Generated banner for Omniverse Realtime Viewer
Generated logo for Omniverse Realtime Viewer

Omniverse Realtime Viewer

NVIDIAOfficialCreated by NVIDIA Omniverse

  • Skill
  • Tool integration
  • Software development
  • Apache-2.0
  • Medium risk

From the author

Use as the top-level router for Omniverse Realtime Viewer USD apps, XR delivery, SPG GPU processing, stylized rendering, and focused viewer reference documents.
— the description by NVIDIA Omniverse

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

The Omniverse Realtime Viewer skill acts as a central routing and architectural framework for developing USD-based 3D viewer applications. It provides a structured approach to composing rendering, scene state, and delivery mechanisms, ensuring that developers can integrate OVRTX rendering, OVStage scene management, and various UI or streaming adapters effectively. The skill emphasizes a modular architecture where the viewer runtime, UI/input adapters, and delivery systems are decoupled, allowing for consistent behavior across local and streamed environments.

What it does

  • Provides a routing architecture for USD viewer apps, XR delivery, and GPU-accelerated rendering.
  • Defines the separation of concerns between viewer runtime, UI adapters, and delivery systems.
  • Offers guidance on integrating OVRTX for high-fidelity 3D rendering.
  • Establishes conventions for camera interaction, scene state management, and input handling.
  • Supports the implementation of specialized features like stylized rendering and CAE/CFD visualization.
  • Provides protocols for agent-to-UI communication and scene control.

When to use it

  • When starting the development of a new USD-based realtime viewer application.
  • When integrating advanced rendering features like stylized visualization or AOV switching.
  • When building XR or spatial editing capabilities into an existing Omniverse application.
  • When you need to standardize the architecture for local and streamed viewer deployments.
  • When implementing chat-driven scene control or agent-based inspection of a 3D scene.

Advantages

  • Promotes a clean, modular architecture that separates rendering logic from UI and transport.
  • Ensures consistent behavior across different delivery methods, such as local windows or WebRTC streaming.
  • Provides clear paths for extending functionality without disrupting the core viewer runtime.
  • Facilitates the integration of complex simulation and visualization data into the viewer.
  • Standardizes conventions for scene state, selection, and camera control.

Usage examples

“How do I set up a basic USD viewer application using the Omniverse Realtime Viewer architecture?”
The agent will guide you to start with the viewer app reference, resolve dependencies, and implement the core runtime owner to manage the OVStage and OVRTX renderer.
“I need to add stylized rendering to my existing viewer. Where should I start?”
The agent will direct you to the stylized visualization references to implement the required USD/Lua/CUDA nodes within your existing runtime owner.
“How can I enable an AI agent to control the viewer scene?”
The agent will point to the A2UI Scene Copilot reference to implement an operation interface that allows the agent to send commands through the existing runtime owner.
“How do I switch between different render outputs like beauty and depth in my viewer?”
The agent will explain how to use the AOV switching reference to map RenderVars and update the RenderProduct's orderedVars through the runtime owner.

Install

  1. Download the skill as a ZIP:

    https://autogineer.com/agent-skills/nvidia/omniverse-realtime-viewer/download
  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip omniverse-realtime-viewer.zip -d .agents/skills

Documentation by NVIDIA Omniverse

<!-- SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. -->

<!-- SPDX-License-Identifier: Apache-2.0 -->

Omniverse Realtime Viewer

Build responsive USD viewers by composing OVRTX rendering, OVStage scene state, and the delivery/UI libraries appropriate to the app. Use this package for viewer architecture and interaction; use each selected SDK's upstream guidance for installation and exact release-specific APIs.

Start With The App

  1. When delivery is explicit, select its recipe through references/routing.md and follow its current phase. Otherwise start with references/usd-viewer-app/README.md. For an existing app, keep its architecture and read only the references for the requested change. Broad app, conventions and UI catalogs are lookups as needed.

  2. Resolve the selected libraries through references/dependencies/README.md. Preserve the project's manifest unless an upgrade is requested.

  3. Build a working frame and camera interaction in the chosen shell, then add the requested capabilities. Read references/conventions.md for shared input, camera, selection, and protocol conventions.

