8(965)-890-30-00
dhlabcreation@gmail.com
dhlab@sfu-kras.ru
Krasnoyarsk, Russia
82 Svobodny pr., building 6
Room 4-02
Go back
April 2021

Basic Principles of Photogrammetry in the Field

The article describes the process of creating 3D models using photogrammetry technology in the field.

Photogrammetry is defined as a scientific and technical discipline which determines shape, sizes, position and other characteristics of objects from their photo images.

Previously, this technology was commonly used in geology and topography. It has become more popular in other spheres (i.e. film industry, game industry, virtual archeology, etc.) due to its affordability based on the increasing of working capacity of the computers.

The whole process of photogrammetry goes through several stages:

  1. Hardware and software selection;
  2. Photographing of the object;
  3. Image processing (when necessary);
  4. Creation of 3D model.

 

Hardware and software

It is general practice by photogrammetry in the field to use a camera and lens only. Tripods, stabilizers, drones, HDRI photography equipment, a color scale and a gray card can be additionally used.

Figure 1 – Hardware
Figure 1 – Hardware

Weather conditions are considered to be an important factor in obtaining high-quality images. The object should be evenly lit, so cloudy weather is primarily suitable for photographing. In other conditions, additional hardware or software may be used to eliminate lighting.

Among the variety software for generating 3D models from photographic images, the most popular are Agisoft Metashape, Reality Capture, Autodesk Recap and Pix4D.

In case it is necessary, there is software for color correction and image enhancement such as Adobe Photoshop Lightroom, Digital Photo Professional, FastStone Image Viewer.

 

Photographing of the object

In most cases objects in the field are isolated, so it is necessary to take the photos of them on three different levels around. The tilt angle of the camera relative to the object is 0°, 35°, 85°.

Figure 2 – Camera positions for photographing an object
Figure 2 – Camera positions for photographing an object

One of the crucial points is the overlap of adjacent frames. The larger the overlap is, the more common points there are in the two frames. Based upon the points, software packages render vertices in a three-dimensional coordinate system in order to create a three-dimensional polygonal model. The recommended minimum overlap of adjacent frames is 30%.

Figure 3 – An example of sufficient overlap of adjacent frames
Figure 3 – An example of sufficient overlap of adjacent frames

Creating a 3D model

Reality Capture is the optimal choice for creating a 3D model based on images taken in the field, as it allows to quickly get accurate results.

The work goes through several stages:

  1. Uploading photos;
  2. Aligning photos;
  3. Setting the area of reconstruction;
  4. Creating a three-dimensional polygonal model;
  5. Generating a texture of the model;
  6. Export and publication of the result.

The first step of the project is uploading photos. In order to do this, one chooses Inputs command (tab WORKFLOW in the toolbar 1. Add imagery) or selects and transfers all the photos to the project.

Then one can proceed to the stage of aligning the photos. For this in the ALIGNMENT tab on the Registration toolbar, Align images command should be selected.

The obtained point cloud allows us to evaluate the quality of the photography and the possibility of further work with the reconstruction of the model. Typically, in Reality Capture, good construction is judged by the number of aligned photos and common points between them.

Figure 4 – Aligning Photos
Figure 4 – Aligning Photos

After finishing photo aligning, one needs to adjust the size and position of the reconstruction area in accordance with the object characteristics.

As a rule, Reality Capture automatically calculates the size and position of the reconstruction area. At the same time, it is recommended to check that the object is entirely inside the reconstruction area, since at the next stage only cloud points inside the reconstruction area are taken into account.

In order to change the size and orientation of the reconstruction area in the tab RECONSTRUCTION, the items from Set Reconstruction Region drop-down should be used.

Figure 5 – Definition of the construction area
Figure 5 – Definition of the construction area

Then it can be proceeded to calculating a 3D polygonal model. In the tab RECONSTRUCTION on the Process toolbar, one selects Normal Detail. Thus, the program will generate a high-poly 3D model. To build a low-poly model, Preview should be selected.

Figure 6 – Calculating three-dimensional polygonal model
Figure 6 – Calculating three-dimensional polygonal model

To generate a texture, one selects Colorize, then Texture on the same toolbar.

Figure 7 – Generating texture
Figure 7 – Generating texture

In case it is required to remove unnecessary areas, one can use the Lasso, Rect or Box tools in the Selection toolbar, then select Filter Selection in Tools.

Figure 8 – Using the Lasso tool
Figure 8 – Using the Lasso tool
Figure 9 – Three-dimensional model after removing unnecessary areas
Figure 9 – Three-dimensional model after removing unnecessary areas

After the intended result is achieved, the three-dimensional model can be exported and published in services such as Sketchfab, 3ddd, Turbosquid.

20 April 2021

Licensing Images from Russian Museums for an Academic Project

/ Read more /