Survey expectations in the point cloud era

[line-weight]

The surveyors who did my survey told me:

 

This is the first time that this has happened with the cloud data, I’ve sent the files to Leica to ask for their help

[Tom W.]

Others will know more but I'm not convinced a laser scan is the best way to survey a landscape anyway. All your screenshots are of external environments. Isn't terrain + landscape features best surveyed with a GNSS antenna to produce spot elevations? Or does your project require a really detailed scan of the environment showing the exact size/shape of all the existing vegetation, etc?

[line-weight]

Yes, that's one of the things I wondered, whether a point cloud is actually the right thing for a topographic survey. There's quite a lot of "stuff" - trees, paths, steps, outbuildings etc so actually it's quite handy to have the point cloud because I can use it to check all sorts of little details but spot heights + positions of main features was all I asked for when I initially got quotes.

 

The survey company seem to mainly do internal floorplans so I wonder if they actually don't quite know what they are doing with this kind of thing. It's quite a tricky site with a steep slope across it.

Edited September 14, 2024 by line-weight

[Poot]

On 9/14/2024 at 10:24 AM, line-weight said:

Yes, that's one of the things I wondered, whether a point cloud is actually the right thing for a topographic survey. There's quite a lot of "stuff" - trees, paths, steps, outbuildings etc so actually it's quite handy to have the point cloud because I can use it to check all sorts of little details but spot heights + positions of main features was all I asked for when I initially got quotes.

 

The survey company seem to mainly do internal floorplans so I wonder if they actually don't quite know what they are doing with this kind of thing. It's quite a tricky site with a steep slope across it.

I am not convinced that laser scans are the best way, outside of small residential sites that are fairly simple to clean up and use, or specific technical projects which require exact precision and have a budget for staff to spend a lot of time cleaning up the models.

Otherwise , I think it is a bit outside our competency.

There is still a large difference between a local surveyor with a small drone, and a proper surveying company that can do aerial lidar surveying (and drones in addition). Professional surveying companies who work regularly with LIDAR, laser, or similar tech  are quite technically involved in cleaning up their raw surveys while maintaining georeferenced accuracy that can be counted on/reliable to build in upon in the real world to extreme accuracy that takes vegetation,etc into account. They don't work in CAD programs, which are not meant for this kind of work, and simply cant handle the massive amount of data in very useful ways.

I am sure there are exceptions...but it is also easier to pay and get cleaned up surfaces, contours, or pointclouds which can be used without dumping half your project budget into making it workable.

[line-weight]

I now want to get my point cloud into a georeferenced file, so that I can start bulding my site model and then my building model around it.

 

The pointcloud itself isn't georeferenced - so I am going to have to do this manually.

 

Any recommended tutorials for how to do this in VW2025? Some of the ones I can find seem to be based on previous versions of VW and I'm aware that some new stuff (Survey Point tool?) has been added in 2025 that is supposed to make the process easier.

 

I haven't really tried setting up georeferencing properly before but want to try it out on this project.

[Tom W.]

1 hour ago, line-weight said:

I now want to get my point cloud into a georeferenced file, so that I can start bulding my site model and then my building model around it.

 

The pointcloud itself isn't georeferenced - so I am going to have to do this manually.

 

Any recommended tutorials for how to do this in VW2025? Some of the ones I can find seem to be based on previous versions of VW and I'm aware that some new stuff (Survey Point tool?) has been added in 2025 that is supposed to make the process easier.

 

I haven't really tried setting up georeferencing properly before but want to try it out on this project.

 

The point cloud needs to include one or more known coordinates in order for you to correctly geographically locate it in a georeferenced file. The Survey Point tool allows you to retrospectively georeference a file but it relies on you entering at least one coordinate.

 

What exactly are you wanting to do? Report geographical coordinates from the model? Import/export georeferenced geometry?

[line-weight]

The point cloud doesn't include any reference to the outside world. So I need to georeference it manually.

 

Then in theory (I think) if I import it into an already georeferenced file it will appear in the right location.

 

I think I have worked out how to georeference it - basically I use the "geolocate" tool, and match the point cloud's position & rotation as best as I can, with the background imagery provided by the geolocate tool.

 

I know this is not very precise ... I am doing it mainly as a proof of concept for now, to get my head around how georeferencing works in VW.

[line-weight]

A question...

 

When setting up a new drawing, what's best practice for the rotation of the geometry?

 

(A) I draw the building in a convenient orientation, so X and Y axes are parallel with the main walls of the building, etc, and then all the rotation necessary to georeference it happens behind the scenes as it were

 

(B) I draw the building such that X & Y axes match North-South and East-West, in which case I guess I'd generally be working in rotated views when actually working on the drawings.

