Original Post
I try to make my question better.
In Quake 3 (I heard) they used up to 10 passes for one single object in the scene. I noticed also (looking at q3ase) that they have support for blending modes like add, modulate, premodulate and so on.
In D3D I couldn''t create an effect like this because in looks like there is no way to get d3d to use the texels on the screen as the second blending argument for the first stage.
Is this only a functionality that opengl has? And do I have then to blend my textures in software to get the desired effects? This would slow down the whole shit a lot (uhhhhh).