  4. Use references/viewer-ux-workflow/README.md for broader interface design. Loading, selection feedback, errors, and input responsiveness belong in the first working app, alongside rendering.

  5. Validate the actual delivery path with references/validation.md. A browser screenshot does not validate a native window or a different transport.

Architecture

OwnerResponsibility
Viewer runtimeOwn the live OVStage, publication clock, OVRTX renderer, scene changes, and render stepping.
UI and input adapterDisplay frames and copied app state; send user intent to the runtime owner.
Storage/query/simulation workersPerform their assigned work and return results through the selected integration boundary.
Delivery adapterPresent locally or stream through ovstream; respect frame ownership and backpressure.

Use OVRTX for USD/3D rendering. Browser apps display its WebRTC video; local apps present its output through their native shell. A missing GPU means rendering is unverified, not a reason to substitute a browser 3D renderer.

Keep viewer cameras, settings, effects, and transient edits in app-owned session or runtime state so the source USD stays intact. Serialize scene mutation and render stepping through the runtime owner. Read references/ovstage-ovrtx-integration/README.md for that loop, then the population or data-plane references when implementing scene loading or attribute writes.

Keep expensive loading and metadata work off the UI and render threads. Keep transport details behind a delivery adapter so camera, selection, properties, and loading state can be reused across local and streamed shells.

Add Capabilities As Needed

Use references/routing.md to select camera/picking, scene management, render settings, cloud assets, simulation, editor tools, deployment, or CAE/CFD visualization. Read the selected capability's implementation flow and its linked contract; the full reference catalog is not a prerequisite for a first frame.

For simulation, use the routing reference's Simulation Ecosystem section plus references/realtime-sim-integration/README.md. For UI-driven forces or control, add references/interaction-round-trip/README.md.

For example scenes, use references/usd-sample-data/README.md. When a runtime failure needs diagnosis, use references/troubleshooting/README.md and the selected SDK's matching release guidance before changing the architecture.

Solid Modeling: Upstream First

For usd-brep, exact BRep, solid modeling, kernel failures, CAD topology, or BrepArray questions, first read references/dependencies/README.md → USD BRep. Follow the owning repository's current agent guidance, skills, operation docs, and package API before selecting local integration references. Discover the upstream router rather than assuming skill names or directory layouts remain fixed. Upstream owns modeling semantics, schema rules, tuning, and diagnostics; this package owns OVRTX display, worker handoff, selection mapping, and publication timing.

GPU Processing And Stylized Viewers

For custom GPU processing, chained outputs and lens sampling, read references/spg-processing/README.md. For toon, ink, technical/X-Ray styles, geometry strokes or full wireframe, continue to references/stylized-visualization/implementation.md. Its USD/Lua/CUDA starting node and geometry/depth/composite build sequence work with the selected viewer recipe; use references/stylized-visualization/style-recipes.md for the preset math and references/stylized-visualization/README.md for supporting contracts. Implement the result inside the existing OVRTX/runtime owner and deliver the published processed output. These instructions do not require a sample checkout or a prior stylized-rendering implementation.

For an optional usd-brep geometry source, first read references/dependencies/README.md → USD BRep and the owning repository's current agent guidance, skills, and operation docs; then read references/stylized-visualization/usd-brep-source.md for the display boundary.

XR Viewers

For headsets, immersive viewers, or spatial editors, start with references/xr-viewer-recipe/README.md. Resolve SDK setup and contracts through references/dependencies/ovxr.md; the library owns OpenXR lifecycle and frame submission. For controllers, hands, navigation, or scene tools, add references/xr-spatial-input/README.md and references/ovxr-python-interactions/README.md. For projection-integrated UI, read references/ovxr-composition-ui/README.md and references/ovxr-composition-ui-validation/README.md.

Files

references/cae-cfd-visualization/ovui-controls.md

<!-- SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. -->

<!-- SPDX-License-Identifier: Apache-2.0 -->

CAE/CFD Visualization: ovui Window Shell And Interactive Controls

Triggers

Use this reference for the ovui window shell plus the interactive header controls that drive a CAE/CFD visualization pipeline: representation (points/surface/isosurface/slice/streamlines) and field selectors, iso-value and slice-position sliders, colormap choice, orbit-camera navigation over a rendered viewport, and sample-preset one-click scenes. It documents the thread-safe path from an ovui callback to a single serialized render loop that owns every stage rebuild and renderer.step.

