[futurebasic] Re: [FB] Re: [XFB] Recursive FN Questions

Message: < previous - next > : Reply : Subscribe : Cleanse
Home   : April 2003 : Group Archive : Group : All Groups

From: Robert Covington <artlythere@...>
Date: Mon, 28 Apr 2003 04:46:09 -0400
On Sunday, April 27, 2003, at 09:24  AM, tedd wrote:

> tedd said:
>
>>> It sounds like a least squares problem to me and then using Bezier 
>>> curves to fit -- that's a lossey function -- in other words, you 
>>> lose data. The Catmul spline is more a lossless function in that the 
>>> data integrity is maintained.
>
> rc replied:
>
>> You don't lose much, and if you want to use a Catmull, you can. I 
>> have said that the points themselves are not as important as the 
>> curve produced. I want to use Bezier. I need -fewer -  points, not 
>> the same or more. Eh.
>
> What I said still remains true

No one disputed any truth, you chef from Curve Hades. :)  That's a spam 
chalupa flavored herring that I refuse to eat without a clean table. 
Waiter!

> : 1) A Catmul spline retains the points -- as such it doesn't lose any 
> points. Also, it is a very special sort of spline -- but I won't go 
> into that; 2) If you use a least squares approach and then apply a 
> Bezier curve to them, then you may be losing points -- as you seem to 
> want.

Arghs. That's why I have said it about 3 times before, including the 
letter you are responding to.

Losing points -is- what I want (once they are figured into the 
curvemaking), and a Bezier is the preferred  curve ( an application 
specific need). I don't need nor want a knot for each sample point. 
Thus this continued whack of the Catmulling catfish (herring) is 
annoying  and non-helpful.

I must now call the Health Dept. upon you for bad fish. But then again, 
you yourself have stated that you have a herring problem. :)

>> Read Graphics Gems article accompanying said FitCurves.c and you will 
>> know what I am talking about, the images show it all.
>
> I believe that I understand what you are trying to do -- been there, 
> done that in trying to lessen the amount of noise in data.

I want simply to have an algorithm that gives the results shown as in 
the image of the Devil or the letter "L", from points to nice cubics on 
page 624-625. Noise is no factor. Noise reduction has not near any 
level of importance as being able to fit the curve. Fitting the curve 
achieves the noise reduction. Or would, if one could enjoy the reality 
of fitting a curve.

Arghh, I just lost my dinner. Waiter!

>>> I don't know what toolbox functions remain in OSX, but I might try 
>>> using some of the joined areas toolbox calls and then unraveling the 
>>> points determined by them and then run those points through the 
>>> Bezier function. Why not use toolbox calls to do the hard work?
>>
>> Because Toolbox calls aren't designed for this type of thing as far 
>> as I know.
>
> Any Toolbox call is designed to do what it is designed to do (by 
> definition) -- but further, you can use the information it provides as 
> you are able (i.e., access its structure record).

I clearly meant a Toolbox call that fitted a curve to digitized points. 
I did not in any way indicate a need for a restatement of the obvious 
and known. ;)

>
>> If you know, please demonstrate. Otherwise, I think you are being 
>> Tedd again with Tedd's Principles (Use PG, Use the Toolbox, ie works 
>> for me will work for you (maybe)  )
>>
>>   :)
>
> Tedd's Principles? Where did that come from? I'm simply taking my time 
> to try to help you. Keep in mind that no good deed goes unpunished. > :-)

Also keep in mind that when someone  puts a smiley after what he knew 
was easy Teddbait, that the Troll was aware of it, and it should be 
ignored. Thus we could have avoided 8 paragraphs in rebuttal. Don't 
rebutt that, or I will rebutt that with more fish jokes.

In summary, reducing a point set or worrying about noise is the least 
of my worries. Any more diversions from the true nature of this problem 
I had is not helpful.

Had, because I have given up on it for yet another year. The thread may 
now die a well due death.

And sorry for a threadbare thread, and bad sardines. Tired, and 
code-lobotomized at this point. A point which remains  
non-parameterized.

I digress out of bad digestion.

Regards-like,

Robert bob