AI Subject Selection
Automatic masking that identifies a subject, sky, background or facial feature using a trained model.
In depth
Trained masking recognises semantic content — this is a person, this is their iris, this is sky — rather than looking for tonal boundaries, so it handles hair, foliage and translucent edges far better than any threshold-based selection. The masks it generates can still be refined by hand, which is usually necessary at fine edges.
Examples in practice
One click produces a mask separating a subject's windblown hair from a busy background in a second or two.