Re: [APML] Histograms

[Jerry Lodriguss <jml@astropix.com>]

> > > As I've been playing around with Photoshop stuff:
> > > contrast/brightness; hue/saturation; color
> > balance;
> > > levels; and curves, I've noticed the histogram
> > changes
> > > from what looks like the silhouette of a hill, to
> > > something that looks like a picket fence if the
> > pieces
> > > of wood were placed about an inch apart and viewed
> > > from about thirty feet away. In other words, lots
> > of
> > > steep, narrow spikes. The images are improving,
> > but is
> > > this what I want?


Alan,

Spikes such as you describe come from manipulating the data.

You have a limited number of tones to work with. When you change the 
brightness and contrast, these tones get stretched, pulled apart, squished 
together, and shifted around.

A gap in the histogram can result when tones are pulled apart to increase 
contrast, resulting in a loss of data.

Sometimes when the tones get squished together a spike will result, which 
means that two previously separate tones have been combined into one, 
resulting in a loss of information.

ANY manipulation of the data will result in a loss of data.  The question 
is, how critical is this loss, and what did you get out of it?

How much can you get away with?  That's the 64 dollar question. At some 
point the data loss will become visible, particularly in posterization.

As Thor has pointed out, we manipulate the data to improve its aesthetics 
and visual presentation, and sacrificing some of the data is how we improve 
the rest of it.

Scan at a high bit depth.
Set a black and white point in levels in Photoshop.
Adjust the mid-tones and do your color correction in curves.
Then convert to 8 bits. This will preserve the most data.

Always archive your original scan before you manipulate it.

Jerry




Astrophoto Web Site: http://www.astropix.com

Photoshop for Astrophotographers Book: 
http://www.astropix.com/APBOOK/0_PROMO/PROMO.HTM

--  APML Archives at <http://astro.umsystem.edu/apml/>  ---
             Unsubscribe at <majordomo@seds.org>