How AR Glasses Work

Every pair of AR glasses solves the same fundamental problem: how do you project a digital image onto a transparent surface and make it visible to the human eye while still allowing the wearer to see the real world behind it? The engineering required to do this in a form factor that fits on your face, weighs under 100 grams, and runs on a tiny battery is genuinely remarkable.

This article explains the core technologies that make AR glasses work. We will cover the two dominant optical architectures (waveguides and birdbath), the display sources that generate the image (micro-LED and micro-OLED), the physics that limits field of view, and why outdoor brightness remains one of the hardest unsolved problems in the industry.

If you are evaluating these tradeoffs as a builder rather than a buyer, our AI glasses guide for developers compares SDK access, input methods, displays, and platform constraints.

The Core Challenge: Light Must Reach Your Eye

A traditional screen emits light directly toward you. A monitor, phone, or TV has pixels that produce light, and that light travels in a straight line to your retina. Simple geometry.

AR glasses cannot work this way because you need to see through them. If you put an opaque screen in front of someone’s eyes, you have a VR headset, not AR glasses. The challenge is projecting a visible image while keeping the lens transparent enough to see the real world.

Two things must happen simultaneously. First, a tiny display source (usually smaller than a fingernail) must generate an image. Second, an optical system must take that image and redirect it so the light enters your pupil at the correct angle to form a focused picture on your retina, while allowing ambient light from the real world to pass through largely unimpeded.

The way this redirection happens is what separates different optical architectures.

Waveguide Displays: The Transparent Approach

Waveguide optics are used in products like the Even Realities G1, Vuzix Ultralite, and enterprise devices like Microsoft HoloLens 2. They are the only current technology that allows truly transparent lenses while displaying digital content.

A waveguide is a thin, flat piece of glass or polymer that traps light inside it through total internal reflection (the same physics that keeps light bouncing inside a fiber optic cable). Here is how the process works:

1. A micro-display (usually micro-LED or LCoS) generates a small, bright image near the temple of the glasses.
2. An input coupler (typically a diffraction grating etched into the glass surface) bends the light from the display at an angle steep enough to enter the waveguide and begin bouncing between its inner surfaces.
3. The light bounces through the waveguide via total internal reflection, traveling from the temple area toward the center of the lens in front of your eye.
4. An output coupler (another diffraction grating, positioned in your line of sight) extracts portions of the bouncing light at each reflection, redirecting it outward toward your pupil.
5. Your eye receives this extracted light and focuses it on your retina, forming the AR image.

The beauty of this system is that the waveguide itself remains mostly transparent. The diffraction gratings are microscopic surface patterns that interact with specific wavelengths of light from the display source but allow most ambient light to pass through normally. The lens looks like slightly tinted glass.

Limitations of Waveguides

Waveguides have significant constraints that explain why current AR glasses have limited capabilities:

Small field of view. The angular range of light that a waveguide can capture, transport, and extract is limited by the refractive index of the glass material. Higher refractive index materials allow wider FOV, but they also increase weight and introduce chromatic aberration. Most consumer waveguide displays achieve between 20 and 35 degrees FOV. For reference, a 25-degree FOV means the virtual image occupies a relatively small portion of your visual field.

Efficiency losses. Not all light that enters the waveguide reaches your eye. Each bounce loses some light, and the output coupler extracts only a fraction at each point. Total optical efficiency for waveguide systems is typically between 1% and 5%, meaning 95 to 99 percent of the light generated by the display source is lost before reaching your eye. This is why brightness is such a problem.

Color uniformity issues. Different wavelengths (red, green, blue) interact with diffraction gratings differently, which can cause rainbow artifacts, color shifting across the display, and uneven brightness distribution. Manufacturers use multi-layer waveguides (one layer per color) or carefully engineered surface-relief gratings to mitigate this, adding cost and complexity.

Eye box size. The “eye box” is the volume of space where your pupil must be positioned to see the full display. Waveguide eye boxes are typically small (8 to 12mm), meaning small shifts in glasses position can cause parts of the image to disappear. Exit pupil expansion techniques (repeating the output coupler pattern) enlarge this but reduce overall efficiency further.

Waveguide Glass Is Part of the Optical Design

Waveguide performance depends on the glass substrate as well as the gratings patterned onto it. A higher refractive index gives optical designers more room to transport light at useful angles, while density, transmission, wafer geometry, and manufacturing tolerances affect weight, color efficiency, yield, and production cost.

