3D scans from post-processing techniques after Smartphone acquisition

Type of course:

Digital learning, Lesson

Language:

Duration:

7 minutes

Workload:

2 hours

Proficiency:

Intermediate

Target:

Professionals

Scanning 3D objects with a smartphone is now possible. The device can capture images that are aligned in space, and if we know the depth of each pixel, we can reconstruct a 3D scene for each image. This depth is the key to good reconstruction. Either you use a depth sensor attached to your device (such as a LiDAR sensor) so that you can scan in real time; or you use specific algorithms that post-process the data coming from the smartphone (mainly images) to obtain a 3D model. This nugget covers two main post-processing techniques: Photogrammetry, which creates a 3D mesh from a set of images, and Neural Radiance Fields (NeRF), which creates a volume for rendering a 3D scene from a set of images. The nugget also discusses the importance of the first step: Structure From Motion, as the quality of the result depends mainly on this step and on the acquisition itself.

Learning outcomes

By the end of the module, learners can apply the methodology to scan properly an object when targeting post-processing
By the end of the module, learners can explain the impact of a bad acquisition on the “Structure From Motion” step, and on the final result.
By the end of the module, learners can justify why and when to use Photogrammetry and NeRF techniques.