Scanner Data Samples

[patmo141]

Ok, I'd like to keep this thread short, on topic and OBJECTIVE...so no bashing. Results are results

Objective: To assess the density of data in various mesh examples from different scanners

Method: I take a small sample ~1cm^2 of a mesh from an stl example file. I calculate the total area of all the triangles, and the average length of all the edges and report

1. average edge length which is indicative but not perfectly equal to vertex spacing

2. Average mesh density = total vertices/total area

In some meshes, the density varies greatly over different parts (eg, prep area vs the opposing dentition). In those cases, I report values for two different places on the mesh.

Discussion: It should be noted that because many of these meshes are "processed" by their own CAD software, these results are not necessarily indicative of scanner accuracy, no matter how much we would like them to be. For example, you will notice the CEREC opposing mesh has more data than the prep mesh! How can this be!? It's because the prep mesh has been "decimated" meaning that data has been thrown away where it is not needed to improve CAD efficiency in later steps. Also, there is great variation across many of the same meshses and unfortunately, I didn't calculate any statistics or uncertainty. So, grains of salt distributed, it's up to you to put them on the food...

Results Summary (this will be edited with all results as I collect more)

CEREC AC Prep
vert spacing: 213 microns
area density: 31.5 verts/mm^2

CEREC AC Opposing
vert spacing: 148microns
area density: 63.3 verts/mm^2

Bruxir Die Model
vert spacing: 63.6 microns
area density: 292.9 verts/mm^2

Bruxir Adjacent Model
vert spacing: 133 microns
area density: 70.5 verts/mm^2

Bruxir Opposing Model
vert spacing: 129
area density: 74.7

3M COS Prep area
vert spacing: 107 microns
area density: 114 verts/mm^2

3M COS Adjacent teeth area
vert spacing: 148.9 microns
area density: 51.7 verts/mm^2

Origin HD(mode)
vert spacing: 66.6 microns
area density: 267.5 verts/mm^2

Origin HD Complex (mode?)
vert spacing: 74.0 microns
area density: 216.5 verts/mm^2

Dental Wings Coping
vert spacing: 156 microns
area density: 53.6 verts/mm^2

3Shape Full Contour Incisor
vert spacing: 199 microns
area density: 31.6 verts/mm^2

3m Coping exp as Rapid Prototype
vert spacing: 77.8 microns
area density: 186.7 verts/mm^2

CEREC Blue Coping
vert spacing: 126.7 microns
area density: 56.35 verts/mm^2



Comments on methodology or for clarification welcome. Also, posting pictures of actual fits on dies (the ultimate indicator) is encouraged.

You can download this .blend file with all the meshes from here.
http://dl.dropbox.com/u/2586482/mesh_analysis.blend

You can send me a file to add to the mix. STL files only please
https://www.yousendit.com/dropbox?dropbox=Dental-CAD-FIles

 

[NicelyMKV]

The origin files were straight from the scanner. No CAD program manipulation.

Thanks Patrick!

 

[DMC]

.......!

 

[Bobby Orr ceramics]

Very interesting Scott !!

 

[NicelyMKV]

Pat,
The "complex" setting just shortens the table rotation between scans and doubles the amount of scans taken. I have received several cases back that were scanned with normal and higher scan data settings and so far see no difference in fits. I did not do two crowns on the same model with the two settings so some may argue how accurate my test was,but.....

I will say that my scanner, at normal settings, seems to have much higher scan data than some of the other systems.

Jason

 

[patmo141]

Cercon Art bridge Framework
vert spacing: 163.2 microns
area density: 54.1 verts/mm^2

(can't edit orginal post)

This mesh and a few of the others above are meshes of the restorations, not a scanned model. Since a restoration has a top which is generate in software, it can have whatever spacing and accuracy we want, so it's not really a reflection of the scanner data except that the internal aspect is "derived" from the scanner data. Usually there are quite a few steps/calculations (undercuts blockouts, cement gap) which may prevent even the internal aspect of the restoration mesh from representing scanner data. Just a heads up when interpreting these things.

 

[DMC]

LOL.

I mean, Pat....Did you get the Four files I sent you a few days ago?

No? Yes? Maybe?

Scott

 

[patmo141]

Yeah brah, the results are up there!