SCHOTT RealView 2.0 lightweight is a September 2026 material-family update aimed at AR waveguides and AI smart glasses. SCHOTT specifies a refractive index of 2.0, density below 4 g/cm³, and blue-light transmittance above 96% at 460 nm across a 10 mm optical path. It is available for sampling and described by SCHOTT as ready for high-volume supply in 150, 200, and 300 mm round wafers, plus rectangular formats up to 310 × 310 mm.

That supply status applies to the optical material, not to a finished consumer product. SCHOTT has not identified a shipping pair of glasses using RealView 2.0 lightweight in this announcement, so it should not be read as evidence of an OEM design win or mass consumer deployment.

Birdbath Optics: The Immersive Approach

Birdbath optics are used in products like the Xreal One Pro, Rokid Max, and many media-focused glasses. They sacrifice transparency for image quality.

The name “birdbath” comes from the curved half-mirror (beam splitter) that sits in front of your eye, resembling the concave shape of a birdbath. The optical path works like this:

1. A micro-OLED or micro-LED display sits above or below your line of sight within the glasses frame.
2. Light from the display hits a curved beam splitter (partially reflective mirror) positioned at approximately 45 degrees.
3. The beam splitter reflects the display light toward a curved reflective surface (the “birdbath” mirror) positioned below or in front of the lens area.
4. This mirror reflects and focuses the light back through the beam splitter and into your eye.

The result is a bright, high-contrast image with a larger field of view than waveguides can typically achieve (45 to 57 degrees is common). However, the beam splitter blocks a significant amount of ambient light, making the lenses appear dark or opaque.

Why Birdbath Works Better for Media

Birdbath systems have higher optical efficiency (more display light reaches your eye), which means brighter images without requiring extremely powerful display sources. They also handle full RGB color uniformly without the chromatic issues that plague waveguides.

The tradeoff is clear: you get a better image at the cost of seeing the real world. This makes birdbath optics ideal for media consumption (movies, gaming) but unsuitable for everyday AR where you need to see your environment.

Micro-LED vs Micro-OLED: The Display Source

Behind every AR or media glasses optical system sits a tiny display that generates the original image. Two technologies dominate this space.

Micro-OLED

Micro-OLED (also called OLEDoS, OLED on Silicon) uses organic light-emitting diodes fabricated directly on a silicon backplane. Sony is the primary manufacturer of micro-OLED panels used in consumer smart glasses.

Advantages of micro-OLED include true black levels (pixels turn off completely), excellent contrast ratios (over 100,000:1), fast pixel response time (ideal for high refresh rates), good color accuracy with wide color gamut, and mature manufacturing processes.

Disadvantages include limited maximum brightness (typically 3,000 to 6,000 nits for the raw panel), organic material degradation over time (though lifespans exceed practical product lifecycles), and power consumption that increases significantly at high brightness levels. Panel sizes are constrained to approximately 0.5 to 1 inch diagonal.

Micro-OLED is currently the preferred technology for media glasses (Xreal One Pro, Rokid Max) where contrast and color matter more than raw brightness, because the opaque or semi-opaque optical design reduces ambient light competition.

Micro-LED

Micro-LED uses inorganic LED elements (typically gallium nitride for blue/green, aluminum indium gallium phosphide for red) arranged in microscopic arrays. Companies like JBD (Jade Bird Display), Porotech, and Compound Photonics are leading micro-LED development for AR applications.

Advantages include extremely high brightness potential (over 1,000,000 nits theoretically, with practical panels achieving 30,000+ nits), inorganic materials that do not degrade like OLEDs, very fast switching times, high efficiency at the pixel level, and better outdoor visibility due to raw brightness.

Disadvantages include manufacturing yield challenges (dead pixels on micro-scale arrays), difficulty with full RGB integration (red micro-LEDs are less mature than blue/green), lower resolution than micro-OLED at equivalent panel sizes, higher cost per panel in 2026, and color mixing requiring careful calibration.

Micro-LED is the preferred technology for waveguide-based AR glasses because its extreme brightness partially compensates for the optical efficiency losses in waveguide systems. When 95% or more of your light is lost in the optics, starting with a million-nit source means you still have tens of thousands of nits reaching the eye.

