Chroma key compositing has a rich history dating back to the 1930s when early film techniques used blue-screen processes based on optical printing and matte photography. The fundamental principle has remained the same across decades of technological evolution: identify a specific color in the image and replace it with transparency, allowing a different background to show through. Understanding the color science and algorithms behind modern chroma keying explains both its capabilities and its limitations.

The reason green and blue are used as background colors is rooted in human skin tones. Skin across all ethnicities contains significant amounts of red and yellow pigmentation but very little green or blue, maximizing the color distance between the subject and the background in color space. Green is more commonly used in digital video because digital camera sensors typically have twice as many green photosites as red or blue (following the Bayer pattern), resulting in cleaner, less noisy green channel data that produces better keying results. Blue screens remain preferred in some film and VFX contexts, particularly when subjects have green elements like foliage or certain costumes.

Color keying algorithms operate in color spaces that separate chrominance (color information) from luminance (brightness information). While the raw pixel data is typically in RGB format, keying is usually performed in color spaces like YCbCr, HSV, or CIE Lab, where the color and brightness components can be analyzed independently. In YCbCr space, for example, the Cb and Cr channels encode the blue-difference and red-difference chroma components. A green screen appears in a specific region of this chroma space, and pixels whose chroma values fall within a defined distance from the target key color are classified as background. The distance threshold determines how strictly the algorithm matches: a tight threshold preserves more of the subject but may leave green remnants, while a loose threshold removes more background but risks eating into the subject edges.

Spill suppression is a critical post-keying step that addresses the green light reflected from the background screen onto the subject. When a person stands in front of a brightly lit green screen, green light bounces off the screen and illuminates the edges of their hair, skin, and clothing with a green tint called spill. Without correction, this spill makes the composite look unnatural, as the subject appears to have a green fringe. Spill suppression algorithms detect and reduce the excess green in pixels near the subject edges, typically by reducing the green channel value to no more than the average of the red and blue channels.

Matte generation is the process of creating the alpha channel (transparency map) from the keying result. A simple binary matte classifies each pixel as either fully foreground or fully background, but this produces harsh, aliased edges. Professional keyers generate soft mattes with partial transparency values along the edges, preserving the semi-transparent quality of hair wisps, motion blur, and fine detail. The matte may also include core matte processing, where the interior of the subject is guaranteed to be fully opaque regardless of any color similarities to the key color, preventing holes from appearing in areas like green clothing or accessories that happen to match the screen color.