View Single Post
Old Oct 3, 2006, 3:20 AM   #27
peripatetic
Super Moderator
 
peripatetic's Avatar
 
Join Date: Nov 2004
Posts: 3,599
Default

Quote:
As usual, the SD14 looks fabulous!
It looks fabulous, but if you had listened to the report you would have heard that the image quality from the prints on display looked quite poor.

That's why you should wait for a proper review.


Quote:
Tell me does the bayer sensor realy invents colors ? if so why pros use it and not the foveon one ?
It doesn't exactly invent colours, it interpolates values. Which isn't quite the same thing.

Each pixel needs an R,G,B value. Consider trying to "guess" a R value for a G site. How do you do it? At random? No, what you do is you look at the values on either side of the G where R was measured. If both of those values are say 125 then it's not a bad guess to assume that the value in the middle will be 125. That sort of guess is very accurate. It gets harder where the two values differ by a lot. Say one value is 100 and the other is 200. This is the sort of trouble you have at object edges. And there are lots of ways of being clever about guessing the right value.

Do an internet search for "Bayer Interpolation Algorithms" there are many different types and the ones that are bettter are usually slower, which is one reason why shooting RAW makes sense; you can use a better but slower algorithm for interpolation because you have a big computer to do the calculation and more time, instead of the little computer inside the camera that has to work really fast.

The reason pros use cameras with those sensors is because actually the Bayer interpolation algorithms are actually very good in general, and although they "guess" what 2 out of 3 values are for every pixel they do it so well that the advantages to be gained from a "true" measurement like the foveon are not very significant compared to the disadvantages.
peripatetic is offline   Reply With Quote