Pop 2 Volumetric Accuracy

So I’m using my Pop 2 to scan the front chassis of a car for reverse engineering. The results look great to the eye (can clearly see weld beads and small imperfections) but it all falls apart as soon as I start making measurements in CloudCompare. The front clip is essentially a U-shape that is 31 inches wide. After finishing my scan, fusing the point cloud and opening it in CloudCompare, I am getting that the width is more like 32 inches. It varies scan to scan from 31.25 to 32. I would be alright with being off like 0.1 inches but 1/4 inch to a whole inch is too much for my purposes. Can the POP2’s volumetric accuracy not handle objects that big? I am using marker mode with a LOT of markers (like 250). There is basically always 4-5 in the frame. Should I be using Body mode instead? Am I assuming the right units are mm in CloudCompare before I convert to inches?


I’ve tried starting from the middle of the chassis to minimize the volumetric accuracy effects, but that gave me worse results that starting from one end. I also tried scanning more of the cross bars to “close the loop” better but that didn’t improve results. The measurement shown in the picture should be 31 1/16".

I’m an automotive engineer and often use a scanner for similar tasks. A few tips:

1 at the first scan, capture all the geometry, the quality is not less, the main thing is to get the points to which you will glue the following scans. You can’t glue two skanas to get the extreme points. Capture all size-critical areas in one scan.

2 Marker mode works disgustingly, use something as geometric labels that you will delete during post-processing - Lego constructor parts are great

3 if geometry allows you to make measurements manually between two holes, you can place two independent scans at the desired distance from each other using geometry and measurement data, then glue the central missing part of the scan. Do it if everything else doesn’t work, it’s a more complicated job that requires professional software for work, such as geomagic wrap.

  1. Most chassis elements are mirrored, so you just need to scan only half, copy it mirrored and sew it using manual measurements
1 Like

@jleal7 the accuracy is based on the scanning distance of 15cm to the scanned object .
Having scans with different modes will change the accuracy , also different distances will affect that .

So scanning in feature mode don’t give you the same accuracy as Body mode .

Try to keep the scanning distance at the same levels and at the same modes possible. .