"deep into HLSL" is a subjective exaggeration -- how far is deep? and with such a simple language, how deep do you have to go before it's a bad thing? but i don't want this to be a torque vs a7 thread either so i won't argue about it from an engine sense, and will just discuss the video.

now, what's in the video? nice art, yes. but there are NO dynamic shadows. the static shadows there are smooth but very low-res. it must be one of two ways: they compiled a static shadow map for it, or they use realtime shadow-mapping, but the latter is unlikely because there isn't anything dynamic about it.

what else is in it? normal-mapping. environment mapping, i guess. dynamic lights for the gun-fire, but like any well-made game they only put as much detail in as they need -- no dynamic shadows cast. people often want every light to cast shadows, but even most games on ps3/360 make as little use of that as possible for the sake of speed and putting resources into visuals that make a bigger impact.

now, not for the sake of argument but for the sake of learning: whats wrong with create_meshes and using shaders on level geometry? i know there have been some issues with tangent, but what besides that? i haven't done any block shaders myself, but the manual makes it seem very simple as to how to combine static shadow maps with shaders. and yes, i know static shadows in a7 atm don't look great. i'm asking from a technical point of view.

julz