Must the screen be green?

No. Blue and other suitable colors can be used depending on the subject and workflow.

Is the matte only black and white?

It can contain intermediate opacity values for mixed or translucent regions.

Does removing spill change transparency?

Spill correction concerns color, although tools may combine operations; it should not be confused with matte generation.

AI Glossary / Computer vision and multimodal AI / Intermediate

Chroma keying

Definition

Chroma keying is a compositing technique that separates a subject from a background using color information, commonly from a green or blue screen. It estimates a transparency matte and usually includes edge and color-spill handling so the subject can be combined with a different background.

What it means

A deliberately chosen background color makes separation easier when it differs from the subject. Pixels dominated by that color are treated as background, while subject pixels remain visible. Mixed pixels around hair, motion blur or transparent material need intermediate opacity rather than a simple keep-or-delete rule.
The matte describes how much foreground contributes at each pixel. A convincing composite also needs suitable foreground color, because reflected screen light can tint the subject and mixed edge pixels contain background color.
Chroma keying is not the same as any automatic background-removal model. It relies on color separation and capture conditions, although modern workflows can combine keying with masks and other methods.

How it works

Capture the subject against a reasonably uniform screen with lighting and color choices that support separation. Select or estimate the screen color, then calculate a matte from color differences and other constraints. Refine uncertain boundaries, use additional masks where needed and remove or compensate for unwanted screen-color spill.
Composite the foreground with the new background using the matte, taking care with straight versus premultiplied color representation. Inspect edges against several backgrounds and through motion. Knowing the screen color helps but does not alone solve every foreground-color and opacity ambiguity; reflections, similar subject colors and transparency can require additional judgment or information.

Three practical examples

A presenter is recorded against a green screen and composited into a virtual studio.
A product video uses a controlled blue background to support replacing the backdrop in post-production.
A visual-effects team combines a color key with hand-drawn masks to preserve difficult foreground details.

Think of it this way

Think of using a distinctive backdrop as a visual label for “replace this area.” At clean edges the label is clear; around hair or reflections it becomes mixed with the subject. The analogy explains why capture planning helps, but keying is not simply deleting a named color everywhere, since the subject may contain that color too.

Worked example

Use a simplified linear-light RGB model I=alpha F+(1−alpha)B. Let foreground color F=(200,50,50), screen color B=(0,200,0) and alpha=0.5. The observed mixed pixel is I=(100,125,25).
If the foreground color and alpha are correctly recovered, compositing over a new blue background C=(0,0,200) gives 0.5F+0.5C=(100,25,125). Merely deleting all green pixels would not produce this correct mixed edge.
In practice F and alpha are unknown, and the observed RGB value supplies only limited information. The known screen color reduces ambiguity but does not generally determine every unknown without further assumptions. The numbers are illustrative in a stated linear representation; applying the same arithmetic blindly to differently encoded color values can produce incorrect appearance.

Illustrative case study

Fictional case: Pine Studio records a presenter wearing a reflective jacket against a green screen. The first key removes the background but leaves green-tinted edges and holes in reflected areas of the jacket.
The editor separates matte refinement from spill correction, uses a holdout mask where appropriate and inspects the result over both light and dark backgrounds. The production team adjusts lighting and wardrobe choices for the next capture. This case illustrates capture and compositing trade-offs; it does not claim that one keyer can perfectly recover all reflective or transparent subjects from a single image.

Common misconceptions and mistakes

Chroma keying is not simply removing every green or blue value. Subjects can contain similar colors, and mixed boundary pixels need fractional opacity. A good-looking matte does not guarantee correct foreground color or spill handling.
Do not confuse spill suppression with making holes in the alpha channel. Compression, uneven lighting and motion blur can make separation harder. Premultiplication errors can create dark or colored fringes. A screen's known color helps, but does not automatically make all foreground and transparency recovery mathematically unambiguous.

Business use

Chroma keying supports reusable video production, virtual backgrounds and visual effects when capture can be controlled. Good preparation can reduce manual cleanup and improve consistency across many clips.
Plan background color, subject materials, lighting, camera settings and compression before filming. Test difficult details such as hair, glasses, reflections and motion. Evaluate the composite over the intended background and at delivery resolution, while also checking alternative backgrounds that reveal edge problems. Record the matte and color-processing steps so revisions remain manageable. The most efficient workflow often combines thoughtful capture with selective post-production rather than relying on an aggressive key alone.

Check your understanding

Hair at the edge of a subject contains a mixture of foreground and screen color. Is a binary delete-or-keep rule always sufficient?

Answer: No. Mixed pixels often need fractional opacity and appropriate foreground-color treatment to composite naturally over a new background.

Related terms

Alpha matte
compositing
green screen
blue screen
colour spill
premultiplied alpha
image matting
holdout mask

Learn next

Learn the compositing equation and alpha representation next. Compare a binary mask with a soft matte around hair or motion blur, then inspect spill separately from transparency. Test the same foreground over several backgrounds to expose edge and premultiplication problems.

Sources

Adobe, Keying in After Effects: https://helpx.adobe.com/uk/after-effects/desktop/animate-in-after-effects/keying/keying.html Foundry, Keylight reference: https://learn.foundry.com/nuke/17.1/content/reference_guide/keyer_nodes/keylight.html Official documentation excerpts consulted. The compositing calculation is original; full colour-workflow and editorial review remain pending.

Previous: Chow - Liu tree Next: Chroma subsampling