From: Milton Aupperle <milton@outcastsoft.com>

Date: July 29, 2008 11:18:11 AM MDT

To: Astro_IIDC@yahoogroups.com

Subject: Re: [Astro_IIDC] Re: Monochrome???

Hi Dave;

On 29-Jul-08, at 9:57 AM, doodlebun wrote:

Thanks Milton,

Next year I will get a DBK and run direct comparisons so this issue

can be sorted out once and for all.

I don't know about "once and for all" as seeing will be one of the main issues.

For example, if you were to shoot Filtered R G B and during your Green session, you got hit with bad turbulence, then your image will be softer in the green channel. If your shot an RGB Bayer movie and got great seeing during that period, will it be better or worse than the Filtered RGB?

Maybe maybe not..

You'll need to have a "blind panel" (by blind I mean they have not seen the images before) of people to look at the undesignated images and then decide which is sharpest. When I did Bayer algorithm and Image processing research for Vitana Corp, I used to render the same image using different algorithms and then assign each image a random number before letting everyone in the company vote on their favorite(s).

I did switch to RED instead of monochrome as we discussed. Did not

notice much speed change.

It won't be huge, but it will be a bit faster when estimating sharpness.

Will drop below 128 x 128 as you recommended.

My current processing speed with my G5 is approximately 1 frame every 1-

2 seconds.

With a large 128 x 128 pixel area, you doing about 1,073,741,824 (1 billion) pixel comparisons per frame processed.

With a 32x32 pixel area, your doing 4,194,304 (4 million) pixel comparisons per frame processed.

Next I decided to tackle the problem of what confidence level is

appropriate. It is discussed in another thread. My results are shown

here...

http://web.mac.com/davidbleser/AncientDocuments/CONFIDENCE_LEVELS.html

One other factor you may or may not be considering is "Matching Tolerance" for the pixel comparison. The "Confidence Level" is the first hurdle to weeding out softer frames and the "Matching Tolerance" is the next hurdle. It decides how close a match to your "Sharpest Frame" each other frame needs to be before it is included in your stack. The higher the value, the tighter the demands and the better it has to match to be accepted. However image noise from high gains will affect the matching tolerance, so you have to be careful in not setting the tolerance too high or you get no matching frames.

Hope that helps..

Milton J. Aupperle

President

ASC - Aupperle Services and Contracting

Mac Software (Drivers, Components and Application) Specialist

#916, 742 Kingsmere Crescent S.W.

Calgary Alberta T2V2H8

1-(403)-453-1624

milton@outcastsoft.com