• Advertisement

Archived

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

portals: automatic scene subdivision

This topic is 6557 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I''ve been implementing little tech. demos for a while in regards to portals. My scenes have been hand-coded box-like cells with 1-2 portals. My rendering scheme is very similar to the "Portals and Mirrors" paper. Anyway, I''ve been having a hard time trying to "auto-portalize" a scene since I obviously don''t want to hand-code levels anymore (I''ve been using Max and trueSpace for a while). What I am looking for is some ideas on how I can take an arbitrary set of polygons and create a set of cells and portals. I have some ideas about this and they are based on the winged-edge data structure but let me know if you have any ideas. Thanks

Share this post


Link to post
Share on other sites
Advertisement
Try out this document, http://www.netmagic.net/~achalfin/Graphics/portal.htm .

Share this post


Link to post
Share on other sites
quote:
Original post by mhkrause

Try out this document, http://www.netmagic.net/~achalfin/Graphics/portal.htm .


Thanks I''ve seen that page before. The restriction of convex cells isn''t neccessary for many cell/portal environments and really limits your worlds (well, otherwise you''d have a ton of cells and portals to split your concave cells in convex ones).

The author needs convex cells for portal determination. That''s a good way to find portals but placing that restriction on your cells just to find portals seems like a kludge to me.

You don''t need convex cells since you can spare some overdraw in hardware systems. For software rendering, that paper is excellent because overdraw kills performance. Z-buffering is free on 99% of h/w so you can use non-exact portal clipping and let the z-buffer sort out the rest. Portals for hardware rendering allow you to not have to send a boatload of geometry to the card.

BTW, the "portals and mirrors" paper is at http://www.cs.virginia.edu/~luebke/publications/portals.html ... It demonstrates a method that uses 2D axial-aligned bounding boxes for portals. It''s a simple method and well-suited to hardware rendering.

Anyway, I guess what I am looking for is a arbitrary polyhedron-hole-finder :-)

Thanks for the pointer though.

Share this post


Link to post
Share on other sites

  • Advertisement