How 3D Reconstruction Can Preserve Cultural Heritage ; My Week 1 with PreserveMy.World

As part of my Week 1 internship with PreserveMy.World, I dove into the world of
3D reconstruction and how it can be used to digitally preserve heritage sites.

What is 3D Reconstruction?

3D reconstruction is the process of capturing the shape and appearance of real
objects or environments and converting them into digital 3D models. For heritage
preservation, this means we can create permanent digital records of sites that may
be at risk of damage, decay, or destruction.

Methods I Researched

COLMAP (Structure from Motion)

Input: Multiple photographs from different angles
Output: 3D point cloud and mesh
Difficulty: Medium — good beginner starting point

NeRF (Neural Radiance Fields)

Input: Photos with camera position data
Output: Photorealistic 3D scene you can navigate
Difficulty: Hard — requires GPU and ML knowledge

Gaussian Splatting

Input: Photos
Output: Real-time renderable 3D scene
Difficulty: Hard — very new technique (2023)

Monocular Depth Estimation

Input: A single image
Output: Depth map showing distances
Difficulty: Easy — great for beginners

Why This Matters for PreserveMy.World

PreserveMy.World aims to use AI to document and preserve heritage sites.
3D reconstruction gives us the ability to let people virtually visit places
that may no longer exist. Imagine being able to walk through an ancient site
in VR, reconstructed from photographs taken before it was damaged.

What I Plan to Try Next

This week I plan to run a COLMAP demo or find a beginner-friendly Colab notebook
to test structure-from-motion on sample images. I'll document the results and
share outputs in my PMW-day1 GitHub repo.