RE: [APML] some more stuff

[Thor Olson <Thor.Olson@EFI.COM>]

 

-----Original Message-----
From: Tony Hallas [mailto:tonyhallas@foothill.net]
... 

 Most "automatic" programs don't work... they can only guess at the proper contrast levels and color levels... after lots of experimenting this is the method that I think works best for me...

 

1.   Keep everything in 16 bits as long as possible. Most programs only combine in 8 bits, so Registar is golden. After reducing the images I align and RGB merge in Registar. Up to now, nothing has been done to the data.

 

2.  Now the 16 bit data is imported into PS, and using curves and levels carefully, the image is compressed without loosing the endpoints. This is vital because there is a lot of information in these areas.

 

But this is where the "magic" is that the automatic programs are trying to achieve. 

Can you figure out what mapping you actually are doing to make this linear-to-visual

conversion?  Keep track of exactly what level and curve adjustments you make.

Will they work on other images, or is every subject unique?  This is an interesting

area that most discussions on this topic don't cover very well, maybe there is a

visual perception principle to uncover here!

 

Thor.

 

Thor Olson                           
web-res astrophotos at:             http:\\www.nightscapes.org

   "Man has ten, computers have two, Nature has e fingers."