Edward Tang | Avegant: How does waveguide "ghosting" force AR designers to warp their industrial designs, and how does non-telecentric LCoS solve it?
00:04:51 - 00:06:28
Other snippets from this talk
Summary of the clip:
How does waveguide "ghosting" force AR designers to warp their industrial designs, and how does non-telecentric LCoS solve it?
Diffractive waveguides suffer from severe back-reflections from their input couplers, returning up to 20% of the projected light directly back into the light engine. Because MicroLED panels are highly reflective (ranging from 30% to 60% reflectivity), this back-and-forth bounce creates devastating ghost images that ruin visual quality and image contrast for the user.
To eliminate these reflections, MicroLED system designers are forced to mechanically tilt the entire engine by at least half of the field of view—for instance, 15 degrees for a 30-degree field of view. This mechanical rotation dramatically swells the outer profile of the glasses, dictating a bulky, compromised industrial design simply to accommodate the angled engine.
Avegant addresses this problem by designing non-telecentric LCoS optical engines that naturally trap and block back-reflected light from escaping the system. This optical trick eliminates ghosting without sacrificing efficiency, freeing industrial designers to place the engine at whatever angle fits the frame aesthetic rather than being constrained by stray light paths.
In this short video, you can learn:
* The mechanics of how waveguide input couplers and reflective display panels cooperate to generate destructive ghost images.
* Why MicroLED engines require a dramatic mechanical tilt of half their field of view to mitigate stray light reflections.
* How non-telecentric LCoS illumination architectures eliminate ghosting while offering industrial designers complete freedom of placement.
📋 **Clip Abstract** Edward Tang details how back-reflections between diffractive waveguides and reflective MicroLED panels create ghost images, forcing awkward mechanical tilting that ruins sleek industrial designs. He introduces Avegant's non-telecentric LCoS systems as a superior optical alternative that eliminates ghosting internally without design compromises.
#NonTelecentricLCoS, #WaveguideGhosting, #DiffractiveWaveguides, #MicroLEDOpticalEngine, #AugmentedRealityOptics, #NearEyeDisplays
This is a highlight of the presentation:
AR’s Display Dilemma - LCoS v MicroLED
More Highlights from the same talk.
01:56 - 03:19
Why is one watt the absolute thermal and power limit for consumer AR glasses?
Why is one watt the absolute thermal and power limit for consumer AR glasses?
Data and research show that consumer AR glasses cannot exceed about 50 grams if they are to be all-day wearable. Above 50 grams, the device becomes a short-term use product, worn for an hour or two, and loses the ability to replace your everyday glasses. Hitting that number is incredibly difficult given all the components that must go into the glasses.
The second constraint is the thermal budget. Devices cannot get too hot: too hot means discomfort or, worse, a safety issue that may burn your skin. Typically you want to stay below 50 degrees, and ideally below about 43 degrees, for the device to be safe and comfortable. Combined with the 50-gram limit, this creates a serious thermal dissipation challenge.
With a 50-gram limit, you cannot put fans in these devices, and you can't really put heat sinks in them either: they are too small and too light, and there is no way to get airflow through the system. That is where the one watt number comes from, as about the maximum thermal dissipation limit achievable through a pair of glasses while keeping them small, lightweight and consumer-friendly.
In this short video, you can learn:
* All-day wearable consumer AR glasses must not exceed about 50 grams, or they degrade into one-to-two-hour short-term use devices.
* Device temperatures must stay below 50 degrees, and ideally below about 43 degrees, to remain comfortable and avoid burning skin.
* Because a 50-gram device cannot use fans, heat sinks or airflow, roughly one watt is its practical thermal dissipation ceiling.
📋 **Clip Abstract** Consumer AR glasses face two hard ceilings: about 50 grams of mass for all-day wear, and skin-safe surface temperatures below 50 degrees, ideally about 43 degrees. Because a device that small and light cannot host fans, heat sinks or airflow, those limits converge on roughly one watt of maximum thermal dissipation.
About the speaker:
* Speaker: Edward Tang
* Company: Avegant Corp.
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#50GramWeightLimit, #ThermalDissipationCeiling, #PassiveCooling, #SkinTemperatureLimit, #AugmentedReality, #SmartGlasses
12:22 - 14:44
How can shrinking the light engine's pupil size more than double waveguide coupling efficiency?
How can shrinking the light engine's pupil size more than double waveguide coupling efficiency?
Pupil size is a really important knob. Light engine efficiency and waveguide efficiency are usually assessed separately, yet what actually matters is milliwatts to the glasses to nits to the eye. That means treating the light engine and the waveguide together, as one system, and accounting for the losses at the interface between them. Co-optimising that interface is where the gains hide.
