RE: filmscanners: film vs. digital cameras - wedding/commercial photography

["Soren Svensson (EUS)" <Soren.Svensson@am1.ericsson.se>]

Title: RE: filmscanners: film vs. digital cameras - wedding/commercial photography

> From: Austin Franklin [mailto:darkroom@ix.netcom.com]
>
> Only the color information is shared amongst multiple pixels
> NOT the edge information.  That does not make the four pixels one pixel.
> Do the geometry.  Each of the four sensors is capable of sensing an
> entirely unique "section" of the image. Why is that so hard to understand?

Because it isn't true. Each sensor has a filter in front of it (R, G or B). That means that you have to use sensors next to it to get a true value of the luminance at each sensor. Each sensor just measures the luminance within a small spectrum.

I think that's pretty clear, isn't it? :-)

> Take the four pixels, a 2x2 box, and say the left two are sensing only
> black, and the right two, only white.  What are the four
> resultant values going to be?

Good point, because that shows the problem. You can't determine what the TRUE value was at each sensor, only guess or treat that 4 sensor group as one pixel. How would you distinguish between the red sensor getting no light or blue light? You can't unless you look at the sensors next to it, and then you're treating them as a group!

> Television works more or less the same way, having some
> fraction (1/4th?) the color information to the edge information.

Well, if you're satisfied with TV picture quality, that method works fine :-)
But there's a reason why high-end digital capture doesn't work that way.

> And I believe human eyes also work similarly, resolving less color
> information than edge information.

Yes, it does (that's how jpeg compresses images).
But as I mentioned in the first paragraph, even the edge information suffers from this "guessing" approach.

Regards
/Soren