This is the interactive-controls surface for the CAE/CFD viewer. For the generic window bootstrap read local-viewer and ovui-local-viewer-recipe; for button normalization and click-vs-drag dispatch read viewer-input-routing; for control widget selection read viewer-control-patterns; for renderer ownership, ordinals, and the write floor read the ovstage-* references and conventions.md.

Ownership Model (Non-Negotiable)

One serialized loop owns every OVStage write and every renderer.step. ovui callbacks run on the UI thread and MUST NOT touch OVStage or the renderer directly. A callback does exactly one of two things:

  • mutates local, numpy-only input/camera state (through the InputController), or

  • enqueues a command onto a thread-safe queue that the loop drains.

Because the loop drains all queued commands — including stage rebuilds — before it ever calls renderer.step, a step can never run while the stage is being rebuilt. This is the same single-owner rule as conventions.md, applied to an interactive CAE control surface.

ovui Shell Considerations

Use these behavior checks with the currently installed ovui API. Do not generate defensive hasattr/alternate-bootstrap branches around them unless a targeted compatibility check proves the wheel changed the contract; if the wheel changed, update this reference instead.

  • ui.init("...", width=W, height=H, max_fps=60) returns None. There is no app handle to hold, and in particular there is no app.step_and_present().

  • Run ui.run_async() alongside the update coroutine in an application-owned asyncio loop. Use run_until_closed from local-viewer to cancel and await both tasks before destroying the runtime, window, and backend.

  • The render-loop coroutine calls runtime.tick(dt) then await asyncio.sleep(1.0 / 120.0) — not sleep(0). A busy spin (sleep(0)) pegs a CPU core and starves the OS message pump, which reads as a hung window. Yield at ~120 Hz instead.

  • Create one fill-window: ui.Window("...", fill_app_window=True, flags=ui.WINDOW_FLAGS_NO_TITLE_BAR). Wrap the flags= construction in try/except and fall back to a no-flags ui.Window(..., fill_app_window=True); some wheel signatures reject the flags kwarg.

  • Viewport = ui.ByteImageProvider drawn with ui.ImageWithProvider(provider, fill_policy=ui.IwpFillPolicy.IWP_PRESERVE_ASPECT_FIT). Push frames with provider.set_data_array(rgba_uint8, [w, h]) where the array is C-contiguous HxWx4 uint8 RGBA (no B/R swap) and the size list is [width, height].

  • Layer a transparent ui.Rectangle with opaque_for_mouse_events = True ABOVE the image, inside the image's own ui.ZStack. The rect captures camera drags over the viewport while the header controls above the ZStack stay clickable.

import asyncio, time
import omni.ui as ui

ui.init("CAE ovui Viewer", width=W, height=H, max_fps=60)   # returns None
provider = runtime.viewport.provider

try:
    window = ui.Window("CAE ovui Viewer", width=W, height=H,
                       fill_app_window=True, flags=ui.WINDOW_FLAGS_NO_TITLE_BAR)
except Exception:
    window = ui.Window("CAE ovui Viewer", width=W, height=H, fill_app_window=True)

with window.frame:
    with ui.VStack(spacing=0):
        build_header()                       # ComboBoxes + FloatSliders (below)
        with ui.ZStack(height=ui.Fraction(1)):
            ui.ImageWithProvider(provider,
                fill_policy=ui.IwpFillPolicy.IWP_PRESERVE_ASPECT_FIT)
            hit = ui.Rectangle(style={"background_color": 0x00000000})
            hit.opaque_for_mouse_events = True   # captures camera drags
            wire_hit_rect(hit, runtime)          # mouse callbacks -> InputController

async def render_loop():
    prev = time.time()
    while True:
        try:
            now = time.time(); dt = now - prev; prev = now
            runtime.tick(dt)                 # advances state, pushes frames
        except Exception as exc:             # never let one frame kill the loop
            print(f"[app] render_loop error: {exc}", flush=True)
        await asyncio.sleep(1.0 / 120.0)     # NOT sleep(0)

# Use the run_until_closed helper from local-viewer in the application.
try:
    asyncio.run(run_until_closed(render_loop(), ui.run_async()))
finally:
    runtime.close()  # app-owned cleanup: stop/join the render owner
    window.destroy()
    ui.shutdown()

Use ui.color("#76b900") or ui.color(r, g, b, a) for style colors instead of packing integers by hand. The helper produces ovui's 0xAABBGGRR format.