 

[paulg100]

Just sent you an STL of a Ineos Red scan pat, looking forward to the info

i wont turn this into a flame thread but the blue already sucks so god knows how the red will compare. popcorn

 

[DMC]

Actually, the Blue is not really that bad.

I must say that after working with them for a while now.

 

[patmo141]

@ paulg

I haven't downloaded it yet (just got home),but I have a prediction....I bet the spacing on the red cam will be THE SAME as the spacing on the blue cam (gasp) and here is why. With structured light scanning , the spacing of the point clouds is determined by the resolution of the camera sensor and the area/volume it is scanning (field of view). Each pixel ends up reporting a depth. Because the Blue Cam and the Red Cam have the same resolution sensor (640 x 480 i believe)....depending on the lenses, field of view etc, they may have similar resolution. The main difference is going to be accuracy, which means we are more confident that each point is where it says it is with the blue cam vs the red cam ....going to do test riiight now.

 

[patmo141]

After I ran the analysis on the small segments I did it on the the entire mesh to see how much variation there was throughout the mesh. It appears to be fairly consistent across the whole model. (I also did this for the Blue Example mesh above). I will say this, my prediction was wrong with regards to vertex spacing. The Red Cam's vertices are actually closer together, yet I don't think anyone will argue that clearly the Blue Cam's data looks better

Cerec Red Tooth
vert spacing: 173 microns
are density: 47.3 verst/mm^2

Cerec Red Implant Abut
vert spacing: 178.5 microns
area density: 42.7 verts/mm^2

Cerec Red Entire Mesh
vert spacing: 176.6 microns
area density: 45.2 verts/mm^2

Cerec AC Entire Mesh (prep side)
vert spacing: 213.2 microns
area density: 30.5 verts/mm^2

**Repeated data from above***

CEREC AC Prep Region Only
vert spacing: 213.2 microns
area density: 31.5 verts/mm^2

Here are the two meshes side by side (red left blue right)

 

[rkm rdt]

No wonder Paul sees "red" when you mention CEREC.

 

[RileyS]

I understand all your digital lingo, but for the rest of the unintelligent people on the forum...will you explain the stats?

 

[Jo Chen]

Were all the scans done on the same model? Just wondering. Scientifically it would be great to have a reference object for future comparison when scanners and software improve. Much like the kilogram object kept in Paris since 1880.

 

[patmo141]

I understand all your digital lingo, but for the rest of the unintelligent people on the forum...will you explain the stats?

Will do as soon as time allows.

 

[patmo141]

Were all the scans done on the same model? Just wondering. Scientifically it would be great to have a reference object for future comparison when scanners and software improve. Much like the kilogram object kept in Paris since 1880.

Indeed, they were not all done on the same model which makes this test anecdotal at best. Alas, the idea of a "master die" has been discussed quite a but and i think some 'standards of assesment' are being laid out by some committee in the sky. A draft of the doc was posted on the forum some time ago.

 

[paulg100]

thanks pat, very interesting, and thats all ill say about that

 

[adamb4321]

Were all the scans done on the same model? Just wondering. Scientifically it would be great to have a reference object for future comparison when scanners and software improve. Much like the kilogram object kept in Paris since 1880.

A bit off topic, the Kilogram object kept in Paris is actually losing weight, albeit a miniscule amount.

 

[patmo141]

Emax Lid from 3Shape

I just received a scan of an "Exam Lid" from a 3Shape scanner (not sure what make and model),I was surprised at the relative large average spacing of the mesh and low area density of points. When I took a closer look, I realized that this is a perfect example of a mesh which has data where it needs it, and not where it doesn't. I also realize that the emax Lid may not be a perfect "master die" because the edges and ridges aren't all that sharp, so a scanner which reduces it's data density based on sharp features will be disadvantaged in our test. Bit, it is also a very good test object because it has regions of different complexity (the top, vs the smooth conical part, vs the ridged sides).

Results:
Emax Lid 3Shape (the whole mesh)
vert spacing: 215 Microns
area density: 32.15 verts/mm^2

Emax Lid 3Shape Dense Area (indicated in yellow)
vert spacing: 160 Microns
area density: 65.8 verts/mm^2

Emax Lid 3Shape Smooth Area(indicated in red)
vert spacing: 310 Microns
area density: 18.5 verts/mm^2



The 3Shape Model


For Comparison

Emax Lid Origin HD Complex
vert spacing: 74.0 microns
area density: 216.5 verts/mm^2


The Origin Model