In a previous post, we addressed the limitations of cameras, and photography in general, and how illuminateAI has fundamentally changed the physics of image capture to accurately read color regardless of lighting conditions. The key to matching color, as we will dive into here, is measuring how objects reflect light.
Reflectivity is the intrinsic property of objects which dictates their color, texture, and how they appear in a given lighting condition, as well as in photographs and videos. In simple terms the chemical bonds and structure of a material (or on a surface) dictate which wavelengths of light are reflected, absorbed or transmitted. Color is how an object appears in a particular lighting condition, while reflectivity is intrinsic to the object (doesn’t change with lighting). Access to reflectivity data makes it possible for AI to measure and predict the color of objects in any light, identify materials and analyze chemical and biological properties.
Color is not exactly what you think: what we think of as color is actually the result of the complex interplay between object reflectivity, lighting and the observer.
Humans and cameras perceive reflectivity as color, which is determined by the underlying object reflectivity and the light shining on the object. For example, an object that primarily reflects light with wavelengths in the 400-500 nanometer range will look blue when illuminated by white light (white light contains a roughly equal mix of all wavelengths). Similarly, objects which reflect light in the 650-750 nanometer range will look red when illuminated by white light. Non-primary colors are created by mixing different colors of light together, an object that reflects both 400-500 and 650-750 nanometer light will appear purple in white light. Where this gets tricky in practice is that objects are rarely illuminated by idealized ‘white’ light – which is why apparent color is not a reliable source of information about reflectivity.
Apparent Color is determined by how an object reflects light, and by the lighting conditions.
In practice, the color we observe and cameras capture is only indirectly correlated with the reflectivity of objects because the lighting conditions vary widely. As shown in Figure 2 below, an object that appears blue in idealized white light (such as D65) will appear blue-green in the warm, red-rich light of sunset or candlelight. The same object will appear dark-blue in dim light (at dusk) and bright blue in the cool blue-rich light from the noonday sun. Under red light the same object will appear to be black. Indoor lighting is similarly variable, notice how different your face looks under fluorescent lights versus when viewed under incandescent lighting. The quality of light sources also has a dramatic impact on appearance. Sunlight contains all the wavelengths of light, but many man-made light sources only provide a subset of wavelengths, as they optimize for energy efficiency and cost. These light sources can make skin look grey or ashy. Fashion and beauty retailers know this and are generally very careful to choose light sources which make their merchandise look its best. Finally, shadows and mixed lighting environments further complicate appearance, resulting in different ‘colors’ on the same object in the same scene. This is why diffuse natural daylight is still viewed as the gold-standard for viewing skin, fabrics, and make-up. Similarly, this is why professional photographers bring their own light sources and/or go to great lengths to measure the lighting in a scene.
Figure2 Diagram illustrating the physics of imaging and how the lighting conditions and the object reflectivity determine the apparent color of an object. The same object, that reflects wavelengths in the 400-500nm (blue) range, will appear blue in white light, blue-green in warm light, and black in red light. As this figure shows, the appearance of an object is dictated both by the light striking the object, and by how that object reflects or absorbs light.
Without knowing the intensity and spectra of light striking an object, humans and AI cannot know the true color or reflectivity of the object.
Reflectivity is the Key to matching color
Shopping for color, and identifying objects with the ‘same’ color is extraordinarily difficult. This is because the apparent color of objects changes with lighting. Two objects that look the same in one lighting condition will often look very different once the lighting changes. This is why matching the color in one location (such as while shopping in a store) is no guarantee that the color will match once you get the object home (when viewed under different lighting). Identifying a true color match requires measuring and matching the object reflectivity, so that objects will look the same in any lighting condition. Objects with the same underlying reflectivity will look the same in bright blue-rich noonday sun, in the softer red-rich light of sunset, and in indoor lighting environments. Unfortunately a photograph or selfie cannot be used to color-match, unless lighting conditions are precisely known. Information about object reflectivity enables humans and AI to avoid the pitfalls of apparent color and make decisions based on accurate color, as well as the chemical and biological properties of objects.