Merging Scans in Revo Scan: How to Avoid Overlap Artefacts?

Discussions
1

Hello,

I am trying to master the Merge function in Revo Scan 6. (My scans are done with a Mini 2, and I’m using Revo Scan on macOS 26.)

Although everything aligns fine, I’m getting some horrible bumpy textures and noise in the merged meshes. Having searched the forums and read around the issue (and got some kind advice from @PUTV in another thread about general scanning tips), I believe the problem areas are caused by where the different meshes overlap, causing extra noise. The recommendations I’ve read are to avoid too much overlap between the different point clouds to be merged, and to clean up the merged point cloud for overlaps. But even doing this, I’m getting some nasty results.

Here are two point clouds I would like to merge:

to-merge
to-merge800×409 82.2 KB

As you can see, I have cut a bunch of stuff away to avoid overlapping too much in some areas, such as the face (I actually cut too much away in some parts, but that’s fine for this test).

Here is the merge preview:

merge1
merge1600×568 91.8 KB

And here is the merged point cloud:

merge2-overlaps
merge2-overlaps800×823 300 KB

As you can see, there are a bunch of areas in the point cloud that are more dense than other areas. These were the edges of the different overlapping areas of the two scans.

But I’ve cleaned up the above point cloud as much as I can - I have run isolation on 80%, and overlap on multiple values between 0.01 and 2.5. Still there are these denser areas.

This is the resulting mesh:

merge3
merge3600×589 42.3 KB

There are nasty artefacts at all the edge overlap points.

What is the correct procedure to avoid this? All the forum threads I’ve read about it seem to end before a definitive solution is given. How do I trim away or clean the different scans to be merged for the best results, and how do I clean up after to avoid all this noise? (As I say, I’ve already used the isolation and overlap tools.)

Many thanks for any advice!

2

How big is this object? Is there a reason you cannot do one scan all the way around it?

image
image801×426 92.1 KB

You should get rid of more of the scan on the left, sharp points like this don’t merge well. Its going more off of geometry, so the cheek piece of the helmet is what you want. The scan on the right is missing a big chunk, so thats not helping you. I assume you were cutting somthing in the back and had the select all on and it took that section out as well.

I’d spend time trying to get cleaner scans to start with before merging.

3

Hi @kayembi

You need to clean the areas after merging ti avoid artifacts , remove the overlapped points and lose points if any before meshing.

4

Many thanks both for the replies, much appreciated.

It is actually very small - about 12mm x 14mm x 14mm, so at the very lower limit of what the Mini can scan. I’m therefore not expecting perfection; I just figure that trying to get good results at this size will be a good way of learning how to get the best out of the scanner. So at this size, I can get a scan all the way around it, but I have the following problems:

  • One rotation won’t capture all angles.
  • If I do one rotation, pause, move the camera to get more coverage, then do another rotation, I still end up with artefacts, especially around the brow and cheeks. This is I believe caused by overlapping areas where the camera couldn’t quite get the data right for certain parts at certain angles.

This may indeed be a bad example. I had cut part of the cheek out because it had a crease line through it and was trying to merge it with a different part of the face.

Thanks. Even with what I think are clean scans, I’m getting these weird bumpy bits at the overlap areas. but perhaps they are just exacerbated by the part being so small.

Thanks, but I’ve already cleaned all overlapped and loose points before merging and before meshing. So:

  • For each individual scan, after fusing at 0.05, I ran isolation at 80% and overlap at 0.01, 0.12, and 0.25.
  • After merging the different scans, I did the same on the merged result before meshing at 0.05.

So I’m not sure what further cleaning I can do to avoid the artefacts.

5

Try running more overlap points. I usually start with my scanpoint distance, and then go up and down in 0.10 increments. and a few more around the distance I scanned at. So If I scan at 0.2, I’l do 0.1, 0.15, 0.2, 0.25, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1. I know it seems like a lot, but it goes pretty quickly and you will clean up more of that.

another option is you can see when you merge where your overlap is. You know where the trouble spots are, so go back into one of the scans and delete those areas of the scan.

Because this part is so small, I’d work on your scanning methods with single point capture. Set up the scanner with a tripod to reduce movement. Make sure you have scanning spray on there. It looked like you did, but it can be hard to get down into some areas. Making sure its all covered will help you.

Its a learning process, and you are working through it. Everyone wants it to be a magic tool these days and doesnt want to put in any work. Keep messing with it and you will get it. The cleaner your scans the easier and more options you have for cleaning up in the software.

1 Like
6

As Nick stated, move the line up and down when cleaning and don’t relay only on what the software tells you .

You can also run simplfyer after meshing to reduce the uneven density still keto all the details .

Of you can export your point cloud to Cloud Compare and clean it there . It is a free software.

7

also @PUTV couldn’t this manual torture be made by an AI function?

1 Like
8

Hi @Archaeoscan.by

Switch to Cloud Compare , you will have cleaning done with one click

Back in the days cleaning if scans took almost one month , compared to this it is nothing :laughing:

But the merge function don’t works the way it should for a while , it should remove all overlapped points after merge , so actually time consuming right now that why I recommend Cloud Compare not only for cleaning but also for merging if high precision is important .