The 3DS screen is effectively 400x240, in 2D mode. 400 pixels are allocated to each eye in 3D mode making it 800x240. At least that's how I understand it.
That's correct. Both in 2D and 3D mode, the 3DS prepares two 2D pictures, one with slight alterations (if in 3D mode), each is 400x240. It interlaces them in a vertical fashion, creating one 800x240 picture, then each vertical line goes through its respective paralax slit and finally to either the left or the right eye. Meaning, the physical screen is 800x240, but it displays two 400x240 pictures, one for each eye.
That however means that the scenes themselves are rendered are 400x240 - otherwise the two screens couldn't be interlaced on an 800x240 screen. The question remains whether the scene/s are rendered once or twice per cycle - is the 3D effect based on rendering a scene twice (more resources needed, I doubt this method is used, there's no need to render twice just to achieve pop-in/pop-out) or via altering the original render (less resources needed since it's just post-processing). It's not uncommon to see several render passes in 3D engines, but doing them twice would be wasteful, I think.
Not that it really matters, what I meant was that the 3D scene itself is 400x240, I just explained it in a rather crude fashion.
@
Skelletonike:
