C2C

[landmeterbeuckx]

YES! YES! A hundred times YES. Until this feature is incorporated, C2C is a non-starter for our work (with light exceptions).
I've mentioned this to countless Faro reps and they all say its coming or they're working on it. I do think that Faro shouldn't overlook the "survey grade" functionalities of Scene in the interest of simplicity. Provides for nice brochure literature to say "no targets, simple one button registration", but that will drive the discerning customers to other solutions.

C2C registration with forced honoring of survey control is a MUST HAVE for Faro and Scene. Rigid C2C bundle best fit is not a solution.

A solution for this is to register in Pointcab if you use targets and spheres. This registration goes the geodetic way. Much more convenient for anyone looking for a survey grade solution.

But most scanning companies don't care about this, it's just the nice images that count, whether they are correctly registered or not.

I seem to remember seeing something about PointCab now doing C2C...anyone got any experience with it and want to comment?

There is some registration possible in Pointcab with planes but for the moment i don't find it ready to replace other softwares. With spheres and checkerboards it runs excellent.

[SAttaya]

Martin,
Please look at viewtopic.php?p=103584#p103584
specifically
"I seem to remember seeing something about PointCab now doing C2C
...anyone got any experience with it and want to comment?"
because of your comment above
"I will keep this thread updated if we have new information or a release schedule."

Did I just miss a post?

a response to the URL above crossreferenceing would make sure more people see it.

[SAttaya]

A solution for this is to register in Pointcab if you use targets and spheres. This registration goes the geodetic way. Much more convenient for anyone looking for a survey grade solution.

But most scanning companies don't care about this, it's just the nice images that count, whether they are correctly registered or not.

I seem to remember seeing something about PointCab now doing C2C...anyone got any experience with it and want to comment?

There is some registration possible in Pointcab with planes but for the moment i don't find it ready to replace other softwares. With spheres and checkerboards it runs excellent.

Apparently PointCab has Cloud-2-cloud, call it scan2scan also "scan to scan" and "cloud to cloud" if you're doing searches.

We still would like to get some real-world experience stories of
someone taking a project with a completed target based registered (spheres, etc, preferably not planes)
and doing the same registration in PointCab's scan2scan/cloud2cloud/etc
(starting with the same input, not the results of the target-based registration (TBR)
and comparing the results from original TBR to PointCab's C2C like was tested/demonstrated
starting with viewtopic.php?p=103456#p103456 and following posts
along with comments about time to register, time to set up / pick up targets
and of course, registration results.

(This isn't in the pointCab section of the forum, because I'd like to see the comparison
to what SCENE does in C2C...which isn't great when you examine clipping boxes of scan overlaps.
Think thickness of cloud on planes in the scan that got the best scan of each (not all) planes
from when looking multiple scans overlaps. That's the best you can get. any scan points outside
of that are by definition "out of registration", although they may be usable and cost effective
for the effort to get a "better" registration.)
I'm sure Martin of PC could comment, but additionally, we'd like to see customers who used/compared the tool.

A great scan to scan registration takes into account that scan data at greater distances and more oblique angles
of incidence either shouldn't be used or give a smaller weighting value at that range and angle, even though
points closer and with good angles of incidence in the same scan should be used.

Interestingly at GEOweek they have a scanner shoot-out, but not a regisration shootout
comparing the quality of the registration .vs. the effort (during scanning and back at office)
...because a great registration is the only registration...everything else results in un-registered point cloud
that just "looks" great depending upon how far away the camera viewpoint is from the scan data, right?

Enough rambling, b2w.

[VXGrid]

I seem to remember seeing something about PointCab now doing C2C...anyone got any experience with it and want to comment?

There is some registration possible in Pointcab with planes but for the moment i don't find it ready to replace other softwares. With spheres and checkerboards it runs excellent.

I'm sure Martin of PC could comment, but additionally, we'd like to see customers who used/compared the tool.

