Source size is a big limiter. Technically, a source that is 0.1µ by 0.1µ can theoretically produce a projection far away the same size. Many factors make this ideal impossible (no lens can ever be perfect, they are made of atoms!)
One of the problems with all these greenies, is that they design it so the beam is focused almost to it's minimum size when it hits the collimation lense. This TREMENDOUSLY limits the ability of the collimation lens to do a good job. To explain, if you were to look VERY closely at a lens at the place where the laser hits it, you will begin to see the imperfections on the surface of the lense, depending on how close you are looking. The smaller the area, the more the beam is effected by imperfections in the lens. In a sense, the effect the surface error has on a beam increases exponentially as the diameter of the beam gets smaller. While it is nice that they are attempting to build the laser to start thin and stay thin, there is a bit of a tradeoff.
So, to improve collimation, you would likely need to get new lenses. One that lets the beam be just a little bit wider when it hits the collimation lens, and then the focal length of the new collimation lens must be much longer, and if you have the beam too wide when it hits the coll., you will end up having to buy a custom lens (ouch$$$$). The only major drawback to this is that the beam will come out of your laser a little wider, but you can make a spot far away that should be significantly smaller than before.
I'ver never tried this but I have studied optics just a little.