The Brightness Math Problem

This is where the physics becomes unforgiving. On a sunny day, surfaces around you reflect approximately 5,000 to 10,000 nits of light into your eyes. For a virtual AR image to be visible overlaid on these surfaces, it needs to be at least comparably bright (and ideally brighter for clear contrast).

A waveguide with 2% optical efficiency needs a display source producing 250,000 to 500,000 nits to deliver 5,000 to 10,000 nits to the eye. Current micro-LED panels can approach these numbers in monochrome (green) but achieving this across full RGB at high resolution remains extremely challenging.

This is why most consumer AR glasses with waveguide displays work well indoors (where ambient light is 100 to 500 nits) but wash out in direct sunlight. It is a fundamental physics constraint, not a software problem or a “next generation will fix it” issue. Progress is being made, but outdoor-readable full-color AR at wide FOV requires display brightness improvements of roughly 5 to 10 times beyond current shipping products.

Why Field of View Is Limited

Human vision spans approximately 200 degrees horizontally (including peripheral vision) and about 130 degrees vertically. The central sharp-focus area (fovea) covers roughly 5 degrees, but the useful field for recognizing objects and reading text extends to perhaps 30 to 40 degrees.

Current AR glasses offer between 20 and 57 degrees FOV. Why not wider?

For waveguides, FOV is constrained by the refractive index of the glass. Light can only be trapped by total internal reflection if it enters the waveguide at angles steeper than the critical angle (determined by refractive index). Wider FOV requires capturing light at shallower angles, which breaks the total internal reflection condition. Higher refractive index materials (n greater than 2.0) help but are expensive, heavy, and introduce optical aberrations.

For birdbath systems, FOV is limited by the size of the beam splitter and focusing mirror relative to the form factor. A wider FOV requires a larger optical assembly, which means bulkier, heavier glasses. The Xreal One Pro’s 57-degree FOV is essentially the practical limit for something that still resembles glasses rather than goggles.

Research approaches to wider FOV include tiled waveguides (multiple stacked waveguide sections covering different angular ranges), hybrid diffractive/refractive systems, and novel materials with extremely high refractive indices. None of these have reached consumer products as of mid-2026. Meta’s Orion prototype demonstrated approximately 70-degree FOV using silicon carbide waveguides, but at a cost not yet viable for consumer pricing.

How Head Tracking Works in AR Glasses

Most AR and media glasses include inertial measurement units (IMUs) consisting of accelerometers and gyroscopes. These detect rotational movement (head turns) with very low latency (under 1ms for sensor data).

More advanced systems add SLAM cameras (Simultaneous Localization and Mapping) that face outward to track the environment and provide positional (6DoF) tracking. Eye tracking cameras face inward to detect where you are looking within the display, enabling foveated rendering and gaze-based interaction. Depth sensors using ToF (time-of-flight) or structured light map the 3D geometry of your environment for accurate placement of virtual objects.

Each additional sensor adds weight, power consumption, processing requirements, and cost. This is why most lightweight consumer glasses (under 100g) include only basic IMU tracking, while full spatial computing requires heavier headset-style devices.

Thermal Constraints

Processing silicon generates heat. Display sources generate heat. Batteries generate heat during discharge. All of this heat must be dissipated from a device sitting on your face, where skin sensitivity to temperature is high.

The comfort threshold for skin-contact electronics is approximately 40 to 42 degrees Celsius. Exceeding this causes discomfort and eventual tissue irritation. This thermal budget severely constrains how much processing power and display brightness can be sustained in a glasses form factor.

It is one reason why most smart glasses offload heavy computation to a phone or external processor. Running an AI model, a camera feed, display rendering, and wireless communication simultaneously in a glasses frame would generate more heat than the form factor can safely dissipate with passive cooling.

What the Future Holds

The next meaningful advances will likely come from higher refractive index waveguide materials pushing FOV toward 50+ degrees while maintaining transparency. Full-color micro-LED panels exceeding 100,000 nits at practical resolutions will help with outdoor visibility. More efficient diffractive coupling designs that improve waveguide optical efficiency from 2% to 5% or above will reduce the brightness requirements at the source. Advanced thermal materials like graphene heat spreaders and vapor chamber cooling in thin frames will allow more on-device processing. Chiplet architectures that distribute processing across the frame will help manage thermal loads.

