Moment-Reenacting: Inverse Motion Degradation with Cross-shutter Guidance

An arXiv paper introduces Moment-Reenacting, a unified framework for reconstructing high-speed video from blurred and rolling-shutter images.

It uses synchronized blur and rolling-shutter captures, a real-world aligned dataset, and a dual-stream reconstruction network.

The method reduces temporal and spatial ambiguities in motion-degraded scenes and improves recovery of clearer fast-motion frames.

By combining two previously separate imaging problems, it could strengthen computational imaging for low-light and complex high-speed capture.