RealityScan: Free Photogrammetry for Everyone

What is RealityScan?

RealityScan is a free photogrammetry application from Epic Games (via Capturing Reality). It transforms photos into 3D models by analyzing images taken from multiple angles and reconstructing the scene geometry.

While it’s primarily designed for creating 3D meshes, I use it for a different purpose: generating camera position data for 3DGS training.

Two Ways to Use RealityScan

1. Traditional Photogrammetry

The standard use case - create 3D mesh models:

Photos → RealityScan → Textured 3D Mesh (.obj, .fbx, etc.)

Great for:

3D printing
Game assets
AR filters
Product visualization

2. Camera Alignment for 3DGS (My Use Case)

Extract camera positions for 3D Gaussian Splatting:

Photos → RealityScan → Camera Data (.csv) + Sparse Point Cloud (.ply)
         ↓
    PostShot / Other 3DGS tools

This is how I use it in my Nankunshen Temple project.

Key Features

Automatic Camera Alignment

RealityScan uses Structure from Motion (SfM) to calculate:

Camera position (x, y, z) for each photo
Camera orientation (rotation)
Focal length and lens distortion

This spatial information is essential for 3DGS training.

Sparse Point Cloud Generation

As a byproduct of camera alignment, RealityScan generates a sparse point cloud - a rough 3D representation of the scene used to initialize Gaussian positions.

Export Options

Format	Use Case
.obj / .fbx	3D mesh for games, printing
.csv	Camera positions (for 3DGS)
.ply	Point cloud (for 3DGS initialization)

Cross-Platform

iOS - iPhone/iPad app
Android - Mobile app
Windows - Desktop application

Workflow for 3DGS

Here’s how I use RealityScan in my 3DGS pipeline:

Step 1: Prepare Images

Extract frames from your video or organize your photos. For the Nankunshen project, I used DaVinci Resolve to extract multi-angle frames from 360° video.

Step 2: Import to RealityScan

Create a new project
Import all images
Wait for automatic alignment

RealityScan will match features across images and calculate camera positions.

Step 3: Review Alignment

Check that:

Most images are aligned (green indicators)
The sparse point cloud looks correct
No obvious misalignments

Step 4: Export Camera Data

Export the camera position CSV and sparse point cloud PLY for use in PostShot or other 3DGS training tools.

Tips for Better Results

Image Quality

Overlap: 60-80% overlap between consecutive images
Consistency: Same lighting conditions throughout
Sharpness: Avoid motion blur
Coverage: Capture from all angles

Challenging Scenarios

RealityScan (and photogrammetry in general) struggles with:

Challenge	Why	Solution
Reflective surfaces	Inconsistent features	Avoid or mask
Transparent objects	No features to match	Use backing
Repetitive patterns	Ambiguous matching	Add markers
Moving objects	Position changes	Remove from scene

Number of Images

Scene Type	Recommended Images
Small object	30-50
Room interior	100-200
Building exterior	200-500
Large site	500+

RealityScan vs Alternatives

Tool	Price	Best For
RealityScan	Free	General use, beginners
Reality Capture	Paid	Professional quality
Meshroom	Free	Open source, customizable
COLMAP	Free	Research, command-line
Polycam	Freemium	Mobile convenience

RealityScan is essentially a simplified, free version of Reality Capture - making professional photogrammetry accessible to everyone.

Integration with 3DGS Workflow

My complete pipeline using RealityScan:

1. Capture (Insta360 X5)
      ↓
2. Frame Extraction (DaVinci Resolve)
      ↓
3. Camera Alignment (RealityScan) ← This article
      ↓
4. 3DGS Training (PostShot)
      ↓
5. Viewing (SuperSplat / Unreal Engine)

See the full workflow in my Nankunshen Temple project.

Getting Started

Download RealityScan from Epic Games
Take 30+ photos of a small object
Import and let it process
Export your 3D model or camera data

It’s free and surprisingly capable for a no-cost tool.

Conclusion

RealityScan democratizes photogrammetry - what used to require expensive software is now free and user-friendly. Whether you’re creating 3D models directly or using it as part of a 3DGS pipeline, it’s an essential tool to have in your toolkit.

What is 3D Gaussian Splatting? - Introduction to 3DGS
PostShot: Train 3DGS Models - Next step after camera alignment
SuperSplat: Edit 3DGS in Your Browser - View and edit results
Nankunshen Temple Project - Full workflow example

Sources

RealityScan Download - Epic Games official page
Reality Capture - Professional photogrammetry software
COLMAP - Open-source Structure from Motion
Meshroom - Open-source photogrammetry

Get in Touch

Have questions or want to collaborate? Feel free to reach out!

dazai.studio

Dazai Chen

dazai.studio@gmail.com