Surface triangulation errors from pointcloud

[Anonymous]

I have been extracting surfaces from pointclouds and I have found in various sections very flat sections where points have not been extracted from pointcloud as shown. (However there is still some slope according to pointcloud stats.)

I have tried using a single scan, multiple scans, different level of detail, different settings for triangle length, different point densities. Nothing works.

All areas of tin where this occurs are very flat so I am wondering if there is a minimum slope issue here. I also note that in statistics there is a 0.00m2 triangle.

Any suggestions?

[TerryDotson]

Right click the Surface name in the ToolSpace panel and go to the Statistics, compare the total points in the surface to the total points contained in your source files (LAS, etc).

[Anonymous]

The scans are MLS so they have 8-12m points. I extract a road strip with distance between points filter set to 1.5m. This gives about 10000 points. For surfaces that extend 1km I can have up to 4 of these flat sections.

The example section above only has about 500 points. 

I have just tested moving the scan to new origin but same thing happens but I am pretty sure Recap doesnt alter the original scan co-ordinates.

[rl_jackson]

Can you provide a data set and version of C3d your using?

[Anonymous]

This is another section showing same thing happening. This section has a bit more slope and has only 1 missing section in 1km segment. As you can see scan is pretty intense and I have 38km of road DTM to extract....

[Anonymous]

Hi, I am using 2019.  

Attached is sample data & drawing

https://www.dropbox.com/s/ivj3fy36b7765g2/MLS%20DTM%20TEST.zip?dl=0

[Anonymous]

Further interesting point, I just extracted 2 separate pointcloud sections that overlapped with common overlap on one TIN missing. The other had missing point section in different area. Time to call in Batman.

[TerryDotson]

... points filter set to 1.5m ...

I don't know much about Autodesk's Point Clouds but that sounds kind of coarse and could be the problem.  There are roughly 137,500 source data points just inside your flat section.  Those points are approximately 0.0185 to 0.065 meters apart.  While I admit the source density is a little overkill, there has to be a better balance of point removal.

Same area with all points used followed by two passes of Civil3D's SimplySurface applied:

Every time I start Recap it scares me how close to MS-BOB things have come.

[Anonymous]

Hi Terry,

The point filter is quite important as i have 38km to extract for engineering packages that dissolve into a puddle at the sight of really large TINs. 

I have extracted 16km length now at 1.5m in 1km sections and the funny thing is the higher the average elevation, the less issues, with the the last few kilometres fine. As well as the fact that the gap can move around depending on which rcs files are used tells me it is a program problem.

The points wizard is actually brilliant so if the bug is sorted out Autodesk could have a killer app.

[enderak]

My apologies for resurrecting such an old topic but I have been having the same or similar issue and I think I found a workaround at least. Hopefully this can help someone else. For reference I am on Civil 3D 2020 and Recap 6.1.0.71.

My LIDAR data was brought into Recap as tiles. Each tile was a separate .xyz file. From this large area (~43 square miles) I needed to create a surface of a corridor 200' wide by 6 miles long. At first it seemed like if my boundary polyline did not have a vertex within one of the tiles, that tile was skipped when making the surface, so I added vertices to make sure each tile had a vertex within it. This seemed to help but did not completely fix the issue. I noticed each vertex had a "bubble" of points around it. So I added more vertices to the boundary, even along the straight sections. I ended up using every 100' (though they probably could have been further apart). With this boundary polyline having vertices every 100' at most I was able to create the surface with no issue.

[King_Winston]

Hi,

I worked around this issue by creating separate tin boundaries in affected areas and then triangulating and merging. I suspect my smaller boundary had more points as your solution alludes to. Sorry to say I have not bid for this type of work or used Civil3D again due to bug. And I haven't eliminated large co-ordinates as being part of the problem, as they have with Autocad since 2.18, here in Australia.

Thanks anyway Justin