[Tom W.]

1 hour ago, line-weight said:

A question...

 

When setting up a new drawing, what's best practice for the rotation of the geometry?

 

(A) I draw the building in a convenient orientation, so X and Y axes are parallel with the main walls of the building, etc, and then all the rotation necessary to georeference it happens behind the scenes as it were

 

(B) I draw the building such that X & Y axes match North-South and East-West, in which case I guess I'd generally be working in rotated views when actually working on the drawings.

 

To date I have used option (B) but with the Survey Point Tool (A) becomes a more accessible option.

 

If your building has few square walls to begin with it's less of an issue...

[Jeff Prince]

3 hours ago, line-weight said:

The point cloud doesn't include any reference to the outside world. So I need to georeference it manually.

Are you sure?  Point clouds can contain georeferencing/positioning information.  Your surveyor needs to publish that in the file and your software needs to read it.  Unfortunately, Vectorworks has an outdated library according to @Dave Donley (IIRC), so it sometimes mishandles/ignores positional data.  I wish Vectorworks would provide precise instructions on this workflow, what libraries and standards should be used, and to resolve common issues.  Point clouds seem like yet another half baked marketing feature when things go wrong.

[line-weight]

48 minutes ago, Jeff Prince said:

Are you sure?  Point clouds can contain georeferencing/positioning information.  Your surveyor needs to publish that in the file and your software needs to read it.  Unfortunately, Vectorworks has an outdated library according to @Dave Donley (IIRC), so it sometimes mishandles/ignores positional data.  I wish Vectorworks would provide precise instructions on this workflow, what libraries and standards should be used, and to resolve common issues.  Point clouds seem like yet another half baked marketing feature when things go wrong.

I don't think there's anything in the file, not that gets recognised in import anyway.

 

It's become clear this survey wasn't really properly done - I'll know better next time around, exactly what to ask for.

[Tom W.]

32 minutes ago, line-weight said:

I don't think there's anything in the file, not that gets recognised in import anyway.

 

It's become clear this survey wasn't really properly done - I'll know better next time around, exactly what to ask for.

 

I have had major issues importing georeferenced point clouds into georeferenced VW files. They come into the file in exactly the right geographical location but with half the points missing due to (I imagine) some variation of the far-out-objects phenomenon. The only way to have the point cloud import with the full complement of points is to have it centred on the User Origin AND Internal Origin then move it into geographical position manually (in my case via a special datum point included by the friendly surveyor). This is obviously majorly annoying + stupid + defeats the whole purpose of georeferencing.

 

I posted about this several times + eventually filed a bug on it back in March this year but haven't had a meaningful response yet (VB-202892 if anyone from VW is reading this...).

[line-weight]

On 12/6/2024 at 4:46 PM, Tom W. said:

To date I have used option (B) but with the Survey Point Tool (A) becomes a more accessible option.

 

I'm curious to understand what the Survey Point Tool changes about this.

 

As far as I can work out, while you can place a Survey Point graphically (eg snap to something in the survey info) and then move everything to the location that you define by entering numerical values, the same is not true for the "reference points". It seems that you can only place them in the drawing by entering numerical co-ordinates.

 

Therefore it seems that once you've placed those reference points in the drawing, the only way to get the drawing to align with them is to rotate the geometry itself within the drawing. I may be missing something here though.

[Tom W.]

5 minutes ago, line-weight said:

 

I'm curious to understand what the Survey Point Tool changes about this.

 

As far as I can work out, while you can place a Survey Point graphically (eg snap to something in the survey info) and then move everything to the location that you define by entering numerical values, the same is not true for the "reference points". It seems that you can only place them in the drawing by entering numerical co-ordinates.

 

Therefore it seems that once you've placed those reference points in the drawing, the only way to get the drawing to align with them is to rotate the geometry itself within the drawing. I may be missing something here though.

 

The triangulation reference points allow you to check that you have the correct rotation when you have applied an Angle to True North.  Note: when you give the Survey Point coordinates + apply an Angle to True North it is the globe in the background (which you see when using Geoimage or Geolocate tools) which you are moving/rotating, relative to the drawing geometry.

[line-weight]

20 minutes ago, Tom W. said:

 

The triangulation reference points allow you to check that you have the correct rotation when you have applied an Angle to True North.  Note: when you give the Survey Point coordinates + apply an Angle to True North it is the globe in the background (which you see when using Geoimage or Geolocate tools) which you are moving/rotating, relative to the drawing geometry.

 

I see. So effectively it lets you rotate the globe in the background by typing in a number, rather than graphically as you might do with the Geolocate tool.

