Debug Drawing Extension API#

Developer Reference

About#

This Debug Drawing Extension API API is used to coordinate groups of lines and points on the screen. Use this API instead of Omniverse’s built-in debug drawing API to have greater control over how the geometry is drawn. The 3D geometry drawn by this remains persistent across frames and is only cleared when desired (unlike the built-in debug drawer).

API Documentation#

See the API Documentation for complete usage information.

Tutorials & Examples#

The following screenshots showcase how the different geometries are drawn:

Points#

Drawing batches of points with different RGBA and radius values:

import random

from isaacsim.util.debug_draw import _debug_draw

draw = _debug_draw.acquire_debug_draw_interface()

N = 10000
point_list_1 = [(random.uniform(-10, 10), random.uniform(-10, 10), random.uniform(-10, 10)) for _ in range(N)]
point_list_2 = [(random.uniform(-10, 10), random.uniform(10, 30), random.uniform(-10, 10)) for _ in range(N)]
point_list_3 = [(random.uniform(-10, 10), random.uniform(-30, -10), random.uniform(-10, 10)) for _ in range(N)]
colors = [(random.uniform(0.5, 1), random.uniform(0.5, 1), random.uniform(0.5, 1), 1) for _ in range(N)]
sizes = [random.randint(1, 50) for _ in range(N)]
draw.draw_points(point_list_1, [(1, 0, 0, 1)] * N, [10] * N)
draw.draw_points(point_list_2, [(0, 1, 0, 1)] * N, [10] * N)
draw.draw_points(point_list_3, colors, sizes)

Lines#

Drawing batches of lines with different RGBA and width values:

import random

from isaacsim.util.debug_draw import _debug_draw

draw = _debug_draw.acquire_debug_draw_interface()

N = 10000
point_list_1 = [(random.uniform(10, 30), random.uniform(-10, 10), random.uniform(-10, 10)) for _ in range(N)]
point_list_2 = [(random.uniform(10, 30), random.uniform(-10, 10), random.uniform(-10, 10)) for _ in range(N)]
colors = [(random.uniform(0, 1), random.uniform(0, 1), random.uniform(0, 1), 1) for _ in range(N)]
sizes = [random.randint(1, 25) for _ in range(N)]
draw.draw_lines(point_list_1, point_list_2, colors, sizes)

Splines#

Drawing splines as filled or dashed between a set of points:

import random

from isaacsim.util.debug_draw import _debug_draw

draw = _debug_draw.acquire_debug_draw_interface()

point_list_1 = [(random.uniform(-30, -10), random.uniform(-10, 10), random.uniform(-10, 10)) for _ in range(10)]
draw.draw_lines_spline(point_list_1, (1, 1, 1, 1), 10, False)
point_list_2 = [(random.uniform(-30, -10), random.uniform(-10, 10), random.uniform(-10, 10)) for _ in range(10)]
draw.draw_lines_spline(point_list_2, (1, 1, 1, 1), 1, True)

Point Cloud Export, Import, and Coloring#

The following standalone examples cover the full point cloud workflow: exporting a Lidar scan to disk, reloading it, and coloring points by a per-point scalar field.

Export a Lidar point cloud to disk

Runs a LidarSensor, captures the first valid scan, and saves it as a point_cloud.npz file that can be consumed by import_point_cloud.py or any NumPy-aware tool:

./python.sh standalone_examples/api/isaacsim.sensors.experimental.rtx/export_point_cloud.py \
    --output-file /tmp/my_cloud.npz

Import and display a saved point cloud

Loads a point_cloud.npz file produced by export_point_cloud.py and renders it in the viewport via an OmniGraph DebugDrawPointCloud node. The --color and --size flags control the uniform RGBA and point radius:

./python.sh standalone_examples/api/isaacsim.util.debug_draw/import_point_cloud.py \
    --input-file /tmp/my_cloud.npz --color 0.0 1.0 0.5 1.0 --size 0.02

Imported point cloud displayed in the Isaac Sim viewport with a uniform teal color.

Color a live Lidar point cloud by distance or intensity

Builds an IsaacExtractRTXSensorPointCloud → IsaacMapScalarsToColors → DebugDrawPointCloud OmniGraph pipeline that colors each point by the chosen scalar field in real time. Use --scalar intensity to switch fields:

./python.sh standalone_examples/api/isaacsim.util.debug_draw/test_lidar_point_cloud_coloring.py \
    --scalar distance

RTX Lidar point cloud colored by distance — near points are blue, far points are red.

Color a live Radar point cloud by radial velocity (Doppler)

The same coloring pipeline applied to an RTX Radar. Target cubes oscillate toward and away from the sensor each frame so the Doppler signal is nonzero. Motion BVH must be enabled (passed via SimulationApp):

./python.sh standalone_examples/api/isaacsim.util.debug_draw/test_radar_point_cloud_coloring.py

RTX Radar point cloud colored by radial velocity — approaching targets are red, receding targets are blue.