Someone summoned me?

Apparently PointCab has Cloud-2-cloud, call it scan2scan also "scan to scan" and "cloud to cloud" if you're doing searches.

We still would like to get some real-world experience stories of
someone taking a project with a completed target based registered (spheres, etc, preferably not planes)
and doing the same registration in PointCab's scan2scan/cloud2cloud/etc
(starting with the same input, not the results of the target-based registration (TBR)
and comparing the results from original TBR to PointCab's C2C like was tested/demonstrated
starting with viewtopic.php?p=103456#p103456 and following posts
along with comments about time to register, time to set up / pick up targets
and of course, registration results.

(This isn't in the pointCab section of the forum, because I'd like to see the comparison
to what SCENE does in C2C...which isn't great when you examine clipping boxes of scan overlaps.
Think thickness of cloud on planes in the scan that got the best scan of each (not all) planes
from when looking multiple scans overlaps. That's the best you can get. any scan points outside
of that are by definition "out of registration", although they may be usable and cost effective
for the effort to get a "better" registration.)
I'm sure Martin of PC could comment, but additionally, we'd like to see customers who used/compared the tool.

So let me real talk this one for once.
In PointCab Origins there is a feature based (checkerboard targets and spheres) and plane based registration.
The feature based registration works fine, if you know what you are doing and what to expect.
There are three steps which need to get done.
a) Find all features
b) Find the right constellation between scans (which sphere in scan A is the same identical sphere in scan B?)
c) Alignment / Adjustment between all features

Automatic feature finding (a)) works okayish for spheres and sufficient for checkerboard targets.
The constellation search (b)) works fine unless there is either a missmatch or bad feature placement.
c) is completely automatic
Today I had a support case with a customer who had some challenges with 70 scans, it took me around 10 minutes for b) and c)


plane based registration in PointCab Origins is currently completely manual.
What does manual mean in this context?
You need to manually match scan pairs.
If scan A and B are overlapping and you like to match them, then you need to procced as follows:

1. Click on the match button
2. Choose the right match
3. Accept

If this does not work (doesn't find the right constellation due to not enough overlapp), you need to rename three planes to identical names.

______________________________________________________________________

We are working on having an automatic constellation search for plane based registration as well, but until this is ready I wouldn't recommend to use or test the plane based scan2scan registration of PointCab Origins for projects whith more than 20 scans.

If you have a project in this size and you like to compare the results, feel free to write me, then I can assist you in the tests.

A great scan to scan registration takes into account that scan data at greater distances and more oblique angles
of incidence either shouldn't be used or give a smaller weighting value at that range and angle, even though
points closer and with good angles of incidence in the same scan should be used.

Interestingly at GEOweek they have a scanner shoot-out, but not a regisration shootout
comparing the quality of the registration .vs. the effort (during scanning and back at office)
...because a great registration is the only registration...everything else results in un-registered point cloud
that just "looks" great depending upon how far away the camera viewpoint is from the scan data, right?

Enough rambling, b2w.

Yup the distance has an effect, not only for plane based or c2c but for feature based as well.
First of all the detection is different because bigger distance means less points on the plane/sphere/target, while the distance measurement has a bigger error, the angle measurement is better.
For planes the surface roughness is important as well.
Should be all accounted for.

[landmeterbeuckx]

There is some registration possible in Pointcab with planes but for the moment i don't find it ready to replace other softwares. With spheres and checkerboards it runs excellent.

I'm sure Martin of PC could comment, but additionally, we'd like to see customers who used/compared the tool.

Someone summoned me?

Apparently PointCab has Cloud-2-cloud, call it scan2scan also "scan to scan" and "cloud to cloud" if you're doing searches.

We still would like to get some real-world experience stories of
someone taking a project with a completed target based registered (spheres, etc, preferably not planes)
and doing the same registration in PointCab's scan2scan/cloud2cloud/etc
(starting with the same input, not the results of the target-based registration (TBR)
and comparing the results from original TBR to PointCab's C2C like was tested/demonstrated
starting with viewtopic.php?p=103456#p103456 and following posts
along with comments about time to register, time to set up / pick up targets
and of course, registration results.

