nvidia-omniverse/ovrtx/render-settings

Claude Code and agent skill by nvidia-omniverse/ovrtx

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Writing render settings to control rendering behavior. Use when user asks to change render settings, set max bounces, configure path tracing, change render quality, enable DLSS frame generation, or modify RTX settings on a RenderProduct.

Install

npx vskill@latest install nvidia-omniverse/ovrtx/render-settings

SKILL.md

Render Settings

When to Use

Use this skill when the user asks to change render settings, set max bounces, configure path tracing, change render quality, enable DLSS frame generation, or modify RTX settings on a RenderProduct.

Inputs

Resolve inputs in this order: existing repository files and referenced snippets, explicit user request, then broader agent context.

  • Target API surface: Python, C/C++, USD, or a combination.
  • RenderProduct prim path and the exact omni:rtx:* setting attribute to author or update.
  • Desired setting value, USD type, runtime/write timing, and whether reset/warmup behavior matters.
  • Existing scene or snippet that already owns related RenderProduct settings.
  • Repository source snippets referenced below. Treat these snippets as the API source of truth.

Prerequisites

  • Use an ovrtx checkout that contains the referenced examples and docs tests.
  • Read the relevant > **Source:** snippet before writing or explaining API usage.
  • Confirm the setting belongs on the RenderProduct prim, not on a Camera, RenderVar, or renderer creation option.
  • Use writing-attributes when the setting is changed through runtime APIs.

Instructions

  1. Identify the RenderProduct path and the exact RTX setting attribute to change.
  2. Write settings on the RenderProduct prim, not on the Camera or RenderVar prim.
  3. Match the value type to the USD attribute schema and preserve existing authored settings that are unrelated to the request.
  4. Apply the writing-attributes skill when setting values through runtime APIs rather than static USD.
  5. When changing code, run the render-settings docs test that owns the setting snippet whenever practical.

Output Format

  • For explanations, cite the relevant API names, source snippets, and caveats.
  • For code changes, summarize the files changed, snippets affected, and validation run.

Scripts

This skill has no scripts.

Limitations

  • The referenced snippets remain the source of truth; update or add tested snippets before documenting new API usage.

Overview

Render settings in ovrtx are written as attributes on the RenderProduct prim. To change a setting, use write_attribute() in Python or ovrtx_write_attribute() in C, targeting the RenderProduct prim path (e.g., /Render/Camera) with the setting's attribute name.

Settings use the omni:rtx: namespace prefix. For example, to set the maximum number of path tracing bounces, write omni:rtx:rtpt:maxBounces as a uint32 attribute.

After changing a render setting, call reset() and run warm-up frames to allow the renderer to converge with the new setting.

Python

Set a render setting on a RenderProduct
Source: tests/docs/python/test_base.py snippet doc-set-render-setting

C

Set a render setting on a RenderProduct
Source: tests/docs/c/test_base.cpp snippet doc-set-render-setting-c

Key Types / Functions

PythonC
renderer.write_attribute(prim_paths=["/Render/Camera"], attribute_name="omni:rtx:rtpt:maxBounces", tensor=np.array([value], dtype=np.uint32))ovrtx_write_attribute(renderer, &binding, &buffer, OVRTX_DATA_ACCESS_SYNC) with uint32 DLTensor

Render Mode Selection

SettingTypeDefaultDescription
omni:rtx:rendermodetoken"RealTimePathTracing"Render mode: "RealTimePathTracing", "PathTracing", or "MinimalRendering"

DLSS Frame Generation (dlss:)

DLSS Frame Generation inserts interpolated frames between rendered ones. It is off by default and is enabled per RenderProduct, like any other render setting.

Source: examples/c/vulkan-interop/src/main.cpp snippet enable-dlssg
SettingTypeDefaultDescription
omni:rtx:dlss:frameGenerationboolfalseEnable frame generation
omni:rtx:dlss:frameGenerationInterpolatedFrameCountint0Interpolated frames per rendered frame, clamped to the hardware maximum (1 on Ada, 3 on Blackwell). 0 or unset means 1, so request 3 to get the hardware maximum

Enabling frame generation changes the shape of the step results: one step returns output_frame_count == N + 1 output frames instead of 1, ordered with the interpolated frames first and the real rendered frame last. Each frame exposes the same single color render var, and only the last frame carries a meaningful accumulation status. Consumers that assume output_frames[0] is the frame to display or inspect need to be updated before turning this on.

The renderer declines frame generation without failing the step, in which case a step keeps returning a single frame. It requires an Ada or newer GPU that drives the display, the RealTimePathTracing render mode, LdrColor or HdrColor as the only render outputs, and a render resolution of at least 126 pixels per side.

Real-Time Path Tracing Settings (rtpt:)

Sampling and Caching
SettingTypeDefault
omni:rtx:rtpt:cached:enabledbooltrue
omni:rtx:rtpt:lightcache:cached:enabledbooltrue
omni:rtx:rtpt:ris:meshLightsboolfalse
Ray Bounces and Shading
SettingTypeDefault
omni:rtx:rtpt:maxBouncesuint3
omni:rtx:rtpt:extraSpecularAndTransmissiveBouncesuint0
omni:rtx:rtpt:maxVolumeBouncesuint3
omni:rtx:pt:fractionalCutoutOpacitybooltrue
omni:rtx:rtpt:maxRoughnessfloat0.3
omni:rtx:rtpt:translucency:virtualMotion:enabledbooltrue
Firefly Filter
SettingTypeDefault
omni:rtx:rtpt:fireflyFilter:enabledbooltrue
omni:rtx:rtpt:fireflyFilter:maxUnexposedIntensityPerSamplefloat3200.0
omni:rtx:rtpt:fireflyFilter:maxUnexposedIntensityPerSampleDiffusefloat3200.0
omni:rtx:rtpt:fireflyFilter:maxPerEmissiveUnexposedIntensityfloat3200.0
Gaussian Splatting
SettingTypeDefault
omni:rtx:rtpt:gaussian:accumulatedDepth:enabledbooltrue
omni:rtx:rtpt:gaussian:accumulatedAlbedo:enabledbooltrue
omni:rtx:rtpt:gaussian:maxGaussiansToAccumulateint48
omni:rtx:rtpt:gaussian:nonStochastic:enabledboolfalse

