Small model limitations

Hi has anyone any close up images of small items with sharp outside and inside edges and flat faces? I am beginning to realise that the pop may not be able to scan my model well enough. I am getting lumpy surfaces and large outside and inside radiuses on what should be sharp edges. The 0.3mm resolution exactly what does that mean (each point can only be placed within 0.3mm?)
I thought I had read that the total dimensions would be within 0.3mm!

My experience is that sharp edges/lines are more accurately scanned when the POP’s sensors are parallel to them. I scanned the cover of a case which has a recessed lip around the top and it was very difficult getting this recess until I re-oriented the POP to better project the pattern into the recess in such a way that the sensors could capture the patterned light.


I know I’m jumping in on this a bit late, but here’s a scan that I did yesterday. It’s quite a small part- only about 2" at max dimension. The narrow part of the curve is .125" wide, and the part overall is .310" thick.

It wouldn’t scan well just sitting directly on the turntable, so I took a terra cotta coaster and put it on the turntable, then took 3-4 scans of the part in various orientations and manually clipped out the coaster and merged the point clouds in Cloud Compare.

You can see the edges all came out pretty well, but it should be noted that there are 2 small holes (.115" each) that didn’t come out, and all the corners on the bit that looks like Texas at the top should be quite sharp.

The .3mm resolution is for dimensional accuracy. So far as I understand it, the dots on any particular frame will be no less than .5mm apart. Subsequent frames will have the dots in different places, obviously, and there will be small variations that lead to some uncertainty in exact dimension, which is why the green outline is kind of fuzzy.


Looks like the trigger guard of a pistol?

1 Like

I’m curious what the scanned object looks like after it has been fused and meshed, because that will clean up a lot of the apparent noise.

As for the tiny holes that did not end up in the scan, first look at the fused and meshed object to verify, but if they’re still not there, then remember that the POP uses Binocular Structured Light, so the pattern that is projected by the emitter must be seen by the two sensors at either end of the POP’s face. If the pattern is projected into a narrow hole, it is probable that the sensors are not going to see much of anything useful.

When I was scanning my car’s engine, I had to remember to orient the POP parallel to the raised portions of the engine block to capture the valleys between the raised portions. When I first started, I kept getting empty areas that had to be filled.

1 Like

That is the fused point cloud, with some cleanup done. Some of the noise comes from a slight radius to the outer faces, but not nearly all of it.

Here’s what the mesh looks like:

Additionally, the process of merging introduced some errors. I suspect that it tried to fit things to the curve and then just got some stuff wrong. The size of the features at both ends is right (within some tolerances for small details and sharp edges), just that the long curve in the middle is wrong.

(The picture came from placing the part on a flatbed scanner, scanning at 300dpi, then inserting the image and scaling it to 731px / 300 dpi.)

Yes, yes. I know how it works. I’m not complaining about losing 3mm holes in an object that’s already smaller than what the manufacturer says is the minimum size to scan. :wink:

A PPK, yes.

1 Like