topic Re: Fit point spline behaving weird. in Fusion Design, Validate & Document Forum

Fit point spline behaving weird.

mr.rickwright — Mon, 25 May 2020 22:22:52 GMT

Apparently Fusion is telling me to take a break!

I drew a fit point spline that I have drawn at least 20 times already it is fully constrained and driven by user parameters. but this time it came out looking weird.

So I drew another one exactly like it using its dimensions. Theoretically it should be exactly the same - but isn't.

what is the hidden meaning? Other than time to take up painting (or needlepoint)!

in the attached the solid fully constrained spline is the first one and is weird. The construction line one is drawn using the first and is the "Correct" shape.

file attached.

Rick W

Re: Fit point spline behaving weird.

TrippyLighting — Mon, 25 May 2020 23:09:06 GMT

@mr.rickwright wrote:

... Theoretically it should be exactly the same - but isn't...

That is not the case in Fusion 360. FOr a Kayak you won't need to sketch symmetric splines either 😉

Re: Fit point spline behaving weird.

mr.rickwright — Tue, 26 May 2020 01:54:10 GMT

Thanks Peter !!- I guess an asymmetrical Kayak will be good at going up wind in a hurricane or tornado 😉

Re: Fit point spline behaving weird.

TrippyLighting — Tue, 26 May 2020 02:31:41 GMT

LOL. Symmetric boats can be created by proper surfacing techniques and mirroring 3D geometry instead of splines. I use and teach those techniques here on the forum all the time 😉

Re: Fit point spline behaving weird.

mr.rickwright — Wed, 27 May 2020 16:48:08 GMT

Peter; you have taught me many things including the benefit of using intersecting surfaces rather than attempting to project spline intersects. But those curved surfaces are extrusions of splines, are they not? I am interested in how you would recommend creating the smooth curves without using splines. I have not explored sculpting directly in 3d as I wanted to stick to a parameter driven model. But if there is a better way - I'm all ears.

Perhaps you misunderstood my issue - I was not attempting to Mirror the spline, as I am following your previous advice of mirroring the lofted surface. I should, however, be able to create a REPEATABLE spline that is driven by parameters. In the example I showed that the two splines had EXACTLY the same dimensions but the first one was absolutely not correct. I see this as a bug that I can repeat. It seems to occur when the length of the spline line is being shortened by applying the dimension.
Again; if there is another way to create a smooth curve driven by dimensions, other than a spline, I am very open to it.

Rick W

Re: Fit point spline behaving weird.

TrippyLighting — Wed, 27 May 2020 16:36:38 GMT

No, sorry, you DO need splines. But you don't need to mirror splines 😉

This video posted very recently by Chad Lockart shows some nice techniques.

I am specifically referencing the tangency ribbons or helper surfaces he creates to get tangency across a mirrored surface.

Re: Fit point spline behaving weird.

mr.rickwright — Wed, 27 May 2020 16:51:58 GMT

sorry you were too quick for me - I edited my post to add clarity.

I really take issue with a spline that is fully defined but doesn't mimic (even closely) another spline defined by exactly same constraints and dimensions. I see this as a pretty major bug.

Rick W

Re: Fit point spline behaving weird.

TrippyLighting — Wed, 27 May 2020 17:10:35 GMT

A few years ago I would have also considered that a bug and @jeff_strater and I had more than one exchange related to this. However, I believe this is simply a unique behavior in how fit-point splines are implemented in Fusion 360.

The scaling of spline handles is unique for each spline and I can imagine a number of reasons for that, now that I have had time to process the things I've learned over the years.

Most people who use a fit-point-spine have really no idea what that really means. If I tell you that a fit point spline in Fusion 360 is a 5-degree multi-span spline, would you know what that means?

How many spline points are needed in a fit point spline to become multi-span?

What is a span?

How does that relate to the tangent handles and what do these do under the hood.

All too often - me included - we have expectations of how something should behave based on a lack of understanding of how these things work under the hood 😉

If you want to control a spline with user parameters then you should use a control-point spline.

Re: Fit point spline behaving weird.

mr.rickwright — Wed, 27 May 2020 17:34:22 GMT

Noted; and I fully agree that nothing replaces years of experience. I started with control point splines but they have other challenges. I guess I have to go back to them and work around those challenges.

BTW yes the horizon modeling video is awesome. I will have to watch it several times to absorb it all.

Re: Fit point spline behaving weird.

mr.rickwright — Fri, 29 May 2020 04:37:23 GMT

Ok I used a Control point spline as you suggested in order to get a fully defined and constrained curve that is Repeatable.

Check out the sketch Called "Bow"

the bottom half is a 3 degree spline and shows as fully defined (Black line)

the top half is another 3 degree spline with different dimensions and as far as I can tell it is fully defined as nothing can move (and it is constrained in the same way as the bottom half)
BUT the top line shows as a blue line indicating it is not fully defined.

Ready to throw in the towel now... I'm obviously not "getting it"

Rick W

Re: Fit point spline behaving weird.

TrippyLighting — Fri, 29 May 2020 12:11:27 GMT

Before continuing to work on this my recommendation would be for you to read through the Autodesk Alias Theory builders.

The spline is fully constrained. It's simply the way yo have constrained it fools the sketch solver in Fusion 360 to believe it isn't.

Have you checked the curvature of the splines you created?

It would look to me that you used a 3-degree spline with 6 control points, which creates a multi-span spline. For best curvature control a CV spline should have n+1 control points with n being the degree of a spline, so for a 3-degree spline that should be 4 and for a 5-degree spline that should be 6.

I created a 5-degree spline with 6 points which very closely matches the shape of you 3-degree spline and have excellent curvature:

Re: Fit point spline behaving weird.

mr.rickwright — Sat, 30 May 2020 16:37:35 GMT

Peter;

Thanks those Theory builders are very informative and I assume fusion works the same way under the hood, just without exposing the options in dialog boxes.

I too can get the curvature looking really nice - see below. Using a Conic (2 degree 😁) that follows the design intent and gives a full range of smooth curves driven by the three parameters of Flare angle, V angle and Curve height (driven as ratio of base aka Rho). But we have established that fusion has no intention of fixing the problems with conic curves. @jeff_strater At least not in this decade.😭

The design intent of this project is for the model to be configurable by changing user parameters.

As soon as you apply parameters to your nice smooth curve and then adjust them the curvature will no longer be so nice and smooth.

Manipulating a 5 degree set of vertices mathematically to mimic a full range of conics is beyond me (even if I did it in a spreadsheet ), and apparently it is beyond most folks as this discussion illustrates.

https://math.stackexchange.com/questions/2060044/b%c3%a9zier-approximation-of-conics

Frankly I am less concerned about curvature as some work with a fairing sander will help work out my frustrations with fusion 😉. I just need the curvature to be smooth enough to Surface Loft.

What I am concerned about is ensuring the model does not blow up when user parameters are changed. A fully defined sketch is the foundation of that.

It seems that the sketch solver is sensitive to the sequence that constraints and dimensions are applied since the lower curve with same constraints is black - which may also make it sensitive to changing user parameters in a specific sequence (since you can't suspend and change multiple). Unfortunately none of the Theory builders I have read so far seem to hint at this underlying sensitivity.

BTW I chose a 3 degree curve because I thought the solver would have less work to do with the math...

I think I will just rebuild (with a 5 degree), apply constraints in a different sequence and deal with the fallout when it blows up after a parameter change.
I've gotten pretty quick at redrawing, managing broken projections etc.