When Characters Stop Acting
The moment a game character dies, the animator's carefully crafted keyframes end and physics takes over. Ragdoll physics — first popularized by Hitman: Codename 47 in 2000 — simulates a character's body as a chain of rigid segments connected by joints with angular limits. The result: every death is unique, every tumble down stairs produces a different pose, and every explosion sends bodies flying in physically plausible ways.
Bones, Joints, and Constraints
A ragdoll skeleton typically has 10-20 rigid bodies representing the head, torso, pelvis, upper arms, forearms, hands, thighs, shins, and feet. Each connection point is a joint with angular constraints that mirror human anatomy — knees bend backward but not forward, elbows have limited rotation, and the spine flexes within a range. This simulation builds a simplified skeleton and lets you tune joint stiffness to see the difference between rigid and floppy ragdolls.
The Verlet Approach
Most ragdoll implementations use Verlet integration with distance and angle constraints, following Thomas Jakobsen's influential 2001 GDC talk. Each bone endpoint is a Verlet particle; distance constraints maintain bone lengths; angle constraints enforce joint limits. The constraint solver iterates multiple times per frame, each pass reducing constraint violations. More iterations mean stiffer, more realistic joints.
Active Ragdoll: The Future
Pure ragdoll produces limp, lifeless motion. Active ragdoll — used in games like Gang Beasts, Human Fall Flat, and Euphoria-powered titles — adds virtual muscles that apply torques to maintain balance, reach for handholds, or protect the head during falls. The blend of passive physics and active control creates characters that feel genuinely alive even as they tumble and stagger.