3
3ShapeSupportGuy
New Member
- Messages
- 15
- Reaction score
- 0
Hi everybody,
interesting thread. I just discussed the topic with some of my colleagues in the 3Shape Scanner Development department.
They had some interesting points that I would like to share with you. See below:
.......................................................................................................
What’s the optimal vertex area density?
Obviously, a few vertices per mm^2 won’t capture all relevant details, but “the more, the better” does not hold, either. An unnecessarily high vertex density is a disadvantage, because it means increased processing time all the way through the CAD/CAM chain.
Instead of simply applying “brute-force” meshing with tiny triangles everywhere, 3Shape have put considerable effort into developing an intelligent meshing algorithm. It preserves detail where needed (in the prep area),but uses few triangles in flat or irrelevant parts. The original poster of this thread has since provided a nice example of the outcome of our algorithm (see http://dentallabnetwork.com/forums/f33/scanner-data-samples-9705/#post57741). Also notice that not only the triangle density, but also their elongation matches the surface curvature.
The vertex density and meshing algorithm that 3Shape have chosen as optimal for their scanners is the outcome of extensive real-world testing, with thousands of units produced in labs in many countries. We always welcome feedback from customers who see poor fits anyhow; other errors in the manufacturing process may be the cause.
What does it all mean for accuracy?
We completely agree with the poster’s statement that vertex density is not indicative of accuracy. The science of high accuracy measurement – metrology – applies reference objects with accuracies much higher than the scanner. These objects can only be manufactured by specialized accredited companies or metrology laboratories. With a reference object mimicking a die, it becomes clear that different scanner brands perform quite differently, and many alarming results are revealed – for example, some scanners can round an edge incorrectly with an error much larger than the oft-stated scanner accuracy of 20 microns.
Figure: Two scanner brands’ measurements (blue line: 3Shape, orange line: Other Dental scanner) of a nominally sharp edge in a custom-made ceramic reference tooth object (middle). Note the relatively large difference in accuracy around the sharp edges.
With an implant bar reference objects, 3Shape tests reveal an error of less than 3 microns along the insertion direction, i.e. in the level of the implants seating surface. This error is on the same magnitude as the reference objects accuracy.
Figure: Left: Measuring a reference object with a high-end coordinate measurement machine (accuracy 3 microns). Right: 3Shape scan of an implant bar reference object with errors in the insertion direction [in microns].
For demanding work, such as long span bridges and implant bars, it is especially important that a scanner is re-calibrated when moved or when the lab’s temperature changes. Therefore, 3Shape scanners come with unique calibration object with a lithographic reference pattern with an accuracy of 2 microns. Software algorithms cannot compensate for highly complex hardware deformations that inevitably follow from transportation or thermal expansion/contraction.
interesting thread. I just discussed the topic with some of my colleagues in the 3Shape Scanner Development department.
They had some interesting points that I would like to share with you. See below:
.......................................................................................................
What’s the optimal vertex area density?
Obviously, a few vertices per mm^2 won’t capture all relevant details, but “the more, the better” does not hold, either. An unnecessarily high vertex density is a disadvantage, because it means increased processing time all the way through the CAD/CAM chain.
Instead of simply applying “brute-force” meshing with tiny triangles everywhere, 3Shape have put considerable effort into developing an intelligent meshing algorithm. It preserves detail where needed (in the prep area),but uses few triangles in flat or irrelevant parts. The original poster of this thread has since provided a nice example of the outcome of our algorithm (see http://dentallabnetwork.com/forums/f33/scanner-data-samples-9705/#post57741). Also notice that not only the triangle density, but also their elongation matches the surface curvature.
The vertex density and meshing algorithm that 3Shape have chosen as optimal for their scanners is the outcome of extensive real-world testing, with thousands of units produced in labs in many countries. We always welcome feedback from customers who see poor fits anyhow; other errors in the manufacturing process may be the cause.
What does it all mean for accuracy?
We completely agree with the poster’s statement that vertex density is not indicative of accuracy. The science of high accuracy measurement – metrology – applies reference objects with accuracies much higher than the scanner. These objects can only be manufactured by specialized accredited companies or metrology laboratories. With a reference object mimicking a die, it becomes clear that different scanner brands perform quite differently, and many alarming results are revealed – for example, some scanners can round an edge incorrectly with an error much larger than the oft-stated scanner accuracy of 20 microns.
Figure: Two scanner brands’ measurements (blue line: 3Shape, orange line: Other Dental scanner) of a nominally sharp edge in a custom-made ceramic reference tooth object (middle). Note the relatively large difference in accuracy around the sharp edges.
With an implant bar reference objects, 3Shape tests reveal an error of less than 3 microns along the insertion direction, i.e. in the level of the implants seating surface. This error is on the same magnitude as the reference objects accuracy.
Figure: Left: Measuring a reference object with a high-end coordinate measurement machine (accuracy 3 microns). Right: 3Shape scan of an implant bar reference object with errors in the insertion direction [in microns].
For demanding work, such as long span bridges and implant bars, it is especially important that a scanner is re-calibrated when moved or when the lab’s temperature changes. Therefore, 3Shape scanners come with unique calibration object with a lithographic reference pattern with an accuracy of 2 microns. Software algorithms cannot compensate for highly complex hardware deformations that inevitably follow from transportation or thermal expansion/contraction.