The simplest way of implementing ellipsoids just might be with isosurfaces. All one needs is the surface for a simple sphere, but because we have vectors that define x, y, and z, independently scalable, those would handle all the distortion needed. If it were considered like an orbital, then perhaps the two solutions will converge -- both require centering, scaling, and (even in the case of orbitals, depending upon one's aesthetics) possible stretching. The remaining piece would be orienting the ellipsoid properly. We could even provide the ortep-like quarter-ellipse cutout if desired.
Q: How are these anisotropic parameters generally described? Does each atom just require three vectors that describe its ellipse? I might play with this a bit. Bob Miguel wrote: >>Greetings, Jmol team! >> >> I've become interested (for various reasons) in displaying >>non-spherical atoms in Jmol. One possible application would >>be to show thermal ellipsoids (i.e. to generate ORTEP-style plots >>from x-ray data). The application I have in mind for ellipsoids >>is to display atoms from a Gay-Berne simulation (the Gay-Berne potential >>is an ellipsoidal generalization of Lennard-Jones that has become >>common in the liquid crystal community. >> >> > >A few years ago someone asked about displaying ORTEP ellipsoids. > > > >>Right now, we're using atom-associated vectors for visualization >>within Jmol, and have modified the PovraySaver to construct >>ellipsoids from the atom positions and the vectors. >> >>In your estimation, how difficult would it be to modify the classes >>in g3d to render an ellipsoid within Jmol itself? >> >> > >Unfortunately, I think that it would be rather difficult. > >The beauty of a sphere is that it looks the same from every direction. >That means that shape & shading information can be precalculated and >cached, allowing relatively high performance rendering. > >An ellipsoid changes shape depending upon the viewing angle. Therefore, >different techniques would need to be used. > > > >>In Povray, an >>ellipsoid is simply a stretched (or squashed) sphere that is then >>rotated and translated into the correct position. I imagine this is >>somewhat more difficult in the g3d rendering scheme. >> >> > >Correct. > > > >>Any ideas if this is a feasible? >> >> > >It could be done ... but I don't think it would be easy. > >Some questions: > >It will be much more expensive than rendering a sphere. But, that may not >be a problem if we are rendering relatively few ellipsoids ... > >Q: Do people want to render proteins with thousands of ellipsoids? Or is >this generally applied to smaller molecules (or small portions of >proteins)? > >Make a wild guess at completing the following sentences: > >Most molecules would have ---- ellipsoids. > >The largest reasonable molecule would have ---- ellipsoids. > >Q: Are all the ellipsoids the same shape, just scaled to different sizes? >Or are some of them more elongated than others? > >Q: Within a given molecule, are there usually multiple ellipsoids that are >the same size & shape? > > > >Miguel > > > >_______________________________________________ >Jmol-developers mailing list >[email protected] >https://lists.sourceforge.net/lists/listinfo/jmol-developers > > _______________________________________________ Jmol-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/jmol-developers
