Spline tension...

[crow]

I'm encountering something that's puzzling me. When I'm alt+dragging on a B-Spline point.
I'm rotoing a pistol at the moment. On the top of the gun the spline curves over the sight at the back and then the next point curves over the barrel at the front. There's a sizable (6cm) gap between these two points. When I alt+drag on the spline over the sight the spline also adjusts itself at the point on the end of the barrel. It's almost as if I might have shift clicked/highlighted both points (if that could be done) at the same time and then alt+dragged, applying the same movement to both of them.
I realise that if I change the tension of the spline at a point the spline will be affected. But, surely any effect that's created on a spline should lessen if the spline is of a certain length? I imagine the 'tension' created at the points is a little like magnetic influence. The 'influence' is strongest close to the point but lessens as we get further away and can be affected by another point (depending on distance from it's neighbour). It seems strange that one point can have an almost equal influence over another point that's 6cms away.
What do you think? :o

[jbills]

I hear ya, crow - have noticed this also. Although it does seem a little "high strung," I'm still trying to remain open minded to the current behavior.

This weighting and magnetism seems to be of help when animating points, but does seem to get in the way when adjusting tension.

I can only speak for myself, but at the moment I prefer the old school way that c*******n handled the behavior between adjacent points, where the tension remained more or less pinned down. Seemed a little more predictable or something. (hard to describe)

But as I said, I'm trying not to jump to any premature conclusions before I see if the advantages outweigh the benefits. Thinking about jumping back into the c-word on a shot in the near future and giving it some thought.

[perryk]

It doesn't matter how far away the surrounding points are, adjusting the tension on a degree three b-spline (what we use) will always affect 3 control points. It isn't a magnetism kind of thing - just pure math and the nature of b-splines.

[adam]

I found it took some adjusting to. I'm quite used to it now, though I have noticed I tend to add a few extra points in places (like the above described gun) to make up for the 'surrounding three points' affect.

[jbills]

curious, perry - what are the advantages you guys cited when deciding on degree 3?

[perryk]

The most prominent reason for choosing degree 3 b-splines is that they are directly convertible to degree 3 Beziers - the 4 pointed kind everybody uses. Another reason is the tradeoff between smoothness and computation speed. Degree 4 takes about 50 percent more computations than degree 3.

[jbills]

ah - those reasons are bigguns. :)

[oatherton]

could there not be an option to choose which type of splines are being used... much like the magnet button within shakes gui for rotosplines. - this simply locks OTHER than the one you are editing where they are.

I am also finding in relation to this, when you add a point along a spline, it can kick the surrounding points out of line throughout the animation. :mad:

[paulm]

Yes, this is annoying, I agree. The problem comes with the position of the control point is not directly related to the curve itself. I've tried this a couple of ways, and both have drawbacks.

I'll try to get it to work again by clicking on the b-spline hull.