Goal: Design and build a device capable of taking photographs in a repeatable and accurate manner for use in photogrammetric analysis, the generation of a three-dimensional model of an object through a collection of photographs.
Results: The team successfully created a photogrammetry device. In the future they plan to continue to project to make a better, 2.0, version. The project involved working with Arduino, basic hardware and tools, and imaging software. The team also compiled what they learned about photogrammetry into an eighteen-page report. Three main points are summarized below.
Three important tricks to improve the overall quality of the generated model:
Specular reflective (shiny) surfaces are not desirable. Reflections change position with respect to the camera and confuse the photogrammetry software. Remedy this by spraying the model with a matte finish and avoid direct light sources.
Capturing as many perspectives of the model is desirable. Only so much data may be captured by rotating in a single lateral ring. Changing the height and angle of the camera for several lateral rings captures essential perspective detail of the model and improves the fidelity of the generated model.
The photogrammetric analysis software uses all information in the images, including items in the background. For the most effective model generation, fill about 60% of the camera view with the model. Items in the background provide useful information helpful to identifying camera location in 3D space.