Evenly spaced points with hermite function?

[breeder]

Hi i posted this question on one other forum but with no responce.

Maybe someone here will know answer.

Run this code is script editor:

{
int $i=0;
vector $start=<>;
vector $end=<>;
vector $tan1=<>;
vector $tan2=<>;

while ($i<=10)
{
float $pointOnCurve[]=`hermite $start $end $tan1 $tan2 ($i*0.1)`;

spaceLocator -p $pointOnCurve[0] $pointOnCurve[1] $pointOnCurve[2];
$i+=1;
}
}

 
So i want evenly spaced locators along this curve.I read little about this , and i think what i need to do reparameterization of this curve(based on lenght).But i don't know where to find info about this.For example maya don't allow to rebuild bezier curve(this actually like hermite as i understand) so i don't know this is possible or not.
 
Maybe there is some python module who allow this type operations on curves etc?

[joojaa]

What kind of performance are we looking at here. The math involved is calculus and linear math so solving this is a bit painfull* but not entirely impossible. Or you could iterate a better distribution, but not really efficient if you intend to use this to dirve something in maya, so you would forego using mel here.

*You see its blending functions arent symmetrical so its harder to find the parameter easily

[breeder]

There is thread on cgtalk regarning particle trails.

So for particle from position in previous frame to next you can only get trails in straight line.

The solution is make simulation  steps higher.

My idea here was to use this curve to distribute particles from frame to frame.

I don't know it is bad idead or not , i wanted to try this , if possible , but as i see , this is not so easy , i am not so "strong" in math etc , so i will probably put asside this idea.

[joojaa]

Its pretty cheap to take more steps. Ive using this kind of strategy for ages. Its pretty easy to store x previous frames wroth of data for curved segments (works well with catmul-rom splines, which are hermites of sorts). Off course yes you can get the data from velocities (better yet accelerations) and positions and velocities are based on moves so they dont really give all that much more info. Personally i use catmul-rom splines and often just normal b-splines for this since its easier just to dump the curve data to the renderer directly, so you dont have to bother with the rasterisation of the data. Anyway depending on your data you can just grow nurbs splines instead too if you wish (works remarkably well with prman).

More of interest is guidance with uniform parametrisation, but its more convenient to do in c++ since its a performance issue if you take a few steps of newton iteration per point.

PS: seems like there s easy trick for this must read a bit more.