RE: [APML]: Stacking more than 2 Images in Photoshop

["Tom Krajci" <krajcit@det5wg57.barksdale.af.mil>]

> From:          Ben Gomes-Casseres <ben@alliancerevolution.com>

> Thanks for your thoughtful reply; I was the one who asked the original 
> question. Your answer makes sense to me for stacking deepsky photos, and 
> Tom has good points too about the S/N ratio.
 
> Still, I suspect that my project may be different from what both of you 
> analyzed. I am stacking 8 images of the total solar eclipse, each of which 
> is properly exposed for a different segment of the corona (i.e. shots taken 
> at 1/4sec, 1/8 sec, 1/30 sec, etc.). On each shot, there is a portion of 
> the inner corona that is blocked out (255) due to overexposure and a 
> portion of the outer corona that doesn't show (<0). Now, by stacking, we 
> are able to patch these pieces together and show a fuller range of the 
> corona. With proper processing, only the very inner band would be at white 
> (255) and the rest would gradually go down to 0 at the outer edge of the 
> corona (i.e. the longest exposure). This technique was described (plus some 
> other blurring/sharpening steps) in S&T Jan 1998 and I used it with only 3 
> images on my 1991 photos (ben@alliancerevolution.com/astro/solar). I have 
> done a trial run of my 1998 slides, but am not happy with the technique I 
> used for stacking, as explained in my previous post.
 
> Do your arguments still apply to this project?

My comments were aimed at stacking identical images, I'd call your 
work more of a montage where different parts of different images are 
combined to make something that's better than any one image.

You're trying to handle a very large dynamic range with a limited 
recording medium - film.  Unsharp masking can do great things with a 
single negative, but the range of brightness of the corona exceeds 
what any one exposure can handle.  I don't have enough experience 
with Photoshop and while I've read some of the solar eclipse impage 
processing articles, and been amazed at the results, I haven't paid 
close enough attention to what they were doing.  I wish I could help 
more.  

I don't know if this idea has been used before, or if it'll work, but 
here's how I'd approach it (until I leaned better ways):  Each 
negative has an area of the corona that it recorded well. . .this 
corresponds to where the negative density is not too high or too 
low. . .where the film response is roughly proportional to the log of 
the exposure.  That's the part of each negative you want to 
save/cut/edit/keep from each scan.  Get rid of the rest. . .it'll be 
no help to try and composite that saturated/underexposed stuff.

Once the pixel values fall above and below some threshold values. . 
.turn them to zero/black, and keep the intermediate values for these 
individual negative scans.

To get rid of large scale brightness gradients in the scanned images 
you'd probably want to do some type of unsharp maksing.  That'll get 
rid of it and enhance small scale detail at the same time.

Then you have to stack them. . .but not "stack" in the traditional 
sense of the word.  Each image will contribute to a different part of 
the final image.  The tricky part comes when two images have common 
areas to contribute. I don't know how to handle that in Photoshop 
using layers, masks, whatever.

Maybe by only using a few (three? four?) good negatives that cover a 
wide exposure range you have less of a problem combining common areas 
from different negatives.

Hope this helps,
Tom Krajci

Capt Tom Krajci
B-52 Intelligence Officer
"Military Intelligence - an oxymoron!"
http://spur.barksdale.af.mil