- Michael Nelson
A rising technology in both the creation of 3d assets is photogrammetry. Photogrammetry is the process through which multiple photos of an object are taken and reconstructed into a 3d model and mesh. This works by identifying the features of the mesh and textures and blending them together and can be improved by making a 3d mesh first using a depth sensor and then overlaying textures gotten from the rgb cameras. Setting up a rig to scan these objects is a tedious processes as it requires good diffuse lighting conditions that are widely spread throughout the target object. Any surface with high amounts of specular reflection such as metals or transparent material like glass are problematic as the algorithms have a hard time identifying landmarks. While there are likely methods to improve this I have thought about two different possible ways to avoid these issues. The first is to coat the material in something which will make the object diffuse enough to not reflect too much light and ruin the image. The second way is to equip a polarized camera which will ignore these reflective light sources to some extent.