VisionCone3D
_{Simple but configurable 3D vision cone node for Godot}

Compatible with Godot 4.4 - see 4.3 branch for Godot 4.3 compatible version

Adds VisionCone3D, which tracks whether or not objects within its cone shape can be "seen". This can be used to let objects in your game "see" multiple objects efficiently. Default configuration should work for most use-cases out of the box.

Features

• Edit range/angle of cone via 3D viewport editor gizmo
• Debug visualization to easily diagnose any issues
• Works with complex objects that have many collision shapes
• Configurable vision probe settings allow tuning effectiveness and performance to your use-case
• Ignore some physics bodies (eg the parent body)
• Separate masks for bodies that can be seen and bodies that can only occlude other objects
• Includes general-purpose ConeShape3D

Installation

Install via the AssetLib tab within Godot by searching for VisionCone3D

Usage

Add the VisionCone3D node to your scene. Turn on debug draw to see it working. Then you can...

Connect to the body visible signals

These signals fire when a body is newly visible or newly hidden.

func _ready():
    vision_cone.body_sighted.connect(_on_body_sighted)
    vision_cone.body_hidden.connect(_on_body_hidden)

func _on_body_sighted(body: Node3D):
    print("body sighted: ", body.name)

func _on_body_hidden(body: Node3D):
    print("body hidden: ", body.name)

Poll the currently visible bodies

func _process(): # doesn't need to be during a physics frame
    print("bodies visible: ", vision_cone.get_visible_bodies())

Performance Tuning

Vision Test Mode

Center

Samples only the center point (position) of the CollisionShape. Most efficient, but least effective as if the center of a shape is obscured it won't be seen.

vision_cone.vision_test_mode = VisionCone3D.VisionTestMode.SAMPLE_CENTER

Sample Random Vertices

Uses CollisionShape's get_debug_mesh to get a mesh representation of the CollisionShape, then samples random vertex points from that mesh. Effectiveness determined by the max body count and max probe per shape count

vision_cone.vision_test_mode = VisionCone3D.VisionTestMode.SAMPLE_RANDOM_VERTICES
vision_cone.vision_test_max_body_count = 50 # Bodies probed, per-frame
vision_cone.vision_test_shape_max_probe_count = 5 # Probes per hidden shape

Collision Masks

VisionCone3D has 2 collision masks, one used for bodies that can be seen by the cone and one for an environment, which can occlude seen bodies but is not itself probed for visibility.

For example, add the level collision layer to collision_environment_mask and the player/enemy/object collision layer to the collision_mask. The player/enemy/object can then hide behind the level, but no processing/probing will occur on the level collision geometry itself.

The Future

This asset is still in development. I have some ideas for further performance tuning options, and I'm open to feedback on the usability and how to improve documentation or workflows.

2D Support?

I am open to adding a 2D version of this addon if there is sufficient interest.

See if VisionCone2D meets your needs in the meantime. No relation.

Upgrading

0.1.0 -> 0.2.0

v0.2.0 has significant performance improvements. Probably should have waited a few days before publishing. It probably doesn't have any users yet, but just in case...

• Use "Change Type..." on your VisionCone3Ds and select Area3D.
• Use new ConeShape3D for all your cone-y collision needs