Compositing

Compositing is the final stage of visual effects where all separate elements — filmed plates, 3D renders, matte paintings, particle effects, cleanup patches — are combined into the final image. There are two fundamental workflows: layer-based and node-based. Layer-based compositing (After Effects, Motion) stacks elements vertically — each layer sits on top of the one below, with blend modes and opacity controlling how they interact. It's intuitive and fast for motion graphics and simple VFX, but becomes unwieldy on complex shots where you need 30+ elements with different blending rules, mattes, and effects chains. Node-based compositing (Nuke, Fusion in DaVinci Resolve, Natron) connects elements as a graph of processing nodes — each node performs one operation (blur, grade, merge, transform), and the tree structure shows exactly how data flows from source to output. Nuke is the industry standard for feature film VFX because it handles deep compositing, 3D space integration, and multi-pass EXR workflows natively. The alpha channel is the foundation of compositing — it defines transparency. Every pixel has four channels: R, G, B, and A (alpha). Where alpha is 1.0 (fully opaque), the RGB values are visible; where alpha is 0.0 (fully transparent), the pixel is invisible. Semi-transparent edges (hair, smoke, glass) have alpha values between 0 and 1. The critical distinction is premultiplied vs straight alpha. Premultiplied means the RGB values have already been multiplied by the alpha — where alpha is 0, RGB is also 0 (black). Straight (unpremultiplied) means RGB values are preserved regardless of alpha — a fully transparent pixel can still carry color information. Most CG renders output premultiplied alpha. Most keyed footage starts as straight alpha. Mixing them up causes dark or light fringing at edges.

The difference between a competent composite and a convincing one lives in the integration pass. Screen and Add (Linear Dodge) blend modes are used to composite light-emitting elements — lens flares, fire, muzzle flashes, particle glows. Screen adds the lighter pixel value, never darkening the base. Add literally adds the RGB values, which can clip to white quickly on bright areas — use it cautiously. Multiply is for darkening — it multiplies the base color by the blend color, useful for shadow passes and ambient occlusion overlays. Overlay and Soft Light are contrast blend modes that brighten where the base is bright and darken where it's dark — essential for adding texture detail like dust, scratches, or grain without replacing the underlying image. Light wrap is the integration technique that separates amateur composites from professional ones. When a subject stands in front of a bright background, the background light wraps around the subject's edges — you see a subtle rim of environmental color on skin and clothing. To simulate this: blur the background plate, isolate the area around the subject's edge, and composite it as a subtle additive glow. The blur radius and intensity depend on the background brightness — a sunset background needs stronger wrap than an overcast sky. Edge blending handles the transition between the foreground element and the background. A hard matte edge is an immediate giveaway. Professional compositors feather the matte edge by 1–3 pixels, add a subtle color grade match between foreground and background, and apply matching grain or noise to the composited element — CG renders and clean plates often lack the camera noise present in the filmed background, which is a dead giveaway at 4K resolution.