That would be ideal indeed. What seems to happen so far is that the TIN is “flattened” along the lowest elevation vertices in an area. This is more like a “quick and dirty” solution as the TIN triangulation remains and even falloffs are not working properly within these: