Hi Joachim,

Here are some comments and thoughts....

I've run Java3D cases with Tex_Coord_3 and Tex_Coord_4 in
conjunction with a Texture2D, and have been able to get
picture-perfect (:)) results with Java3D - Doug and Kelvin
have seen some results. HOWEVER, there is one important thing
that I cannot verify for you with my testcases. And that is,
in my cases I needed q to be equal to *UNITY* uniformly
everywhere. So, division or multiplication errors, if any
in Java3D, by q qould not have been detected in my cases.

In the process I have also verified that the q will be
taken to be unity by default if you don't mess around
with it. Therefore, I needed to use just a Tex-Coord_3
finally for my cases. I had actually enquired about this
several months ago but received no replies :(. So had to
work out things on my own by trial-and-error.

Some very quick thoughts....

I think if your concern (from your previous e-mails) is
whether or not s and t are divided by q, you could give the
following a shot (note that I haven't really tried it myself
though, and may not be aware of all ramifications)....

1. Take any Texture2D problem for which you know Java3D
   produces correct results.
2. Set Tex_coord_4 bit in your GeometryArray
3. For textureCoordinates do:
   tc1 = tc1_original * w
   tc2 = tc2_original * w
   tc3 = 0
   tc4 = w

   where w is any sensibly arbitrary number (say Math.PI)

4. set your texture as Texture2D as you did in the original
   case.
5. Run this modified case.


If tc1 and tc2 are indeed divided by tc4 before the Texture2D
is indexed, then I would expect the original and modified
cases to produce identical results. This would at least verify
non-auto texture coordinate cases.

Don't blame me if you end up wasting your time...This is my
2 cents worth based on a 1 sec. flash.

Regards

Raj ~~~~~~

===========================================================================
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".

Reply via email to