April 26, 2018
In this article, we'll provide some tricks, how to capture the object surface is too thin to add any extra features to track. Let’s use a key to show how to make scanning these objects easier.
Possible problems and their solutions
- Problem: the area of the key edges is too small for a 3D-scanner to capture.
- Solution: To use a proper background. As long as the scanner “sees” a discrete surface behind the edges, it will be possible to capture a thin object, placed on this background. This article contains some small tips about the proper background.
- Problem: the key has similar geometry on both sides, and the edges lack features for effective alignment.
- Solution: Normally we recommend adding texture to the object, but since it’s too small in this case, the best approach is to do an all-around scan in one go. To do it, you will need to fix the key, for example, with pliers, and use a sheet of paper as a background.
3. Lack of data
- Problem: The global registration algorithm lacks enough data for accurate registration for this key in particular.
- Solution: We already have the sheet of paper as a background to fine-tune the tracking, so let's use it for registration, as well. Keep in mind, that the paper should have text or some other texture features on it.
An example of the setup for 3D scanning a key
More examples: 3D modes of thin objects
We once scanned a razor blade, using Artec Spider — it was the thinnest object we’ve yet captured.
As when capturing the key, one trick is to hold such a thin object above a sheet of paper with coloured text, which will be helpful for registration during both 3D scanning and processing.
We used a similar method to capture a dagger (link to 3D model on Sketchfab): we held it fixed above a paper with printed text.
Capturing thin edges
In addition, you may come across tricky cases, when you scan a thin object and you'd like to collect as much information of its edge as possible for future alignment process.
A credit card is used as a reference on a photo below, arrows and numbers illustrate the main steps:
- Scan1: capture a front part of the object and part of the background (by "background" we mean any colourful background sufficient for texture tracking)
- Scan2: capture a rear part of the object together with part of the table
- Scans 3 and 4: capture an edge of the object from back to front (or vice versa). You’ll then receive a scan, which contains an edge, both object sides and the background. The most important thing here is to capture the edge with background from two points
Now you can align all your captured data, run global registration and enjoy a perfect thin edge on your 3D mesh, after the sharp fusion algorithms finishes its calculations.