RE: [cgmo-webcgm] Error in nurbs01 test

[Robert Orosz <roboro@AUTO-TROL.com>]

Lofton,

Attached is the PNG file that I used.  It's in my local WebCGM 2.0 test
suite directory dated 1/25/2006, so it must have been on the FTP site at one
time.  I quickly glanced through your recent emails to the group, and I
don't think you ever emailed it directly to us.  In any event, the image in
my NURBS01.png file looks perfectly circular to me.  It sounds like it might
be the same PNG file that you describe as having in your working directory.


To answer your last two questions, I do expect to see the extra bulge that
you describe.  However, as I've told you before, I'm not a NURBS expert, so
I have no idea how *much* bulge should be present.  It is difficult to
discern in the reduced-size version that you sent.  I had to print both
files at exact size, hold them up to my window, and look through them one on
top of the other to see the difference. 

If this PNG file is accurate (I'll leave that up to the NURBS experts), then
I don't think this is a very effective test.  Users viewing the test through
the test harness will likely think the test passed regardless of whether the
viewer applied the weights or not.  We need a test where the weights make a
dramatic difference on the shape of the curve.  Note that the nurbs02 test
has all of the weights equal to 1.0, so it is a NUBS in NURBS clothing.

Regards,

Rob

-----Original Message-----
From: Lofton Henderson [mailto:lofton@rockynet.com]
Sent: Thursday, March 09, 2006 5:06 PM
To: Robert Orosz; 'cgmo-webcgm@lists.oasis-open.org'
Subject: Re: [cgmo-webcgm] Error in nurbs01 test


Rob,

Checking this out...

At 02:03 PM 3/8/2006 -0700, Robert Orosz wrote:
>[...]
>
>There is an error in the nurbs01 test in the test suite.  It appears that
>the interpreter/viewer used to generate the reference PNG image ignored the
>weights in the NURBS element.

I just looked on the FTP site, and there is no NURBS01.PNG there!

So for starters, what are you looking at for the reference PNG?

In my working directory, I have a NURBS01.PNG that looks pretty 
circular.  Dave has just sent me an earlier version that looks lopsided 
circular -- there is a little extra bulge at 0, 90, 180, 270.  I have 
attached a reduced-size (400x400 instead of 2300x2300) version.

Does that look right to you?

If not, what do you *think* the PNG should look like?

Regards,
-Lofton.

>To illustrate my point, I've generated an
>equivalent file (without the boilerplate text labels) containing a NUBS
>element which is unweighted by definition.  The knot vector is identical to
>the nurbs01 test, and I've placed the control points at metrically
>equivalent positions in VDC space.  This file views identically to the
NURBS
>element in the NURBS01.cgm file as depicted in the PNG image.  Applying
>weights to the spline should significantly alter the shape of the curve.
>
>You could do an equivalent test by using your convert program to remove the
>weights in the NURBS element (i.e. set them all to 1.0) in the NURBS01.cgm
>file.  You will see similar results.
>
>Regards,
>
>Rob
>
>  <<NUBS01_rob.cgm>>

NURBS01.png