Curves Intersection

[klingon]

I got 2 curves. They are intersecting themself in one point I named P.
On first curve I need to determine 2 sequential EditPoints that the point P.

For example:
curve1 has 20 EPs. I need to know $i so that P is between EP[$i] and EP[$i+1] .
I mean: if, following the curve path, we start from EP[$i] going to EP[$i+1] we will encounter the point P.

Is this possible to know that $i ?

TIA
Marco

[failure]

If curve has uniform parametrisation then integer part (floor) of the point's u parameter equals the value you need.

CODE
// select a curve point
string $s[] = ls -sl;
string $u = match( "[0-9]+\.[0-9]+", $s[ 0 ]);
int $i = ( int ) $u;
print $i;

[klingon]

Unfortunately my curve hasn't uniform parametrisation (it comes out from a fitBSpline on a 1 degree curve I've built reading its editpoints from a file) and I verified that EPs are not positioned at integer values of U.

But maybe ther's another way. The curve point comes from a curveIntersect so I know its U value. If I could read the U values of the editPoints of the curve I will solve my problem with a simple loop.

How can I query to obtain the U value of an editPoint?

Many thanx for your help
Marco

[failure]

Just rebuild the curve to uniform parametrisation

[klingon]

I can't do that. I need to preserve the exactly EPs I have on the curve (they come from a 3D-scanner).
I've noted that rebuilding the curve results in a similar but not exactly equal curve and I'm not sure the rebuilded one fit the scanner points (the EPs on the original one).

No hope to obtain the U coord of EPs in a curve?

Marco

[failure]

Well then, detach the curve at selected point and count the ep's on one of the remaining segments.

CODE
string $c[] = detachCurve;
$i = getAttr ( $c[ 0 ] + ".spans" ) - 1;

btw. why do you need this info anyway?

[klingon]

Great
I will duplicate the curve then detach it and count the EPs.

I'm writing an UI in MEL for helping people in reconstruction of surfaces from 3D-scanned data.
At the moment I got pseudo-circle curves requiring a loft action (pseudo-conical objects) and I need to rotate them to avoid a twisted loft.
The user will draw an additional curve (snapped to the others from the scan) to inform which is the vertical profile.
The info I've asked is for recreate the curves starting from the nearest EP to the intersection point with the vert.profile. This will assure no-twisted loft.

I'm an aged programmer but newbe on Maya.
I like Maya a lot but maybe it's not the right tool for working with 3D-scanned data.
NURBS are based on CVs but I need to handle EPs and I find it quite difficult.

Anyway this one is done.
Thanx for your help
Marco

[failure]

nurbs are always based on cv's - because underlying math demands it. Working with eps in kind of reverse engineering

[joojaa]

theres another way of doing this. Select all cvs and querry their paramnettervalues then you can use the method forst outlined by failure.

Yes EPs dont represent the curvbve they just represent one special case of the span. the cvs define the curve. Infact onbe can have a unlimitted number of curves with with exactly the same EPs thet are wildly different in shape.

But a cvs define ONE solution so same cvs allways result in same curve. Same EPs dont.

Thstwahy the curve could live without eps but not without cps

butyes this sint hard at all, all you need i to querry the intersection points. Do you realy need to knwo what divisor you are at? Or si it because you wnat to leave the curve somehow intact at renages taht arent touched.

[klingon]

---

theres another way of doing this. Select all cvs and querry their paramnettervalues then you can use the method forst outlined by failure.

---

how can I achieve this?

---

Yes EPs dont represent the curvbve they just represent one special case of the span. the cvs define the curve. Infact onbe can have a unlimitted number of curves with with exactly the same EPs thet are wildly different in shape.

But a cvs define ONE solution so same cvs allways result in same curve. Same EPs dont.

Thstwahy the curve could live without eps but not without cps

butyes this sint hard at all, all you need i to querry the intersection points. Do you realy need to knwo what divisor you are at? Or si it because you wnat to leave the curve somehow intact at renages taht arent touched.

---

The former you've said.
Consider a conic-like object that is scanned by capturing some circle-like curves (it's a touch 3D scanner) along the object height.
With Maya I rebuild the conic-like surface by lofting the curves but I must ensure they are aligned on their first extreme. If not I will get a twisted loft.
To align the curves I require that a user draw a new curve, snapped to the circles curves coming from scanner, and this new curve will determine which will be the starting point of all the circle curves.
Now if the circle curve C1 has 7 EPs : E1, E2, ..... E7
and I find that the intersection point with the user drawn curve is between E5 and E6, I will regenerate the curve C1 by doing:

curve -d 1
-p E6x E6y E6z
-p E7x E7y E7z
-p E1x E1y E1z
-p .....
-p E5x E5y E5z;

and then

fitBSpline;

This will assure the new curve are fitting the EPs I've red from scanner file and that the loft will be not twisted.
I can't see other ways to achieve that.

Marco

[failure]

Why don't you just move the seams of the curves to the point of intersection?

[klingon]

Why don't you just move the seams of the curves to the point of intersection?

You mean by rotations ?
If I rotate 1 curve in order to move its seam to the intersection point, then I will need to apply the same rotation to all of the other curves. But one of the goal of my script is to rotate each curve of a different degree.

Marco

[failure]

No, by:

CODE
moveNurbsCurveSeam;

But you won't be able to do this if curve is non-uniformly parametrised. You'll need to rebuild sooner or later if you plan to produce nice and usable surfaces.

Anyway, you can query the parameter of edit points using curveInfo node:

CODE
file -f -new;
curve -d 1 -p 0 0 0 -p 1 0 1 -p 1 0 2 -p -1 0 2;
fitBspline curve1;
createNode curveInfo;
connectAttr fitBsplineCurveShape1.worldSpace curveInfo1.inputCurve;
getAttr curveInfo1.kn;

It actually queries the knots so you'll get some extra data that represent knots responsible for looping. Just ignore every value that's outisde of parameter range as well as possible duplicate values of multi knots.

[joojaa]

theres a script under mayaInstDir/scripts/unsupported called nurbsCurveKnots.mel that can get you the parametric values of yoru knotpoints (eps) chek it out

[klingon]

Thanx a lot failure and Joojaa for your suggestions. They are very appreciated.
I will ask if the minimal variation introduced by rebuilding the curves can be ignored. If so I will rebuild them.
Very usefull your info on how query the parameters of curves.

Happy New Year to you guys and to all the Maya forum people.

Marco