i have option of both use of
Automatic Texture Generation option
of opengl or classsically passing it
to Texture Coordinate Array ?
Does someone know which one may work faster?
this depends on the driver you’re using so a test would be needed for a certain answer.My bet would be with the second though.
i think it also depends on how you render your geometry. if u use display lists then passing texture coordinates epxlicitly is faster. for vertex arrays i can’t tell, but i think zen is right.
If you’re geometry throughput bound (say, bus bandwidth limited) then texgen will be faster on cards with hardware transform and lighting. Especially object linear texgen, because the card can often bypass eye space and go straight to projected space.
Anyway, if you have a specific situation you’re concerned about, then benchmark it both ways.