You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If there is a DEM that is fully valid everywhere in its domain except for a little hole in the center where data is missing, and one attempts there to blend with another DEM using --priority-blending-length, the blending should happen only there. Yet, apparently, the DEM edge (pixel row row and col are either 0 or image size) is also treated as a place where there is no data because it can't see further pixels and blending is attempted there too. I think this comes from the fact that the weights we use naively decay to zero each time one does not see in all directions from a given pixel, and the true image boundary (as opposed to the image continuing but with no data in those pixels) needs to get special handling.
The testcase is on lunokhod1 in /home/oalexan1/projects/dem_mosaic_blending.
The text was updated successfully, but these errors were encountered:
If there is a DEM that is fully valid everywhere in its domain except for a little hole in the center where data is missing, and one attempts there to blend with another DEM using --priority-blending-length, the blending should happen only there. Yet, apparently, the DEM edge (pixel row row and col are either 0 or image size) is also treated as a place where there is no data because it can't see further pixels and blending is attempted there too. I think this comes from the fact that the weights we use naively decay to zero each time one does not see in all directions from a given pixel, and the true image boundary (as opposed to the image continuing but with no data in those pixels) needs to get special handling.
The testcase is on lunokhod1 in /home/oalexan1/projects/dem_mosaic_blending.
The text was updated successfully, but these errors were encountered: