Strange results near NULL areas

[lrntct]

First of all, thanks for this module, it is just what I needed.
I tried to use it to patch an ALOS DSM with a LiDAR DTM over Acapulco, México.
It seems to work well at the interface of the two maps, inland.
However, on the shore, where the ocean is NULL, the results seems a bit strange:

Here is the ALOS DSM on a small part of the shore:

The LiDAR:

The results of
r.patch.smooth -s input_a=dtm_lidar@Acapulco input_b=AW3D30@Acapulco output=AW3D30_lidar transition_angle=1

It looks like the information of ALOS is passed to the new map, instead of using the LiDAR, and the transition is coarse outside the cape.

[petrasovaa]

I think I know what's the problem. I was always working with DEMs where input_a was smaller and completely inside input_b DEM, so I didn't need to handle the edge case, which requires different behavior. I will look into that. Thanks for your interest!

[petrasovaa]

In 783bc67 I tried to address this issue by adding a new parameter blend_mask where you specify which edges should be excluded from the blending. This problem is actually more difficult to solve than I expected...

[smathermather]

Hi @petrasovaa -- will this be an issue with blending elevation models overlapping as so?:

We are looking to use r.patch.smooth through WebODM for DEM merging for the Split-Merge approach, and I wonder if this is appropriate.

[petrasovaa]

R.patch.smooth was designed for little bit different use case (merging UAV dataset into larger lidar data), so I don't think it would work out of the box. However, the method is fairly general, so with some tweaking for this use case, I think it's doable. Could you possibly send me sample data so that I can test the approach?

[smathermather]

I should have something for you first thing next week. I finally got it processing... .