New Kodak sensors see well in dark
- — 15 June, 2007 12:47
Shooting in low light situations can be a source of frustration. Kodak aims to mitigate that with its new image sensor technology, expected to be available in products in 2008.
The company describes the technology as a "groundbreaking achievement," and if it works as described, that might not be hyperbole: The new technology makes CCD and CMOS image sensors more sensitive to light. That means sharp pictures, even in low-light environments.
The newly introduced technology could transform digital photography for both professional and casual shooters alike.
"We have a new technology for how to manufacture image sensors in a way that increases the sensitivity of the device to light by two to four times what's available today," says Mike DeLuca, Eastman Kodak market segment manager for professional and applied imaging.
That translates to one or two photographic aperture stops, meaning photographers can expect better performance and reduced motion blur when taking pictures in low lighting.
The company is first developing CMOS sensors -- common in digital still cameras and camera phones -- but this technology is expected to be applied to CCD sensors as well.
"Our first sensor with this technology will be available for sampling to camera manufacturers in the first quarter of 2008," says DeLuca.
With current technology, photos taken in challenging lighting are "either noisy, or you have to leave the shutter open for a long period of time," DeLuca says. "This can result in a blurry image if you shoot a picture of mobing images." While a flash can sometimes be used to add light to the scene, it may not always be available or appropriate.
Kodak's new approach builds on the image sensor technology widely in use today, which is based on the "Bayer Pattern," an arrangement of red, green, and blue pixels introduced by Kodak scientist Dr. Bryce Bayer in 1976. The pixels work together with the image sensor to collect light.
"It's the de facto standard in the industry," DeLuca says of Bayer. Half of the pixels on the sensor collect green light, while the remaining pixels collect red and blue light; software later reconstructs the a full-color signal for each pixel.
Enter Kodak's new high-sensitivity image sensor.
"We're introducing a fourth pixel in addition to red, green, and blue," DeLuca says. "This one is clear--panchromatic--so all wavelengths of light go through and are detected by the pixel. Those panchromatic pixels are more sensitive, because they do not filter out any light [striking the sensor]. We can then use those panchromatic pixels to increase the sensitivity of the sensor, and use the colour pixels to collect the colour information that ends up in the final image."
Currently, a colour filter rests near the top of a pixel. "Light would come from above, and then go through this colour filter to go out, so only the red light or blue light or green light passes," DeLuca says. "And then the remaining light is detected by the pixel structure."
But Kodak is changing the filter. From a manufacturing perspective, it's not a dramatic fix. "All we need to do is to change the configuration in that color filter layer, leaving the rest of the pixel unchanged," DeLuca says.
Kodak's new image sensor technology has a software component as well. The high-sensitivity sensor, with its new approach to patterns, requires revised software algorithms to generate a full colour image from the raw information coming off the sensor.
"There's additional work that needs to happen to integrate the new algorithms--which will continue to be developed--into the camera," DeLuca says. "We're not changing the fundamental structure of the silicon, which gives us the opportunity to deploy this broadly. This technology has the opportunity to become a new standard."
Kodak has applied for more than a dozen patents in this space, tied specifically to the high-sensitivity sensor technology. DeLuca says the patents include "the concept of using panchromatic pixels plus red, blue, and green to make a full colour image; the specific patterns of red, blue, and green pixels; and software algorithms and techniques for processing the data."