Controlling randomly generated points along curve

I'm doing some experiments with the Random component, and currently I'm trying to generate a series of points along a selected curve, basically 'dividing' it into segments of UNequal length.

I have done this by using the Evaluate length component on a curve and having a Random component controlling the 'length factor for curve evaluation.' Easy enough, but I have found that the points created tend to form clusters of points along the curve. I don't like this

I am assuming that I shall have to use the cull components to remove points that appear too close to one another, but this is where I get stumped. Any help please?? thanks!

Replies to This Discussion

Permalink Reply by David Rutten on September 23, 2012 at 4:40am

The parameter density of Nurbs curves is not necessarily equal throughout the curve. Think of it this way; if you were to travel along the curve domain at a constant speed, your actual speed through space once mapped to the 3D curve geometry would not be constant. You'd accelerate and decelerate all the time. This is caused by control-point positions, control-point weighting and the curve knot-vector.

I assume you're evaluating your curve at random parameters, which would account for the clumping. You see, when the parameter density is high, you'd be travelling relatively slowly and you'd get more points closer together.

A way around this would be to evaluate the curve at normalised lengths rather than curve parameters, as length evaluation does not suffer from this acceleration/deceleration effect.

If you're only worried about two random points coming too close together, you should take a look at the [Populate Geometry] component which also places random points but tries to keep any two neighbouring points as far away from each other as possible.

David Rutten

david@mcneel.com

Poprad, Slovakia

Permalink Reply by Daniel Conradie on September 23, 2012 at 4:47am

Oh! I understand what you mean. I found it odd that a random point generation would create this clumping. I did indeed evaluate the curve at random parameters. Thanks!

RSS