Hi Daniel,
This is due to numerical inaccurancy when the ray is tangent
to the triangle. A suggest fix is modified the utilities
PickResult (source is included in the distribution)
static boolean intersectRay (Point3d coordinates[],
PickRay ray, PickIntersection pi)
and change
if ((uCoor[i] > 0) && (uCoor[j] > 0)) {
etc.
to
if ((uCoor[i] > -delta) && (uCoor[j] > -delta)) {
where delta = 1e-12
Bug 4712222 is submitted for this.
Thanks.
- Kelvin
----------------------
Java 3D Team
Sun Microsystems Inc.
>Date: Wed, 03 Jul 2002 12:37:37 +0200
>From: "Römer, Daniel" <[EMAIL PROTECTED]>
>Subject: [JAVA3D] Bug? Picking not 100% accurate
>To: [EMAIL PROTECTED]
>MIME-version: 1.0
>Delivered-to: [EMAIL PROTECTED]
>
>Hello!
>
>I found out the Picking classes are very unreliably. For that i created a
>little test case which is attached as source. I tried picking all vertices
>of a primitive (Sphere with 1 Shape3D, 1 GeometryArray). There are 3
>possible results. First is a true hit where the aiming point equals the
>geometry intersection point (red color). It is also possible that it is not
>equal because of shadowing effects, which means aiming point at the backside
>and intersection point at the front side (blue color). The last possibility
>is that there is no hit at all at the primitive which normally could not
>happen (yellow color). To understand the meaning of the colors I also
>attached some screenshots of my scene.
>
>So the procedure for the testcase is to get a vertex point, do picking and
>then decide for one of the three possible results. After that a little
>sphere with the right color is attached at the picked vertex coordinate. The
>green sphere represents the focal point of the picking. After a full cycle
>you can see that the backside was correctly picked (blue color) because of
>the shadowing. But at the front side there are many vertex points which have
>a yellow color which meas that there is not hit! This not correct because
>picking at a vertex coordinate must be always a hit.
>
>Now I don´t know If there is a reason for this and I am doing something
>wrong or if it is a bug. Please someone would be so kind to take a look at
>this?
>
>Many thanks,
>Daniel
>
==========================================================================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".
