Digital BW, The Print

It's a bit more subtle that that. It has to do with the information 
content in the image. Here the term information is used in it's 
mathematical sense. For example if you were taking a picuture of a 
smooth 18% grey card, the amount of information would be small and you 
only need the number of pixels to capture this information. This image 
could be interpolated to your hearts content (assuming good robust 
interpolation procedures were used ) and get what ever resolution (above 
the min required to capture the information) with no ill effect.

However, if the initial sensor is not sufficient to capture the 
information in the image, no amount of  processing afterwards can 
generate that information. It's the same as filtering out all the 
frequencies above 8 kHz in an auido system sampling at 20 kHz and then 
interpolating up to 50 kHz and turning up the gain on an equalizier to 
try to get it back the information above 8 kHz - all you get is high 
frequency noise.

Interpolation may work on one image, but may fall apart on another. In 
general there is some "slight of hand" in all digital cameras. They list 
the raw pixel but actually it really divided in thirds - each third 
capturing a specific color. This data is now "interpolated" to produce 
color information at each pixel. The new stacked sensors will change 
that some but they are trying to claim that their real resolutions is 
not 3 megs but 3 times that. So there is a lot of "numbers games" being 
played.

NASA did an experiment with color image  resolution some time back  They 
flew two sensors in Landsa. One was a color sensor using similar methods 
as the current crop of digital cameras to produce a color image. The 
other was the black and white sensor without the color mask - to 
captured only energy of the photons. NASA defines resolution to be the 
closest objects that can be detected and differentiated on the ground. 
There are standard ways to mesure and test ranges that are used in the 
testing and calibration of sensors.

Calibrated targets were put on the ground to measure the resolution and 
the images were processed through standard procedures to measure 
"resolution."  The resolution of the color sensors was on the order of 
17 meters and the black and white on the order of 10 meters. There was 
also a large variance (3 meters if memory serves me right)  in the 
measured resloution of the color sensor based on the acutal target used. 
The variance of the black and white sensor was less that 1/3 of a meter. 
Because of this Landsat uses black and white sensors and IR sensors for 
high resolution imagery and uses the color sensors to fill in details, 
e.g., live foliage, dead foliage, muddy water, clear water, etc. The 
resolution you can get with film at the same orbit with the same optics 
is less than 6 inches - with better dynamic range.

With digital cameras it's buyer beware of the hype. They are to some 
extent selling a bill of goods. They are fine for point and shoot and 
the new crop is fine for photojournalism. Some of the medium format and 
large format digital backs may be approaching the quality you can get 
with 35 mm film. But it's going to be a long before we see the digital 
35 mm replace film in images where quality is absolutely critical.

Truman

Bob Frost wrote:

> Anthony,
>
> Surely every pixel on a digital camera (except for a few exotica) is
> interpolated? The skill of the camera maker is surely in choosing the best
> size and number of sensors, their physical arrangement, and the
> interpolation techniques.
>
> For example, a Nikon D1x has a chip with 4028 x 1324 sensors, but the 
> image
> produced is 3008 x 1960 pixels - all interpolated !!
>
> As to 6 mpixels being all one needs, I can see the difference between my
> current scanned and printed images with 24 mpixels and my previous 
> ones with
> 'only' 11 mpixels. With just 6 mpixels an A3+ print would only have 
> 167 ppi
> ! With 24 mpixels I can send 330 ppi to the printer - in the recommended
> range.
>
> Bob Frost.
>