Miguel,

Oops, my mistake:  You were on the right track: my Z scaling was
exaggerated...I was mentally trying to achieve more precise depth buffering
by using more Z bits, while forgetting that Z can't be scaled arbitrarily
when you've got geometry with fixed Z dimensions...silly me!

That fixes the sphere intersection problem and also (to my surprise) greatly
improves the surface quality -- presumably because the normals are now more
correct.

Here are two new comparison images in which I've fixed PyMOL's light source
to match that of Jmol.  Much better:

http://delsci.com/jmol/pymol040104b.jpg

http://delsci.com/jmol/jmol040104b.jpg

The specular power is still mismatched but otherwise it looks as good as can
be expected -- this can definitely work!

As far as performace goes, this is a one-shot demo, so I don't have any
sense of that yet.  We'll need to for some code-level integration for that.
However, it is clear to me that the existing JOGL-dependent PyMOL JNI
extension will need some significant work in order to "if" or "#ifdef" out
all the OpenGL call and then efficiently pass all the geometry back up to
Java in an efficient manner (say for example, in a display list).  Once
that's done, PyMOL could simply feed a stream of integers that G3D or Jmol
interprets as primitives for rendering.

What fun!

Cheers,
Warren

--
Warren L. DeLano, Ph.D.                     
Principal Scientist

. DeLano Scientific LLC  
. 400 Oyster Point Blvd., Suite 213           
. South San Francisco, CA 94080    
. Biz:(650)-872-0942  Tech:(650)-872-0834     
. Fax:(650)-872-0273  Cell:(650)-346-1154
. mailto:[EMAIL PROTECTED]      
 

> -----Original Message-----
> From: [EMAIL PROTECTED] 
> [mailto:[EMAIL PROTECTED] On 
> Behalf Of Miguel
> Sent: Tuesday, January 04, 2005 9:15 AM
> To: [email protected]
> Subject: Re: [Jmol-developers] PyMOL geometry in Jmol
> 
> 
> > Miguel,
> >
> > Okay, with Sascha making it look so easy, I couldn't resist playing 
> > around with g3d a bit tonight in order to explore just how 
> feasible it 
> > would be to use PyMOL as an optional JNI extension to generate 
> > molecular geometry inside Jmol (or g3d).
> >
> > Here's what I've been able to accomplish in a few hours, simply by 
> > having PyMOL directly write out Java source code to message Jmol 
> > (contrived yes, but nevertheless a quick way to test the concept 
> > without creating any real glue).
> >
> >>From the original geometry in PyMOL...
> >
> > http://delsci.com/jmol/pymol040104.jpg
> >
> > The following can be obtained inside of Jmol (well, actually just 
> > org.jmol.g3d.Graphics3D inside a JPanel).
> >
> > http://delsci.com/jmol/jmol040104.jpg
> 
> Very good!
> 
> > No color yet, and the lighting is different, but otherwise it is 
> > remarkably close.  However, the spheres aren't clipping each other 
> > quite right and the triangles are flat shaded.  Questions:
> >
> > 1) I've set the slab and depth values to 1000 & 4000 respectively, 
> > which covers the full Z range of the scene, but still the spheres 
> > aren't intersecting one another correctly.  The fillSphereCentered 
> > method is used to draw them one by one.  If that's okay to 
> do, then what's missing?
> 
> slab values in RasMol/Chiime/Jmol run from 100 to 0
> 
> 100'% visible, 0% visible
> 
> > 2) Is there a way to provide a normal vector for each 
> triangle vertex?
> > That's pretty much essential for HQ surface rendering.  
> Clearly Jmol 
> > can do smooth shading correctly with the ribbons, but unless I'm 
> > overlooking something obvious, that isn't possible right now for 
> > individual triangles.
> > What's the outlook for this?
> 
> There is currently no mechanism to handle this ... so don't 
> waste any time looking for it.
> 
> I plan to implement some type of phong shading across the 
> faces of triangles some time with the next two weeks.
> 
> I have been thinking about it for a few weeks and have a 
> high-performance design planned out.
> 
> Q: How is performance?
> 
> > By the way, when you said that g3d was clean and standalone...you 
> > weren't kidding!  It took me just 15 minutes to get that first 3D 
> > sphere drawn in a simple Java app, sans OpenGL.  Fantastic work!
> 
> Excellent!
> 
> > I've also discovered a convenient little coincidence:  PyMOL's 
> > built-in ray tracer uses input geometry in almost exactly the same 
> > form as g3d.  So while I am currently taking PyMOL ray tracing 
> > geometries into Jmol, the reverse should also be quite 
> easy: with just 
> > a few changes, PyMOL could be used as a photorealistic rendering 
> > engine for Jmol -- though of course only in cases where Jmol has 
> > access to native extensions, such as when running under WebStart.
> 
> Very good.
> 
> I must say that I am impressed and quite excited that you 
> were able to slap this quick-and-dirt-demo together so easily.
> 
> > Cheers,
> > Warren
> >
> > --
> > Warren L. DeLano, Ph.D.
> > Principal Scientist
> 
> This bodes well for potential collabaration between PyMOL and Jmol.
> 
> Miguel
> 
> 
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues 
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Jmol-developers mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/jmol-developers
> 




-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
_______________________________________________
Jmol-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jmol-developers

Reply via email to