Over the years, I have experienced with many different terrain systems in my games. I started with chalking rivers and roads on a table surface, gradually adding styrofoam hills. Later on, my brother and I painted our wargames table limegreen (we had some leftover paint from a home-decoration project), and for bigger battles we used the ping-pong table. From that period I also have some 30cmx30cm home-made terrain tiles lying around ... we never got more then 10 tiles ;-)
Then a friend and I bought a Geohex master set. We considered this the ultmate system, but in the end it was considered to be too fragile for easy and fast setup and storage. Then I bought a big green flocked felt mat (back to basics). This was now the perfect system due to ease of setup, and transportable! So I acquired new styrofoam hills that go with the mat. These days, I'm using a Kallistra terrain set (hex-based), and I now consider this the ultimate system, especially because of the many terrain attributes available. All these previous terrain systems are still stored in my house, of course. And I didn't even mention some of the very specialized terrain boards for convention games and the felt mats I have for starship combat, air combat and naval combat. Oh yes, I also have a hex-gridded map made out of brown packing paper for desert games.
Sigh. How many terrain systems do you need for 'generic' games? Right now, I feel that the Kallistra set is the most flexible, but maybe this will change again in a few years? Should I sell or throw out the previous ones? Existential questions ... !
But let me go on a bit and tell you why I think the Kallistra system is the best. It is high-quality material, light and has a good visual appeal. But the main advantage is that it speeds up gameplay because it is inherently hex-gridded; i.e. it superimposes a hex-grid (about 10cm side-to-side) on your playing surface. And this is the best advantage of them all.
I have become a big adept of hex-based games over the past 2 years. Hexes allow for easier movement (no more measurements), range calculations etc., and so in general, significantly speed up a game. I used to think that no serious wargame should use hexes, because we need those 1" accurate movements and ranges. But, movement and ranges in most game systems comes in multiples of 3" or 4" anyway (I checked a few systems to make sure). So, if you would divide your distances by 3" or 4", you get the range expressed in hexes, and nothing is lost. So, for movement and range calculations, hexes are a big win.
What about visibility? Visibility is always a problem in miniature wargames, and is often eye-balled. Traditional hex-based board wargames assume that visibility is 'calculated' from hex-centre to hex-centre, and one is inclined to do the same for a hex-based miniature wargame. That is fine, but not strictly necessary. Visibility can still be determined on a figure by figure basis, even though you compute ranges expressed in hexes. And thus, skirmish-based games such as Western shootouts bacome perfectly possible, with all the expected detail of hiding behind a single tree or bush. And eye-balling the visibility the traditional way is actually faster than drawing a line from hex-centre to hex-centre. And for games involving precise orientation of figures or regiments, this is also perfectly possible within the hex, since the lines of sight are not determined using the hex-grid, only their length is.
Now, the Geohex system also come in hexes, but mich bigger ones, and this is not useful to translate into a gaming system. A Kallistra hex can contain one unit of let's say 10 25mm figures, but a Geohex hex is much bigger than that. So, although Geohex is also hex-gridded, the grid is not very useful for games. Being a computer-graphics person, I can 'visualize' all sort of finer resolution grids on top of the Geohex (e.g. all the centre-points and corner-points of a hex-grid also form a hexgrid at double resolution), but I think that most people will not be able to make those abstractions, such that gaming speed will slow down again.
My main objections against hex-gridded systems has always been twofold: visual appeal and accuracy of gameplay. The first one is solved by Kallistra; the second one is not an issue when using small enough hexes, and being able to move away from the inch-crunching mentality.