Startup Light “reimagines photography” - Will phones finally compete with cameras for image quality?
posted Saturday, April 18, 2015 at 12:53 AM EDT
Smartphones have pretty much devoured the point & shoot camera market, but their somewhat wide-angle lenses and inability to zoom has left room in the market for pocket cameras. Looking upmarket, smartphones' (usually) itty-bitty sensors haven't been any kind of a challenge to higher-end cameras when it comes to image quality.
Image quality and zoom might not be such big stumbling blocks in the near future, though, thanks to technology from a startup company called Light, who are only just now beginning to talk about their technology publicly.
Light's approach is built around two key concepts: Make up for the small size of smartphone sensors simply by having more of them, and substitute an array of different focal length "prime" lenses for the digicam's zoom. Shots from multiple camera modules are stitched together into a higher resolution image via fancy software, bringing not only significantly better image quality, but also some capabilities reminiscent of Lytro's "light field photography."
If this is all sounding a little familiar to you, it might be because of similarities to the technology developed by LinX, an Israeli imaging startup that was just purchased by Apple earlier this week.
While the LinX approach is also based on the idea of stitching images from multiple camera modules into higher-quality results, Light goes a step or two further, providing multiple focal lengths (35, 70, and 150mm equivalents) across the array of modules. LinX uses a conventional lens/sensor stack, with the lens directly on top of the sensor, while Light relies on folded optics to achieve the longer focal lengths while still maintaining a slim profile. In Light's scheme, "zooming" is accomplished simply by taking image data from modules with longer or shorter focal length lenses built into them.
Camera makers have used folded optics for many years now, as a way of shoehorning zoom lenses into very slim camera bodies. With folded optics, the sensor/lens system is essentially rotated 90 degrees relative to the front of the camera. The lens/sensor assembly is located up one side or across the top of the camera body, with a 45-degree mirror or prism reflecting/refracting light entering the front of the camera body down into the rest of the optics/sensor module.
To the best of my memory, the first digital camera using folded optics was the Minolta Dimage X, released in March of 2002, although the concept was first introduced in the Tessina sub-frame 35mm camera, first made in 1960. (I'm sure our astute readers will correct me if I've remembered any of that incorrectly ;-)
The photo at right shows one end of the Dimage X, with a marketing decal in place showing how the sensor and optical components are arranged inside. It was a pretty unique approach at the time, so Minolta wanted to show it off a little. The red line traces the path of light from the front-panel lens opening at top right down to the sensor at the bottom.
Comparing the length (height) of the lens assembly in the diagram to its width, it's easy to see how much can be gained in terms of camera body thickness.
Full details about Light’s folded optics design aren’t yet clear, but in speaking to MIT Technology Review, Light cofounder and CTO, Rajiv Laroia, revealed that individual camera modules will be able to become active or inactive based on the needs of the image being captured. He further explained the mirrors inside of the modules will be capable of moving independently to better record the data required to capture the best image.
We found that last bit about movable mirrors particularly interesting. Traditionally, the mirror or prism in a folded-optic camera has been fixed, but the ability to tilt it from side to side or up and down would have obvious application for image stabilization. Given that you have the mirror there already, adding a piezoelectric actuator for image stabilization seems almost a no-brainer, given how little incremental space would be required for one.
Laroia himself didn't mention IS, but rather talks about being able to "better record the data required to capture the best image." We're not sure what that might mean; perhaps making sub-pixel shifts between sequentially-captured images, similar to what what we've seen in the Olympus E-M5 Mark II's super-resolution mode? Perhaps changing the "aim" of the separate lenses so their lines of sight will converge at the primary subject's focal distance? It's also possible that the ability to angle the view of each module slightly could help capture depth information, to allow for post-capture refocusing, as is possible with Lytro cameras today.
We don't know, and it's of course likely that he was deliberately being a bit vague, but we'll try to learn more and report back to our readers on what we find.
According to MIT Technology Review, Light will announce a deal next Tuesday with Taiwanese electronics manufacturing giant Foxconn. The deal is said to be a licensing arrangement between the two companies that will allow Foxconn to implement Light’s technology into mobile devices, in exchange for an undisclosed investment in the company itself.
Of course, nothing comes for free, and Light CEO Dave Grannan told Technology Review that Light's technology would require a cell phone incorporating it to be about as thick as a current model in a case. This is a bit at odds with current trends, and of course, you'd want to put your Light-equipped phone in a case as well, making it thicker yet.
It seems that Light isn't focused (no pun intended :-) solely on smartphones though; they referred to the 16-module array shown above as something that might be used in a self-driving car or a home security camera. Still, Light has the words "reimagine photography" emblazoned on their home page, so it seems that photography is a big part of their plans.
Overall, we're very intrigued by what Light's up to with their tech; hopefully we'll be able to bring you more details on it sooner than later. Stay tuned...
(Special thanks to Ganon Burgett for his help on this piece!)