isaaclab_ov.renderers

Sub-module for OVRTX renderer backend.

Additional Public Classes

The following classes are part of the public isaaclab_ov.renderers API.

OVRTXRenderer	OVRTX Renderer implementation using the ovrtx library.
OVRTXRendererCfg	Configuration for OVRTX Renderer.
map_attribute_for_warp_writes	Map binding for CUDA writes and yield its buffer as a Warp array; commit after the fill.

class isaaclab_ov.renderers.OVRTXRenderer

Bases: BaseRenderer

OVRTX Renderer implementation using the ovrtx library.

This renderer uses the ovrtx library for high-fidelity RTX-based rendering, providing ray-traced rendering capabilities for Isaac Lab environments.

Methods:

__init__(cfg)
__new__(*args, **kwargs)

__init__(cfg: OVRTXRendererCfg)

classmethod __new__(*args, **kwargs)

class isaaclab_ov.renderers.OVRTXRendererCfg

Bases: RendererCfg

Configuration for OVRTX Renderer.

The OVRTX renderer uses the ovrtx library for high-fidelity RTX-based rendering. width, height, num_envs, and data_types are obtained from the CameraRenderSpec when create_render_data() is called (same pattern as Isaac RTX).

Methods:

__new__(*args, **kwargs)
__init__([class_type, renderer_type, ...])

classmethod __new__(*args, **kwargs)

__init__(class_type: type[OVRTXRenderer] | str = <factory>, renderer_type: str = <factory>, temp_usd_dir: str | None = <factory>, log_level: str = <factory>, log_file_path: str = <factory>, enable_shadows: bool = <factory>, colorize_semantic_segmentation: bool = <factory>, colorize_instance_segmentation: bool = <factory>, colorize_instance_id_segmentation: bool = <factory>) → None

isaaclab_ov.renderers.map_attribute_for_warp_writes(binding: Any, device: wp.Device | str, dtype: Any) → Iterator[wp.array]

Map binding for CUDA writes and yield its buffer as a Warp array; commit after the fill.

The caller fills the yielded array with Warp work enqueued on device’s current stream (the default for wp.launch/wp.copy). On exit – error or not – the mapping is unmapped with that stream as the CUDA sync, so OVRTX’s commit of the mapped data waits for the fill on the GPU instead of racing it. OVRTX has no discard path (unmap always commits), so a failed fill still publishes whatever landed in the buffer.

Parameters:

• binding – OVRTX attribute binding (from bind_attribute) whose buffer is written.

• device – Warp CUDA device, or its string alias, on which the fill work runs.

• dtype – Warp dtype the mapped tensor is viewed as (e.g. wp.mat44d).

Yields:

The mapped buffer as a zero-copy Warp array, valid only inside the with block.