fobru

02-19-2003, 03:37 AM

Hallo,

Has anybody experience with BSP??

Can you help me with some code??

Any help is good!!

Thanks in advance

Has anybody experience with BSP??

Can you help me with some code??

Any help is good!!

Thanks in advance

View Full Version : Binary Space partitioning

fobru

02-19-2003, 03:37 AM

Hallo,

Has anybody experience with BSP??

Can you help me with some code??

Any help is good!!

Thanks in advance

Has anybody experience with BSP??

Can you help me with some code??

Any help is good!!

Thanks in advance

Eric

02-19-2003, 03:46 AM

Jan2000 already gave you interesting links in your previous topic:

http://www.opengl.org/discussion_boards/ubb/Forum3/HTML/008798.html

Have you read it all and digested it already???

Regards,

Eric

http://www.opengl.org/discussion_boards/ubb/Forum3/HTML/008798.html

Have you read it all and digested it already???

Regards,

Eric

M/\dm/\n

02-19-2003, 03:46 AM

That should be posted on super GURUs board even ogl.org doesn't have. http://www.opengl.org/discussion_boards/ubb/biggrin.gif

But anyways, I tried Enabling/Disabling depth_test on GF2GTS and got no FPS increase (although there was different story on TNT2), AFAIK BSP trees are made upon this idea, you can simply disable depth_test and render everything from back to front.

Although I may be wrong, as this is subject for me tu study & I can't find time for it.

But anyways, I tried Enabling/Disabling depth_test on GF2GTS and got no FPS increase (although there was different story on TNT2), AFAIK BSP trees are made upon this idea, you can simply disable depth_test and render everything from back to front.

Although I may be wrong, as this is subject for me tu study & I can't find time for it.

antipop

02-19-2003, 04:02 AM

Originally posted by fobru:

Hallo,

Has anybody experience with BSP??

Can you help me with some code??

Any help is good!!

Thanks in advance

I've used Binary Space Partitioning for raytracing, however, that's probaly a bit outside the scope.

But, since the view frustum is a convex hull, it should be fairly easy to use a BSP tree to determine which leaves in the tree which are potentially visible.

What do you want to do with your BSP-tree?

Hallo,

Has anybody experience with BSP??

Can you help me with some code??

Any help is good!!

Thanks in advance

I've used Binary Space Partitioning for raytracing, however, that's probaly a bit outside the scope.

But, since the view frustum is a convex hull, it should be fairly easy to use a BSP tree to determine which leaves in the tree which are potentially visible.

What do you want to do with your BSP-tree?

fobru

02-19-2003, 04:22 AM

The final step is to make "Constructive Solid Geometry" (CSG)

Like making a hole in a cube (subtracting one object from another).

In a previous question one of the answers was that BSP is a first step to get there!

So I want to start with that.

Thanks to everybody who wants to help

Like making a hole in a cube (subtracting one object from another).

In a previous question one of the answers was that BSP is a first step to get there!

So I want to start with that.

Thanks to everybody who wants to help

antipop

02-19-2003, 04:27 AM

Originally posted by fobru:

... BSP is a first step to get there!

So I want to start with that.[/B]

Aha, I see the clue. In the process of constructing a BSP-tree, some of the triangles are split. However, after the tree is finished no triangles intersect, and thus, finding CSG-intersections, differences etc. should be a matter of rejecting triangles.

I think there is a FAQ on BSP-trees somewhere, at least, it was one some years ago.

To build such a BSP-tree do the following: Given a set of triangles, find a suitable splitting plane (one of the triangles for example). Then make two new sets P and N. For each triangle: if the triangle is completely inside the positive part of the half-space defined by the splitting plane, put the triangle in P. Otherwise, if the triangle is completely inside the negative half-space, put it in N. If the triangle intersects the clipping plane, split the triangle (into three I guess) into parts which are either completely positive or completely negative.

Do this recursively on P and N until P and N only contains one triangle.

This is just from the top of my head --- I guess I have forgotten something.

[This message has been edited by antipop (edited 02-19-2003).]

... BSP is a first step to get there!

So I want to start with that.[/B]

Aha, I see the clue. In the process of constructing a BSP-tree, some of the triangles are split. However, after the tree is finished no triangles intersect, and thus, finding CSG-intersections, differences etc. should be a matter of rejecting triangles.

I think there is a FAQ on BSP-trees somewhere, at least, it was one some years ago.

To build such a BSP-tree do the following: Given a set of triangles, find a suitable splitting plane (one of the triangles for example). Then make two new sets P and N. For each triangle: if the triangle is completely inside the positive part of the half-space defined by the splitting plane, put the triangle in P. Otherwise, if the triangle is completely inside the negative half-space, put it in N. If the triangle intersects the clipping plane, split the triangle (into three I guess) into parts which are either completely positive or completely negative.

Do this recursively on P and N until P and N only contains one triangle.

This is just from the top of my head --- I guess I have forgotten something.

[This message has been edited by antipop (edited 02-19-2003).]

zeraiam

02-20-2003, 11:13 AM

you might find this interesting:

http://www.opengl.org/developers/code/bspfaq/index.html

regards

http://www.opengl.org/developers/code/bspfaq/index.html

regards

Powered by vBulletin® Version 4.2.3 Copyright © 2016 vBulletin Solutions, Inc. All rights reserved.