Jump to content

Point Cloud questions


Recommended Posts

Last week I got (somewhat unexpectedly) into point clouds and am running into some issues/questions.

I've imported two point clouds into a single Vectorworks drawing, one of 57+ million points and one of 188+ million points. The import went ok, file size is a bit over 800 megabytes. The point cloud files were .xyz files containing only coordinates and nothing else. Currently only the point clouds are in the drawing.

However, Vectorworks limits the display of point cloud objects to 35 million points, even though it is said to import all points of the point cloud.

How does VW sample the 35 million points from the complete set?

Even though it seems to look representative of the full cloud there will obviously be some gaps in the representation and it makes me wonder if I may be missing crucial things if e.g. on 35 million of 188 million (approx. 20%) of the data set is used.

To differentiate between the two (overlapping) point clouds I've set up classes, one with red pen and fill and one with green pen and fill, and assigned one class to one point cloud object and the other to the other point cloud object.

When in top/plan view the assigned class colour shows properly, when in OpenGL it all renders as black/grey. Is this a bug or is it something that can be solved somehow and if yes, how? Or does the sheer number of points interfere with the display in 3D?

The other question, for those having used large point clouds in Vectorworks, is what are your experiences with large point clouds, any particular issues to keep in mind, workarounds/solutions?

Link to comment
  • Vectorworks, Inc Employee

Looks like currently the selection of points is completely randomized, with the main issue being the fitting of the 35million points into memory. There are plans to add user control over this in the future.

Apparently in 3D at the moment, the points are given their color as assigned from the point cloud directly, which is what allows full color scans to appear as they do, but I am unaware of a way to override this 3D color.

Link to comment
  • 3 weeks later...

Jim,

Thanks for the reply, I'll add a wish list request for more control over colour of the point cloud after import.

On another forum it was suggested to break up the point cloud files into multiple parts of 35 million points maximum and then import each part in order to get the full number of points.

I had considered that too as a possibility but it would depend a bit on how the lidar scan was done and put into the file as it could be possible to end up with a point cloud that only covers the extreme ends on both sides and nothing inbetween after splitting the original point cloud file in multiple parts.

Given a distance of several hundred meters between the two ends, how much of an impact would such a point cloud object have on Vectorworks' performance?

Link to comment
  • Vectorworks, Inc Employee

It probably depends on how much other geometry is in the document, but since a point cloud with 35million points is still effectively 35million objects as far as Vectorworks is concerned I suspect trying to double or triple that number would be difficult without exceptional hardware, and even then only 3D views that are handled by your GPU will be capable of keeping up.

That's only a guess, but I suspect the limit in points was made exactly because of that existing limitation.

Link to comment

Thanks for the info. I guess I'll just have to try and figure out. The person who suggested splitting up the point cloud ended up with 6 point cloud objects of 35 million points each and it seemed to work trouble free for him.

Just to make sure regarding the point cloud ends, if in a single point cloud object one cloud half is on one end and the other half of the points is e.g. 300 m away would this make a difference compared to both halves being next to each other with no separation distance or will it not make much difference because it are only the points that matter?

Link to comment
  • Vectorworks, Inc Employee

"Trouble" very much depends on your hardware in this situation I think. Many engineering decisions have to be based on the minimum specs, and I don't think very highly of the minimum specs ;)

You will see performance changes the further any geometry is from 0,0 in the document. But 300m shouldn't be too very much. I tend to see the more severe issues in the multiple kilometer and mile range, which comes from Vectorworks using mm as its base units for performing math, regardless of the document units.

Link to comment
"Trouble" very much depends on your hardware in this situation I think. Many engineering decisions have to be based on the minimum specs, and I don't think very highly of the minimum specs ;)

Yes, minimum specs often seem to be the absolute minimum rather than usable minimum, but my experience is that it also depends on the software itself. In Vectorworks' case it appears to be an efficiency issue across the board regardless of specs as it used to slow down a lot faster than other CAD software I am using. It has improved a bit over the years but there is still room for considerably more improvement. (It's on my wish list)

You will see performance changes the further any geometry is from 0,0 in the document. But 300m shouldn't be too very much. I tend to see the more severe issues in the multiple kilometer and mile range, which comes from Vectorworks using mm as its base units for performing math, regardless of the document units.

In this case I meant that the distance between the two ends of a single point cloud's extremes may be 300m. The actual location can be several hundreds of kilometres away from the 0,0 in the document as it the point cloud is georeferenced to a national grid.

Link to comment

Art V - one option to try would be to subsample the point cloud file using the open source software CloudCompare prior to bringing it into Vw.

CloudCompare gives you control over how the subsampling is performed:

  • Random - max number of points;
  • Spatial - minimum distance between points;
  • Octree - level of subdivision simplification.

It also will allow you to modify the point cloud's colors.

HTH.

Link to comment
  • 7 years later...

Hi.

I am opening this discussion again, as I try to import a point cloud with 498+ million points. 

However vectorworks seems to limit at 100.000.000 points. 

 

I dont seem to find any recent information about how many points vectorworks is able to import at a time. 

Do any of you have the knowledge if it is possible to import larger clouds than 100.000.000 points? 

 

I get this dialog when importing: 

 

Skrmbillede2023-10-26kl_10_41_42.png.dd023701315d6818b05b4e43361a4864.png

 

Thanks in advance. 

 

// Mathias 

 

 

Link to comment
13 minutes ago, mhersland said:

Hi.

I am opening this discussion again, as I try to import a point cloud with 498+ million points. 

However vectorworks seems to limit at 100.000.000 points. 

 

I dont seem to find any recent information about how many points vectorworks is able to import at a time. 

Do any of you have the knowledge if it is possible to import larger clouds than 100.000.000 points? 

 

I get this dialog when importing: 

 

Skrmbillede2023-10-26kl_10_41_42.png.dd023701315d6818b05b4e43361a4864.png

 

Thanks in advance. 

 

// Mathias 

 

 

 

That's weird in Help it says the limit is 35 million not 100 million...

Screenshot2023-10-26at09_56_37.thumb.png.1c1464ca698d57ff1adc8af59177177c.png

 

Can you ask whoever supplied the point cloud to divide it into several smaller ones...?

 

 

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...