Re: curve fitting

--- In CAD_CAM_EDM_DRO@y..., batwings@i... wrote:

>So now as software
> writer you have to program the method of generating the equation in
> computational memory space, no small task with big data sets. Then

you have

> to evaluate the equation for each increment of X or Y (one vs the

other

> depending on instant slope of tangent) and then you write the

output to the

> machine as lines or arcs anyway. For all that you get maybe a few %

more

> precision around your curve.
>
> OTOH, stepwise arc method is simple in concept and code, easy on

CPU hence

> fast, and results in essentially the same thing at work. What am I

missing?


That is the gist of my question. Has anyone seen a more efficient
and (maybe) better thought out algorithm than the 3 point stepwise?
Application of quadratics, regression techniques, cubic spline
techniques, etc?

My interest is less in the direct generation of G-code arcs, than in
the generation of something that is vector (little V) based like a
DXF format, that can be further processed in a cad or cam program.

Best Regards,

Fred Smith
IMService