Re: perfect rgb references
Re: perfect rgb references
- Subject: Re: perfect rgb references
- From: "Bruce J. Lindbloom" <email@hidden>
- Date: Mon, 04 Feb 2002 11:20:40 -0600
Daniel (email@hidden) wrote:
>
Maybe someone could clear things up to me what all the magic is that
>
can be done with these images.
I actually made these images for a purpose quite different from what Tom is
using them for. They are free and well documented, and you may use or not
use them as you see fit.
The reason I originally made the images came from a very frustrating and
futile exchange I had with Dan Margulis on his Color Theory list last
summer. As many of you know, Dan has a long-standing "16-bit challenge"
saying that 16-bit color correction has no quality advantage over an 8-bit
counterpart (please, let's not start up that topic again -- anyone taking up
his challenge is a fool because his test is seriously flawed, IMO). I
neither agreed nor disagreed with Dan, but rather, tried to understand the
parameters of his test (e.g. what is the role of gamma, what is the role of
noise, what is the role of quantization, what roles do Photoshop's tricks
play such as 15- (not 16-) bit representation, color dithering, etc.).
This lead me to write programs that actually *measure* things about digital
images, such as given an N-bit image, how many bits are actual image data
and how many bits are noise, how many unique levels are present in each
channel (even for 16-bit images), how many unique colors does it have, etc.
As a reference, I wanted a "natural" image (as opposed to computer generated
"Chiclet" charts and gradients) that contained full, uncombed histograms, no
noise, no aliasing, etc. That is what these images are (the same scene
rendered in both 8- and 16-bits, at gammas of both 1.0 and 2.2, making a set
of four total). I decided to share these images with the general public.
That is why I placed them in the category of "Imaging" instead of "Color."
But again, if you don't find them useful, then for heaven's sake, don't use
them.
--
Bruce J. Lindbloom
www.brucelindbloom.com