I mean a culling of faces , not just whole object when its out of sight.I heard 3dgamestudio has this feature - every face thats not in sight or is behind a wall is invisible.I d-loaded some 3dgs demos and they are pretty fast.If that was in blender , blender would become a beter engine than 3dgs(atleast I think so , because the modeler in 3dgs isnt that good and 3dgs is not exactly “for everyone” like it was said in their site - not like blender,blender is easyer , but that doesnt mean that blender can’t do stuff that 3dgs does and the modeling in blender is perfect,its really comfortable and stuff).Well,i was just wondering if this kind of culling would ever be in blender.What do u think about it?
I noticed that when I pass out of camera range in certain areas, stuff ‘disappears’ and when I turn the camera back, it reappears. I think it already does this. In the scene I am working in now, one of the characters vanishes if I get too close and turn at a certain angle.
i might be mistaken, but it seems like there was a script for that in python and plugins somewhere.
Yes the culling of whole objects is there, but the culling of faces isnt.For instance,you have a big level,its a one object - the problem is the level would be culled only when you totaly dont see it.But what if you want to cull only the part of the level that is not in your cameras view.Since its a joined object it wont be doing the culling untill you dont see the level completly.Thats why we need not just a object culling but a face culling too , that would give extreeme speedup.
That’s why we need more programmers
if it does, you’re doing something wrong
more complicated culling would put more of a load on the cpu, which blender is using too much as-is
also, all culling techniques that are more complicated place restrictions that blender does not. They require you use a particular editor technique for generating your levels [bsp] or significant compile times [other spatial partioning] for minimal benefit [again, the graphics card is not blender’s bottleneck]
if you have a scene which is graphics card limited, you have waay too many polys or waay too many alpha faces [eating fill rate]. This would be a problem in any engine regardless of culling technique.
z3r0 d you seriously sound like a game programmer =D
always helping out. thanks for the explination now i get it 
I guess it wont be in blender.Thanx for your replies.
Sometimes depending on hardware - it’s not worth culling backfacing polys.
It’s quicker for the hardware to draw both sides than to figure out which side to cull.
[quote]that would give extreeme speedup.
if it does, you’re doing something wrong[/quote]
If your culling gives a speed improvement you did something wrong? 
Hehe, thats a first.
All professional quality 3d engines at least cull polygons facing against the camera. Check some out.
Ketsji is a very flexible engine so to keep it that way only two methods of polygon culling need to be added:
- Backface culling. (One sided polygons facing against the camera.)
- Frustum culling. (Polygons outside of the camera’s view frustum.)
This would give a good speed improvement, no need to send polygons you can’t even see down the rendering pipeling.
if you have a scene which is graphics card limited, you have waay too many polys or waay too many alpha faces [eating fill rate]. This would be a problem in any engine regardless of culling technique.
It will still run faster in an engine with culling than not.
It won’t make your engine be able to handle a billion polygons but, no one is expecting that. Just a good speed improvement.