← Learn library

Unreal Engine & Blueprints

Setting Up Rag Doll Physics

Channel: CGI Nerd Unreal Engine ·

Written guide

Original CGI Nerd Unreal Engine video thumbnail
Setting Up Rag Doll PhysicsCGI Nerd Unreal Engine · Video hosted on YouTube

Playing loads YouTube content. YouTube processes data after activation.

Watch on YouTube

If the player is blocked or unavailable, use Watch on YouTube.

Software: Unreal Engine

This lesson adds a reusable Death event to the side-scroller character and uses a temporary keyboard input to activate ragdoll physics. It works with the existing character skeleton and physics asset, enabling collision and simulation on the mesh rather than authoring a new rig or physics-body layout.

Health-driven death is discussed as a future trigger, not implemented here. The exact Unreal Engine release is unidentified. The final demonstration also leaves the original camera and character-control behavior active, so the visible ragdoll result is not a complete gameplay death state.

Create the event and a temporary test caller

At 00:01–02:03, the instructor opens the SideScrollerCharacter Blueprint and creates a custom event named Death. The number-five keyboard event calls it so the physics behavior can be tested before environmental or health logic exists.

The event name organizes the operation; it does not automatically give the engine a complete death system. The graph still needs explicit instructions for collision and simulation, and later lessons can replace the test caller while retaining those instructions.

Compile and confirm that the key reaches the intended character Blueprint before interpreting a missing physical response as a problem with the skeleton.

Separate collision permission from physics simulation

During 02:03–04:01, collision is enabled on the character’s Mesh component. Pressing five at this stage produces no visible ragdoll because simulation has not yet been activated.

The instructor opens the existing skeleton and physics asset to show the bones and capsule-like physics bodies. The asset’s simulation preview illustrates the kind of body response the gameplay event will use. This inspection does not create or tune those bodies; the template already supplies them.

Collision and simulation are separate responsibilities. Enabling collision allows the mesh’s physical representation to participate in contact, while enabling simulation lets the selected bodies move under physics. One setting does not replace the other.

The source does not read out every collision-mode or response-channel setting. Inspect the actual mesh configuration if contacts differ from the demonstration instead of applying an invented complete preset.

Enable the bodies below the pelvis hierarchy

At 04:01–06:00, the instructor searches through several similarly named nodes before choosing Set All Bodies Below Simulate Physics for the Mesh component. A literal Name value supplies pelvis as the bone name, and New Simulate is enabled.

“Below” refers to the skeleton hierarchy, not simply the physical area below the waist. Descendants of the chosen bone can include much of the character, depending on the skeleton. Inspect the hierarchy shown in the source asset before assuming the same name covers the same bodies on another character.

The transcript identifies the pelvis name and enabled simulation but does not establish every optional pin, including the precise include-self choice. Preserve that limitation when reproducing the setup; do not confuse the clearly named starting bone with a fully specified body-selection configuration.

The intended execution order is to enable the mesh’s collision and then activate simulation for the selected hierarchy. The initially considered blend-weight node is not the final simulation node used in the demonstrated graph.

Test moving and airborne transitions, then inspect the camera problem

During 06:00–07:49, pressing five changes the character into a simulated ragdoll. Tests include standing, moving toward a wall, and jumping before activation. The body’s existing motion affects how it falls and collides after simulation takes over.

The camera does not automatically become a ragdoll-following camera. It remains associated with the original character setup, can lose sight of the simulated body, and is still controllable. The instructor explicitly leaves that issue for another video.

Do not interpret a successful collapse as proof that movement input, camera control, or the rest of the character lifecycle has been disabled. No health threshold, repeated-event guard, respawn, or complete animation-to-physics recovery is demonstrated here.

The reusable result is the mesh simulation event. Casting from an environmental Blueprint later replaces the temporary key caller with a spike interaction, and Switching Cameras addresses the related viewing problem. Those are separate stages rather than behavior already supplied by this event.