The Dense Point Cloud Ground Classification tool is great, but I find myself classifying, resetting, classifying, resetting, over and over and over again. While I understand what the 3 variable do (mostly), I always seem to have to tweak them to try and get the best results for ground classification.
A great feature to have would be the ability to sub-select a portion of your point cloud, then be able to apply nearly instantaneous changes to the cell size, angle, and vertical change variables to see how the ground classification changes based on the small changes. Once the low points are calculated based on the cell size, you could have sliders for the other 2 variables to watch how the changes affect the ground classification in real time. Then you could apply those variables to the entire dataset.
I know that the movement is towards automatic classification, but that appears to be a long way away at the moment. Having specific control over the point selection would certainly be a valuable tool to have.