What is Photogrammetry?

Take dozens or hundreds of overlapping photos around an object. Software finds the same features in different photos, works out where each camera was, and triangulates millions of 3D points. Those points become a dense mesh, textured with the photos themselves.

What it's great at

• Real textures and true proportions, since everything is measured rather than guessed
• Natural surfaces: rock, bark, ruins, statues and food
• Museum digitisation: several scans in our library, like the netsuke and the boulder captured from 164 photos, were made this way

What it struggles with

Shiny, transparent and featureless surfaces (glass, chrome, plain white walls), moving subjects, and anything you can't walk around. The output is also very dense and needs retopology or decimation before real-time use.

Photogrammetry vs AI image-to-3D

AI image-to-3D uses one photo and infers the hidden sides, so it's fast but approximate. Photogrammetry needs many photos and more time, but measures what's actually there. Our comparison guide goes into when to use which.