Weber and contrast ratios
according to Timo


Timo claims that the ability of the visual system to discern intensities - the Weber fraction - is 2%. The step from code 50 to code 51 corresponds to an intensity step of 2%. Codes less than 50 will be liable to exhibit banding (contouring).

Timo insists upon coding linear light, even in 8 bits. If the highest intensity is code 255, then the contrast ratio between white (255) and the darkest code free of banding (code 50) is 5:1. That's equivalent to a contrast ratio limit of 5:1.

If the display contrast ratio was 10:1, then codes as low as 25 could be distinguished from black. But in linear-light intensity coding, the step from code 25 to code 26 is a 4% jump in intensity. This is larger than Timo's Weber fraction, and is bound to be perceptible.

Even if you assume that vision degrades to 4% contrast sensitivity near black, there is no way that linear light coding in 8 bits will achieve acceptable performance in viewing environments (or display media having contrast ratios in excess of 10:1: Eight-bit linear coding cannot work in television, cinema, projected transparencies in a darkened room, or even high-quality printing.



For Timo, we need to add a row in the table, for contrast ratio 5:1. If Timo gets acceptable image quality with 8 bit linear light coding, I conclude that the contrast ratio in his display environment is about five to one.

Contrast sensitivity
 2%  1% 0.5%
Contrast ratio   
5
(Timo's environment)
81
(~6.3 bits)
   
10
(office environment)
116
(~7 bits)
231
(~8 bits)
462
(~9 bits)
30
(living room, television)
172 342 682
100
(cinema theater)
232
(~8 bits)
463
(~9 bits)
923
(~10 bits)

Timo demands linear-light coding. But equivalent performance could be obtained from a nonlinear system having only 81 codes, easily accommodated in 7 bits.

Main thread:

Timo thread:

See also: Gamma FAQ - Linear and nonlinear coding - Weber and contrast ratios - Timo and linear coding

Charles Poynton - color links - Gamma FAQ - Linear and nonlinear coding
Copyright © 1998-03-11