Kodak boosts digital camera sensitivity
Kodak says digital cameras will work better in dim conditions with a straightforward but significant change to image sensors. Photos: Kodak's highly sensitive sensor Podcast: Help for the camera klutz
- Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
The new technology, to be unveiled Thursday and used in products in 2008, increases light sensitivity of existing image sensors by a factor of two to four, said Mike DeLuca, marketing manager for Eastman Kodak's image sensor solutions group.
Translated into photography terms, that means a camera's shutter speed could be cut in half or a quarter, helping cut camera shake or motion blur problems. Alternatively, it could let photographers shoot in low light with less image "noise"--the pesky multicolor speckles that degrade photographs.
"That's the real bane, when you think about it. There's just not enough light to collect," said IDC analyst Christopher Chute. Of Kodak's new method, he said, "It's pretty revolutionary."
Light sensitivity has become a serious problem in digital cameras, particularly as higher megapixel counts have increased noise levels in image sensors.
And unlike some efforts to improve digital cameras, the new Kodak technique can be applied to any existing image sensor, leading Kodak to hope it will be able to license the high-sensitivity technology far and wide.
"We absolutely feel there is a big opportunity for this...to become a new standard in the industry," DeLuca said. "We really want to propagate this out as far as the market feels it should be taken."
Kodak's new method better reflects how human eyes actually work, separately registering color and brightness information--and devoting more pixels to brightness, where the human eye is sensitive to detail.
The company's technology doesn't require any new fundamental changes to the heart of the image sensor, where a grid of electronic detectors converts incoming light first into electric signals and then digital information. Instead, the new technology adds some neutral "panchromatic" pixels to the usual array of red, green and blue pixels in the grid, then uses a different software algorithm to reconstruct the full-color images from the sensor output.
Pixel patterns
In nearly all of today's digital cameras, each sensor pixel detects either red, green or blue, with those particular colors placed in a quasi-checkerboard arrangement called a Bayer pattern, after the Kodak engineer, Bryce Bayer, who developed it. Every second pixel registers green light, while the remaining pixels capture either red or blue.
With a 12-megapixel Bayer-pattern sensor, 6 million pixels are green, 3 million are red and 3 million are blue. Software reconstructs the full-color image so each pixel has a red, green and blue component through a process called de-mosaicing.
In the high-sensitivity pattern--as yet lacking a formal name--half the pixels capture red, green or blue color information, while the other half are panchromatic pixels that capture only the brightness. So a 12-megapixel sensor would have 6 million panchromatic pixels, 3 million green pixels, 1.5 million red pixels and 1.5 million blue pixels.
Different software algorithms, which typically run in an image processing chip within the camera, must be used to reconstruct a full-color image, DeLuca said. Today that software is a little larger than the Bayer demosaicing software, but with optimization that Kodak now is working on, it should become "comparable."
As a way to deal with low-light shooting problems, Chute said, Kodak's technology is a compelling alternative to expensive image stabilization, which moves lens elements or the sensor to counteract camera shake, or post-processing, which uses software to try to reduce image noise.
From prototype to product
Kodak plans to release the technology in several ways. First, it will build it into its own cameras, DeLuca said. Second, it will offer it and the accompanying software with the image sensors it sells to other camera makers--both the CCD (charge-coupled device) sensors it builds itself and the CMOS (complimentary metal oxide semiconductor) sensors that IBM and Taiwan Semiconductor Manufacturing Company (TSMC) build for it.
And it could license the technology to other companies. "We will engage in conversations with other companies as appropriate," DeLuca said.
The first prototype sensors intended for production use are scheduled to arrive in the first quarter of 2008, he said. Typically, production-quality chips arrive three to six months after, he added.
Kodak's technology has the potential to spread widely, Chute said. "But first it needs to prove itself. It can't just be a lab rat." And another possible obstacle is intellectual property. Kodak hasn't been afraid to file patent infringement suits against camera makers Olympus and Sony.
It's good that it's relatively easily to add the technology to existing cameras, Chute added, but Kodak shouldn't expect its technology to spread like wildfire. "Camera manufacturers and the photo industry tends to be very conservative. They go with what they have and make it better."
Click here to hear Stephen Shankland in an interview with News.com's Charles Cooper: