From: Robert Horton 
To: James Lerch 
CC: atm_free{at}yahoogroups.com, ATM List 
Reply-To: Robert Horton 




Hi James,

One thing I noticed in the images you posted was a significant change in
color quantization when going from the "Normal" webcam shots to
the new camera.  Also, with the blue led you don't get any red or green
components which effectively divides the number of unique grey values by
three (when you convert from RGB to Greyscale).

If you count the number of unique colors across a large portion of one of
the new images, you get somewhere in the range of 25-35 levels.  This
doesn't provide a lot to work with and will limit the spatial resolution
due to large areas being the same "grey" level.  Also, it looks
like something in the analog-to-digital conversion drops the low-order
three bits of the eight bit color value which will limit you to just 32
unique levels (5-bits).

If you do the same thing with the webcam images you get thousands of
individual colors which smoothly blended into 250 or so unique grey values
when converted to greyscale.  This helps provide a very narrow zone where
the values match on the left and right sides of the mirror.

I'm not sure why you changed cameras when the webcam was working fairly
well.  Was it to allow the wider light cone and avoid vignetting that you
mentioned in your original note?  If you pull up the same zone's raw images
from the normal webcam set and each of the other exposure settings, you'll
see a significant difference in "harshness" of the images that
may be due to a combination of the camera, the slitless source (as you've
pointed out) and maybe even some pre-processing done by the camera
electronics prior to sending it out.

Note that this stuff is on top of any skew due to optical/physical problems
such as those pointed out by Nils.

Anyway, just some observations on a very interesting project...

Bob Horton

--- BBBS/NT v4.00 MP