Drone topo cleanup workflows

[jmaeding]

I discovered CC last week, and wow.
I have a fair amount of experience now with Agisoft Metashape, Trimble Business Center, and Carlson P3D.
I've drone flown about 30 projects and my co-workers another 50, so we have lots of data.
The pt clouds are typically 30 to 120 million points.
We compile photos to pt cloud using Agisoft Metashape, and get what I consider great results.

The workflows we use to isolate ground points are always:
1) classify using the built-in classify tools
2) add or remove "non-ground" points from results by hand, polygon windows typically.

The problem is those tools do not allow selection by color (that I know of). TBC selects by intensity, not RGB though.
I have found with CC it is super useful to try to remove vegetation using color.
Then run the SOR filter to clean a bit, then run CSF to get ground, using 0.5 resolution.
Then I can add back in anything I want that CSF rejected, using color filter again.

I looked at CANUPO, but heard it needs hours to run on an 80 million pt set of data.
So its filter by color, clean noise, and CSF that CC offers, from what I see.

Are there other tricks anyone can mention?
One I am thinking of, is points more than 0.5 foot above or below the tin surface I could build with the CSF results.
So the key there is scalar data from difference between a tin and the points in question.
I love how I could then filter using the slider, then split from those values, that is genius.
I actually was going to write some of this myself but CC already has it most likely.
thanks in advance.

[daniel]

If it's just a distance between the cloud and the TIN, then maybe the simple "Cloud to Mesh" distance computation would do? (it's not exactly the vertical distance though...).