Much has changed in camera design over the years, but snapping photos and shooting video still invariably requires a lens to capture light and focus on a subject. But if a camera could somehow replicate this process digitally, making relatively chunky lens attachments completely unnecessary, what would be left to look at? Well, going by new research underway at Rice University, not really much at all. Engineers have produced a functional camera that is thinner than a dime, raising the possibility of tiny, flexible versions that could one day be embedded in everything from your wallpaper to your credit card.
Aptly dubbed FlatCam, the device is the handiwork of electrical and computer engineers at Rice University who are looking to challenge what an imaging device looks like. While smartphones have brought about a huge reduction in the size of cameras, they do still require a lens, which necessitates a three-dimensional cube-like shape. This ties the possible surface area of the sensor to the thickness of the device, and because sensor size dictates the ability to collect light, means performance is often compromised in smaller cameras.
This is one of the design challenges the team is seeking to overcome, looking to harness the enhanced light collection of a larger sensor, while maintaining a compact form, albeit in a different shape. The way the've approached this is by turning to one of the camera's very early iterations, the pinhole camera design. Where these first cameras had a single hole, allowing light to pass through and project onto a surface, the FlatCam has a mask over its off-the-shelf sensor dotted with a grid-like pattern of different aperture sizes.
These allow different light data through onto the sensor, which is then relayed to a desktop computer where algorithms use the information to construct an image. In its current form, FlatCam can produce with a resolution of 512 x 512 in seconds, though the researchers say that as manufacturing techniques and the algorithms progress, so too will the resolution of the images.
While the images snapped with FlatCam won't be winning any photography contests, the researchers say it is still early days and the concept opens up all kinds of possibilities, with work already underway on the next generation.
"We can make curved cameras, or wallpaper that's actually a camera," says Richard Baraniuk, professor of electrical and computer engineering at Rice University. "You can have a camera on your credit card or a camera in an ultrathin tablet computer."
Other potential uses for super-thin cameras include disaster relief and security applications. And as there is no lens involved, the costs associated with manufacturing and then attaching them to a camera would be avoided, making the FlatCam a cheap alternative.
You can read a paper outlining the research online here.
The video below features the researchers giving an overview of the device.
Source: Rice University
I have often wondered why one couldn't use a parallel or fan beam multiple hole collimator like what is used in Nuclear Gamma cameras.
When I was working in Nuclear Medicine (and had these Ideas) over a decade ago, this was (probably) beyond the realm of commercial possibility, but so were 42 Megapixel CMOS sensors.
Like the above Idea about using Graphene (that may not work because the graphene is way smaller than the wavelength of light causing severe diffraction artefact) with today's technology in the realm of micromachining, it should be possible to make a highly efficient parallel/convergent/divergent collimator at the same scale as a high resolution camera sensor.
The benefit with collimation would be that the image processing time should be simpler than using the method with multiple pinholes as reported in this article.
Obviously IF a lytro effect were desired, a high resolution sensor could be divided up into multiple effective "micro-collectors"
BUT Collimation MAY allow infinite depth of focus.
For industrial applications, the benefit would be all depths of field would simultaneously be in focus. With maximum light collection.
For a 360x360^o Spherical view the sensor, and integrated collimator would be manufactured as a sphere, perfect image registration, with simple mapping for flat images.