///Optimizing Glass AI Network
Loading Data .  .  .
Name: Glass Imaging®
Business Genre: Artificial Intelligence
Value: Delivering DSLR quality to smartphones
Location: Los Altos, California USA

XR

Problem
Extended reality devices face severe imaging constraints. Passthrough cameras must be small, wide-angle, and low-latency, while operating across unpredictable lighting conditions. The result is often soft, noisy, or distorted imagery that breaks immersion and limits the usability of mixed reality experiences.
Our Solution
GLASS AI corrects optical distortions and enhances clarity in real time, enabling passthrough quality that matches the human expectation of presence. By reversing lens aberrations and suppressing noise at the RAW level, GLASS AI helps XR devices deliver visual fidelity without adding size, weight, or latency.
Before
After

Implementation

Edge
Server
Cloud
Hybrid

Edge

GLASS AI is optimized for edge AI-enabled chipsets and integrates directly into OEM imaging pipelines. Running natively on-device, it delivers real-time photo and video enhancement within the native camera app.

Server

GLASS AI can be deployed on OEM-defined servers or workstations to support local workflows. Captured RAW data is transferred and processed with full-fidelity models, giving OEMs a flexible architecture tailored to their performance and workflow requirements.

Cloud

When local compute is limited, GLASS AI can be deployed in the cloud. RAW image data is securely uploaded, processed at scale, and returned with enhanced quality—unlocking advanced imaging capabilities on lightweight devices

Hybrid

OEMs can combine edge, server, and cloud deployments to fit their specific use cases. Hybrid architectures leverage on-device compute alongside cloud or server resources, ensuring flexibility and consistent imaging performance across conditions.

GLASS AI Photo Modes

Neural ISP

Our Neural Image Signal Processing block receives single-frame or multi-frame burst RAW data and reverses lens and sensor imperfections to deliver unmatched image quality, even in the most demanding conditions.

Neural Zoom

Our Neural Zoom block fuses temporal information from a burst of RAW frames to reconstruct high-resolution detail beyond the limits of optical magnification. By extracting sub-pixel data across frames, it delivers consistent sharpness and clarity at any focal length.

Neural HDR

Our HDR block extends dynamic range by fusing multiple RAW exposures while preserving the sensor's native color response and noise characteristics. The result is accurate tone and highlight recovery without compromising texture or realism.

Neural Night

Our Neural Night block combines temporal fusion with physics-based denoising to extract maximum signal from low-light RAW captures. The result is remarkably clear images in the darkest conditions while preserving natural color and texture.

Neural UDC

Our Neural Under Display Camera, UDC, correction block models and reverses the optical degradation introduced by the display layer, restoring detail, clarity, and color fidelity while preserving the seamless device design that under-display cameras enable.

GLASS AI Video Modes

Neural Video

Our Neural Video block processes RAW video streams in real time, applying physics-based correction to reverse lens and sensor imperfections frame-by-frame while maintaining smooth, artifact-free playback.

Neural Zoom

Our Neural Zoom for video accumulates temporal information across frames to resolve detail beyond optical limits, delivering stable, high-magnification footage without the noise penalty of digital zoom.

Neural Night

Our Low Light video block leverages temporal information to suppress noise while preserving motion and fine detail, producing clean footage in challenging lighting without blur or artifacts.

Neural Video

Our Neural Video block processes RAW video streams in real time, applying physics-based correction to reverse lens and sensor imperfections frame-by-frame while maintaining smooth, artifact-free playback.

Neural Zoom

Our Neural Zoom for video accumulates temporal information across frames to resolve detail beyond optical limits, delivering stable, high-magnification footage without the noise penalty of digital zoom.

Neural Night

Our Low Light video block leverages temporal information to suppress noise while preserving motion and fine detail, producing clean footage in challenging lighting without blur or artifacts.
We train device-specific models and deploy them wherever your workflow requires: on-device for real-time capture, or in post on GLASS-hosted infrastructure or in your local pipeline.

Co-Designed AI + Optics

We unlock smaller camera modules, lighter optical assemblies, and better imaging performance that wouldn’t be achievable without GLASS AI. The result is more immersive experiences in sleeker, more comfortable headsets.
Contact Us
GLASS AI