info-inventor
[Top] [All Lists]

Re: CSG, boolean operations nodes...

To: Jarek Rzepecki <jarekr@xxxxxxxx>, Inventor List <info-inventor@xxxxxxxxxxx>
Subject: Re: CSG, boolean operations nodes...
From: Nigel Stewart <nigels@xxxxxxxxxx>
Date: Tue, 21 Aug 2001 15:25:35 +1000
References: <Pine.GSO.4.20.0108200948370.9159-100000@xxxxxxxxxxxxxx>
Sender: owner-inventor@xxxxxxxxxxx
Jarek,

There was recently a similar enquiry on the Mesa mailing list.
Here is my reply, in full.

PS: I havn't seriously looked at supporting image-space CSG
using OSS Inventor, but seems like something that could be
done, and I've got source for a few different algorithms
including Goldfeather (also used by Weigand).

Nigel

> Hi!
>   I need a boolean operations on objects in OpenInventor. I know that OI
> himself doesn't support them... 

-------------------------

Antonio,

You may be interested in my PhD research:
http://www.nigels.com/research/

There is some source code there, although quite
an old snapshot, since I'm due to put some papers
out...   

>         - What pattern of classification it is used to group
>           the surfaces on a product?

The Goldfeather algorithm uses a stencil test to isolate 
each surface of a primitive, then uses a stencil test to
clip that surface in the z-buffer before merging the z-buffer
result with all the other clipped surfaces.

>        - How to implement restore or save depth buffer?

Two choices - 

glReadPixels/glDrawPixels 

Tends to be slow, sucking data across the bus between graphics
hardware and main memory.

glCopyPixels

Subdivide the viewport and use half of the framebuffer as your
second z-buffer.  Still fairly slow, though.

Also, depending on your graphics hardware and driver, copying
z-buffer pixels may not be totally reliable.  On NVIDIA hardware
for example, "far" values tend to fail.  See:

http://www.nigels.com/research/scs/scs6.jpg

Last time I tried Onyx2 (a few years ago) the z-buffer copy
failed entirely:

http://www.nigels.com/research/scs/scs8.jpg

>         I was wondering if anyone can help me with those topics
>         or where can I find some source code of the algorithms described
>         on that paper.

If you find other source code, let me know, I'd be interested
in it too...
 
>         I also would like to know if any of you has
>         T.F. Wiegand e-mail.

I don't think he's at Cambridge anymore, but you should be
able track down some people that might know where he's gone...

So, what kind of application are you looking at?

Nigel
--
                     Nigel Stewart    
     Email: nigels@xxxxxxxxxx  WWW: http://www.nigels.com
         Research Student, Software Developer.
      Just two stops down the line at Zoo Station.

<Prev in Thread] Current Thread [Next in Thread>