Question about the seemingly unnecessary complexity of NetworkSrf


#1

I have created a NetworkSrf in the following file:

Example.3dm (588.5 KB)

As you can see it has a million isocurves. Of course working with that surface is going to give me a lot of trouble. But here comes the weird thing; I used RebuildUV to uniformally space the V-direction curves (that surface is also in the file) with 20 V-curves. When I then use Pascal’s SrfDeviation script found here, I can see that the surfaces are pretty much identical.

Why does NetworkSrf create such a complicated surface if it it able to recreate it with so little curves?


(Pascal Golay) #2

Hi Luc - it would be better to have the curves, but, I notice a couple of things - the dimensions are pretty large and NetworkSrf has a tolerance, maybe comparatively small, set in its dialog- that will come into play in how complex the result is. But NetworkSrf does a refitting of all the inputs- it is not a command to use if you want to mess with the surface afterwards as a rule. Anyway, post the input curves and maybe we can tell more about what to do differently…

-Pascal


#3

These are the input curves and the input surface that surrounds it;

Example_Curves.3dm (331.6 KB)


(Pascal Golay) #4

Hi Luc - I’m not sure what to tell you other than as far as I can see the surface seems perfectly consistent with what NetworkSrf does and is designed to do… BTW, looking at your rebuilt surface side by side and CurvatureAnalyisis shows the rebuilt one has lost some definition, which is what would be expected, looking at the input curves…

-Pascal


#5

Thanks. I think I see what’s happening now. :slight_smile: