Skip to content

Scanning and processing 360 videos to 3DGS models

By Olli Huttunen · more summaries from this channel

14 min video·en··58848 views

This is an AI-generated summary of Scanning and processing 360 videos to 3DGS models — a 14 min YouTube video by Olli Huttunen, published July 17, 2025. It condenses the full transcript into 9 key takeaways with clickable timestamps.

Summary

Ollie demonstrates his custom "magic wand" 3D scanner, detailing the camera settings, a unique Blender-based image extraction process, and the post-processing workflow for generating high-quality Gaussian splatting models from 360 camera footage.

Key Points

  • The video showcases a custom "magic wand" 3D scanner setup utilizing three Insta360 X5 cameras for capturing environments. 
  • Optimal camera settings for the Insta360 X5 include 8K 30fps video recording with adaptive tone, as higher resolution interval or time-lapse modes introduce processing complexities. 
  • The post-processing workflow begins with importing raw video files into Insta360 Studio for conversion to full 360 equirectangular format. 
  • A custom Blender add-on is then used to extract specific camera views, creating 18 angles from three different heights, with adjustments to avoid the operator and ensure comprehensive coverage. 
  • For image alignment and Structure from Motion (SFM), the creator compares Postshot and Reality Scan software. 
  • Postshot is generally preferred for aligning images extracted from 360 videos, especially for outdoor scans, as it more accurately solves pinhole camera images despite lengthy processing times. 
  • Reality Scan can be a faster and effective option for indoor scans that feature more distinct geometric shapes. 
  • The demonstrated "magic wand" setup successfully converts 360 camera captures into valid 3D data and stunning radiance fields, proving its effectiveness for Gaussian splatting models. 
  • The creator plans to finish coding and fine-tune his Blender plugin to make this 360 scanning process more accessible to others. 
Scanning and processing 360 videos to 3DGS models

Scanning and processing 360 videos to 3DGS models

Ollie demonstrates his custom "magic wand" 3D scanner, detailing the camera settings, a unique Blender-based image extraction process, and the post-processing workflow for generating high-quality Gaussian splatting models from 360 camera footage.

Key Points

The video showcases a custom "magic wand" 3D scanner setup utilizing three Insta360 X5 cameras for capturing environments.
Optimal camera settings for the Insta360 X5 include 8K 30fps video recording with adaptive tone, as higher resolution interval or time-lapse modes introduce processing complexities.
The post-processing workflow begins with importing raw video files into Insta360 Studio for conversion to full 360 equirectangular format.
A custom Blender add-on is then used to extract specific camera views, creating 18 angles from three different heights, with adjustments to avoid the operator and ensure comprehensive coverage.
For image alignment and Structure from Motion (SFM), the creator compares Postshot and Reality Scan software.
Postshot is generally preferred for aligning images extracted from 360 videos, especially for outdoor scans, as it more accurately solves pinhole camera images despite lengthy processing times.
Reality Scan can be a faster and effective option for indoor scans that feature more distinct geometric shapes.
The demonstrated "magic wand" setup successfully converts 360 camera captures into valid 3D data and stunning radiance fields, proving its effectiveness for Gaussian splatting models.
The creator plans to finish coding and fine-tune his Blender plugin to make this 360 scanning process more accessible to others.
Summarize any video — free
Summarizer.tube
Copy All
Share Link
Bookmark

Summarize any YouTube video, free

You just read an AI summary of this video. Paste any other YouTube link and get the key points with clickable timestamps in seconds — no signup, 5 free a day.

More Resources

More Summaries