REFLECTED LIGHT

Reflected light is one of the hardest problems to deal with in color graphics design. Light from the workspace is reflected in varying degrees from the front and interior surfaces of the display. This light "dilutes" the controlled light from the graphics, reducing luminance contrasts and color purities from the designed values. If the amount and location of the reflected light are particularly bad, the legibility, color discrimination, and sometimes even the detectability of the graphics can be reduced.

For most applications the reflected light in the users' display environments will differ from yours and from each other. This makes it very hard, often impossible, to design luminance contrasts that will look exactly the same on all users' displays as on the designer's.

So what can the designer do to ensure the usability of the design? The viewing environment is in the user's control rather than the designer's. We outline the physical problem, then discuss the designer's options.

Viewing Geometry

There are two geometrical layouts to understand:
1) the geometry of the reflected light, i.e., the position and shape of the reflected light within the viewed graphics and
2) the geometrical layout of the viewing environment, i.e., where the light sources are with respect to the screen and the viewer.

Geometry of the reflected light. The reflected light interferes with perception of the graphics in two main ways, depending on the shape of the reflected light and where it is within the field of view.

Geometry of the viewing environment. Which light sources will be reflected on the viewer's screen depends on the layout of the viewing environment. The basic geometry is the same throughout, but the result is different for the two main types of reflection, diffuse and specular.

Reduced Luminance Contrast and Color Purity due to Reflected Light

Here is an example of how light reflected from the display surface reduces the luminance contrast of the graphics:

Let's begin with a display in a dark room, with a "white" background of 100 nits and consider a symbol with a luminance of 10 nits. The Michelson contrast of the symbol is Lm = (100 - 10) / (100+ 10) = 81.8%. Now we turn on the room lights and there is a reflection of my white t-shirt on top of the symbol and its immediate surround. If the reflection is 10 nits of additional light on both symbol and surround, then the Michelson contrast is now Lm = (110 - 20) / (110+ 20) = 69.2%.

If sunlight falls directly on the surface of a display, the diffuse reflection can be much worse. In a recent measurement, an actual cockpit LCD display was illuminated by a sunlamp simulating direct sunlight through a cockpit window. Without the sunlamp the white and black luminances were 260.7 nits and 12.6 nits, respectively, giving Lm = (260.7 - 12.6) / (260.7+ 12.6) = 90.8% for black symbols on white background. With the sunlamp on the white and black luminances were 329 nits and 80.9 nits, respectively, giving Lm = (329 - 80.9) / (329 + 80.9) = 60.5%.

The effects on color purity can also be substantial. The reflected light in the simulation was white, and when it is added to the chromatic lights on the display, it dilutes them. The red primary of the display, for example, has CIE x and y of 0.605 and 0.341, respectively, without the sunlamp. With the sunlamp on x = 0.411 and y = 0.326, significantly closer to the color of the white background. Excitation purity was reduced from about 78% to about 22%. With the sunlamp on the "red" primary appeared barely colored. Under these conditions discrimination of color labels could become unreliable. The problems might be much worse: CRT displays often have much higher reflectances than this specialized LCD.

There is some evidence, however, that visual adaptation can reduce the impact of this kind of general loss of color purity on color recognition Post & Calhoun. When all of the colors in a laboratory visual scene were diluted by reflected light the color names given to colors that were good examples under more favorable conditions (e.g., in the center of the chromaticity region called "red") were little affected. While this is somewhat reassuring, the laboratory conditions that allowed this good recognition performance may not be comparable to some common real work environments. Users often scan back and forth among displays and printed materials under different illumination conditions.
More about visual adaptation.

What the Designer Can Do About Reflected Light

Ideally the graphics will be used on appropriate physical displays in well-designed, predictable viewing circumstances. Under these conditions the designer can measure the expected reflected light and take it into account in designing the luminance contrasts of the graphic elements. However, even under these ideal conditions, if the luminance range of the users' reflected lights is too great the intended perceptual salience hierarchy will be difficult to maintain. If there are high luminance reflections the luminance contrast of the least important data must be increased above that for normal viewing conditions if they are to remain visible. Then the same graphics may be too salient when there are lower luminance reflections, distracting attention from the more important data.

The best solution is to reduce and stabilize the reflected light by changes in the viewing environment, use of anti-reflection displays, or both. If the viewing geometry can't be controlled in that way, it may be necessary to provide for the users to control the overall luminance contrast of their display. This gives the user control over the salience hierarchy, a two-edged sword. For a user who thoroughly understands how to maintain the salience hierarchy this is ideal. For less aware users it introduces an opportunity for lost information and error hazards. We have observed one applied setting in which the operators were allowed to adjust the luminance contrast of map and other context information separately from the contrast of target symbols and labels. While most operators turned the context information down to barely-visible levels, others turned it to undetectable levels. Should they need the context information (and realize that they needed it) they would be required to turn it back up.

Related Topics:

About This Site

Home | Design Process | Color Graphics Topics | Color Tool | Guidelines | Aerospace | Color Science | Utilities

  Site Curator: Richard Mogford | Responsible NASA Official: Richard Mogford | Privacy Statement