Re: [opendx-users] elliptical tubes (Followup)

[Chris Pelkie]

Someone asked if I could send them a working 'elliptical tubes' project. The answer is, essentially, no. I looked back and there were macros and nets and data and it was at least 2 yrs ago. A real tangled mess, and not sufficiently documented for sending off to the great unknown. Sorry.

I think the essential thing to know about FancyRibbon is that the "data" is the radius. In my case, I was just using a scalar (circular profile), then converting this to a 2-vector by duplication (Compute [a, a]). So you need to provide the two elliptical radii for major and minor axes as a 2-vector in "data" at each point along the line.

Furthermore FancyRibbon requires 'tangents', 'normals', and 'binormals' components. These give respectively the 'direction' vector at each point on the backbone, a normal to that direction vector (perpendicular to flow direction, if you will), and the cross-product vector (binormal) to those prior two vectors. This fully describes the 'twist' for each position along the backbone. If your direction vectors are more or less parallel and there are no sharp changes in direction, you can (I have) faked up the normals and binormals reasonably well; the direction vectors are faked by just subtracting point n-1 from point n's position. If, however, you have a twisty tube (say, a protein), you have to calculate reasonable vectors, or maybe your chemistry program generates these for you.


_______________________________
Chris Pelkie
Scientific Visualization Producer
622 Rhodes Hall, Cornell Theory Center
Ithaca, NY 14853