A balloon bursts in an instant. A glass shatters into dozens of fragments. A golf club strikes a ball at a tremendous speed. By the time a conventional camera records its next frame, the most revealing part of the event is already gone.

This was the challenge that led to the idea behind Quanta-SloMo.

Traditional video frame interpolation tries to fill in missing moments by estimating what happened between recorded frames. That works for smooth motion, but reality isn’t that predictable. Objects bend, break, deform, and change direction too quickly for conventional methods to reconstruct accurately.

Our researchers behind Quanta-SloMo approached the problem differently.

We designed a Single Photon Avalanche Diode (SPAD) camera, a next-generation sensor capable of capturing extremely high-speed visual information. While SPAD footage is noisy, it provides thousands of additional observations between the frames captured by a conventional CMOS camera. Quanta-SloMo combines the strengths of both sensors to reconstruct what actually happenedbetween frames, even during highly dynamic events.

Reconstructing Moments Cameras Never Captured

In experiments involving bursting balloons, shattering cups, and spinning discs, Quanta-SloMo achieved a 21% improvement in PSNR at an extreme 100x interpolation factor, outperforming leading video interpolation methods.

From sports analytics and filmmaking to industrial inspection, robotics, and scientific discovery, this work points to a future where cameras don't just capture reality; they help recover the moments we never had the chance to see.

And if we can now reconstruct what happens between frames, what other hidden details of our world are waiting to be revealed?

Slow Motion Beyond Human Vision: How Quanta-SloMo Reconstructs 100x Faster Reality

{{ pageEducationalSlot.headingContent[index] }}