From Images to Gaussian Splats in Unreal Engine | RealityCapture + PostShot Workflow
In this tutorial, I’ll show you how to convert images into point clouds and then into Gaussian splats, and finally import them into Unreal Engine for real-time rendering.
We’ll start by using RealityCapture to align images and export a binary sparse point cloud, then move to PostShot to train and generate Gaussian splats, and finally import and clean them inside Unreal Engine 5 using the PostShot plugin.
This workflow is perfect for:
Photogrammetry beginners
Real-time rendering in Unreal Engine
Gaussian splats for games, VR, cinematics, and visualization
Lightweight, high-detail 3D assets
-What you’ll learn in this video:
Convert images to point cloud using RealityCapture
Export sparse point clouds correctly (binary PLY)
Train Gaussian splats in PostShot
Adjust training steps for better quality
Import Gaussian splats into Unreal Engine 5
Enable and install the PostShot Unreal plugin
Clean up unwanted splats and optimize results
-Tools Used:
RealityCapture
PostShot (Jawset)
Unreal Engine 5
Gaussian splatting is becoming a powerful alternative to traditional meshes, especially for high-detail scans with low performance cost. If you’re working with photogrammetry, UE5, VR, or real-time environments, this workflow will help you a lot.
If this helped you, don’t forget to like, subscribe, and share.