On Dec 24, 2007 11:38 PM, Bob Hanson -- Jmol <[EMAIL PROTECTED]> wrote:
> OK, from the exchanges of 12/14/07 and since, you guys are pretty
> fast. I'm impressed.
We haven't modified jmol itself at all yet. All we (= Robert
Bradshaw, mainly) have
done is figure out how to embed jmol in the sage notebook and create scripting
files so that jmol will display all of our 3d primitives so that all
3d pictures we can
already draw can be rendered using jmol. We've run into some issues, e.g.,
http://trac.sagemath.org/sage_trac/ticket/1537
involving "triangulation and shadow weirdness" whose fix might involve having
to actually change jmol.
I think we (the Sage developers) still barely understand what jmol is capable.
> Allow me to introduce myself. http://www.stolaf.edu/people/hansonr.
> Professor of Chemistry, St. Olaf College; Current Principal Developer,
> Jmol.
Nice to meet you. I'm William Stein (http://wstein.org), Associate Professor
of Mathematics, University of Washington (Seattle, WA).
> What can I tell you about Jmol? Sounds like you have found the
> documentation and experimented a bit with at least writing some pmesh
> files that Jmol can read. Also take a look at the HTML files in
> http://chemapps.stolaf.edu/jmol/docs/examples-11, particularly:
>
> http://chemapps.stolaf.edu/jmol/docs/examples-11/pmesh.htm
Oh, cool, you can do translucent surfaces! Nice. The moveto is
also frickin' cool. And that this is also scriptable from javascript
is great -- this is exactly what I've been looking for.
> http://chemapps.stolaf.edu/jmol/docs/examples-11/isosurface.htm
Wow, those isosurfaces are gorgeous. I'll have to figure out how
to define them...
> http://chemapps.stolaf.edu/jmol/docs/examples-11/new.htm
> http://chemapps.stolaf.edu/jmol/docs/examples-11/new2.htm
Wow, this is amazing, especially example 30.
> http://chemapps.stolaf.edu/jmol/docs/examples-11/new0.htm
>
> We are just about to deliver Jmol 11.4.0, so a lot of recent work has
> been going into detailed testing of functionality. I see you are
> talking about pmesh. And if you like that, fine. It works great, but
> the script interface is a bit slow because of all the string parsing.
> We could set up a direct Java interface, or even a binary file format
> if that would help.
I think it's just the first thing we thought of that would work somewhat
for drawing surfaces. Should we be using something else?
> Realize that there are two Jmol components -- the stand-alone
> application and the applet.
Right now in Sage if you draw a 3d image and render it from the command
line it pops up the stande-alone application. If you render from the notebook
it uses the applet.
> The Jmol application has some interesting
> features, like command-line creation of JPG and PovRAY output, and
> generic "exporter" capability that is easily expanded. The applet has
> tremendous interactive capability, but not so much with pmesh objects
> as of now. Still, that could be possible, and the 3D annotation
> capabilities are really outstanding. (You can create 3D annotations
> ("echo"s) that can serve as hypertext links, for example, and you can
> create points in space that visible or not can be clicked on and will
> allow callbacks to the web page. The whole thing is connected with
> LiveConnect and scriptable via JavaScript.
Can you create callbacks like that using the java applet? This would be
great for Sage, I think.
> A very interesting capability of Jmol that has arisen in the past year
> is the capability to save and restore the current state of the applet
> - everything. So if you set up a "scene" in the application you can
> just write the state file to disk and open it with the applet, and you
> will be right where you left off. This has been used in a variety of
> contexts to good effect. I've been interested in exploring it in terms
> of collaborative visualization. (I do something, save a state, send it
> to you, you load it and see the exact same thing.) You might be
> interested in that.
Very cool. Sage also uses tachyon3d for ray-traced rendering and
it would be nice to have a way to find out what angle the user selected
from the applet, so we can render the same scene using tachyon.
Also, it would be good that if a user closes the sage notebook then
re-opens it, that all there 3d images are restored to exactly the same
state.
