Slashdot Mirror
The Future of Digital Camera Technology
An anonymous reader writes "CNet News has an interesting look at where digital camera technology is headed now that the megapixel buzzword can be put to rest. From the article: 'In compact cameras, I think that the megapixel race is pretty much over,' says Chuck Westfall, director of media for Canon's camera marketing group. 'Seven- and eight-megapixel cameras seem to be more than adequate. We can easily go up to a 13-by-19 print and see very, very clear detail.'"
I lay out books and magazines for a living, and the vast majority of images that come to us are 300 dpi jpegs, or tiffs and eps's converted FROM jpegs. We routinely print oversized glossy material, which uses trim sizes greater than 8x10 in virtually all cases. We have had no quality issues, and I speak from a production environment.
Resolution is more important than compression method. Ten times out of ten I guarantee you couldn't tell the difference between a RAW file and a Fine JPEG image.
The color problems you speak of are caused by the camera, not jpeg itself. The jpeg file format is capable of rendering in any color space, and provides excellent color reproduction. Problems can arise from the internal jpeg engine in the camera, which in a less expensive model may not accurately convert the raw data from the sensor.
--They say only a fool looks at the finger pointing to the sky...
I thought we were pushing the theoretical limit for that - there are only so many photons impacting the sensor surface, and it's not possible to catch many more with much more accuracy than we already are.
Actually, even if you had a theoretically "perfect" CCD or CMOS, you can catch about two-to-four times as many photons.
The problem lies in the way the photosites capture light. Most designs are variants of the every other location is green with red and blue alternating the others. Something like:
RG
GB
Green gets twice the representation as human eyes are more sensitive to green and thus more detail in that part of the spectrum is considered desirable.
A recent trick to squeeze out more is to turn the photosites at 45 degrees to the grid you actually capture. You're then forced to interpolate more but the theory is that you get a smoother response.
Regardless though, any given location can catch red OR green OR blue parts of the spectrum. If green falls, 50% of it is lost. If red falls, 75% is lost - same with blue. You're always throwing away half to three quarters of your photons simply by having photosites dedicated to individual colors.
With Foveon they try tackling things differently. By exploiting the fact that different wavelengths can penetrate silicon to different depths, they figured you can have a three layer deep photosite that captures red AND green AND blue - none of this ignoring chunks of the spectrum and throwing away data.
Of course, for all it's a cool idea, it's proprietary, has only made it in to a few cameras and doesn't seem to be hitting its full potential yet. My guess is there's still quite a bit left that can be squeezed out of CMOS (Canon's 10D got noisy at-or-just-after 400 ISO wherease the 20D, 18 months later, could handle 800) and we'll see them follow that technology for a while whilst waiting for Foveon to move out of patent protection.
Still, in the future, I'd imagine we'll see Foveon or something different but exploiting some similar concepts replace individual colored photosites. Until that point, no matter how good things get, there's always a full stop of light's worth of extra quality sitting and waiting.