And here I go again :
f you have - for example 2 trees - `tree A` is 3x bigger than `tree B`, if you want to make them equal dense you have to raise the density - this leads to that the B trees will be scattered less than A or vice versa, the solution will be to raise the density until the smaller of trees will be at desired density, but then the density of A trees would be too much - the solution is the collision detection, but collsion detection is very slow when using on large landscapes, so is there another way to make equal density of trees with variable sizes ?
EDIT:
I propose another distribution mode, based on bounding box of geometry.
So it will work like this : for B tree use distribution map with these params : pixels=100 / units=1000m, and for the A tree would be set pixels=33 ( because the A tree bounding box is 3x bigger than B tree ) and units=1000m