Active liveness
Active liveness detection asks the user to perform an action on camera — turning their head, blinking, or following an on-screen prompt — and checks that the response matches what a live, responsive person would produce.
Passive liveness
Passive liveness detection works without asking the user to do anything, analyzing signals in a single capture — texture, depth cues, and other characteristics of a real face versus a flat reproduction — to make the same determination.
The trade-off between them
Active methods tend to be harder to spoof with a simple photo but add a step to the user's flow. Passive methods are faster and more frictionless but depend more heavily on the sophistication of what they're analyzing. Many systems combine elements of both.