<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">On 2/19/2016 3:44 PM, John Carlson
wrote:<br>
</div>
<blockquote
cite="mid:61E5CD7D-64A5-48F9-BEAA-1E94D9E5FD05@gmail.com"
type="cite">
<blockquote type="cite">
<pre wrap="">On Feb 19, 2016, at 5:28 PM, Leonard Daly <a class="moz-txt-link-rfc2396E" href="mailto:Leonard.Daly@realism.com"><Leonard.Daly@realism.com></a> wrote:
John,
You wrote that you do not want a line, but a cylinder.
* Is there a reason you need 3D geometry?
</pre>
</blockquote>
<pre wrap="">
If I were to be showing molecules, would you be arguing for 2D geometry? If I were trying to show the effects of an N-body problem would you be arguing 2D geometry? I don’t *need* anything in 3D. Do we even *need* computers? So obviously no one should code for it. Sorry for trying to do something cool in a mix of declarative 3D and imperative languages. I will go play with three.js and PlayCanvas. At least I can actually modify most of my shader uniforms in PlayCanvas.</pre>
</blockquote>
<br>
I did not state my question very well. I understand why the demo is
in 3D and the round shapes are spheres and not circles.<br>
<br>
I wanted to know if there are reasons for using 3D geometry for the
connecting rod between the spheres? And if so, does it need to be a
cylinder, (round cross-section) or can it be rectangular or
triangular cross-section? <br>
<br>
A little research determined that constructing a line with width in
WebGL is not an easy thing to do. :-( See
<a class="moz-txt-link-freetext" href="http://mattdesl.svbtle.com/drawing-lines-is-hard">http://mattdesl.svbtle.com/drawing-lines-is-hard</a> for a description.
It sort-of looks like it needs to be geometry.<br>
<br>
<br>
<br>
<blockquote
cite="mid:61E5CD7D-64A5-48F9-BEAA-1E94D9E5FD05@gmail.com"
type="cite">
<pre wrap="">
</pre>
<blockquote type="cite">
<pre wrap="">
* If you do need 3D geometry, I think what would generally be useful is a connecting geometry node that goes from the origin of one Transform to the origin of another.
</pre>
</blockquote>
<pre wrap="">
Yes, I believe this is what doug sanden essentially provided us. You can put any geometry you like in there, and it will stretch it between two nodes.
I am just trying to use *something* adequately supported in X3D V4. If I have to use something in the CAD profile, that would work too, as long as it works in my X3D player.</pre>
</blockquote>
<br>
Having a node that creates geometry between two other Transforms is
potentially very useful. I think it is part of the layering
component, though I never investigated that.<br>
<br>
Would it server your purpose if the structure was part of X3D V4?<br>
<br>
IOW, something like:<br>
<br>
ConnectingShape {<br>
end1: DEF-name of one node (probably Transform)<br>
end2: DEF-name of another node (probably Transform)<br>
geometry {...}<br>
appearance {...}<br>
}<br>
<br>
The system would handle the geometry creation and coordinate
transformations to keep everything aligned. It might need to be a
limited type of geometry, e.g. just a cross-section.<br>
<br>
This would (in essence) be a very simple Extrusion with just the end
points and no change in cross section. I think this would solve your
problem, but would it be too limiting?<br>
<br>
<br>
<blockquote
cite="mid:61E5CD7D-64A5-48F9-BEAA-1E94D9E5FD05@gmail.com"
type="cite">
<pre wrap="">
</pre>
<blockquote type="cite">
<pre wrap="">
* The disadvantage of a cylinder is the triangulation that is required to produce a nice surface can vary depending on the perceived curvature from your viewpoint. This may cause the need for additional resolution of the triangle mesh. That can be expensive if the cylinder is animated.
</pre>
</blockquote>
<pre wrap="">
So basically, you’re throwing out Extrusion, and now Cylinder, what next? Is it okay to use Box for this? I guess what you’re trying to do is provide a high quality standard. I’m trying to produce an easy to use and understand standard. I don’t need 60fps…that’s where I’ll take the performance hit.</pre>
</blockquote>
<br>
Nope, nope, nope. I was just commenting on using a cylinder to
emulate a circular cross-section rod.<br>
<br>
<br>
<br>
<blockquote
cite="mid:61E5CD7D-64A5-48F9-BEAA-1E94D9E5FD05@gmail.com"
type="cite">
<pre wrap="">Essentially, what I think you’re advocating is using IFS' and doing all the mesh animation and ultimately geometry in shaders. I agree this is a good approach. Let’s make it happen. How do we take current declarative technology and produce vertex and geometry shaders?</pre>
</blockquote>
<br>
Not quite, because the content creation tools are not advanced
enough to make that workflow easy and smooth. It might be a good
place to get to and design of this version should keep it in mind,
but I don't think this is a good starting point.<br>
<br>
<br>
<div class="moz-signature">-- <br>
<font class="tahoma,arial,helvetica san serif" color="#333366">
<font size="+1"><b>Leonard Daly</b></font><br>
3D Systems & Cloud Consultant<br>
X3D Co-Chair on Sabbatical<br>
LA ACM SIGGRAPH Chair<br>
President, Daly Realism - <i>Creating the Future</i>
</font></div>
</body>
</html>