01Tools
CoSTAR National Lab has released StarLightGS, a plugin for Unreal Engine 5.7 and later, as a beta under the MIT license. It renders 3DGS into the GBuffer and relights it with the engine's standard lights and shadows.
Why it mattersIn games and video production, 3DGS captured from real scenes can now be placed in dynamic lighting. For prototypes such as day-night switching or virtual production with moving lights, this adds the option of treating 3DGS as part of the scene rather than as a baked background image.\n\nBecause it is published as an MIT-licensed plugin, it is easy to embed in in-house tools or modify. However, the training side that produces the relightable PLY is limited to non-commercial use, so commercial projects will need to consider a separate way of producing the assets.
Source: github.com +1 moreRead more
02Tools
WallGS is a Gaussian Splat renderer for Unreal Engine 5.3 to 5.8, with PLY and SOG support, 4DGS, Quest support and collision generation. The higher-tier WallGS Pro adds features such as relighting.
Why it mattersIn real-time production with UE, a practical question is whether scan results can be placed in the same scene as ordinary meshes and characters. WallGS provides depth of field, integration with Lumen reflections, and collision generation. This adds options for testing prototypes of games, virtual production and VR experiences that use real-world scans.
Source: wall.gs +1 moreRead more
03Research
speridlabs has released "Idol" on Hugging Face, a model that generates a 3D Gaussian Splatting .ply file from a single image of an object. It runs with 7.2GB of VRAM, and an SDK is provided.
Why it mattersFor game, XR, and prototype production, this adds an option for creating 3DGS assets from a single image in a short time. Because the output is a standard .ply, it is easy to pass to existing splat viewers and conversion tools.
Source: huggingface.coRead more
04Research
Yibin Zhao and colleagues have presented OctreeSplat, a method that allocates Gaussians according to local complexity without per-scene optimization. The authors say it reduces the number of primitives.
Why it mattersIf the number of Gaussians can be kept down, it may reduce data size and rendering load. This is a direction worth watching for web and mobile delivery.
Source: arxiv.orgRead more