Dammit, I thought you would.What does De Boor's algorithm do? I'm reading about it, but it's all in maths, and I can't quite tell what it's for. Maybe it does the thing I just said.
(,
Thu 25 Jun 2009, 13:26,
archived)
well if we're reading the same wikipedia article it does stuff...and it does it well apparently.I know stuff all about this sort of curve fitting.
edit: do you just want to fit a degree n polynomial to a set of n points?
(,
Thu 25 Jun 2009, 13:31,
archived)
edit: do you just want to fit a degree n polynomial to a set of n points?
If I understand the question, no.I want a smooth string of beziers with one control point each (are those "degree 1 polynominals"?), which put together go through all the points in my list.
(,
Thu 25 Jun 2009, 13:46,
archived)
If it's a single control point, and you're not actually going through those,The control points for the curves on either side of each point should form a straight line. The angle of that line is probably half way between the adjacent point lines.
Maybe.
(Maybe.
Bloop Fri 16 Jul, 22:10,
Thu 25 Jun 2009, 13:58,
archived)
So, they lie somewhere on some line with that angle. :)Progress!
(,
Thu 25 Jun 2009, 14:04,
archived)
no that doesn't work for zigzags i give up
( Bloop Fri 16 Jul, 22:10,
Thu 25 Jun 2009, 14:04,
archived)
Hmm, how would a zigzag go
The middle spline would have to be straight, and the other two severely hook-shaped in order to join on to it smoothly. So their control points would have to be way off to the side relative to their start and end points.
My advice would be to find a computational geometry page, not a mathematics onesolving problems like this in code is very different from solving it on paper, in maths...
(
The Alchemist king of the needlessly complicated,
Thu 25 Jun 2009, 13:36,
archived)
The problem is that when computer scientists get advancedthey start using maths, the traitorous bastards. The ones who are still capable of explaining stuff in English don't know the answers.
There's this page: www.css.tayloru.edu/~btoll/s99/424/res/mtu/Notes/Lab/curve/tracing.html
I think that means De Boor is just for drawing a curve with a lot of control points. Didn't occur to me that was much of a problem, but apparently so.
I have a question going on here: en.wikipedia.org/wiki/Wikipedia:Reference_desk/Mathematics#Bezier_curve_with_width
(,
Thu 25 Jun 2009, 13:42,
archived)
There's this page: www.css.tayloru.edu/~btoll/s99/424/res/mtu/Notes/Lab/curve/tracing.html
I think that means De Boor is just for drawing a curve with a lot of control points. Didn't occur to me that was much of a problem, but apparently so.
I have a question going on here: en.wikipedia.org/wiki/Wikipedia:Reference_desk/Mathematics#Bezier_curve_with_width
they fucking well do as well, and that.I've been struggling for months with mathematical 'pseudocode' algorithms that have just a bit too much "assume an infinite set of things" to be feasibly implemented as code :(((
are these algorithms mean to take a curve, and produce the control points, or take some points, and produce the control points of the curve that fits them?
(are these algorithms mean to take a curve, and produce the control points, or take some points, and produce the control points of the curve that fits them?
The Alchemist king of the needlessly complicated,
Thu 25 Jun 2009, 13:50,
archived)
I need the second of those things.The algorithms, I think, judging by the pictures, are just for quickly drawing a point on a curve defined with lots of control points (what k3b/-\b would call an "degree lots polynominal" I think). Therefore useless to me. Really I just want to know how to find the edge-path for a curve with a width, so I can fill inside and thereby draw a nice fat curved line with an even width, for the flash renderer I'm writing.
(,
Thu 25 Jun 2009, 13:57,
archived)