The input to a waveguide is not uniformly efficient. One side is very inefficient, because light has to travel multiple bounces through that input pupil, and every time it touches it, a lot of light is lost. Other areas are highly efficient: light enters the input pupil and couples immediately into the waveguide without secondary interactions. Shrinking the pupil lets the engine target the efficient regions.
A typical microLED system has a pupil of three and a half to four millimetres. Assuming about 1,000 nits per lumen at that pupil size, simply reducing pupil size lifts waveguide efficiency by 50%; moving from unpolarised to polarised light adds another 50%. That means more than doubling waveguide efficiency. And a smaller pupil allows a thinner waveguide, cutting weight and freeing budget for lighter glasses or more battery.
In this short video, you can learn:
* Reducing pupil size raises diffractive waveguide coupling efficiency by 50%, and polarised light versus unpolarised contributes another 50%.
* Waveguide input pupils are non-uniform: one side loses light to repeated bounces while other regions couple immediately with no secondary interactions.
* A smaller light engine pupil also permits a thinner waveguide, reducing weight and leaving budget for lighter glasses or more battery.
📋 **Clip Abstract** Pupil size controls how much light the light engine actually delivers to the eye, because waveguide input pupils are unevenly efficient. Co-optimising the light engine pupil with the diffractive waveguide — shrinking it from three and a half to four millimetres and polarising the light — can more than double waveguide efficiency while thinning the waveguide.
About the speaker:
* Speaker: Edward Tang
* Company: Avegant Corp.
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#WaveguideCouplingEfficiency, #PupilSizeReduction, #DiffractiveWaveguides, #MicroLEDLightEngine, #AROptics, #NearEyeDisplays
08:29 - 10:16
How does average pixel level (APL) expose color microLED power limits versus LCOS?
How does average pixel level (APL) expose color microLED power limits versus LCOS?
The graph tracks display power consumption against average pixel level. Color microLED systems, because of inefficiencies, thermal efficiencies, EQ issues and etendue problems, can really only operate at low APLs, typically well under 10% APL if they must stay within the right power constraint or power budget. LCOS today is efficient enough to operate across all APLs within the needed power budget, which helps explain why companies outside China, such as Meta or Snap, already ship LCOS displays.
Traditional LCOS systems do not benefit from low APL in the same way. They illuminate the entire image very uniformly, so power is pretty much flat with content. That creates an intersection point between the two display technologies on the power-versus-APL graph. This power intersection point typically happens under about 10% APL. Below that level, color microLED power constraints are severe; above it, LCOS can keep operating within the required budget without APL limitations.
The good news is that efficient LCOS technology is already a shipping answer. For a high-resolution, full-color display with no APL limitations, LCOS is available today. It avoids the low-APL restriction that constrains color microLED light engines under the power budget. That makes LCOS the practical choice when small, lightweight glasses require performance across content. This is not a future promise but an existing display path for full-color images.
In this short video, you can learn:
* Color microLED systems can typically operate only well under 10% APL to stay inside the required power budget.
* LCOS operates across all APLs within the needed power budget and delivers full-color images with no APL limitations.
* The power intersection between LCOS and color microLED typically happens under about 10% APL, and Meta and Snap already ship LCOS displays.
📋 **Clip Abstract** Average pixel level determines whether a display stays inside the power budget: color microLED systems are limited to low APLs, typically well under 10%, due to inefficiencies, thermal efficiencies, EQ issues and etendue problems. LCOS operates across all APLs, and companies outside China such as Meta or Snap already ship LCOS displays for full-color images with no APL limitations.
About the speaker:
* Speaker: Edward Tang
* Company: Avegant Corp.
* Event: Eindhoven 2026
* Location: High Tech Campus, Eindhoven
#AveragePixelLevel, #LCOSDisplays, #ColorMicroLED, #PowerIntersection, #ARDisplays, #LightEngines
About this talk
→ AR’s Display Dilemma - LCoS v MicroLED · Edward Tang · MicroLED Connect + AR/VR Connect
More highlights from Edward Tang
- ▶ Why is one watt the absolute thermal and power limit for consumer AR glasses? · MicroLED Connect 2026
- ▶ How does average pixel level (APL) expose color microLED power limits versus LCOS? · MicroLED Connect 2026
- ▶ How can shrinking the light engine's pupil size more than double waveguide coupling efficiency? · MicroLED Connect 2026
- ▶ Why is shrinking LCOS pixel sizes below three microns critical for achieving larger fields of view in AR? · MicroLED Connect 2026

