my scanning intention is documenting archaeological finds, especially neolithic pottery sherds. I want to produce intersections and good 2D illustrations. Only taking pictures is offen difficult because of small details of incised decorations, whose visibiity is strongly depending on light conditions.
So i had the chance of comparing my mini to a really big Hexagon industrial scanner with professional software (in the background of the picture):
First we had to change lenses, then to recalibrate. The whole process of scanning and processing took longer than with mini, and the results were better, but not much better as you see on the second picture. Mini is not so sharp and detailed, but it did a very good job. The object is about 6 cm in diameter.
The only problem with mini (and also Pop3) i have is that merging several scans produces “seam lines” and different surfaces. It seems that the volume of the single scans is slightly different so that merging doesn’t snap scans exactly overlapping together. In this case the solution was to do one single scan in different angles with the pause and resume function. But this will not be possible in all cases. Suggestions anybody?
Hi Daniel , I will check today if the software produces bad seams or if it has to do with something else .
I experienced something off yesterday , maybe the merging have some bug , I will find out .
In Main time I would suggest you try to merge both scans using Cloudy Compare and see if you get the same seam results , Cloud Compare have very professional approach regarding merging and registration of scans .
Wow, that’s interesting! Because Pop3 scans were even worse to merge than mini ones. My other workaround except scanning all in one for this problem was to reduce fusing accuracy in advanced mode a little but, then it got better.
there are a couple of things that you can do to avoid or reduce seams in Revoscan
Both scans to merge must have the same rebuild density
You have to have a good overlap edge, around 20 to 40 percent for a good registration
Sometimes using markers mode registration yelds better results
the edge of the models (where the data ends) on each scan should be removed since there is usually some noise generated at the fringe of the scanned object
Alternatively and for more precision it is better to use Cloudcompare. You can then reimport the pointcloud to RevosStudio and remesh it there
For the best precision and detail try to have the scanner the closest possible to the object with the highest scan range setting set to the lowest possible value.
From my research and understanding of the way Revopoint scanners work, your limit is the XY capture grid resolution that the depth cameras have and the 16bit Z data capture.
The closer you are and the less depth range set, the better the detail in XYZ
In my case just the latest Win version.
Less fluid and slow. Nearly always freezes on exit.
I have the last 6 releases so I can roll back.
All previous Win versions work fine for me with Mini and Range.
The seam problem is all caused because the way how Revopoint processes scans. Imagine that you’re scanning something in two “chunks”, you scan the top and the bottom. And after each scanning what happens? Of course it’s the “fusion” step, this is the culprit why we get the seams. The two scans don’t “know” any information about the other but only themselves thus the global registration only takes into account all the frames from that single scan, because of that the two scans can have different alignment.
Now the worst part is the fusion process, as the name suggests it first finely registers all the frames and then fuses (or merges) them resulting in one single point cloud. If you now try to align these two fused point cloud fine registration can’t happen because these point clouds lost the memory of which frames are which - making them just two big separate chunks of data and thus fine alignment is impossible.
If Revopoint changed the way fusion process works, which would be to just make it finely align the frames and NOT merge them then global registering of multiple point clouds would be possible.
Let’s say merging multiple scans would be like adding frames from one scan to the other
That’s not how it works , both scans are already fused before merging into editable point clouds .
The Merge functions do not access any frames or data from main folder , it only align the second scan based on features or hand placed markers to the first scan you choice to merge .
Seams showing up when both scans differ with accuracy , scanned at different distance or fused previously at a different pitch point .
The Revo Point have only global simple registration function , it is not as advanced anyway .
How higher the accuracy of the scans how more chance if a seam .
I guess the best option would be to have manual alignment merge option like we got in Revo Studio but if your scans differ , there will be always a seam , that why merging 2 scans from different Scanners will always produce a seam .
In the latest case of Revo Scan 5.2.1 it is obviously a bug that need to be fixed .
That’s is not all true at all , you only merge fused point clouds ( chunk of data ) . No matter in Cloudy Compare or other high end software .
The results don’t need access to any frames, the point cloud have already the all nesesery point cells frames ordered and proper fused , this is made up scenario here my friend , things don’t work this way .
Cloud Compared can process very fine point cloud registration from Revo Scans .
I just believe Revo Scan don’t have that level of precision to do the same job .
No matter chunk of data or not …
Also @Archaeoscan.by each scan made using 5.2.1 create slightly different accuracy despite the fact the distance was the same resulting in shifting of details , so please Daniel , install 5.2.0 for now and do not bother with 5.2.1 until this problem is fixed .
There is a new RS 5.2.2 released today , I did not tested it yet , but I will … you can check the new version in main time , it is in Download Center on Revopoint3d.com
