ll.
On the first picture, you see the value x (relative to the world system) which i wanted to extract and turn into the curve system (t 0.0 - 1.0). However it seems that the decimals are being rounded too soon (only 3 decimals in the sine function) and thus leading to imprecise results.
In order to translate these values i did the following:
multiplied the sine with a real world length, and then divided the result with the curve's length.…
something on the transfer bugs out the results as well. I believe that the amount of branches played a huge role for the lag there as well.
So why does the top num component has the profiler? all 3 of these components only have internalised data…
c curves (open curves and closed curves that have a kink at the seam) that the first few and last few knot values are identical. In fact, you need a few identical values at the start and end to ensure the curve reaches all the way to the first and last control-point. This is called "clamping". The number of repeated (or "multiplicity" or "valence") knots must be equal to the degree. So with the above example of a degree 3 curve with 7 control-points, a valid knot vector would be:
0;0;0;1;2;3;4;4;4
9 values, with the first three and last three being identical. You are never allowed to repeat the same knot value more often than the degree, and you are never allowed to decrease the values either.
Weights are single numbers attached to individual control-points. Weights are most often used to create Conic Section curves (arcs, ellipses, parabolas, hyperbolas) and these are always degree=2. You can also assign weights to freeform curves if you want of course. When you choose to do so, you must supply an equal number of weights and control-points. Furthermore, weights are relative only to each other. If you assign a weight value of 1.0 to all control-points, you get the exact same curve as you assign a weight of 4.9 to all control-points.
The difference between "Curve control points" in NurbCrv and "Control point count" in the Knot component is that one actually represents a list of points, whereas the other is merely an integer representing how long that list is.
In order to create a valid knot vector, all I need to know is the degree, the number of control-points involved and whether or not the curve is supposed to be periodic or not (periodic curves are a special case of nurbs curve, best to stay away until you're comfortable with the regular nurbs curve type). It doesn't matter where the control-points are, I only care about how many there are.
--
David Rutten
david@mcneel.com
Poprad, Slovakia…
Added by David Rutten at 3:05pm on February 3, 2012
t and L,
so (L+(L/(3*π)))/2
and so on
I guess I could make theoretically make this value an input parameter.
With all the different features and controls people are asking for though, I'm concerned that if I keeping on adding them to one component it makes it an unwieldy 50 input monster.
I think it will be better to make several components each geared towards particular usage. For this it would be helpful to hear from all of you about what you are using or want to use it for.…