In the screenshot below you can see floor plan with 3 overlapping scans that are not registered correctly.
The envisioned workflow. If we look at the screenshot above. It seems that it would be very quick if in that view I could just define which is the wall by which I want to align all 3 scans, say the horizontal wall, and then tell how much to shift each scan so that they match. For example, I click 6 times to define the points that define a plane like this:
At this point I don't care about the z axis. For example, I hit toolbar start button and click two times and see how wall of the yellow scan becomes horizontal, same thing for the gray and white scan. Now rotations of all scans match to horizontal line. The next thing I need is to shift each scan along a horizontal line to match the side walls, so I would click 3 times to get three points:
and then select those points and click another button. Now all 3 scans will match from the birds eye view. Then I need to match the z axis, so I would create one more point on each point cloud on some horizontal surface and then say 'match the z value' or something like that.
This seems to be so much faster than messing with the rotation and move manipulators. I know that this is exactly what the C2C or target registration is meant for, but after I complete the target registration, I get 36mm error despite
my target registration effort at the 9th Scan Manager Tension Table which shows only 3.02mm error:
The scans that overlap in that area are 6, 5, 10:
And you can see that I have spent quite a bit of effort to add and match planes. Plus this is the end result, there was soooo much time that went into rotating and zooming and clipping (and resetting the view from Ortho to Perspective to View All to get rid of 3D view camera clipping bug etc.) to inspect the results, saving steps because sometimes when I set Anti Correspondence I would not be able to get back the previous 'good' "Dist. Error". So while struggling with this I figured that I could register the scans manually in a controlled manner much faster.
In some cases, I find that Scene is little finicky about how many references it needs to register scans using sphere targets or planes and it keeps asking for more references. Some times I feel that it would be faster if I ignored the target tensions table of the scan manager and would just manually match and rotate each scan in the birds eye view to get the walls to match, and then just matched them vertically but it just takes way too many clicks to do that in the correspondence view.Plugin to match two scans manually in the birds eye view without Correspondence View manipulators
-
- I have made 50-60 posts
- Posts: 52
- Joined: Fri Mar 18, 2022 8:23 pm
- 2
- Full Name: Las
- Company Details: Sfee
- Company Position Title: General
- Country: USA
- Has thanked: 6 times
- Been thanked: 2 times
Plugin to match two scans manually in the birds eye view without Correspondence View manipulators
You do not have the required permissions to view the files attached to this post.
- Leandre Robitaille
- V.I.P Member
- Posts: 542
- Joined: Sat Aug 03, 2019 1:53 am
- 4
- Full Name: Leandre Robitaille
- Company Details: Cima+
- Company Position Title: Civil Technician - Surveyor
- Country: Canada
- Has thanked: 61 times
- Been thanked: 246 times
Re: Plugin to match two scans manually in the birds eye view without Correspondence View manipulators
This is just a poor understanding of how scene works and how to distribute and control your errors.
First. Planes are horrible for placing scans. Angular error on these are to high and should only be used to roughly allign something for another registration( i.e c2c 2 scans together; do a plan + a point, target based, then c2c)
Second use more spheres.
Third, you need to understand how cluster work. They lock your configuration and prevent error distribution outside a cluster. Hence I try to have everything on the same level as much as possible.
How to fix this project; bring all scans back to the same level. Disband all clusters (right click, registration, disband all cluster). Also delete all plans and points.
Now all scans that see sphere between one another are part of the traverse, they stay on that top level you just created. You need to then bring in clusters scans that do not see spheres and c2c fit those clusters, each cluster need to be the smallest possible, less scan as possible, but all of those need to have spheres in a FS / BS way and your path needs to loop, similar to closing a traverse when doing total station.
A normal scan would be lets say a 3 sphere link with the previous and 3 sphere link with the next scan.
If you have 2 scans that do not link with sphere to then link with the next scan you need to do a c2c of those 2 scans together.
First. Planes are horrible for placing scans. Angular error on these are to high and should only be used to roughly allign something for another registration( i.e c2c 2 scans together; do a plan + a point, target based, then c2c)
Second use more spheres.
Third, you need to understand how cluster work. They lock your configuration and prevent error distribution outside a cluster. Hence I try to have everything on the same level as much as possible.
How to fix this project; bring all scans back to the same level. Disband all clusters (right click, registration, disband all cluster). Also delete all plans and points.
Now all scans that see sphere between one another are part of the traverse, they stay on that top level you just created. You need to then bring in clusters scans that do not see spheres and c2c fit those clusters, each cluster need to be the smallest possible, less scan as possible, but all of those need to have spheres in a FS / BS way and your path needs to loop, similar to closing a traverse when doing total station.
A normal scan would be lets say a 3 sphere link with the previous and 3 sphere link with the next scan.
If you have 2 scans that do not link with sphere to then link with the next scan you need to do a c2c of those 2 scans together.
-
- I have made 50-60 posts
- Posts: 52
- Joined: Fri Mar 18, 2022 8:23 pm
- 2
- Full Name: Las
- Company Details: Sfee
- Company Position Title: General
- Country: USA
- Has thanked: 6 times
- Been thanked: 2 times
Re: Plugin to match two scans manually in the birds eye view without Correspondence View manipulators
Thank you a lot for your input, Leandre, but I am still interested in hearing if anyone has written any plugins that would allow to match the scans without the finicky Correspondence View manipulators.
Regardless, what did you mean with
Regardless, what did you mean with
Leandre Robitaille wrote: ↑Tue Apr 04, 2023 2:51 am ..but all of those need to have spheres in a FS / BS way and your path needs to loop, similar to closing a traverse when doing total station.