Original Post
Hi , i got a problem , i use a sort of occlusion culling , first i project the points building up the triangles using a cache for avoiding redundat computing, then i perspective project these buffered points and pass the projected triangles to a software z buffer asyncronous ( spelling ?? ) with the opengl rendering code, i was wondering if all this stage could be well replaced using vertex arrays and then using the feedback buffer to obtain the device coordinates to pass to the software rasterizer, is it a good way to go ? would i loose the asyncronicity beetween the software mapping and the opengl rendering code ? i would like to use this second solution for 2 reasons , first i would use hardware accelleration, second , the rpojected coordinates sometinems go far out the projecting plane giving awfull articatf ( division by zero like ), anyone already tried this before ?
thanks in advance