Compositing is the process of combining visual elements from separate sources into a single image or shot so that they appear to belong to the same scene. In motion graphics & VFX for video, it is the stage where live-action plates, computer-generated renders, matte paintings, graphics and practical elements are layered, matched and integrated into the final frame.
A compositor works with mattes, color, light and lens behavior rather than with the original shoot or 3D scene. The goal is photographic believability for visual effects, or clean, controlled design for motion graphics, and in both cases the viewer should not notice where one source ends and the next begins.
How compositing combines image layers
Compositing rests on the alpha channel, a per-pixel value that defines how opaque each pixel of a layer is. The basic over operation places a foreground on a background by weighting each pixel by its alpha, and nearly every complex composite is built from many such operations.
Most professional work uses premultiplied alpha, in which the color values are already multiplied by the alpha. Mixing premultiplied and straight alpha incorrectly is a classic cause of dark or bright fringes around edges, so compositors track the alpha state of every element.
Mattes come from several sources, each suited to different footage. The table below lists the main ways a compositor obtains them.
| Matte source | How it is produced | Best suited for |
|---|---|---|
| Chroma key | Isolating a green or blue backing color | Planned studio shots |
| Rotoscoping | Drawing animated shapes by hand | Shots without a clean backing |
| Luminance key | Separating by brightness | Fire, smoke, sky against dark areas |
| Render passes | Alpha output from 3D software | Computer-generated elements |
| Machine learning segmentation | Automatic subject detection | Fast roughs and garbage mattes |
Layer-based and node-based compositing
Compositing software follows one of two models: a layer stack, familiar from motion graphics tools, or a node graph, standard in feature-film VFX. Both produce the same mathematics, but they scale very differently as shots grow in complexity.
| Aspect | Layer-based | Node-based |
|---|---|---|
| Structure | Stack of layers, top to bottom | Graph of connected operations |
| Typical work | Motion graphics, titles, design | Feature VFX, complex plates |
| Reuse of results | Precomps nest layers | Any output feeds many branches |
| Readability at scale | Becomes hard beyond dozens of layers | Stays traceable in large shots |
| Learning curve | Gentle | Steeper |
In practice, many studios use both models: motion designers build animated graphics in a layer tool, and compositors finish photoreal shots in a node tool. The choice depends on shot complexity and team pipeline, not on image quality.
Making elements match: the integration toolkit
A composite looks real only when every element shares the same light, color, lens and motion characteristics as the background plate. Most of a compositor’s time goes into this integration rather than into the initial layering.
| Task | What is matched | Typical tools |
|---|---|---|
| Color matching | Black level, white point, saturation | Grade and color correction nodes |
| Edge treatment | Softness and spill around the matte | Edge blur, despill, light wrap |
| Grain and noise | Sensor texture of the plate | Grain analysis and regraining |
| Lens effects | Distortion, chromatic aberration, defocus | Lens distortion and defocus nodes |
| Motion blur | Shutter behavior of the camera | Vector blur from motion data |
| Camera movement | Position and perspective over time | Motion tracking and camera solves |
Motion tracking and optical flow provide the data that locks inserted elements to the plate, while keyframing controls changes that cannot be tracked automatically. A static logo on a moving wall, for instance, needs a planar or 3D track to stay in place.
Color management and image data
Modern compositing is done in a scene-linear working space, where pixel values are proportional to light. Blending, blurring and defocus only behave like real optics when the math happens in linear light, not in a display-encoded gamma space.
Plates and renders are commonly stored as 16-bit half-float OpenEXR files, which hold values above 1.0 for highlights and can carry multiple render passes in one file. A color management system such as ACES or OpenColorIO converts each source into the working space and back to the delivery format.
Common compositing mistakes
Most failed composites share a short list of recurring causes, and experienced reviewers spot them within seconds. The table pairs each symptom with its usual cause and fix.
| Symptom | Likely cause | Fix |
|---|---|---|
| Dark halo around subject | Alpha premultiplied twice | Unpremultiply before color work |
| Element looks pasted on | Black levels do not match | Match blacks and highlights to the plate |
| Green tint on hair edges | Uncorrected spill | Apply despill and edge color correction |
| Element looks too sharp | Missing grain, blur or defocus | Regrain and match lens softness |
| Element drifts against plate | Weak or partial track | Refine the track or solve the camera |
In everyday language
Compositing is like making a collage that does not look like a collage. Pictures from different places, such as an actor filmed in a studio and a city skyline created on a computer, are cut out and stacked on top of each other.
The hard part is not the stacking but making everything look like it was filmed together. The compositor adjusts colors, shadows, blur and film grain until the pieces blend, so the viewer simply sees one believable scene.
Seen in the real world
A car commercial is shot on an empty mountain road on a gray day, but the director wants a sunset sky and a city skyline in the distance. The compositor receives the 4K plate, a computer-generated skyline render with separate light and shadow passes, and a matte painting of the sky.
The compositor first tracks the moving camera, then pulls a luminance key to separate the overcast sky from the mountains and refines the edges with rotoscoping around trees. The skyline is placed into the tracked scene, graded to match the warm sunset, softened with atmospheric haze and given the plate’s grain. After a few review rounds, the final shot shows a car driving toward a glowing city, and nobody watching can tell that the sky and city were never there.