Header Controls (ComboBoxes + FloatSliders)

The header carries: a representation ComboBox (points/surface/isosurface/slice/ streamlines), a field ComboBox (populated from the loaded dataset), an iso-value FloatSlider (0–1 fraction of the field range), a slice-position FloatSlider (0–1 along the slice axis), and a colormap ComboBox. Each control's callback ONLY enqueues a command — it never touches OVStage or the renderer. Wrap every callback in try/except; an unguarded ovui callback exception can tear down the app loop.

  • ComboBox: combo = ui.ComboBox(default_idx, *items), then combo.model.add_item_changed_fn(callback). Read the selection inside the callback with model.get_item_value_model().as_int and index into the item list; guard the index and fall back to a sensible default when out of range.

  • FloatSlider: slider = ui.FloatSlider(min=0.0, max=1.0), seed it with slider.model.set_value(initial), then slider.model.add_value_changed_fn(callback). Read with model.get_value_as_float().

  • Populate the field ComboBox from real dataset field names. Pre-load the dataset once before building the header (the reference calls runtime._ensure_dataset() and reads runtime.fields) so the dropdown shows actual fields rather than a guess; fall back to the current field name if the dataset has not loaded.

def _on_representation_changed(model, *_):
    try:
        idx = model.get_item_value_model().as_int
        mode = REPRESENTATIONS[idx] if 0 <= idx < len(REPRESENTATIONS) else "surface"
        runtime.enqueue_representation(mode)          # enqueue only
    except Exception as exc:
        print(f"[app] representation callback error: {exc}")

def _on_iso_changed(model, *_):
    try:
        runtime.enqueue_iso(float(model.get_value_as_float()))   # 0..1 fraction
    except Exception as exc:
        print(f"[app] iso callback error: {exc}")

The enqueue methods (enqueue_load, enqueue_colormap, enqueue_representation, enqueue_field, enqueue_iso, enqueue_slice) each put a (kind, payload) tuple on the runtime's queue.Queue. They are the only UI-thread-safe way to change pipeline state.

The Single Serialized Loop

The UI thread enqueues intent; the owner coalesces it and dispatches bounded scientific work. It continues camera interaction and rendering while computation runs, then publishes copied results for the current request and generation. Classify the result with control-to-publication routing: colors and geometry arrays update existing prims, while type/composition changes replace affected structure. A new dataset uses the scene-replacement path.

tick(dt):
    accept and coalesce compatible UI intent
    dispatch scientific work if its bounded slot is free
    accept completed current-generation results
    publish changed arrays or commit a prepared scene replacement
    commit camera changes and render the completed publication

Keep local control drafts until the corresponding accepted state arrives; an intermediate result must not overwrite a newer drag target. Reset and preset replacement discard superseded drafts. Fit on a new dataset only when that is the selected camera policy; field, colormap and iso/slice edits preserve the view.

_render_current() enforces the write-floor ordering from conventions.md and the ovstage-* references: get the row-major camera xform, commit_camera_pose writes omni:xform at a new ordinal and advances the write floor, then step_extract(ordinal=...) renders at that ordinal — never above the floor — and the resulting RGBA is pushed to the viewport provider.

Camera Navigation

Implement pure NumPy camera math (OrbitCamera in the suggested camera.py) using camera controls, so it works before the OV stack is installed. Implement the InputController in the suggested input.py using input routing; it translates ovui mouse events into camera moves and flags the runtime camera-dirty; it never writes OV state.

  • Gestures: orbit (spherical azimuth/elevation), pan (move target in the camera right/up plane), dolly (exponential distance change), zoom (wheel), and fit (frame an AABB: target = center, distance from max extent and FOV with a 1.5× margin, plus derived near/far clip).

