Hello Simon,

Simon Haegler wrote:
> osgHello
> 
> how do the following options affect the render performance? especially the 
> culling threshold is of interest, as i'm making heavy use of occlusion 
> culling.
> 
>   _renderAction->setSortTrans(0);
>   _renderAction->setZWriteTrans(0);
>   _renderAction->setLocalLights(0);
>   _renderAction->setCorrectTwoSidedLighting(1);
>   _renderAction->setOcclusionCullingThreshold(0); // hierarchical multi frame
> 
> are there any other options i should consider for the optimization of 
> rendering performance?

I can not really comment on this, maybe Andreas, who implemented the OC 
support for 1, can share his experiences ?

> my scene (citymodel) contains about 100k nodes, 16mio (transformed) vertices 
> and makes heavy use of instancing (reuse of geocores) and material groups. 

ok, that should keep the number of display lists/vbos and state changes 
low, the number of nodes could be a problem though. Is there a chance 
that you can collapse the scene more, without loosing too many 
possibilities for occlusion or frustum culling ?

> with the camera on eyelevel and activated hierarchical multiframe occlusion 
> culling the framerate is about 24fps on a nvidia 8800 ultra. render 
> resolution is 800x600.

what happens at higher resolutions, i.e. are you reaching/at the fill 
limit?

> do you have general optimisation ideas or is this the maximal framerate i can 
> expect? when does the use of VBOs make sense? displaylist caching?

For a (mostly) static scene like a city and on your hardware that has 
plenty of memory on the graphics card, VBOs sound like a good idea.

        Hope it helps,
                Carsten

PS: It would be great if you could give some feedback on this list, what 
effect if any the optimizations you might try give you. Thanks.


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users

Reply via email to