Equipment and procedures for photogrammetry
As photogrammetry is the most useful method for getting started on creating 3D models quickly, easily, and at a reasonable price, we will only describe this technique and its associated equipment below.
The basic equipment for creating 3D models using photogrammetry consists of a digital camera, tripod, computer, and software. The work can be facilitated by a turntable in order to systematically take series of photos with sufficient overlap, as well as lamps and some form of light tent in order to create good lighting.
Camera and lenses
It is possible to create acceptable 3D models with the aid of a simple compact camera or mobile camera with a resolution of at least 5 megapixels, but a system camera gives you a better end result. A fixed lens of approximately 50 mm works best for object photography using a system camera. Avoid wide-angle and fisheye lenses, as these distort the images and make them unusable for the software. Fixed focal length is preferable, but zoom lenses can also be used.
Tripod and remote trigger
A tripod is an essential part of the photogrammetry equipment. Choose a heavy, stable tripod which can be raised and lowered easily. Use a remote trigger or the camera’s own built-in trigger, with a delay of at least 5 seconds to avoid blurring.
The pictures should be taken with a 50-60 per cent overlap to ensure that you capture the entire surface of the object. Large objects and buildings are circumnavigated with the camera, but smaller objects can be put on a turntable. Any sort of flat, turning tray can be used as a turntable. Ten degrees is a suitable rotation between photos.
For the best possible results, the lighting should be diffuse, flat, and as even as possible, so that shadows and contrasts do not affect the calculation of the object’s structure. For the best indoor results you should use a light tent and real photo lamps. Taking photos outdoors on an overcast day is also okay, if no other options are available.
Photography, image quality and camera settings
Select the highest possible resolution and image quality, even if the image will be compressed for online use at a later point. High resolution images make it easier to do post-processing where needed, and high-resolution originals can also be used in print later on.
Where possible, take RAW format photos which can then be saved as uncompressed TIFF files (supported in Agisoft Photoscan and other 3D rendering software).
Use the lowest ISO possible in order to minimise image noise. For a good, crisp image of the entire object, use a small aperture between f/8 and f/16.
Make sure to photograph the entire surface of the object, with 50-60 % overlap between photos, as anything not in the photos will not be included in the final model.
To get the best possible result you need to be systematic in taking photos. The object must be photographed from several heights, and on each level the object (or camera) should be rotated approximately 10 degrees between each photo, so that the photos cover the full surface structure of the object.
It is possible to create 3D models using a simple office computer, although the process will take hours or possibly days of rendering. The best computer for this is one with the specifications of a powerful gaming computer. There are also cloud services which create 3D models from photographs.
From photo to 3D model
There are numerous pieces of software, commercial as well as free, which can create 3D models from photographs. Sometimes several different pieces of software are used to process the photographs and create and process 3D models. Regardless of the software, the work process remains the same.
The processing of photos into 3D models mainly consists of the following four steps:
• Align photos (build sparse point cloud) – The cameras’ (photographs’) positions are identified. All images are adjusted so that shared points between adjacent photos are identified and arranged into a “sparse point cloud”. At this early stage, irrelevant parts of images such as backgrounds and supports can be discarded. You can either use the programme’s own tool to do this, or do this part of the process in advance using, for example, Adobe Photoshop.
• Build dense cloud – the points identified in the previous step are arranged, compiled, and given X, Y and Z coordinates in a three-dimensional space. You can also discard irrelevant points at this stage.
• Build mesh – the points are linked into a composite “mesh” of triangular shames (polygons). The mesh can then (in most cases) be edited, either directly in the programme, or by being exported to another piece of software for editing, and then re-imported.
• Build texture – Finally, the original two-dimensional images are used to create a texture, i.e. a photographic surface on the mesh, and give the final 3D model its photorealistic surface.
The various calculation steps of the process require various resources from the computer, and the time needed is affected both by the number of images, the content of the images and the resolution, and the capacity of the computer’s graphics card, RAM, processor, and hard drive.
The finished model can then be uploaded directly to Sketchfab through the software. Uploading to Wikimedia Commons requires the model to be saved as a .STL file, as this (as of April 2018) is the only file format supported by Wikimedia Commons.
Practical examples of working with photogrammetry:
Digitalisering i 3D – Kom igång med fotogrammetri (The Swedish Exhibition Agency)
Fotogrammetri för nybörjare – Erik Lernestål och Fredrik Andersson, Digikult 2017
The Royal Armoury on Digitaliseringsbloggen: http://digitaliseringsbloggen.lsh.se/2017/05/varmotet-2017-om-var-workshop-kring.html#.WpPCGGrOWpq
The Royal Armour on photogrammetry, and the work on Gustav Vasa’s helmet: http://digitaliseringsbloggen.lsh.se/2017/01/fotogrammetri-for-nyborjare.html#.WpPCqmrOWpo
Three Swedish museums experiment with photogrammetry: https://blog.sketchfab.com/three-sweedish-museums-experiments-photogrammetry/