Keying & Compositing Basics

Compositing is the core VFX skill — everything else is a variation on it. At its simplest, you have a foreground element (your keyed actor), a background (the new world), and your job is to stack them so the audience's brain accepts them as a single photograph of a single moment. Pulling the key — telling the software "make the green transparent" — is genuinely the easy part now; a modern keyer does it in a click or two if your green screen was shot well. The hard part, the part that separates convincing work from the "obviously green screen" look everyone recognizes, is everything you do after the key to marry the two layers together. Beginners pull a key, drop it on a background, see the actor appear, and call it done — and it always looks pasted-on. Pros spend most of their time on the integration: the color, the edges, the light, the grain. Let me walk through both halves.

Pulling a clean key

The keying workflow, in plain terms:

• Apply the keyer and pick the green. Use the eyedropper on your green to make it transparent. In Fusion, After Effects (Keylight), or HitFilm, the first pass gets you most of the way if the screen was lit evenly.
• Check the matte, not the picture. View the black-and-white matte: your subject should be solid white, the background solid black, edges crisp. Grey areas mean a soft or holey key — fix those before anything else.
• Despill. Remove the green light that wrapped onto your subject's edges and hair. Almost every keyer has a despill control; without it, you get a tell-tale green fringe that screams "composite."
• Refine the edges. Shrink or soften the matte edge slightly so there's no hard green line or chewed-up hair. A hair's-width of feather reads far more natural than a razor cut-out.
• Garbage-matte the junk. Draw a rough mask to cut out light stands, the edge of the screen, or anything that isn't your subject, so the keyer only deals with clean green.

Making it believable

Now the real work — integration — and this is where you should spend your time, because a technically perfect key still looks fake if the two layers don't share the same world. The human eye is astonishingly good at catching a mismatch even when it can't name it, so you have to consciously unify the layers. First, color and light: grade your keyed subject to match the background's color temperature, brightness, and contrast — a person shot under cool studio light dropped onto a warm sunset will never fool anyone until you warm them up and dim them to match. Look at where the light comes from in the background and make sure your subject's lighting agrees; if it doesn't, at minimum add a matching edge light or shadow so they don't feel lit from the wrong side. Second, edges and wrap: real objects pick up a little color from their surroundings, so a subtle "light wrap" that bleeds a touch of the background color around your subject's edges glues them in. Third, shared texture: your background plate and your foreground almost certainly have different amounts of grain, noise, and sharpness, and matching them — usually adding a little grain and a hair of blur to the too-clean layer — is one of the biggest believability wins there is. Finally, grounding: give your subject contact — a shadow on the floor, feet that meet the ground, a reflection — because nothing says "pasted on" like a person floating with no connection to their world. Do all of this and even a modest green-screen shot on a budget camera will hold up. Skip it and the most expensive key in the world still looks cut out. Compositing rewards patience and observation more than technical wizardry: look hard at your two layers, ask "what's different between them?", and fix each difference one at a time. With keying and compositing under your belt, the next chapter tackles the other essential comp skill — locking your effect to a moving camera through motion tracking.