Got any idea how the function decompress works from his engine here it is: void world_t::decompress() { for(int i = 0; i < bsp->leaf_count; i++) { int_vec visible; dleaf_t& leaf = bsp->leaves; int v = leaf.visofs; const byte* vis = bsp->vis; for(int c = 1; c < bsp->models[0].visleafs; v++) { if(vis[v] == 0) { v++; c+=(vis[v]<<3); } else { for(byte bit = 1; bit; bit<<=1,c++) { if(vis[v] & bit) { visible.push_back(c); } } } } leaf_visibility.push_back(visible); } } I think tha it expands all leafs so that PVS can be found. Haven''t got that much of a clue though. Does each leaf have much face data which needs to be used to point in the edges array PLEASE HELP COULD YOU OUTLINE HOW BASICALLY THE BSP FORMAT WORKS PSEDUO CODE WOULD BE GREAT i have seen the quake 2 documentation on Flipcode and Bart Sekura''s half-life loader but still abit confused as i''m quite new to the world of 3D graphics and openGL Thanks ALOT Cobera Steve