MIT researchers crack SPAD 3D camera limits with new method

A team of researchers from MIT and Boston University has developed a breakthrough computational imaging method that overcomes critical limitations of single-photon avalanche diode (SPAD) 3D cameras. These cameras, known for their extreme sensitivity and time resolution, have struggled in real-world applications due to two major challenges: 'pile-up' distortions in high-photon-flux conditions and severe data bottlenecks from raw photon data generation.

The researchers' solution combines free-running capture with an analysis-by-synthesis software pipeline. This approach reliably captures scene distance and reflectance across a wide range of illumination conditions, even in bandwidth-constrained scenarios. Their work, presented at IEEE ICCP 2026 where it won Best Paper Award, demonstrates that SPAD cameras can now operate effectively in bright daylight conditions that previously caused data corruption. The method enables high-resolution imaging in real-world applications like autonomous vehicles, robotics, and medical imaging where traditional SPAD cameras failed due to lighting constraints.