Archived

This topic is now archived and is closed to further replies.

toe99001

Portals and PVS

Recommended Posts

toe99001    122
I have just implemented a PVS compiler based on the article “Visibility Preprocessing For interactive Walkthroughs” written by teller and sequin (divide the geometry into cells and portals using a BSP-tree and determine what is visible). Ok it is nice to see whole chunks of geometry disappear without any geometrical calculations but I am not satisfied I want more. My question is if it seems to be a good idea to use a portal culler that performs on the cells/portals that are said to be visible due to the lookup in the PVS-list or should I just stick with a traditional frustum culler. I ask because it feels like it will be a pain in the ASS to combine the PVS system with a portal culler and I do not want to start with something that is not worth the effort. Tobbe

Share this post


Link to post
Share on other sites
Guest Anonymous Poster   
Guest Anonymous Poster
Well, the question is, do you save as much time by not rendering the geometry as you spend by trying to cull it? How detailed is your geometry? Do you expect this cull step to weed out a large number of tris that PVS didn''t already catch?

If you think you _are_ going to get a lot from culling, time the algo and see. You can always cut the stuff out if it doesn''t work, and you''ll have the beginnings of a realtime portal engine to use later on down the line.

Share this post


Link to post
Share on other sites