There’s a lot of buzz around the metaverse, but one of the challenges in enabling it is getting the right balance of performance, size and power consumption in the headsets and glasses that provide eye tracking and face tracking.
To address this, Omnivision has launched a BSI (backside illumination) global shutter (GS) image sensor, the OG0TB, which it said is the world’s smallest image sensor for eye and face tracking in AR/VR/MR and metaverse consumer devices. Aimed at some of the smallest and lightest battery-powered wearables, such as eye goggles and glasses, the new CMOS image sensor comes in a package size of just 1.64mm x 1.64mm, featuring a 2.2µm pixel in a 1/14.46-inch optical format (OF), and features 400×400 resolution and ultra-low power consumption.
The key to getting to this small size is the use of three-layer stacking, built on Omnivision’s PureCel Plus-S stacked die technology. In an interview with embedded.com, Devang Patel, the company’s marketing director for IoT and emerging segments, said, “For the new OG0TB image sensor, we are leveraging two of our technologies – CameraCubeChip and Nyxel – to deliver a product with small form factor and NIR sensitivity. We use three layer stacking to make the world’s smallest form factor global shutter image sensor which pushes the envelope on performance and size.”
The result is, according to the company, the world’s smallest ‘ready-to-go’ module with design flexibility to put the camera in the most ideal placement on some of the smallest and slimmest wearable devices. Ultra-low power consumption is critical for these battery-powered devices, which can have 10 or more cameras per system; the company said its OG0TB BSI GS image sensor consumes less than 7.2mW at 30fps.
The Nyxel technology enables a quantum efficiency (QE) at the 940nm NIR wavelength for sharp, accurate images of moving objects. The sensor’s high modulation transfer function (MTF) enables sharper images with greater contrast and more detail, which is especially important for enhancing decision-making processes in machine vision applications. The sensor supports a flexible interface, including MIPI with multi-drop, CPHY, and SPI.
Patel said that AR/VR headsets will be the primary application for the new image sensors. “Our AR/VR partners can use this off-the-shelf and connect directly.” He added, “All cellphone brands have a vision to bring AR glasses to market. In fact, it is said that your glasses will be your cellphone in the future. I do believe in that vision.”
The worldwide market for AR/VR headsets grew 92.1% year over year in 2021, with shipments reaching 11.2 million units, according to data from the International Data Corporation (IDC) Worldwide Quarterly AR/VR Headset Tracker. New entrants as well as broader adoption from the commercial sector will propel the market further as headset shipments are forecast to grow 46.9% year over year in 2022. In fact, IDC expects this market to experience double-digit growth through 2026 as global shipments of AR/VR headsets surpass 50 million units by the end of the forecast, with a 35.1% compounded annual growth rate.
Patel said the OG0TB GS image sensor is sampling already to lead customers and in mass production at the end of 2023. He expects actual end-products in market probably at the end of 2024.
The 1mm x 1 mm holy grail
According to Patel, Omnivision is already looking at developing a 1.2mm x 1.2 mm pixel size sensor. “The holy grail that the industry is looking for is 1mm x 1mm. It is doable, but there are things that need to be solved at system level – for example, you need pins for power, trigger and shutdown. So there are two phases involved in order to get to the holy grail: first, get below 1.6mm, and then work with partners to solve the system level challenges.”
- Enabling the hardware for the metaverse
- embedded news: Hot Chips, face authentication at IFA, and more
- Fourth-generation global shutter explained, and why embedded image sensors need better performance metrics
- FRAMOS integrates GMSL3 in embedded vision