This paper introduces a neural space-time model (NSTM) for computational imaging reconstructions from multiple sequentially captured measurements. The model jointly estimates the scene and its motion dynamics without data priors or pre-training, mitigating motion artifacts and resolving sample dynamics from raw measurements. NSTM's efficacy is demonstrated in differential phase-contrast microscopy, 3D structured illumination microscopy, and rolling-shutter DiffuserCam, showcasing its ability to recover subcellular motion dynamics and reduce misinterpretations of living systems caused by motion artifacts.

Ruiming Cao, Nikita S. Divekar, James K. Nuñez, Srigokul Upadhyayula, Laura Waller