by [TC]²
A Note on Color Inconstancy
By Niraj Agarwal
Technology Manager
Clariant Corporation
What is color inconstancy ?
Color inconstancy is the change in color of a single sample under different
lights. The magnitude of color inconstancy can be defined by DE CMC of the sample between two lights.
The ISO standard for shade (in)constancy is called CMCCON and it refers
to the DE CMC for a sample between D65 and a second light source. Given
the CIE L*a*b* values for a sample in D65 and in a second light source
it is possible to calculate the CMCCON for a shade.
 |
|
Figure 1: A color named
POPPY viewed in 3 different lights (Ultralume, D65, CWF)
|
As shown in Figure 1, product color can change significantly in going from store light to daylight. This change can be unexpected and could leave the consumer less than satisfied with their purchasing decision. Therefore, color inconstancy is a problem that needs a more complete analysis.
How is it different from metamerism ?
Shade inconstancy is the change in color of a single sample. Metamerism,
on the other hand, is the change in color difference between
a pair of samples. Normally, luminant metamerism is of greatest
interest although factors such as observer and geometry can also play
a role.
 |
|
Figure 2: Frequency distribution
of CMCCON values for palette colors shows that 8% of the colors
shift more than 8 DE CMC between D65 and Inca
|
The magnitude of metamerism can be measured by the change in DE CMC for a sample pair between any two luminants. So there are
two DE CMC values involved. A metameric pair could have a DE CMC=0 in one light and a high DE CMC in the second light.
Creating shades with a low inconstancy in itself does not improve the
chances of getting matches with a low metamerism. It is entirely possible
to get high metamerism matches for color constant shades and it is also
possible to get non-metameric matches for color inconstant shades.
Metamerism can only be controlled by using the similar colorants in
the color standard and the sample. Attempting to reduce the color inconstancy
of a color standard does not in any way reduce the possibility of metamerism.
What is the extent of color inconstancy ?
By analyzing 2300+ colors in a textile palette covering a wide gamut
it was found that 10 to 20% of the colors have a very high color inconstancy
in going from D65 to incandescent light with color shifts of DE CMC greater than 6 units. Most colors showed a color shift of
more than 2 color units.
Calculations with other color palettes and light sources showed similar results.
 |
 |
| Figure 3: Colors in the palette that show the highest inconstancy between D65 and artificial light (INCA) | Figure 4: Colors in the palette that show the highest inconstancy between daylight (D65) and TL 84. |
Which colors show the highest inconstancy ?
Upon taking a closer look at the shades with high color inconstancy
between D65 and incandescent light we see that red-oranges and bright
blues particularly those of medium depth show the most pronounced color
inconstancy. Figure 3 shows the top 100 colors with the highest
values for CMCCON.
 |
|
Figure 5: Normalized reflectance
for top 100 inconstant colors. This also shows two broad color
families that have shade constancy issues.
|
Further insight is gained by considering the reflectance
curves for the colors with a high inconstancy. If we normalize their
reflectance curves (divide by maximum reflectance for each curve), these
curves fall into two broad families as shown in Figure 5. These
are colors that show a large proportional change in reflectance between
the low and high wavelength limits of the visible range. The spectral
energy distribution for incandescent light shows a monotonic increase
over the visible range. Therefore, colors with the highest change in
reflectance are the most inconstant between D65 and incandescent light.
The exact set of colors that show the highest color inconstancy depends
on the pair of light sources being used. Figure 4 shows the results
for D65 and TL84. Here also we see that colors covering a certain region
of the color space show the highest inconstancy although this region
is somewhat different from what we saw for D65 and incandescent lights.
The results obtained here are not a characteristic of the particular
color palette used in this study. Calculations with other color palettes
showed qualitatively similar results.
Color inconstancy therefore depends primarily on the type of color and
the luminant pair. Metamerism, on the other hand, depends on the type
of colorants used and the luminant pair. Color constancy depends on
the broad shape of the reflectance curve for a color whereas metamerism
depends on smaller differences in reflectance curves.
Can shade inconstancy be controlled ?
Yes and no. Modest changes in shade inconstancy can be made by altering
the recipe used to make a color. Inconstancy is primarily determined
by the color itself and only secondarily by the colorants used. Some
colors with certain characteristic curve shapes will ALWAYS be color
inconstant regardless of the recipe used.
In attempts at reducing shade inconstancy for the top 100 inconstant
colors we found that only modest gains were possible by altering the
recipe. There is some room to maneuver by recipe changes but for the
most part it is true to say that shade constancy is mostly determined
by the shade at hand not so much by the recipe itself.
Further, a less shade inconstant recipe does not in any way reduce the
chances of metameric matches.
September 2003
To search for vendors of color measurement and management systems, search our buyer's guide database.