> Recent work you might also be interested in is in the area of
> synchonization -- multiple applets can be synchronized so that
> commands or mouse actions on one drive the others. This is working
> perfectly now, and it is just one step away from being able to script
> remotely so that, for instance, multiple people viewing the same page
> could see the same real-time 3D manipulations of a mathematical
> object.
That's very cool. This could be relevant to when multiple people
view the same sage worksheet at once (currently we have no
interesting synchronization in this case, but we plan on it).
> I'd be very interested in your feedback relating to the JVXL format
> for isosurface compression. We think we have a winner there. It really
> makes isosurfaces deliverable over the web for the first time. This
> has started to see use for the delivery of protein surfaces. See
> http://chemapps.stolaf.edu/jmol/docs/misc/JVXL_format.pdf
The correct link is:
http://chemapps.stolaf.edu/jmol/docs/misc/JVXL-format.pdf
I didn't know about it until just now. In fact, I'm not even
completely sure what
an isosurface is or how we can use them for sage yet. I'll read the above
article soon.
> I put together a little page for you guys to experiment with. It's at
> http://chemapps.stolaf.edu/jmol/docs/examples-11/math.htm
> Try a few things there; see what you think. The example uses 100,000
> quadrilaterals, and it's pretty fast in terms of rendering.
Awesome!
> Just to set the record straight -- the graphics engine was written by
> Michael Howard, a software engineer, not a chemist. Everyone who has
> looked at this cannot believe this is Java, and it's because Michael
> did such an outstanding job with the rasterization methods. No Java3D
> here; no OpenGL. Just bare-bones write-the-pixels coding.
> Unbelievable.
Yep.
> I came on board a couple of years ago as a chemist with lots of
> software experience, but by then the Java primitives for graphics
> rendering were all in place. Credit for that goes exclusively to
> Michael (aka Miguel). I'm now the principal developer. My focus has
> been on expanding Jmol's capability within the molecular visualization
> world. But I'm thrilled that you folks are interested in it as well.
> Very cool.
Excellent -- I'm glad you see Jmol as something that should grow into
other areas.
>
> Q: What do you have now in the area of Java applets?
Essentially nothing. The Sage notebook is an Ajax app -- lots of javascript
and server-side code (in Python, using Twisted web2). We basically
don't use java applets at all in Sage, though it's always been my plan
to do so.
> Q: Is pmesh the right delivery mechanism? Or is there something else
> that can be delivered to Jmol, such as primitive function that let
> Jmol query for the values of x(u,v), y(u,v), z(u,v) and thus build its
> own pmesh internally? Something like:
>
> Point3f xyzFromUv(float u, float v);
>
> Could you build that into an accompanying applet, so the two could
> talk to each other? That could be fun.
The Sage notebook is an ajax app, and evaluating a function like this to
have the mesh built in jmol would potentially be very slow:
1. Jmol requests evaluation of a value
2. Javascript notebook code queries the Sage server for the value
of the function
3. The sage notebook server queries a running Sage compute process for
the value of a function.
4. The Sage compute process calls some library function, maybe defined
in PARI or Maxima.
5. The sage notebook server gets the value of the function
6. The javascript notebook gets the value, then returns that two
the Jmol applet.
This would not be good for building a mesh. If the Sage notebook were a java
app and Sage were written in Java, it would be a different story, but that's not
what Sage is.
> Q: Are you most interested in command-line/file-based interaction with
> the Jmol application? Or are you more interested in the possibilities
> involving the applet?
I am personally much more interested in the applet. For command line
users there are other potentially very good options for 3d visualization,
e.g., vtk with mavayi2, whose development http://www.enthought.com/
helps support. But that vtk-based stuff is of no use for the Sage notebook.
> Q: What are you using right now for the cool graphics I see at your
> site?
For 3d graphics, mostly Tachyon3d:
http://jedi.ks.uiuc.edu/%7Ejohns/raytracer/
It's small, lightweight, very fast, but completely non-interactive.
--William
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
