Re: Usable Tones in Different Gammas
Re: Usable Tones in Different Gammas
- Subject: Re: Usable Tones in Different Gammas
- From: Rolf Gierling <email@hidden>
- Date: Sat, 09 Dec 2000 14:46:00 +0100
Warning, math!
This is what I did to get the visible greys on a monitor:
First I am ssuming my monitor is as good it shows Lightness with a range
from 0 - 100 in Lab.
Second grey consists of equal Red, Green and blue components.
When I translate the data RGB to an virtual visibleRGB the monitor shows
up, it would be:
virtualRGB = 255 * (RGB/255) pow (gamma)
Now I can use monitor matrices math to convert this virtualRGB to XYZ
values, where finally
Y is the luminance. As the human vision is nonlinear, we have to
translate that luminance to
Lightness in the Lab colorspace.
So I can make a table what Lighness is produced by a given RGB greytone.
Assuming that a
lightness change of minimum 1 is needed to be distinguable, I counted
the number of greytones
as followed:
Output Values
Gamma 0-63 64-127 128-191 192-255 Total values
--------------------------------------------------------------------------
2.2 23 21 22 21
87
1.8 28 21 19 16
84
So looking just at the values a gamma of 2.2 would be better.
The actual number of greys in real monitors will more decrease, because
real monitors dont have that ideal gamma but a more s shaped tone
reproductions.
On the other side Robin Myers is right when he says that looking on the
bit coding
a gamma of 1.8 is better. For example when you want to print the images.
Conclusion: Where have all that beautiful graphics cards gone that had
lookup tables
with more than 8 bits per channel? With more than 8 bits we could easily
have 100
visible greytones on our screen and still use a "better" gamma 1.8 image
coding.
Rolf Gierling