Omnivision’s TheiaCel® image sensors integrate with NVIDIA DRIVE AGX
Press Release, 7 January 2026
At CES 2026, Omnivision announced that its TheiaCel® CMOS image sensors the 8-MP OX08D10 and 3-MP OX03H10 are now supported on NVIDIA’s DRIVE AGX Hyperion autonomous vehicle platform. The collaboration aims to accelerate the development of Level 4-ready, AI-powered autonomous vehicles by providing high-performance imaging solutions that meet the industry’s strict safety and cybersecurity standards.
The OX08D10 is Omnivision’s first sensor with 2.1-µm TheiaCel® HDR technology, combining LOFIC (lateral overflow integration capacitors) and DCG™ high dynamic range to capture LED lights without flicker across all driving conditions. Its superior HDR performance enables detection ranges up to 200 meters, making it ideal for automotive exterior cameras. The OX03H10, a 3-MP split-pixel sensor with TheiaCel® technology, delivers exceptional low-light performance and unrivaled LED flicker mitigation, providing clear, reliable imaging for parking cameras and other vehicle viewing applications.
Both sensors are now part of the NVIDIA DRIVE AGX Thor platform, which features NVIDIA Blackwell GPUs, Arm Neoverse V3AE CPUs, and the DriveOS 7 software stack. The platform supports multiple cameras, radars, lidars, and essential interfaces like GbE/10GbE and PCIe, ensuring robust system integration for advanced driver-assistance and autonomous functions. By combining Omnivision’s imaging capabilities with NVIDIA’s high-performance compute platform, automakers and developers can speed up design, testing, and deployment of next-generation autonomous systems — from research to commercial production.
“Omnivision’s TheiaCel® sensors offer excellent performance, and we’re proud to be part of the NVIDIA development platform,” said Dr. Paul Wu, Global Director of Automotive Product Marketing at Omnivision. The OX08D10 and OX03H10 sensors are available to order now, providing the imaging foundation for safer, more capable autonomous vehicles.
Compiled using AI
