A lightweight Nuke gizmo plus a native C++ binary to denoise noisy renders or plates directly in Nuke. The release package ships with a prebuilt C++ binary and the gizmo, so you can drop them into your setup and start working.
Screenshots will be added later.
The latest release includes the prebuilt C++ binary and the gizmo, arranged like this:
C:.
├───bin # Prebuilt C++ binary (required at runtime)
└───gizmos # Nuke gizmo(s) and assets
└───imgs # Icons / images used by the gizmo UI
Important: The C++ build binary is already compiled and included under bin. You do not need to build it yourself.

- Nuke 12.0+ (tested on newer versions as well)
- OS compatible with the provided binary in bin
- Permission to load external binaries on your machine/pipeline
Choose one of the following setups.
-
Download the latest release and extract it somewhere convenient (e.g.,
C:\tools\Nuke-AI-Denoiser). -
Make the gizmo discoverable by Nuke:
- EITHER copy the
gizmos/folder into your Nuke profile directory:- Windows:
C:\Users\<you>\.nuke\gizmos\ - macOS:
/Users/<you>/.nuke/gizmos/ - Linux:
/home/<you>/.nuke/gizmos/
- Windows:
- OR keep the repo anywhere and add the path via your
~/.nuke/init.py:import nuke, os nuke.pluginAddPath('gizmos') # If the repo lives inside ~/.nuke # OR explicitly: nuke.pluginAddPath(r'C:\path\to\Nuke-AI-Denoiser\gizmos')
- You can also use NUKE_PATH to point to the repo root:
- PowerShell:
$env:NUKE_PATH = "$env:NUKE_PATH;C:\path\to\Nuke-AI-Denoiser"
- bash/zsh:
export NUKE_PATH="$NUKE_PATH:/path/to/Nuke-AI-Denoiser"
- PowerShell:
- EITHER copy the
-
Restart Nuke.
-
Most important step: set the binary location in the node UI
- Create/Select the AI Denoiser node.
- Open the “bin” tab.
- Set “Bin Path” to the folder that contains the binary from the release, i.e. the bin directory you extracted.
- Example (Windows):
C:\path\to\Nuke-AI-Denoiser\bin
- Example (Windows):
- A screenshot of this step will be added later.
If you maintain centralized tool repos:
- Keep the release structure intact on a network path, e.g.
//server/tools/Nuke-AI-Denoiser. - Add
.../gizmosto Nuke’s plugin path via your studio’sinit.py. - In the gizmo’s “bin” tab, set “Bin Path” to the shared
.../bindirectory. Consider locking or templating this knob for artists if desired.
- Add the node:
- Press Tab and type “AI Denoiser” (exact name may vary based on your menu setup).
- Connect inputs:
- Plug your noisy Beauty render or plate into the main input.
- If the gizmo exposes optional passes (e.g., albedo/normal), connect them for improved detail preservation.
- Point to the binary:
- In the node’s “bin” tab, set “Bin Path” to the release’s bin folder (see Installation step 4).
- Adjust controls:
- Start with defaults; then tweak Strength/Mix and any pass toggles.
- Use the Mix/Blend knob to A/B between original and denoised to avoid over-smoothing.
- Render:
- Proceed with your usual Write node. On farms, make sure the Bin Path is accessible from render machines.
Tip: For CG, auxiliary AOVs (albedo/normal) can significantly help maintain edges and textures. For live-action plates, subtle denoising blended back often yields the most natural results.
- Node doesn’t appear in the Tab menu:
- Verify
gizmos/is on Nuke’s plugin path (print frominit.pyto confirm). - Ensure no typos in the path and that you restarted Nuke after changes.
- Verify
- Binary not found / fails to load:
- Confirm the “Bin Path” in the “bin” tab points to the correct bin folder from the release.
- Check OS/architecture compatibility and that you have permission to load external binaries.
- Ensure antivirus/security tools aren’t blocking the binary.
- Slow performance:
- Reduce Strength or preview a smaller ROI.
- Cache heavy upstream nodes or write temp EXRs for iteration.
- Add screenshots:
- Setting “Bin Path” in the “bin” tab
- Node UI overview and before/after comparisons
- Presets / quality profiles
Specify a license here if applicable.
Thanks to the Nuke community for best practices around gizmo packaging and distribution.