Summary
This explores how Circle Optics and Deepnight are combining panoramic imaging and artificial intelligence to deliver continuous situational awareness across daylight, dusk, and nighttime conditions. Host Jennifer Sertl speaks with Circle Optics Director of Product Steve Feller and Deepnight CEO Lucas Young about the evolution of AI-enhanced low-light imaging, computational photography, and the future of autonomous perception. The conversation highlights how this partnership is helping defense, security, robotics, and critical infrastructure customers see more clearly, make faster decisions, and operate more effectively in challenging environments
Circle Optics is working with Deepnight to explore how AI based low light enhancement can extend the operational capabilities of panoramic imaging systems. The collaboration brings together Circle Optics’ wide field of view imaging technology with Deepnight’s computational approach to low light imaging, with potential applications in counter UAS, defense, security, autonomous systems, and infrastructure monitoring.
This article draws from a recent episode of 360 Pulse, the Circle Optics podcast, featuring Steve Feller, Director of Product at Circle Optics, and Lucas Young, founder and CEO of Deepnight. Listen to the full conversation and subscribe to 360 Pulse, available wherever you get your podcasts.
Extending Panoramic Imaging into Low Light
The technical challenge behind the collaboration is a fundamental one: cameras need light.
“Circle Optics builds these wide field of view panoramic cameras that try to capture the whole scene, but they require light,” said Steve Feller, Director of Product at Circle Optics. “That’s really a limiting factor if you’re looking at security applications, military applications, defense, things like that.”
Circle Optics has evaluated different approaches to extending its imaging capabilities into low light environments, including infrared technology. The company began working with Deepnight as another potential approach to improving performance when available light is limited. Deepnight uses algorithms to enhance the visual signal captured by digital cameras, allowing more useful information to be extracted from low light imagery.
Applying Lessons from Computational Photography
Deepnight CEO Lucas Young came to night vision after working in consumer imaging.
Young previously worked on camera technology for augmented reality devices at Meta and later on camera algorithms for Google Pixel smartphones. Those experiences shaped his thinking about how imaging performance can be improved through a combination of optics and computation.
Smartphone cameras operate with significant physical constraints. Their lenses and sensors must fit into compact devices, yet computational photography has enabled them to produce images that would be difficult to achieve through optics alone.
Young saw an opportunity to apply a similar principle to night vision. Traditional military night vision commonly relies on image intensifier tubes, which amplify available light through an analog process. As Young explained during the conversation, the light entering a conventional image intensifier is not encoded as a digital signal. That limits the ability to apply the type of computational processing now common in consumer imaging. Deepnight is pursuing a different architecture based on digital sensors combined with algorithmic processing.
Young describes the approach as bringing together optics and computation to extract better performance from limited available light.
Low light computational imaging can reveal visual information that is difficult to distinguish in the original camera image. This before and after example demonstrates how algorithmic processing can improve the usability of imagery captured under limited lighting conditions.
Expanding the Operating Environment
For Circle Optics, the immediate interest is extending the environments in which panoramic imaging can provide useful information. “What their algorithms bring to the table is an existing system with its limitations,” Feller said. “We can apply specialized algorithms and analytics to provide a better result than we could without them.”
There are engineering tradeoffs.
Additional processing can improve image performance but may also increase computational requirements or latency. Depending on the application, engineers may prioritize image quality, processing speed, power consumption, or other system requirements.
Feller sees those decisions as part of a broader shift toward designing imaging systems for machines as well as people. He has worked extensively in computational imaging, including the development of optical systems optimized for digital consumption. As cameras become increasingly integrated with AI, autonomy, and machine vision platforms, the relationship between the optical system and the software processing its data becomes more important.
Potential Applications in Counter UAS
Counter UAS is one area where Circle Optics and Deepnight are working together. The companies are deploying an integrated Counter UAS capability across U.S. Air Force installations, combining Circle Optics’ panoramic imaging with Deepnight’s low light computational imaging technology. The deployment addresses the need for surveillance capabilities that can operate across changing lighting conditions. Read the full press release: Circle Optics and Deepnight Deploy Integrated Counter UAS Capability Across U.S. Air Force Installations.
Young sees cost and manufacturing capacity as important limitations of conventional night vision. Image intensifier tubes are highly specialized components, and according to Young, only two foundries in the United States produce them. Deepnight’s approach combines digital sensors with algorithmic processing, creating another potential path for applications where conventional night vision may be too costly or difficult to deploy at scale.
Small UAS illustrate the challenge. Young noted that a Gen 3 image intensifier tube can cost approximately $3,000, making conventional night vision difficult to incorporate into UAS platforms whose total target cost may be only a few thousand dollars. Beyond UAS and Counter UAS, he sees potential applications in automotive systems, autonomy, surveillance, and other environments that need to operate effectively in darkness.
For Circle Optics, the collaboration is about extending the environments in which panoramic imaging can provide useful information. “It’s not just that our systems work better,” Feller said. “It solves the needs of the customers we have and the customers we’re going to have.” By bringing panoramic capture and computational low light imaging together, Circle Optics and Deepnight are exploring how to provide meaningful situational awareness across a broader range of operating conditions.
