Interp3R is the first method to estimate depth and camera poses at arbitrary time instants by interpolating pointmaps using asynchronous event data.
arXiv · March 17, 2026 · 2603.14528
The Takeaway
It bridges the gap between discrete image frames and continuous-time geometry, allowing for 3D reconstruction during the 'blind time' between shutter clicks. This is critical for high-speed robotics and XR applications where motion blur or gaps in vision cause tracking failures.
From the abstract
In recent years, 3D visual foundation models pioneered by pointmap-based approaches such as DUSt3R have attracted a lot of interest, achieving impressive accuracy and strong generalization across diverse scenes. However, these methods are inherently limited to recovering scene geometry only at the discrete time instants when images are captured, leaving the scene evolution during the blind time between consecutive frames largely unexplored. We introduce Interp3R, to the best of our knowledge the