Use rep.distribution.uniform() and rep.distribution.choice()

Omniverse Synthetic Data Generation (SDG)
1

Hello,

I want to position one object in one of 3 areas randomly.
For the implementation, I combined rep.distribution.uniform() and rep.distribution.choice(). But the program just stops without an error message.

To reproduce that issue, I created the following example:

import omni.replicator.core as rep

import omni.usd

with rep.new_layer():

    camera = rep.create.camera(position=(1000, 1000, 1000), look_at=(0,0,0))

    render_product = rep.create.render_product(camera, (1024, 1024))
   
    torus = rep.create.torus(semantics=[('class', 'torus')] , position=(0, -200 , 100),count=3)

    plane = rep.create.plane(scale=10, visible=True)
    
    

    def get_shapes():
        shape = rep.get.prims(semantics=[('class', 'torus')])

        with shape:
            position1 = rep.distribution.uniform((0, 0, 0), (1, 1, 1))
            position2 = rep.distribution.uniform((1, 1, 1), (2, 2, 2))
            position3 = rep.distribution.uniform((2, 2, 2), (3, 3, 3))
            
            positions = rep.distribution.choice(choices=[position1, position2, position3])
            
            rep.modify.pose(
                rotation=rep.distribution.uniform((0, 0, -180), (0, 0, 180)),
                position=positions)

            rep.randomizer.color(colors=rep.distribution.uniform((0, 0, 0), (1, 1, 1)))
        return shape.node


    rep.randomizer.register(get_shapes)

    with rep.trigger.on_frame(num_frames=10,rt_subframes=10):
        rep.randomizer.get_shapes()
        
        
    # Initialize and attach writer
    writer = rep.WriterRegistry.get("BasicWriter")
    writer.initialize(output_dir=f"/path/to/Data", rgb=True, bounding_box_2d_tight=False, bounding_box_2d_loose=True, semantic_segmentation=False)
    writer.attach([render_product])

Using the scatter function is not an option, as the objects rotatation won’t vary.

Happy about any information.

2

Hello @hugo.81243 ,

Unfortunately, randomizing distributions is not currently supported. Rest assured we have this feature high on our roadmap and understand its importance. Please let us know if this is a blocker for you and we can try to work together to determine a workaround that you can use in the mean time.