150,000 frames per second in 3D: Record-setting imager for fast, one-shot events

A new imaging technique can record the 3D shape of a moving object at up to 150,000 frames per second, fast enough to follow a projectile striking a turbofan blade, a falling maple seed, or the vibrations of a loudspeaker. The authors say it sets a new world record in 3D profilometry.

The method, called C-BLIP, short for chromatically band-limited illumination profilometry, was developed by a team led by Jinyang Liang at the Institut National de la Recherche Scientifique (INRS) in Quebec and announced Nov. 12. Profilometry measures an object’s surface shape, and C-BLIP applies it to events that unfold too fast and too unpredictably for older tools.

Why fast 3D is hard

Many fast events, such as collisions, can’t be repeated on demand, so a camera has to capture each one in a single take. Most 3D profilometry works by projecting patterns of light onto an object and reconstructing its shape from how the patterns distort. Researchers describing related systems note that this approach is capped by two things: how fast the projector and camera run, and how many patterns each 3D frame requires.

The INRS team adds that existing ultrafast 3D systems are often held back by slow acquisition, computationally heavy image reconstruction or the high cost of specialized equipment.

A cheaper, lighter approach

To get around those limits, the team paired a cost-effective optical design with eye-safe LED illumination and a lightweight reconstruction method. The researchers say that combination makes ultrafast 3D imaging more practical and accessible for a wide range of research environments, not just specialized labs.

What C-BLIP captured

In experiments, C-BLIP recorded a projectile colliding with a turbofan blade, the aerodynamics of a falling maple samara (a winged seed that spins as it drops), and the vibrations of a loudspeaker membrane. The team also detected ultrasonic vibrations on an object’s surface and extracted acoustic information from them, recovering a spoken message.

“What excites us most is the ability to observe phenomena that were previously invisible,” said Cheng Jiang, a postdoctoral researcher at INRS and the study’s first author.

Putting 150,000 in context

The record claim is specific to 3D profilometry. Earlier ultrahigh-speed systems in this field reported about 100,000 frames per second, including a previous band-limited illumination system presented by Liang and a 2025 deep-learning approach from another group. Imaging methods built for other purposes, such as those that capture light in flight, run at billions or trillions of frames per second, but they don’t measure surface shape the same way.

Where it could be used

The researchers see uses in mechanics and aerodynamics, microfluidics, high-rate material testing and additive manufacturing.

“With C-BLIP, we can continuously capture fast, nonrepeatable three-dimensional phenomena that were previously difficult to observe and quantify,” said Liang, who holds the Canada Research Chair in Ultrafast Computational Imaging.

The team is now working to optimize the system by adding GPU-accelerated processing, which could eventually allow real-time visualization of these fast events.