Glossary · Processes
Photogrammetry
A 3D model made from many ordinary photos: software calculates the shape from images that show the object from all sides.
Explained in more detail
In photogrammetry, an object is photographed from many directions. The software looks for common points in the images, uses them to work out the camera positions, and calculates the spatial position of each point from the sight lines of several images. Any camera or a smartphone with apps such as RealityScan, Polycam or KIRI Engine is enough, and All3DP mentions 50 to 1,000 images. The method is the cheapest but less accurate than scanners with their own light source, and it suits figures and artistic models better than precisely fitting replacement parts. Shiny, transparent and uniformly smooth surfaces offer too few features.
Where does the term come from?
From Greek phos, “light”, gramma, “something written, image”, and metron, “measure”: measuring with light images. The architect Albrecht Meydenbauer gave the method its name; it appeared in 1867 as the title of an article in the Wochenblatt des Architektenvereins zu Berlin.
In practice
- Photograph with strong overlap between the images so that the software finds common points.
- Even, soft light without harsh shadows and reflections improves the result.
- Include a scale bar or an object of known length in the picture. From photos alone, the software does not know the real size.
- Shiny or smooth surfaces can be prepared with a matting scanning spray, as with optical scanners.
Not to be confused
Photogrammetry is not LiDAR: LiDAR actively measures distances with light, photogrammetry calculates the shape from photos alone.
Frequently asked questions
Can I scan 3D models for printing with my phone?
Yes, photogrammetry apps produce usable models of figures and larger objects. The accuracy is usually too low for precisely fitting spare parts; a scanner with its own light source is better suited for that.
How many photos do I need?
That depends on the object and the app, and All3DP gives 50 to 1,000 images. More important than the number is that every spot is visible in several images from different angles.