New materials help ensure that what you see in the viewfinder is finally what you get in the image
Katie Thrasher
It was the moment of truth. Ethan Tseng and a team of scientists from Princeton University and the University of Washington were testing their camera, based on novel ideas and technology in the form of metamaterials. Then disappointment: the first images were useless.
Tseng and a few other scientists joined forces on the concept in 2019. They wanted to know, “Can we make a truly tiny camera a reality?” Such a camera – if it worked – could, for example, let doctors explore the innards of a living human without relying on complicated invasive procedures. Given their tiny size, such a camera could cut the expense and effort involved in sending telescopes to space.
Tseng and the team toiled away for years, seeking a camera the size of a grain of salt that could capture clear, color images. While there is no shortage of small cameras — in smartphones, medical endoscopes and more — image quality drops in proportion to size. New ideas were required.
The metasurface option
One such idea was the use of metasurface optics.
In normal lenses, the thickness and material of the lens changes the speed of light and causes it to bend – or refract.
In contrast, metamaterials rely on different properties. Tseng’s team redirected light by relying on very tiny, repetitive nano-antennas whose sizes are less than the wavelength of light. A thin sheet of these nano-antennas could be designed to bend light in a similar way as a bulky glass or plastic lens.
But metamaterials often exhibit aberrations that distort images, making them less effective than traditional lenses for producing images with a wide field of view – the kind seen in regular cameras. Tseng and team tackled this problem by using new computational methods. They used learning algorithms, the same kind used to train neural networks for modern artificial intelligence, to co-design both the metasurface and the software post-processing.
Because the camera that Tseng and his team built uses only a single metasurface as the lens for directing and focusing light onto the camera sensor, they were able to build a camera the size of a grain of salt.
Many a slip
Asked during a chat with Truly Curious if he had a “eureka” moment in the development of the camera, Tseng deadpanned, answering, “I don’t know if eureka has ever really happened since Archimedes.”
He believes science does not usually happen in that way. Like the space race, their project faced disaster and delays and long days and nights of frustration. While their attempts did not go up in flames, or crash, one of their initial failures was that they did not think beyond the center of the field of view.
Their first series of simulations showed good image quality results, but when they physically created the first lens and tested it, the images that came back were blurry around the edges. This taught them that simulations do not always match reality.
But the team’s most epic fail occurred at the very end of their research. They knew that the lens and the software worked, that they had produced amazing pictures, but they wanted to do one more test. However, they were dealing with variables beyond their control.
Each camera test requires a newly fabricated lens. The fabrication process time varied due to COVID-related delays from a lack of material availability to plant shutdowns. And it turns out, the final product can vary in quality as well. Tseng compares it to baking a cake: “You have the same recipe and the same ingredients, but unless the conditions are exactly the same each time, you get a slightly different cake.”
That had an effect on the imaging results from that final test.
“It was so bad. The images were unusable,” Tseng said.
Pulling it all together
The team wrote their research paper based on their best run and the images that they had. It had more than enough material to show that a camera the size of a grain of salt could produce clear images in full color. That’s 400 nm to 700 nm, which is the visible spectrum. Besides, it had a 40-degree wide field of view, an f-number of 2, and a maximum aperture of 0.5 mm. While not quite ready to take on the Hasselblads of the world, their camera was solid proof of concept.
Tseng knows his work is not complete, but their research had taken a crucial first step.
“It all goes back to how and why you do science,” he said, “You never really know until you finally see the results. You give it your best and you hope that it yields great things.”
Well, he has at least one tiny great thing.
Katie Thrasher is a writer, editor, researcher, and student in the science writing program at Johns Hopkins University. She also holds a degree from the College of Human Ecology at the Ohio State University.
Click here for the original paper.
