Ovrtx Sensor Capture

Customize and validate OVRTX sensor declaration, selection, stepping, and native readback for one or multiple render products. Use when developers add camera-like sensors or debug sensor identity and output routing in the add-on. This skill probes capabilities and does not claim every sensor family is already implemented.

NVIDIA-Omniverse ba8600c 2 files · 1.8 KB Updated

File contents

OVRTX sensor capture

Use the public client in addon/ and its tests. Sensor identity is the USD product path, not a Blender object name or output filename.

  1. Run the focused tests/test_ovrtx_runtime_client.py and tests/test_ovrtx_session.py controls, then inspect a compatible native probe under scripts/ before adapting it.
  2. Supply an explicit fixture, worker, native-client path, sensor paths, matching RenderVar paths, dimensions, time, and caller-owned output JSON.
  3. Require successful worker start, simulation creation with every requested sensor, one bounded sample step, exact per-RenderVar reads, decoded frames, finite dimensions/data, deletion, and shutdown.
  4. Run test_ovrtx_runtime_client.py and test_ovrtx_session.py; retain tests for deduplication, changed-sensor session identity, selected-product routing, iterator pages, and terminal errors.

The current public probes cover LdrColor camera products. For depth, semantic, LiDAR, or another sensor schema, first probe native capabilities and then use the corresponding extension skill. Do not synthesize output from mesh sampling or postprocessing while labeling it native sensor readback.

nvidia-omniverse/omniverse-labs/tree/main/projects/ov-blender-example/skills/ovrtx-sensor-capture commit ba8600c397

Frequently asked questions

npx skillmds@latest add nvidia-omniverse/ovrtx-sensor-capture