Dale (and Summerwind),thanks for your replies,
I fully appreciate your comments below, but getting back to the
crux of my question, what makes a DSLR require more post-processing
than a digicam. (remembering that my original post was in response
to the general consensus on this post that a Rebel shot image would
require more post-processing than a Pro-1). Is the in-camera
processing different in both, and if so, how in particular?
(I am looking for Phil's articles on the subject, any links would
be helpful.)
Regards,
Harrow.
P.S. I see you are also an engineer, so we can talk the same
language here ;-)
Yes... That is correct, and hope I can be helpful to you.
There are lots of things that make these cameras behave
differently, but lets go right to the basics first. As an
engineer, you will appreciate the calculations...
The pixel to pixel spacing on the Rebel 300D sensor is 7.4 microns.
To resolve a dark line / white space pair then two pixel positions
would be required, making 2 X 7.4= 14.8 microns for a line pair.
(The dark line would be on one pixel, while the white space would
be on the next one, and then the pattern repeated.)
Since there are 1000 microns per millimeter, the calculation now
becomes:
1000
14.7= 68.02 line pairs per millimeter as the maximum possible
sensor resolution. To prevent Aliasing then, a fairly strong
spatial AA filter needs to be placed in front of the sensor to
remove all frequencies above 68 lpm. Canon calls this a low pass
filter, which it is, but remember it is spatial frequency not light
frequency.
This also tells us that a high resolution lens with resolving power
greater than 68 lpm has the high frequency data removed by the
filter. All that is required is a lens with at least 68 lpm of
resolving power. But what is really important is since this is a
digital senosr, that the lens deliver the maximum possible MTF at
60 lpm! That takes really good L quality glass to get the best
image.... and you
can see the difference. Don't be fooled by the
relatively low resolving power.
It also tells us that the photosites are fairly large at 7.4
microns, and therefore the dark current, and brownian noise effects
are a much smaller percentage of the total charge, so the sensor
has low noise.
The sharp cutoff AA filter softens the image somewhat however, and
so without resharpening the image, it appears soft.
Now let's look at the Pro1. Its pixel to pixel spacing is 2.7
microns. By the same set of calculations as above... a line pair
requires
2 x 2.7=5.4 microns. And 1000/5.4= 185.18 line pairs per mm!! Wow!
As a result a much higher spatial frequency and less aggressive AA
filter is used here to cut off frequencies higher than 185 lpm to
avoid moire and aliasing... Now the opposite requirement is placed
on the lens. It is very difficult to design a lens that can
resolve 185 lpm, and the image circle needs to be reduced in order
to achieve that resolution. Canon and Olympus have succeeded in
making lenses of this caliber for the pro1 and the 8080.
You can see that the image is natively sharper without post
processing because of a weaker AA filter than the 10D and 300D.
However, the smaller photo-diode size means that dark current and
quantum noise effects are a larger percentage of the total charge.
There is therefore, more noise, and less possible ISO sensitivity.
The gain can't be turned up quite so high on the ISO amplifier.
Both of these cameras have a Bayer color pattern overlaid on the
sensor surface to effectively make an array of sites that are
sensitive to Red,Green, and Blue. (There are some sensors made for
CMY complementary colors, and Canon used to use one, but that is a
different story.) The raw data out of the sensor looks pretty
horrible.
The digic processor that Canon has developed is responsible for
demosaicing these sites into a proper color image. There is an
RGBG pattern and Canon uses the Green sites to obtain luminance
data, since the green receptors closely match the luninance
sensitivity curve of the human eye.
From the above discussion it is apparent that the DSLR gives a
smoother more noise free image, but more post processing is
required to bring back the sharpness after being softened by the
aggressive low pass AA filter. The smaller sensor requires less
sharpening, but some is applied anyway, to make the image pleasing.
Hope this helps some. There is a lot more....
kindest regards
Dale
It is possible to make a small fortune in photography, you just
have to start with a large fortune.
