I’m using a Revopoint Inspire with Revo Scan 5.3.4, and the quality in individual scans is excellent, and the ability to merge two scans of an object is excellent… except, wherever the two scans overlap, the resulting mesh becomes bumpy, instead of the high-quality smoothness in the areas that weren’t captured in both scans. This is frustrating, because it means I need to aim to have as LITTLE overlap as possible between the two scans, except, that makes a proper alignment less likely! I can’t even trim the second scan after alignment; editing the scan breaks the alignment… I’ve tried the three-point manual alignment, but the alignment quality is much worse compared to the good alignment of two scans with a lot of overlap.
I know that I can “smooth” the mesh in a separate step, but, I already had a smooth scan before merging; I don’t want to take a clean scan, and then dirty it, and then automatically ‘smooth’ it and lose details in the process.
Is there any way that I can achieve one of these:
A) Merge scans without any dirtiness
B) Merge scans while limiting the dirtiness to a smaller ‘boundary’ area, so the amount that needs to be cleaned is minimal
C) Erase parts of scans AFTER alignment, while retaining the alignment, so that I can minimize the overlap manually while retaining the high-quality alignment that the large overlap originally gave me
The best is to clean the scans well before fusing , removing parts that overlap too much , clean the edges ( trim) , overlapped points and use Advanced mode for fusing , after merging please use cleaning options like removing lose point and most important overlapped points before meshing .
When meshing please use the proper Grid settings according to you early fuse settings , for example scan was fused at 0.2mm , use meshing Grid 0.2 or 0.19 ( never higher than 0.2 .
It is better to have higher areas of overlapping for merge , as I like to trim the edges manually to get the sharp solid edges before merging … there is always lots of lose and overlapped points at the edges if the scanned model was not scanned complete .
Also pay attention that both scans are scanned at the same distance possible , having 2 scans scanned at different distance or with different accuracy will create a seam what is normal behavior.
Since different distances create different accuracy …
If you really need a pro result , I would suggest you to try free Cloud Compare software after you prepare your scans in RS , it have very superior registration tools …
I made a simple tutorial how to merge the scans in CC , highly recommend to try it out …
Hi Daniel , you see when you scan and don’t finish completely the edges the accuracy my be little less at the end of the edges , a proper scan need full angles to keep all 3D data , most of the edges don’t have it . That why I prefer to scan higher amounts of overlap and trim it before merging to have better results after .
The same thing you get when scanning with 2 scanners and merge them together, the edges will be bumpy or have visible seam because of accuracy differences.
After merging the point cloud should be simplify at 1% , there is s new feature coming on computer software that will help with that and simplify it based on geometry .
Aha, your tip about the grid and fuse sizes has helped a lot. I had been accepting RevoScan’s default settings, but it chose two different “point distance” values for my two scans (0.51 and 0.47), and a default grid size of 0.57! I went back and performed fusion with a point distance of 0.4 on both scans, and then meshed with a grid size of 0.38, and now the noise is greatly reduced. It’s still noticeably bumpier where the scans overlap, but no longer unusably so at least.
I’m going to experiment some with different grid sizes, but I’d still appreciate any advice about getting my merged scans to look as clean as my single-shot scans (or at least, minimizing the area where the merging artifacts appear). It feels like such a waste to be able to make such clean scans, only to have them ruined in merging (even if the “ruining” isn’t as severe now).
I did try CloudCompare btw, but I found my results were very similar to what I got in RevoScan? (This was before re-fusing at 0.4 however.) And CloudCompare’s SOR filter completely garbled my color data.
That is still something that should not happen, the edges of the scan should be not bumpy and I noticed it happening, so probably a bug that need to be investigated , and that is something wrong while capturing …
If you merged 2 scans with different fusing settings you will get bumpier the scan that have lower fusing settings , so pay attention both have the same fusing settings regardless of the accuracy .
Yes please experiment as each scan need specific attention based on its features … but try to not get the settings below the fusing point , for example Fused at 0.2 keep the grid from 0.20 to 0.15 max , never above 0.20 as it will induce artificial points and more noises and bumps
Yeah , if you fuse at 0.4 there will be not much of color data left … since each point represents each color pixel .
Experimenting with different settings will allow you to learn how much you need to use resource for different types of scans … as not all of them need highest settings at all . As I said before many times , a cube needs 8 points too be a cube , not 3 millions … if you know what I mean …
3 milion of points will not make it more accurate than 8 points …
