View Single Post
  #30  
Old February 9th 09, 10:34 PM posted to aus.photo,rec.photo.digital,rec.photo.digital.slr-systems
John A.[_2_]
external usenet poster
 
Posts: 1,551
Default Faking and expensive tilt-shift lens

On Mon, 09 Feb 2009 15:28:12 -0500, ASAAR wrote:

On Mon, 09 Feb 2009 18:24:05 GMT, John A. wrote:

On a Bayer-filtered sensor 1/2 pixel = 1 photosite. In your sequence
remove step "0" (which is pretty much meaningless) and replace "pixel"
with "photosite" and you've pretty much got what I've described.


You still don't get it. First of all, the "0" step was only used
to identify the surrounding 4 pixels (sensels). Traversing from any
pixel to the next requires stepping the distance from one pixel to
another, which is *not* 1/2 the distance/size of a pixel. If it'll
help, place four hamburgers on a plate. How far is it from the
center of one hamburger to either of its adjacent neighbors (not the
diagonally opposite burger). If you say it's 1/2 of a hamburger's
diameter, you've bitten off more than you can chew.


I get it. You don't.

If a sensor is going to be moving like this its initial position will
already be where your step "0" is intended to put it, so that step is
not necessary. We can safely assume the sensor will be designed to
start out exactly where it needs to be for its first exposure. It's
just good engineering. Otherwise it's just added motion that only
serves to wear the mechanism. Seriously, why would it have a "home"
position somewhere other than where it needs to be for one of the four
exposures?

And a pixel IS made of 2x2 photosensors on a Bayer-filtered sensor
array. The sensors are arranged something like this:

G R
B G

Together, those four photosensors make up one pixel.

Ahhh... But I think I see what you're talking about. That's in a
single-exposure Bayer system. In the theoretical four-exposure system
we are discussing we would get four three-color synthesized pixels
from each 2x2 Bayer array. (Minus a row and a column for the image, I
believe, assuming we'd only keep the overlapping portions of the four
exposures.) That was what you meant, right?

So maybe we both get it, and don't get it. =D