HI David,
Thanks for the feedback and suggestions....
The intent was for the SweptVolumes to support a number of rays being
cast in each direction, thats why the rays are specified as an array
of Vectors. However the VPDefaultCollision class does not support
multiple rays in the current implementation.
I like the idea of adding the intersection Normal to the RayCollision, I'll
look at adding this before the next release.
On a separate note, for anyone trying to implement SceneGraphIO interface
to use the IO API for their own classes, there is a bug in the current release
which will cause problems. The bug is that the method createNode() in
com.sun.j3d.demos.utils.scenegraph.io.state.javax.media.j3d.SceneGraphObject is
abstract, it should not be abstract and instead have a no-op implementation.
This bug is fixed in the next release.
Rgds
Paul
----------------------------------------------------------
Paul Byrne Email : [EMAIL PROTECTED]
Sun Microsystems Phone : (650) 786 9926
Visualization Software Group Fax : (650) 786 5852
----------------------------------------------------------
>From: David <[EMAIL PROTECTED]>
>Subject: [JAVA3D] J3dfly: Attn Paul Byrne
>To: [EMAIL PROTECTED]
>
>I just spent a few hours reading the code in the j3dfly example program. I
>found it very enlightening. In particular the collision detection technique
>is quite nice and flexible.
>
>Question1: The SimpleSweptVolume uses single rays from the eye location. If
>I am correct in my understanding of this, the limitations are that since the
>rays are from the eye, its possible the edge of the view will intersect
>geometry, even though the center of the view never will. To apply this to
>either an avatar or to make the view collision "better" I would need to two
>implementations:
>
>For avatar: Using an avatar bounding box I would use 4 rays in each
>direction, set at the corners. This would be a "conservative" check, as the
>bounding box would encompass space where there was no real geometry.
>
>For View (first person): I would need 4 rays front and back and the rest
>would stay the same. The four front and back rays would take into account
>the field of view to calculate the proper span.
>
>I would also need to improve VPDefaultCollision to loop through all the rays
>assigned to each direction, rather than just using [0]. Is that right?
>
>Question2: If I wanted to get the normal infomation from the collision to
>calculate a better "bounce" I would need to add a normal vector to
>RayCollision, and add the PickResult.getPointNormal() info to the
>RayCollision.
>
>Thanks for the excellent demo!
>
>Dave Yazel
>
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
