To create the 3D model of the training head, I used a technique called Structure-from-Motion (SfM), a method within photogrammetry that generates 3D models from multiple photographs by analyzing key geometric points. I placed the training head in a light box, allowing me to take photos from every angle.
Creating the 3D Model From a Photo Collection in Metashape
Knowing that the training head was made of different materials with varying textures, I focused on capturing the face, as its matte texture was ideal for SfM. The base, however, posed challenges due to its reflective white plastic, which interfered with the SfM process. As a result, I decided to manually model the base while using SfM for the face and head.
The Skin and the Base of the Training Head
After capturing the images, I uploaded them into Metashape, a photogrammetry tool by Agisoft. The initial model created was promising, but imperfections remained on the skin of the head due to missing geometry. While Metashape has some tools for refinement, it wasn’t enough to fully correct the issues.
To refine the model further, I used Blender, an open-source 3D modeling software. Blender allowed me to correct the skin’s imperfections with precision and manually create the base. Once the base was completed, I combined it with the head, resulting in a finished 3D model of the training head, mirroring the physical object.