sorta techish (and maybe ignorant) question on normalmaps
 

I watched the normalmap tutorial philip's made, but besides it being way cool, and a good tutorial, there's something I'd like explained: why? (i'm thinking games)
I'm not sure I fully understand why normalmaps are smarter than hi-poly models. Sure you can some excellent details in there, but isn't rendering a normalmapped model as taxing on your system as rendering a hi-poly model? Again, thinking games. I'm thinking that a hi-def map would be as tough to calculate as a super fine mesh; getting rgb-values for every pixel must be pretty hard for the hardware...

I guessing I'm dead wrong since normalmaps are hot :)

Well normalmaps aren't very hi-def maps. They are basically used as texture maps to give that detail in there. In some ways hi-poly models are better but you would need the latest video cards and a lot of extra stuff upgraded on your computer so it would pull the game. With normalmapping the character is very low-poly yet it looks like a hi-poly with the normalmap applied to it.

Side Note: The way i see it is that rendering a normalmapped model doesn't even come close to taxing your system as a hi-poly model would. That's my opinion on this subject.

Indeed. The whole point of producing a normalmap is to fool you into seeing stuff that isn't really there.

True, a normal map will be more taxing on the system than generating a bog standard texture map, in the same way that bump mapping is. However, it's an awful lot easier for the machine to calculate a bunch of surface normals than to calculate a 3D transform for several thousand extra polygons, send them to the video card and have it render them.

that makes sense.
Thnx guys, I am that much wiser. I'd forgotten the 3d transformations are the tricky bit. I kept thinking "well, can't the computer calc the normals of a big bunch of polys, nearly as fast as it can parse some rgb-values into normals?" and well, maybe it can, but that's really not the issue...

thnx again