Researchers Found a Way to Extract Detailed Slo-Mo Videos From a Single Blurry Photo

By Andrew Liszewski on at

The Phantom high-speed cameras used by YouTube channels like The Slow Mo Guys to capture mesmerising slow-motion footage are obscenely expensive. But soon, thanks to a new technique that can extract slo-mo videos from a single photo, the smartphone in your pocket could outperform even a high-speed camera that costs well over $100,000 (£75,000).

Researchers from Switzerland’s École Polytechnique Fédérale de Lausanne’s (EPFL) Engineering Mechanics of Soft Interfaces Laboratory worked with another team from Harvard University’s SMRLab to develop a way to analyse blurry photos and generate step by step frames representing detailed slices of the original motion that previously appeared to be frozen in time.

Simply being out of focus is one cause of a blurry photo. But more often than not, thanks to excellent auto-focus technology on modern cameras, blurred results are caused by a shutter speed that’s not fast enough to freeze the action in front of the lens. Quickly wave your hand in front of a camera with a seemingly fast one-second shutter and you’ll be left with a smeared image of your hand as the entire waving motion is captured in a single shot. An image sensor’s ability to translate a second of motion (or longer) into a single frame is the key to how this new Virtual Frame Technique (VFT) imaging method works.

But there’s a catch. That blurry photo of your friend’s dancing at the club last night can’t be reverse-engineered into a short video of everyone having a good time. The VFT technique is limited to capturing and deciphering only high-contrast black and white images.
Parts of an image that appear blurry actually contain sufficient data to reproduce multiple frames of a phenomenon in motion. Photo: EPFL (The Optical Society)

By limiting the colours being recorded to just black or white pixels, the researchers were able to take advantage of a sensor’s bit-depth (its ability to see thousands of different intensities of a given colour) to dramatically boost the frame rate they could capture in a single image. The VFT technique also requires the subject’s lighting to be almost perfectly uniform and intense for optimal results, limiting its use to studio or laboratory environments where the conditions can be precisely controlled.

Slow-motion footage of tape being peeled off a surface.GIF: EPFL (The Optical Society)

The resulting footage isn’t as captivating as a spectacular slow-motion explosion, but it can be still quite valuable for researchers who are often trying to study only a very specific part of a given phenomenon—such as how the adhesive on cellophane tape behaves as it’s being peeled off a surface, as demonstrated in the GIF above. The researchers believe the camera technology in a modern smartphone, or even just a basic point-and-shoot, could be used to capture phenomenon at over a million frames per second using this technique, eliminating the need to spend a big chunk of funds on elaborate photography gear. [The Optical Society via EurekAlert!]