  • Button remap (matches conventions.md): {0: orbit, 2: pan, 1: dolly}; wheel = zoom (positive steps zoom in). A press becomes a drag only after motion exceeds a 1.0 px threshold, and the first post-threshold step is measured from the down-point so it is not jumpy.

  • get_camera_xform() returns a row-major float64 4×4: row 0 = right, row 1 = up, row 2 = -forward, row 3 = eye. State is sanitized before matrices are built — distance/elevation clamped, non-finite lanes scrubbed — so the loop never commits a non-finite pose.

Sample-Preset Pattern

To reproduce a named sample scene in one click, keep a small registry of presets, each a dict of {dataset, representation, field, colormap} (optionally iso/slice fractions). Applying a preset is a single enqueued command that the loop drains and applies as one configuration through the same publication routing as the individual controls, so no preset code touches OVStage directly. A preset button resolves the named entry and enqueues it. The owner updates existing arrays or replaces structure according to the requested changes. A new dataset follows scene replacement with the selected camera policy.

# presets.py — a registry; the app is adding this module.
PRESETS = {
    "StaticMixer temperature": dict(dataset="static_mixer.npz",
        representation="surface", field="Temp", colormap="viridis"),
    "Disk isosurface": dict(dataset="disk_out_ref.npz",
        representation="isosurface", field="Pres", colormap="turbo"),
}

def apply_preset(runtime, name):
    runtime.enqueue_preset(dict(PRESETS[name]))  # app-owned typed command

Commit a preset as one coherent requested configuration so intermediate field and representation choices cannot publish into the replacement dataset.

Gotchas

  • Buffered stdout when redirected: Python buffers stdout when it is not a TTY, so piping logs to a file can hide [status]/[render] lines until exit. Run python -u -m <app> (or flush=True on prints) to see logs live.

  • Friendly field names: resolve a display name like "Temperature" to the real dataset field (e.g. "Temp") against the actual field list on every rebuild, so coloring never silently falls back to the wrong field. Implement this as an application-owned case-insensitive alias map, then require a visible selection or error when no unambiguous field exists.

  • First-frame RTX shader compile: a fresh process's first renderer.step compiles RTX shaders and can take minutes. The GPU may look idle and the window may look frozen — this is not a hang. Allow a long first-frame timeout (for example, 300 s) and poll rather than assuming failure.

  • Never let one frame kill the loop: wrap tick, command application, and every mouse/control callback in try/except; log and continue so a single bad frame or callback does not tear down the window.

Suggested Application Modules

  • app.py — ovui shell: ui.init/ui.Window, header ComboBoxes + FloatSliders, ZStack image + transparent hit rect, update/presentation task ownership.

  • runtime.py — serialized ViewerRuntime: enqueue methods, coalesced last-wins queue drain, tick, and write-floor commit + step.

  • camera.py — orbit/pan/dolly/zoom/fit and a sanitized row-major camera xform.

  • input.py — button remap, drag threshold, and camera-dirty state.

  • viewport.py — ByteImageProvider wrapper and letterbox mapping.

  • config.py, __main__.py — fixed render size, interactive/test modes, and preflight env order.

See also: local-viewer, ovui-local-viewer-recipe, viewer-input-routing, viewer-control-patterns, camera-controls, conventions.md, ovstage-runtime, ovstage-ovrtx-integration, ovrtx-rendering.

Requirements and compatibility

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Needs Python, Needs Node.js

The author says: Orchestrator skill. Downstream focused references may require NVIDIA GPUs, ovrtx, ovstream, ovui, OpenUSD, Python, Node/React, Tauri, Electron, C++, or cloud GPU deployment access depending on the selected viewer path.

Security

What our scanner noticed· Medium risk

  • Large base64-encoded blobs
  • Instructions that use the network

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Created by
NVIDIA Omniverse
Publisher
NVIDIA (Official)
Source
NVIDIA/skills · skills/omniverse-realtime-viewer · commit fce3cca
Licence
Apache License 2.0 (redistribution allowed)
  • LICENSE (added by AutoGineer from LICENSE-APACHE)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by NVIDIA Omniverse or NVIDIA.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download omniverse-realtime-viewer.zip