I have been trying to understand how point sets work in Variable fillet. I have gone through the documentation and it's clearly stated that the added points control radius only at each individual point (I was under the impression that points were defining a length of the specific radius before carefully reading the Help file)

So, if I understand correctly, the 4 points of Set 1 in the 03 example define a radius at each point and in between the fillet engine blends with the original radius. If that's the case, is there a way of defining a specific steady radius in a section of the edge being filleted?

Finally, I see some different behavior of the fillet engine when defining a single point on the meridian line(term?) of the merged sphere compared to another random point on the circumference. Why is that? (trying to understand how this works! not saying it's a problem)

Any tips or things to keep in mind while using variables will be highly appreciated! :)

Sphere is some sensible with fillet because it's a volume who i yet naturally filleted! :)

Better to train with cubical volumes first

Sometime works, sometime not! :)
Sometime rotate the sphere for no intersection or intersection or align with "curvelines generator" can help...

So variable fillet with spheres is some hard! :)

Hi nameless,

> I have gone through the documentation and it's clearly stated that the added points
> control radius only at each individual point (I was under the impression that points
> were defining a length of the specific radius before carefully reading the Help file)

So the radius will be that value only at that particular point but as the fillet travels away from that point it will blend between that value and the next neighboring one.

Basically there will be "stations" along the edge to be filleted. At the start of the edge if there is not any point set there it will have the regular edge radius there (in the future there will also be "edge sets" so you can set different radius values for an entire edge but currently there is only one global edge set).

At a spot between 2 stations the radius at that spot will be a blend between those 2 station's radius values. The blend is a cubic ease-in/ease-out type process, like this (unless you have enabled the "linear blend" option):



Vertical direction in the above function graph is the radius value that will be used, horizontal direction is the location along the edge.

> So, if I understand correctly, the 4 points of Set 1 in the 03 example define a radius at each point
> and in between the fillet engine blends with the original radius.

So no, not quite - the 4 points of Set 1 will make for 6 stations along that edge - the start and end station will be the regular edge radius, and there will be 4 internal stations of the Set 1 radius value. Then the radius used at a particular point on the edge will be a blend between the 2 nearest stations. If the 2 nearest stations have the same radius value then the radius will be that same value throughout that span (a "span" in this sense is the region between 2 stations).

So it will be like this:


So one thing to note there is that 2 stations in a row with the same radius value will result in that same radius throughout that span. So if you have 4 stations all of the same radius then it's no different than if you had just the outer 2 stations of those 4. What I mean is if you take out the internal 2 stations above, you'll have:



It's the same graph shape as the 4 station one above, when 4 stations have all the same radius value in a row.


> If that's the case, is there a way of defining a specific steady radius in a section of the edge being filleted?

You've already done that in your 03 example above, if you have at least 2 stations of the same radius value in a row the radius will hold at that value in that span.

So in your case this area of the fillet is holding at a steady radius value in this blue region:



> Finally, I see some different behavior of the fillet engine when defining a single point on the meridian
> line(term?) of the merged sphere compared to another random point on the circumference. Why is
> that? (trying to understand how this works! not saying it's a problem)

It's because the start/end "seam" location of the closed edge has influence, it gets the regular edge radius at that spot if you have not placed a point on it.



> Any tips or things to keep in mind while using variables will be highly appreciated! :)

Try to avoid a whole lot of radius change within a relatively small distance.



Hope this helps!

- Michael

Hey Michael, you do not joke around about clarity when you provide answers do you. Everything is perfectly clear now, the "station" concept brought the way this works home. So, the "weird" behavior is because the two ends of an edge are considered stations in that aspect not because the seam of the sphere is an issue.

You are a real treasure : )

Indeed that works better today! :)

Hi Pilou! What did you change? Was this thread educational for you too? :))

I change the process of puting the points ! :)

But i am always perplex of the result of a single normal Fillet when there is crossing curve generator!



It's for that i try to rotate this one for avoid that! :)