Mask

You need to edit a specific area of an image without affecting the rest – that's precisely what a mask is for in compositing. It isolates areas pixel by pixel, whether for rotoscoping, color correction, or compositing layers. On set, you don't think in terms of masks; in editing and especially in the VFX pipeline, it becomes part of your daily work.

A mask functions like a digital stencil. You draw or generate a shape – a polygon, a Bezier curve, or you can have it calculated from luma and chroma values – and place it over your layer. Everything outside becomes invisible, everything inside is editable. In rotoscoping, you trace masks frame by frame to isolate figures or track objects. In keying, you use masks to refine green screen areas. For color correction, you limit contrast, saturation, or curves to a face, a wall, a car – without global adjustments.

In practice, you work with multiple mask types in parallel: a rough roto mask isolates the figure, a refined matte works out the hair edges, an additional mask allows for local color adjustments. Modern compositing software – Nuke, After Effects – also offers intelligent mask tools with motion tracking and AI-powered auto-roto, but manual control remains essential for precision. The mask edge is critical: too hard looks unnatural, too soft loses control. This is where feathering, erosion, and dilation play their role.

A common mistake: creating masks without planning for motion. If your mask remains static for four seconds, you'll immediately see that something is wrong. For dynamic shots, you need keyframed masks or tracking-based solutions. In motion tracking, a defined mask is carried along with the movement. This saves you hundreds of frames of manual work – provided the track is accurate. Regular preview renders with different background colors help to spot mask errors early, especially at transitions and in fast cuts.