Hi Matthew, How about the case where you want float32 instead of float64 to save memory? You would need to go through the dataobj and handle the scaling yourself (including all the special cases)? -Brendan