[Tom W.]

2 minutes ago, line-weight said:

 

I see. So effectively it lets you rotate the globe in the background by typing in a number, rather than graphically as you might do with the Geolocate tool.

 

I have only used the Geolocate Tool to reposition the Internal Origin in an already-Georeferenced file. I believe you can use Search Mode to enter coordinates if you want to.

 

The Survey Point tool as far as I understand lets you easily Georeference a file after the fact i.e. you've drawn a building in a file with the Internal Origin + User Origin at 0,0 + you've drawn it square to the X/Y axes + then want to correctly align/orientate the background globe relative to the model. You can keep the model as it is + keep the User Origin in the same place relative to the model but reposition/rotate the globe in the background so that Stakes will return the correct E/N coordinates + surveys will import in the correct geographical location + with correct rotation. 

[line-weight]

1 hour ago, Tom W. said:

The Survey Point tool as far as I understand lets you easily Georeference a file after the fact i.e. you've drawn a building in a file with the Internal Origin + User Origin at 0,0 + you've drawn it square to the X/Y axes + then want to correctly align/orientate the background globe relative to the model. You can keep the model as it is + keep the User Origin in the same place relative to the model but reposition/rotate the globe in the background so that Stakes will return the correct E/N coordinates + surveys will import in the correct geographical location + with correct rotation. 

 

I believe you can do this with the Geolocate tool too.

 

In my case, I just have a survey without any external reference. So, I can import this, and (if I want) rotate it such that the building I'll be drawing is square with the X & Y axes. Then I can use the geolocate tool to view my survey/drawing with real-world satellite (or map image) behind it, and then move and rotate that background image until I get the best fit possible with my survey info.

 

I think the survey point tool is only really useful if you have a survey with some point on it where the exact real world co-ordinates are known.

 

Next time I get a survey done, I'll be making sure I ask that it includes some real world reference, so I don't have to do all the manual work I'm needing to do with the one I've got for this project, which is a rather ropey one.

[Tom W.]

1 hour ago, line-weight said:

 

I think the survey point tool is only really useful if you have a survey with some point on it where the exact real world co-ordinates are known.

 

Absolutely. But you would always expect a survey to include this.

 

Equally, if you georeference your file first then import the survey it will come in at the correct geographical location automatically + no further action required.

[line-weight]

19 minutes ago, Tom W. said:

Absolutely. But you would always expect a survey to include this.

 

I'd say not always - for example sometimes I'll have done a simple survey myself and that's all that's needed for some projects. Or maybe I've been given some form of survey info that was done in the past before georeferencing became a thing.

[Jeff Prince]

On 12/6/2024 at 12:59 PM, line-weight said:

I don't think there's anything in the file, not that gets recognised in import anyway.

 

It's become clear this survey wasn't really properly done - I'll know better next time around, exactly what to ask for.

when you setup your vectorworks file, are you using the same coordinate System as the point cloud?  If not, that can cause a lot of issues.

[line-weight]

32 minutes ago, Jeff Prince said:

when you setup your vectorworks file, are you using the same coordinate System as the point cloud?  If not, that can cause a lot of issues.

How would I find out what coordinate system the point cloud is using?

 

It's just an .las file and when I import it into VW it appears more or less near 0,0.

[Jeff Prince]

1 minute ago, line-weight said:

How would I find out what coordinate system the point cloud is using?

 

It's just an .las file and when I import it into VW it appears more or less near 0,0.

Ask the surveyor what coordinate system they published the info in.  That info could be in the header if the file.  

When you setup up your georeferencing, select the same coordinate system as what the surveyor says they used.  Then, when you import that data, it should come in very close to your expected location.

[line-weight]

1 hour ago, Jeff Prince said:

Ask the surveyor what coordinate system they published the info in.  That info could be in the header if the file.  

When you setup up your georeferencing, select the same coordinate system as what the surveyor says they used.  Then, when you import that data, it should come in very close to your expected location.

 

Tbh I have my doubts that these surveyors know what they are doing enough to have published it in any co-ordinate system. In this case, I've managed to get things into an orientation that is good enough for what I need. It's something I'll look out for in future projects though.

[Poot]

15 hours ago, line-weight said:

 

Tbh I have my doubts that these surveyors know what they are doing enough to have published it in any co-ordinate system. In this case, I've managed to get things into an orientation that is good enough for what I need. It's something I'll look out for in future projects though.

For these reasons, I have made it a point to request survey files/drawings in the specific coordinate system I would like, and with the file prepared as needed - so it is easy for us to use and not have errors in placement/orientation, etc. that can lead to problems later on.