Randomizer Nodes

Each randomizer node changes one aspect of the scene. A node receives the Scene when the tree executes and mutates it in place. Nodes are appended to a Randomizer in execution order: instances, object poses, appearance, lights, then the camera.

Import

python

from telekinesis.illusion.randomizer.randomizer_node import (
    RandomizerNode,
    ObjectInstanceRandomizer,
    ObjectPoseRandomizer,
    MaterialRandomizer,
    BackgroundRandomizer,
    LightRandomizer,
    LightPoseRandomizer,
    CameraPoseRandomizer,
)

target_objects is matched by prefix throughout, so passing "part_1" selects every instance registered under that name.

Nodes

ObjectInstanceRandomizer

Samples how many instances of each model are visible in the scene, and reveals them.

ObjectPoseRandomizer

Samples 6-DoF poses for the visible target objects, rejecting poses that collide.

MaterialRandomizer

Assigns a random material to each target object, sampled from the requested material types.

BackgroundRandomizer

Samples an HDRI background, which also drives the scene lighting.

CameraPoseRandomizer

Samples one camera pose per view, so a single randomized scene yields several images.

LightRandomizer

Samples validated light properties such as color, power, size, and cone angle.

LightPoseRandomizer

Samples named light poses with clearance from visible geometry.

Custom Nodes

Add any other axis of variation by subclassing RandomizerNode and implementing randomize().

Next Steps

Workers

Drive a complete, sharded dataset run from a spec YAML instead of assembling the tree in Python.

Flying Things

Build a randomizer tree with one node per axis of variation and render a labeled COCO dataset.

Open tutorial →