The timeline for these advances reaching affordable consumer products is difficult to predict. Some (like improved waveguide materials) are already appearing in enterprise products. Others (full-color high-brightness micro-LED) remain 2 to 4 years from mass production.

Frequently Asked Questions

Why can I see through some AR glasses but not others?

The optical architecture determines transparency. Waveguide-based glasses (Even Realities G1, Vuzix Ultralite) use transparent glass with embedded diffraction gratings, allowing you to see the real world clearly. Birdbath-based glasses (Xreal One Pro, Rokid Max) use semi-opaque beam splitters that block most ambient light to improve display contrast. The choice between them trades image quality for real-world visibility.

Why are AR glasses displays so small compared to VR headsets?

Field of view is constrained by optical physics. VR headsets use large, close-mounted lenses to fill your vision (90 to 120 degrees), but they are bulky because the optics require physical size. AR glasses must remain thin and lightweight, using waveguides or compact mirrors that cannot achieve the same angular range without violating physical constraints of total internal reflection or form factor limits.

Will outdoor AR glasses ever work well in sunlight?

Eventually, but it requires display brightness improvements of 5 to 10 times beyond current technology. The issue is that waveguide optics lose 95%+ of display light, so the source must be extraordinarily bright to compete with sunlight. Micro-LED technology is progressing toward this goal, but full-color panels at the required brightness and resolution are still years from consumer readiness.

What is the difference between micro-LED and regular LED?

A regular LED is a macroscopic light source (the kind in room lighting or indicator lights). A micro-LED is a microscopic LED element, typically 3 to 10 micrometers in size, arranged in arrays of millions to form a self-emissive display. Each micro-LED is an individual pixel. The “micro” refers to the physical size of each light-emitting element, not the overall display.

Why do some AR glasses only show green text?

Monochrome green displays are dramatically brighter and simpler to manufacture than full-color displays. Green light falls at the peak sensitivity of human vision (around 555nm wavelength), meaning green pixels appear brightest for a given power input. Products like the Even Realities G1 use monochrome green micro-LED because it maximizes readability with minimal power consumption and optical complexity. Full color requires either three separate LED types (RGB) or color conversion layers, adding cost and reducing efficiency.

Do AR glasses damage your eyes?

Current evidence suggests that AR glasses do not cause permanent eye damage when used as directed. The light levels reaching your retina are comparable to looking at a phone or monitor. However, prolonged use of any near-eye display can cause temporary eye strain, dry eyes (from reduced blinking), and focusing fatigue. The long-term effects of daily multi-hour AR glasses use over decades are not yet studied because the products have not existed long enough.

Why are AR glasses so expensive?

The display sources (micro-OLED and micro-LED panels), precision optics (waveguides with nanometer-scale diffraction gratings), and custom silicon for spatial processing are all manufactured in relatively small volumes compared to smartphones or monitors. The technology is early in its production curve. Costs will decrease as manufacturing scales, but in 2026, these components remain expensive to produce at consumer quality levels.

How does prescription correction work with AR glasses?

Most AR glasses accept custom prescription lens inserts that clip or snap into the frame. The AR display optics are calibrated for a specific focal distance (usually optical infinity or 2 to 4 meters virtual distance), and the prescription insert corrects your vision to that focal point. Some frames (like Even Realities G1) are designed from the ground up to accept standard prescription lenses from an optician.

Sources

• JBD: Roadrunner platform announcement — Official announcement of 2.5μm pixel pitch MicroLED microdisplays for AR glasses
• JBD: Roadrunner II polychrome projector — Full-color MicroLED projector announcement for consumer AR glasses (June 2026)
• Meta: Orion AR Glasses technical overview — Details on silicon carbide waveguide lenses and approximately 70-degree FOV
• SCHOTT: RealView 2.0 lightweight glass wafers — Official material specifications, sampling status, and supported wafer formats (September 2026)
• TechRadar: RayNeo X3 Pro review — Independent review confirming 6,000 nits peak brightness from MicroLED waveguide display
• Even Realities G1 product page — Specifications for monochrome green micro-LED waveguide display at ~40 grams
• MicroLED-info: JBD Roadrunner platform details — Technical analysis of 10,160 PPI resolution and mass production timeline (H2 2026)