Observation Skills and CG - An Introduction to Computer Graphics for Artists - page 43

objects. However, to say that a fi re engine is “red” cannot be exactly true when the
sun refl ects sharply off the corner of a “red” fender. As any camera will show, the
brightness of such a refl ection is so strong that it will appear to be near-white even
if it is “red” paint. In the same way, if the fi re engine is parked in a dark alley at
night, it might look red in comparison to surrounding objects, but a photo would
have deep purple pixels instead of red because of the color and low intensity of light
at night. A group of cars in a parking garage will all appear to be the same color if
mercury lighting is used in the garage. Red and blue cars sitting beside each other
will both appear to be different shades of yellow.
Color is relative. This means that the idea of any object possessing an absolute
color is false. Despite this, objects do possess qualities that are absolute. A car may
have rust on a chrome fender. The rust refl ects light one way and the chrome refl ects
it another. The difference between how each material refl ects light is consistent and
measurable. The actual “color” of either object may be impossible to ascertain with-
out light, but a reasonable estimate may be made using the tools provided in any CG
application. To determine the diffuse color of an object, one must fi rst keep in mind
how light affects color and then strip out those effects as if they are contaminants.
The color that remains will be a good approximation of the diffuse color that should
be assigned to the object for it to render realistically.
The correct observation of color requires rigorous analysis of the factors that
contribute to the colors we see on objects. For instance, in a photograph of a
curved whitewashed wall, one might wonder if the wall is truly white. By placing
a white sheet of paper over any part of the photo, it will be clear that the white
paper is more “white” than any part of the “white” wall in the photo. If the indi-
vidual pixels of the wall are sampled in a digital imaging program, there will be
a variety of non-white colors due to the curvature of the wall. This is because the
curves have the effect of changing the angle of the light to the object as it curves
away from the light. Are any of the pixels the right color? Probably not. Is your
paper actually white? Again, probably not. If you hold it up to the light, you will
see a highlight along the edges facing the light. Those edges are whiter than the
face of the paper, so the face cannot be truly white. What of the edge, is it white?
Again, no. The edge cannot be brighter than the light source, and the light source,
even if it is the sun, is not white (the sun projects blue light.) With some highly
refl ective surfaces, it is possible to create refl ected colors that appear to be
brighter than the light source, but this is due to the lack of atmospheric occlusion
at a close distance.
To arrive at a reasonable estimate of an object's true ambient color, one must
remember that light is made of photons that strike an object, while shadows are
areas that receive fewer photons and are thus less affected by the light source.
This indicates the possibility that colors in shadow areas are more accurate
representations of the true diffuse color of an object than the areas in light.
The problem is that our eyes are less sensitive to dark colors than light ones. The
colors may be a better representation of the object's intrinsic diffuse value, but
we cannot see it accurately because of physical limitations in our eyes. Cameras
have a similar problem because they require an accumulation of refl ected light to
create a picture.