spec table.

Audio, microphones and the small parts

Speakers and microphones are part of the tracking problem, not an accessory. Spatial audio needs head tracking to stay anchored to the world, and Vision Pro adds audio ray tracing and a six microphone array. Steam Frame puts dual speaker drivers on each side of the strap, and Quest 3 keeps a 3.5mm headphone jack.

The rest of the inside is adjustment hardware: the IPD mechanism (58 to 70mm of lens separation on Quest 3, an automatic 51 to 75mm on Vision Pro), the facial interface that sets eye relief, and the heat spreaders or fans that keep the chip under its thermal limit.

What the parts list explains

Three things follow from the teardown.

First, resolution alone does not describe a display. Pixels per degree, panel type, local dimming and lens quality decide whether text is readable and whether dark scenes hold up.

Second, the tracking hardware constrains the software. A headset without eye tracking cannot do eye tracked foveated rendering, and a headset without color cameras cannot do color mixed reality, which is why multiplatform releases have to scale features per device.

Third, physics still sets the price. A headset has to fit a display, optics, cameras, a chip and a battery within a few hundred grams on a face, and every choice trades against another.

What is still missing

The biggest unsolved problem is focus. Every shipping consumer headset renders at a single focal distance, so your eyes converge on an object at one distance while focusing at another. That vergence accommodation conflict measurably increases the time it takes to focus and is a known source of discomfort, and software can only work around it while the optics stay fixed.

Field of view has also plateaued near 110 degrees horizontally on mainstream headsets, battery life is still measured in a couple of hours, and micro-OLED remains expensive enough to define the top price tier. Those three constraints, not resolution, are what the next generation has to beat.


Sources

Frequently asked questions

What is inside a VR headset?
A display panel and a lens for each eye, a ring of outward facing cameras with an infrared depth sensor, an IMU, a mobile chip with memory and storage, and a battery, plus speakers and microphones. The strap and facial interface exist to hold those parts in place.
What is the difference between LCD and micro-OLED in a VR headset?
LCD panels use a backlight behind color filters, so black levels depend on dimming, and Quest 3's panels have no local dimming and cannot show true black. Micro-OLED pixels emit their own light, which is why Vision Pro fits 23 million pixels into two panels with much higher contrast, at a much higher cost.
Why do VR headsets use pancake lenses?
Pancake optics fold the light path three times using polarizers and a half mirror, so the display can sit close to the lens. Quest 3's visor is 62.3mm thick against 73.9mm for the Fresnel lens in Quest 3S, at the cost of light: a conventional pancake design delivers roughly 25 percent of the display's output to the eye.
How many cameras does a VR headset have?
It varies by design. Quest 3 has two color cameras, four tracking cameras and a depth projector, Vision Pro has twelve cameras including four internal eye tracking cameras plus LiDAR, and Steam Frame has four monochrome outward cameras, an infrared illuminator and two internal eye cameras.
Does a VR headset have a GPU?
A standalone headset has an integrated GPU inside its mobile system on chip. Quest 3 runs a Snapdragon XR2 Gen 2, Steam Frame a Snapdragon 8 Gen 3, and Vision Pro an M5 with a separate R1 chip handling sensor data. A PC VR headset has no GPU and uses the graphics card in the PC.
Why is the battery in the strap or in a separate pack?
To move mass off the front of the face. Steam Frame puts a 21.6Wh battery in the rear strap for 435 grams total, and Vision Pro uses a 353 gram battery on a cable while the headset itself is listed at 750 to 800 grams. Quest 3 keeps the battery in the visor at 515 grams with about 2.2 hours of average use.

Related reading