Photogrammetry: a 3D model from photographs

How photogrammetry works

Photogrammetry is measuring from photographs. Software finds the same points in the images, works out from their positions where each photo was taken, and calculates the shape from that. You typically need 40 to 150 photos around the object, with plenty of overlap. Colour carries into the result as well, so the method suits textured models. It is a cheaper alternative to 3D scanning with a scanner.

Getting a good model

• Even light. An overcast sky is better than harsh shadows.
• A matt surface. Shiny, transparent and plain-coloured objects confuse the software.
• Sharp photos with at least 60 % overlap, taken from several heights.
• Scale. You cannot tell the size from photos, so put a ruler next to the object.

The result is usually less accurate than from a laser scanner, with smooth surfaces coming out wavy and details soft.

From photos to print

A photogrammetry model is a large textured mesh. Before printing you need to trim off the surroundings, simplify it, close any holes and set the scale, see model repair. The texture is lost with an ordinary STL. If you want colours, the model has to be turned into a multicolour 3MF where the colours are assigned by hand, and we cannot print more than 4 colours at once.

Upload an STL, 3MF or OBJ up to 50 MB and you see the price per part straight away. For objects based on other people's photographs and scans we do not check rights, and the responsibility is yours. We do not offer photogrammetry or scan editing.

Software and process

You can use Meshroom for free, and on a phone there are apps that guide your shooting and compute the model in the cloud. The process is similar: you photograph the object all the way round at two or three heights, the software builds a rough model, and you trim it, simplify it and save it in a format suitable for printing. It is best to put the object on an untextured surface that you remove later.