Kodak announces a new digital imaging filter that would replace Bayer filtering and give 2 to 4 times greater sensitivity in low light:

http://apnews.excite.com/article/20070614/D8POIKM80.html

Irv

... but presumably sacrifice chrominance resolution. The original Bayer array weighted the resolution data towards green as a compromise, while this approach (and there are several patterns of RGBY being discussed) weights the resolution towards luminance (for improved sensitivity) while losing color resolution. It'll all be in the algorithms to extract the image see here for a Q&A.

It's not immediately apparent how this will work out in practice, given how the human eye/brain combo actually works, and the inevitable suitability for one approach for some situations and another for others. A recent Scientific Americanarticle(subscription required or hard copy!) shows how science is unravelling the processing (into "movies") that happens in the eye before any info is sent to the brain - the brain doesn't get anything like RGB per pixel as its input!

Andy

That makes sense. So it's possible (or even probable) that Kodak's panchromatic filter system(s) will not provide as good a color reproduction as the current Bayer system does. However, could this new filter system be a step towards that long-awaited "B&W" digital camera? Many of us have wished for a monochromatic digital camera that devoted all of its pixels to luminance and had no color filtering at all. I know that the Kodak filtering system isn't monochromatic, but at least it's 50% panchromatic (unfiltered), which is a lot closer than the Bayer system of 0% panchromatic. Could we expect to achieve better B&W resolution and tonality with this new system than current chips provide?

Irv

I'm a total layman when it comes to Bayer Arrays etc. But it sounds to me that tech just adds another set of cells into the array, luminance : "Kodak’s new proprietary technology adds panchromatic, or “clear” pixels to the red, green, and blue elements that form the image sensor array." Doesn't this mean that instead of RGB we would get RGBL(uminance). IOW, no loss in color rendition, but a gain in luminance. Or am I missing something? Smaller pixel size resulting in higher noise, maybe?

Here is a nice blog entry on it....

http://theonlinephotographer.typepad.com/t...s_brillian.html

Feppe, In the new array there are sensor elements that have no color filter above them and these measure luminance (Y) [ where they say clear, actually they mean colorless - the others are clear(meaning transparent), but colored; but I digress ]. Where these colorless filters are is where there used to be color filters - so there are now fewer sensor elements that detect specific colors, and color information will be more dependent on interpolation than it was before (i.e. more "guessing" of values rather than measuring). The usual Bayer array is RGBGRGBG with 4G, 2R and 2B per 8 pixels. In the new array it's RGBGYYYY, so there's only 2G, 1R and1B per 8 pixels (or 1/2 as much color information.