currently teaching some lectures about filleting, details, transitions.
neither _filletEdge with variable Handles nor _variableFilletSrf nor _filletSrfToRail are my friends.
this post is about _variableFilletSrf
i would expect to be able to build a surface like the violet one at the screenshot at the top.
the biggest Problem is, that _variableFilletSrf does only look at the intersection of the 2 surfaces involved.
if the resulting surface is extending the initial intersection curve (which is always the case, if the surfaces do not intersect at 90-degree), this extension part is out of control for the user.
this screenshot shows the issue - trying to connect the blue R15 to the blue R5 Fillet.
left to the R15 handle - no control …
interpolation
but the interpolation of the Radius is not nice.
so to get a nice transition in this case, it is additionally nescessary to workaround by double the Handle at Start and End (2xR15, 2xR5)
… not sure how other uses integrate _variableFilletSrf in their workflows - but I fail most of the times, trying to solve it with filletEdge, pipeTrim, or custom filletSrfToRail Script.
Yes one would expect that it would produce a surface that would fit with other fillets, but it never does. Its been pointed out many times since the command was first introduced that this makes the command close to useless.
FilletSrfToRail, however, has a lot of potential for creating the needed surface.
If you run FilletSrfToRail using the edge of your fillet that wraps around the vertical 20mm fillet you get a surface that is a little less complex and a little more accurate than your purple surface. How was the purple surface made?
The trick with FilletSrfToRail is creating a rail curve that produces a surface that has good tangency with the ends of neighboring fillets
I was able to create the red curve using history on FilletSrfToRail and modifying the red curve until good tangency was reported by EdgeContinuity on both ends.
_filletSrfToRail is not behaving as expected, i rarely get the surfaces i want / need, and i get many crashes.
I gave up using it.
this might be a mac os x thing… (you re on windows i guess)
i use a custom script for the violett surface - and yes i still have to optimise to get a less dense surfaces…
other than that, I totally agree that in its current form VariableFilletSrf is hard to use. @Tom_P did you sent crash reports when you were using FilletSrfToRail?
I’ve often had excellent results with FilletSrfToRail. I can’t recall any crashes. I’ve used it to do exactly what you are doing, connecting fillet strings of different radius sizes. It does occasionally fail to produce a complete fillet but usually I can modify the rail to get around that. An example of an incomplete fillet is when you use the edge of your purple fillet that is on the planar surface as the rail.
FilletSrfToRail makes a much more accurate and smooth fillet than your homemade method. Below is a screen shot of the file I posted yesterday. I made the FilletSrfToRail fillet using the edge of your purple surface as the rail. I turned CurvatureGraph on for both surfaces. As you can see the purple surface is very wiggly while the FilletSrfToRail fillet is nice and smooth. Your purple surface is more accurate than the FilletEdge version in your file but less accurate than the FilletSrfToRail fillet.
The fillet edge version is useless to me. Its not accurate or smooth. That causes downstream problems. The main one being that when I export models for approval to other solid modeling programs they do not recognize the surfaces as fillets.