Path Tracing Settings (pt:)

Path Tracing
SettingTypeDefault
omni:rtx:pt:samplesPerPixeluint512
omni:rtx:pt:samplesPerIterationint1
omni:rtx:pt:adaptiveSampling:enabledbooltrue
omni:rtx:pt:limits:maxBouncesuint4
omni:rtx:pt:limits:extraSpecularAndTransmissiveBouncesuint2
omni:rtx:pt:maxVolumeBouncesuint15
omni:rtx:pt:limits:maxFogBouncesuint2
omni:rtx:pt:fractionalCutoutOpacitybooltrue
Denoising
SettingTypeDefault
omni:rtx:pt:denoising:enabledbooltrue
omni:rtx:pt:denoising:optix:temporalboolfalse
omni:rtx:pt:denoising:blendFactorfloat0.0
omni:rtx:pt:denoising:optix:denoiseAOVsbooltrue
Sampling and Caching
SettingTypeDefault
omni:rtx:pt:radianceCache:enabledbooltrue
omni:rtx:pt:lightCache:enabledbooltrue
omni:rtx:pt:ris:meshLightsboolfalse
Firefly Filter
SettingTypeDefault
omni:rtx:pt:fireflyFilter:enabledbooltrue
omni:rtx:pt:fireflyFilter:maxUnexposedIntensityPerSamplefloat3200.0
omni:rtx:pt:fireflyFilter:maxUnexposedIntensityPerSampleDiffusefloat3200.0
omni:rtx:pt:fireflyFilter:maxPerEmissiveUnexposedIntensityfloat3200.0
Spectral Rendering

Enable the process-wide spectral material and light backend before creating the first renderer, then enable spectral transport on its PT RenderProducts. Run RGB and spectral comparisons in separate processes; the startup choice cannot change after initialization, including after destroying all renderer instances.

Source (Python startup): examples/python/spectral-rendering/main.py snippet doc-spectral-config-python
Source (C startup): examples/c/spectral-rendering/main.cpp snippet doc-spectral-config-c
Source (USDA): tests/docs/usd/data/spectral_rendering.usda snippet doc-spectral-authoring
Source (USD Python): tests/docs/usd/test_spectral_rendering.py snippet doc-spectral-authoring-python
SettingTypeDefault
omni:rtx:pathtracing:spectral:enabledboolfalse
omni:rtx:pathtracing:spectral:wavelengthMinfloat10.0
omni:rtx:pathtracing:spectral:wavelengthMaxfloat10000.0
renderingColorSpaceuniform token"lin_rec709_scene"
omni:rtx:pathtracing:spectral:responseCurveuint0

renderingColorSpace is an attribute of the RenderSettings prim, not the RenderProduct, and has no omni:rtx: prefix.

Non-Uniform Volumes
SettingTypeDefault
omni:rtx:pt:ptvol:enabledboolfalse
omni:rtx:pt:volumes:transmittanceMethodtoken"biasedRayMarching"
omni:rtx:pt:volumes:tracking:maxScatteringStepsint1024
omni:rtx:pt:volumes:tracking:maxShadowStepsint32
omni:rtx:pt:limits:maxVolumeBouncesuint2
Multi-GPU
SettingTypeDefault
omni:rtx:pt:multigpu:enabledbooltrue
omni:rtx:pt:mgpu:autoLoadBalancing:enabledbooltrue
omni:rtx:pt:mgpu:compressRadianceboolfalse
omni:rtx:pt:mgpu:compressAlbedobooltrue
omni:rtx:pt:mgpu:compressNormalsbooltrue
omni:rtx:multiThreading:enabledbooltrue

omni:rtx:pt:multigpu:enabled is read from the RenderSettings prim, not the RenderProduct.

Global Volumetric Effects
SettingTypeDefault
omni:rtx:pt:ptvol:raySkyboolfalse
omni:rtx:pt:ptvol:raySkyScalefloat1.0
omni:rtx:pt:ptvol:raySkyDomelightboolfalse
Anti-Aliasing
SettingTypeDefault
omni:rtx:pt:pixelFilter:filtertoken"triangle"
omni:rtx:pt:pixelFilter:radiusfloat1.0

View Lighting Mode (Camera Light)

View Lighting Mode replaces all scene lights with a single camera light. By default the camera light is a distant headlight aligned with the view direction; it can be switched to a spot light positioned at the camera. The spot-only settings (radius, cone angle, cone softness) have no effect when the light type is "distant"; the angle setting has no effect when the light type is "spot".

Source: tests/docs/python/test_base.py snippet doc-set-view-lighting

This is the start of the file. Read the full SKILL.md on GitHub.

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
220
Repository forks
33
7-day trend score
27

Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 16 Sep 2026, a month ago.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/nvidia-omniverse/ovrtx/render-settings/badge)](https://verified-skill.com/skills/nvidia-omniverse/ovrtx/render-settings)

Something wrong with this skill or its page? Tell us.