Crate 2.0.2 Brings Gaussian Splat Conversion Into Nuke

Michael Rubloff

CrateTools has released Crate 2.0.2, an Apache 2.0 asset browser that previews and converts Gaussian splats inside Nuke.
Nuke has handled splats natively since 17.0 shipped with a Gaussian splat toolset built on GeoImport and SplatRender, and the 17.1 open beta added dynamic splats and basic relighting. Crate works the step before that, browsing a studio's libraries as a folder tree and previewing geometry and splats in 3D before anything lands in the Node Graph.
Conversion is handled in the same application. Crate bundles PlayCanvas splat-transform as one of six external engines, together with a Node.js runtime to execute it, so a splat can be converted between formats without leaving Nuke. The other five engines are F3D, OpenImageIO, FFmpeg, ImageMagick, and MediaInfo, covering thumbnails, animated previews, metadata extraction and color managed image processing.
Thumbnails are generated as an artist browses, including from geometry and splats, and cached once and shared, so the first person to open a folder pays the cost and everyone else reads the result. Libraries and their caches are configured separately, which lets a facility keep previews on SSD or NVMe while the raw elements sit on high capacity storage.
Access is split into two roles. Curators configure libraries, users and engines, and Visitors browse and import, with write privileges and interface options set per role. Crate 2.0.1 claims a 90 percent speed-up in library loading time for image sequences.
Crate runs on Nuke 15, 16 and 17 across PySide2 and PySide6, with development and testing focused on Nuke 17. Crate is Windows only, developed and tested on Windows 10 22H2 and working on Windows 11. macOS and Linux are not supported yet. Every engine path is configurable in crate_engines.txt, which is the route to native builds on other platforms.
Crate itself is Apache 2.0, though the bundled FFmpeg is the gyan.dev full build under GPL v3, invoked as a separate process. It is available now on GitHub, though the engine bundles, including splat-transform, are distributed as Google Drive links.
- TRENDINGLoading...
- TRENDINGLoading...
- TRENDINGLoading...
New to Gaussian Splatting? Start here