(This isn't in the pointCab section of the forum, because I'd like to see the comparison
to what SCENE does in C2C...which isn't great when you examine clipping boxes of scan overlaps.
Think thickness of cloud on planes in the scan that got the best scan of each (not all) planes
from when looking multiple scans overlaps. That's the best you can get. any scan points outside
of that are by definition "out of registration", although they may be usable and cost effective
for the effort to get a "better" registration.)
I'm sure Martin of PC could comment, but additionally, we'd like to see customers who used/compared the tool.

So let me real talk this one for once.
In PointCab Origins there is a feature based (checkerboard targets and spheres) and plane based registration.
The feature based registration works fine, if you know what you are doing and what to expect.
There are three steps which need to get done.
a) Find all features
b) Find the right constellation between scans (which sphere in scan A is the same identical sphere in scan B?)
c) Alignment / Adjustment between all features

Automatic feature finding (a)) works okayish for spheres and sufficient for checkerboard targets.
The constellation search (b)) works fine unless there is either a missmatch or bad feature placement.
c) is completely automatic
Today I had a support case with a customer who had some challenges with 70 scans, it took me around 10 minutes for b) and c)


plane based registration in PointCab Origins is currently completely manual.
What does manual mean in this context?
You need to manually match scan pairs.
If scan A and B are overlapping and you like to match them, then you need to procced as follows:

1. Click on the match button
2. Choose the right match
3. Accept

If this does not work (doesn't find the right constellation due to not enough overlapp), you need to rename three planes to identical names.

______________________________________________________________________

We are working on having an automatic constellation search for plane based registration as well, but until this is ready I wouldn't recommend to use or test the plane based scan2scan registration of PointCab Origins for projects whith more than 20 scans.

If you have a project in this size and you like to compare the results, feel free to write me, then I can assist you in the tests.

A great scan to scan registration takes into account that scan data at greater distances and more oblique angles
of incidence either shouldn't be used or give a smaller weighting value at that range and angle, even though
points closer and with good angles of incidence in the same scan should be used.

Interestingly at GEOweek they have a scanner shoot-out, but not a regisration shootout
comparing the quality of the registration .vs. the effort (during scanning and back at office)
...because a great registration is the only registration...everything else results in un-registered point cloud
that just "looks" great depending upon how far away the camera viewpoint is from the scan data, right?

Enough rambling, b2w.

Yup the distance has an effect, not only for plane based or c2c but for feature based as well.
First of all the detection is different because bigger distance means less points on the plane/sphere/target, while the distance measurement has a bigger error, the angle measurement is better.
For planes the surface roughness is important as well.
Should be all accounted for.

Martin,
Where is this match button to be found (for the plane registration)?

[VXGrid]

Martin,
Where is this match button to be found (for the plane registration)?

You need to open the registration.
At the bottom between scans is a button called C2C scans
when clicked it will search for planes in the opened scans (if no planes are in there yet) and then will show below the panoramas the matching.
Looks like this:

registration.png

[Leandre Robitaille]

I think there should be a forum rule against these types of post highjacking. There is a thread on the pointcap section called C2C, you should have all taken the pointcab discussion there.

This post was a cry of the hearth to have changes done to the Faro scene c2c algorythm. It was to be used as a reference point to highlight issues the community has with the current state of the c2c in Faro Scene.

I would appreciate if a moderator could move all the pointcab post from the second page to that C2C thread on the pointcab section. All of first page is still on topic.

Thank you,
Best regards,

[landmeterbeuckx]

Leander,

I'm not saying you're not right but it happens all the time. If someone asked which Faro instrument they should buy, another will say an rtc isn't bad as well. Then the thread moves into a Leica one...