It is well known that a lot of people play without shadows, because they take too many resources. Could DBG perhaps comment on what it is that makes it so? Is it not processed at the gfx card? And is there a wish to fix it?
My guess is that they just choose an inefficient algorithm for shadows. To me it looks like those old "stencil-buffer" shadows from the GLquake-era, rather than the more modern "shadowmap"-approach. I play with them disabled, so im not sure, but iirc it looks like a simple top-down projection on the floor rather than something that takes lightsource-positions into account. A cheap alternative for low-end machines could be to just draw a slightly darker semi-transparent "blob" on the ground below every 3d-model(they can likely use the target-ring code for this). This would still give a visual cue to the player, it can sometimes be hard to tell(is he levitating and shrunk or just far away?). Generally speaking: fast real-time correct shadowing on large-scale raids will probably not be feasible, unless DBG actually put serious effort into it...With that many models onscreen they should probably look into deferred screen-space techniques.
