Extrude Between 3D geometry artifacts

Question

I am trying to make 3D multipatches of geologic units. I have a TIN of the top and base for each unit, and a polygon that represents each x,y areal extent. I have ensured that the z geometry of the base of each unit is always underneath the top. For each unit, I use "Extrude Between" to make a 3D volume. I am running ArcGIS Pro 3.3.1, but I have been having this same issue on all versions of ArcPro. All of the data is on my local drive. All of the data have defined spatial reference.

I am getting unpredictable artifacts in the output. Sometimes the output is completely empty, and sometimes it creates only a portion of the polygon area. Oftentimes it returns a multipatch with line artifacts. It seems to happen more often with more complex geometries, but not always. Here are some screenshots of the line-y outputs I am getting:

Here is a screenshot of a 2D view of the multipatch and associated polygon area where it created a volume just fine in some areas, completely failed to make a volume in some areas, and created line artifacts in other places.

These are the things I have tried:

* Exploding all multipart features in the input polygon. This sometimes fixes the empty output problem, but not always.
* Buffering the polygon so that the boundaries are not coincident with the TIN. I have tried buffers that are both larger and smaller than the TINs, of varying distances. This does not seem to have any effect at all.
* Simplifying the polygon geometry. This does seem to help, but only when I simplify the polygon so much that it no longer resembles the original.
* Splitting the polygon into smaller chunks. This sort of helps, but I only with some areas of the polygons. Some areas create a 3D volume just fine, and some need to be split into the tiniest chunks to avoid the line-y output. The amount of splitting that I need to do seems random and unpredictable.
* Decimating TIN nodes for the top and base TINs. This has no effect.
* Writing the output to a new file geodatabase. This has no effect.
* Creating new versions of all input files in new file geodatabases. This has no effect.

This issue seems to consistently come up with data from many different projects. Any help would be greatly appreciated!

Trisarahtops · Answer

I've been doing more testing, and making the TINs non-coincident with each other seems to be working to solve the line artifact issue. (yay!) However, a major issue I'm having is that sometimes the Edit TIN tool takes an unworkable amount of time to finish. For one TIN, it took 4.5 hours. All data was on my local drive, so it is not a server issue. I also have a fairly high-powered computer.

I though it might have something to do with the complexity of the buffered polygon, so I tried clipping to a simple rectangle around the data. This worked for some TINs, but the edge of the TIN did not change for other data, even when trying both soft and hard clip. For those, I had to buffer the TIN domain (as @Khalid suggested) and then clip to that boundary instead. Although this workaround works for individual TINs/multipatches, it is not a workable solution for us because each TIN/multipatch will need to be created one at a time. There is not a single workflow that seems to be working for all of the data. We need to make these into 3D visualizations, normally with over 50 different units. It is not workable for us to do all 50+ units individually.

Additionally, I am still getting "Warning 050205" with some (but not all) of the data. This problem remains yet to be solved. Unfortunately, shapefile format does not fix it. Is there a hidden z-extent parameter somewhere? I tried to set the min and max Z environment setting, but it didn't change the result. I also tried making sure both TINs have a vertical datum (NAVD88). It didn't seem to make a difference.

Trisarahtops · Answer

Thanks for the ideas. I've tried TINs with non-coincident boundaries. I'm still getting the same unpredictable results. Sometimes I get line artifacts, and sometimes I get the "Warning 050205: Unable to process geometry" error, and several of the polygons areas have no multipatch geometry. I've also tried writing to shapefile instead of feature class. Again, the results are all over the place. Here's a summary of what I just tried:

* One TIN clipped (soft clip) to the extrude polygon, the other TIN left as a rectangle over the whole project area. Results: no line artifacts, but failed to create geometry in one of the polygons.
* Both TINs clipped (soft clip) to the extrude polygon. TIN boundaries and polygon boundaries are all coincident. Result: line artifacts all over the place, except for one polygon where it looked good. Interestingly, the polygon that succeeded is the one that failed to create geometry in the previous attempt.
* Both TINs unclipped, left as large rectangles. Some of their boundaries are coincident-ish, but only in a few places. Result: no line artifacts, but several of the areas failed to create multipatch geometry.
* One TIN clipped (soft clip) and one TIN left as large rectangle. Wrote to shapefile instead of feature class. Result: created line artifacts AND failed to create geometry in some polygons.
* Both TINs unclipped, projected to WGS1984 (per recommendation from ESRI support). Extrude polygon also projected to WGS 1984. Result: no line artifacts, but failed to create geometry. Tried both shapefile and feature class, the result was the same.

It seems that the issue may have something to do with TINs having coincident boundaries, but I cannot get the "Warning 050205" problem to go away. Additionally, I cannot figure out any patterns for why it fails in some places but not others, or why it draws line artifacts in some places but not others.

I have been working with ESRI support on the issue. Projecting to WGS1984 (from NAD83) seems to remove the line artifacts in some cases, and it seems that making the TIN boundaries not coincident may also help. But, the issue with failure to create geometry in some places is yet to be solved.

Khalid · Answer

Hello, sorry for the trouble you've experienced with this tool! It seems like you have tried many steps. Do the two TIN surfaces have coincident boundaries? If so, try modifying the boundary of one of the TIN surfaces so that it differs slightly (e.g. make it bigger than the other). Coincident boundary issue isn't just a problem that applies when the polygon matches the TIN(s), but also when the TIN surfaces match each other. One way to modify the boundary:

* Run "TIN Domain" on one of the surfaces and output the polygon option.
* Buffer this domain polygon by a small distance (e.g. 6 inches/15 centimeters)
* Edit the TIN so that this buffered polygon is added as a z-less clip polygon
* Run the Extrude Between operation again with the non-coincident TINs and polygons.  The output will be clipped to the intersection of all the inputs, so you will hopefully get what you're looking for.

Also, are you able to generate the proper output when the result is a shapefile? The reason I ask is that certain properties that are carried by the spatial reference, like an unintentionally strict XYZ domain that may have been defined in the TIN or the output coordinate system environment setting, could prevent the creation of an output feature when the destination is a GDB feature.  Shapefile has no mechanism for enforcing such constraints, so you'll get all features, but a GDB feature class will reject such features.