Can I Use Structure Sensor (Mark II) for automotive scanning?
Automotive scanning for designing custom floor mats, seat covers, and so on, is an appealing use case for Structure Sensor (Mark II).
However, this type of automotive scanning presents several challenges that the user must consider before purchasing.
Black and Reflective Objects
Black objects pose a particular difficulty for our sensors.
In order to capture depth, our sensors project an IR speckle pattern to the object the user intends to scan. This pattern distorts as it intercepts objects in the field, and the object is calculated based on the distortion. For more in-depth information on how Structure Sensor (and all of our sensors work), please check out this video:
Black objects absorb the IR lasers, causing the objects to “disappear” to the sensor.
Reflective objects scatter the IR erratically, causing random artifacts and distorted meshes.
Some users have mitigated these problems by applying a temporary paint or powder to the objects in question assist the sensor in capturing depth information. This may not be a good option for scanning car seats or mats.
In order to generate an entire scan of an object, you need 360-degree access around the object. Additionally, the minimum focal length of Structure Sensor (Mark II) is around 40 cm, which in tight car spaces, is not enough to capture the scan.
The precision of Structure Sensor (Mark II) can be found on this chart. However, if you’re trying to scan 3D objects or track motion paths, the quality of results will vary widely depending on software used. Make sure you’re using the latest Structure SDK and have tried a variety of different apps to gauge precision and accuracy of results achievable.
We are still collecting data and running case studies but, from what we have seen working with professionals in the field is that, most measurements are within 1-2% when verified against a manual tape measure or existing blueprints.
You can find more of the technical specification of the Structure Sensor on the following page: https://support.structure.io/article/367-what-is-the-difference-between-structure-sensor-and-structure-sensor-mark-ii
This may not be enough for certain automotive projects.
Please view the sample scan of a car dashboard below to help assist you with your determination:
Scanning Large Objects
As a rule, object scanning is limited to the bounding box of your software. While bounding boxes can be expanded fairly large, the resolution of the scan within that bounding box will decrease.
This means that scanning objects with an extreme bounding box size (like a truck, for example), requires a different strategy.
You have two options to capture this scan: stitching together several scans, or using Canvas in an experimental fashion.
Stitching Scans Together
The first option requires you to capture your intended object in multiple scans. You will need to mentally create a 3D-grid of your object and scan each section of the grid as fully as possible.
Once all of the scans are captured, you will need to import the resulting meshes into applications like Blender or MeshLab and combine them together. For a tutorial on how to do this, please check out this video: https://youtu.be/YLoKjR8H-JQ
Using Canvas Experimentally
If you would rather capture your object with a single scan, you can potentially do so by using Canvas. While Canvas is intended as a room scanning app for architects, contractors, and interior designers, it allows for high-resolution scanning without the use of a bounding box. As such, it offers the potential to scan large objects in a single mesh.
Please note: this is an experimental use of Canvas, and as such, we do not guarantee results; nor will Canvas's Scan To CAD service work with your resulting mesh.
First, position your object so you have 360-degree access. Begin your scan, scanning your object with a paintbrush-like motion, not covering the same area twice. Work your way around your object until you complete your scan.
You will then need to perform a sort of self-Scan To CAD to retrieve the resulting mesh. To do so, please check out this article: https://support.canvas.io/article/33-can-i-do-scan-to-cad-myself