Here is an all-fillets version - no fudgy surfaces or point manipulation - all true arcs, same radii. The initial blue surface is 'ExtrudeCrvAlongCrv:
Corner_AllFillets.3dm (307.9 KB)
-Pascal
3 Likes
Pascal, how did you managed to create the 2nd step? I extended the upper vertical surface and made a “! _VariableFilletSrf” transition between it and the initial blue surface, then I did the same between the latter and the upper fillet (using 0,1 mm radius for both). However, there is about 4,34 degrees lack of tangency between the new fillet surfaces. 
Hi Bobi - I used FilletSrf at .1 throughout. Also, I used InfinitePlane (IP + Enter) as one target for the green fillet and cleaned up the plane surface there after.
-Pascal
1 Like
For some reason ! _VariableFilletSrf gives a different result, hence it was incapable of doing what’s achievable with the basic ! _FilletSrf command. Kind of reminds me how the ! _OffsetSrf and ! _VariableOffsetSrf differ to each other even when using the exact same setting for distance. 
However, I found a BUG that’s probably the main reason why I my attempt to repeat what you showed above has failed. Looks like the ! _FilletSrf command is extremely sensitive to where the surfaces are located relative to the CPlane origin. I used your first (left) example as a base to build the rest surfaces. In some instances the fillet surface had some unwanted warping caused by adding some extra control points, and the tangency was hugely compromised. I measured more than 9 degrees deviation between the lower pair of edges. That looked suspicious, so I tried to repeat the command on the third (right) example and this time it was just perfect. Then I copied the same surfaces from the 3rd example to the 1st example and the ! _FilletSrf command failed again. I also recorded that on a video.
Thanks, I’ll see if I can repeat…
-Pascal
1 Like
Thanks Pascal. This is beautiful. I’ve never knew IP option existed. I managed to reproduced what you did even replacing red blend with variable. One edge wouldn’t match so I extended it and than trimmed back. It seems so easy now…
I couldn’t reproduce the bug Bobi was showing. I wonder if this has something to do with his history being enabled…
Norbert, I agree with your point but in my case multiblend saved me a lot of time where no high surface continuity is required. And I get these all the time…
I tried that particular scene with History turned off. The problem lies in the position of the surfaces relative to the CPlane origin.
The active History in my video was used to show the different angle tolerances and shape of the resulting fillet surface depending on the position of the original surfaces in the 3d space.
in general i like “all rolling ball fillets”
there is a second solution for this:
this solution shows the blue ball rolls first
then the green ball rolls second:
(to get a nice result i tightened the tolerance to 0.00001)
in pascals version the hierarchy is reversed.
but as all 4 initial radii are the same (R0.1) i would guess the design intention does not have this hierarchy…
( from a design point of view i would vote for the VSR solution posted above)
and to show how immense the influence of this hierarchy is for the design - a similar corner - problem solved by _filletEdge (not so nicely)
kind regards -tom
2 Likes
Could not agree more! MultBlend was the tool that seemed so amazing when I first started using VSR, but over time I came to realize just never made surfaces I was happy with. It’s better than Patch for quickly plugging holes when you’re trying to whip out a concept, but that’s not exactly something to get excited about.
1 Like
Pascal, a quick follow up on a different note. My original file is almost 2.1 GB. Yours, with 3 times as much of geometry is only above 300 KB. Why is this. What causing this extra load… I used purge before posting yet the size remains excessive…
Tom, thank you for looking into it. I’m going to try your rolling stones on few corners and see if I can make it work for me.
As for the multiblend I’ve never worked for auto or aerospace industries, I’m not good enough, my non bias opinion, but it works in my little corner of the CAD world… and two thumbs up for the great videos you posting Sky. I doubled my skills after watching them.
1 Like
Hi Adam -
Running the Audit3dmFile command on this file might throw some light. It is a bit cryptic but you can paste it here if it doesn’t tell you much.
-wim
MB, not GB… but yeah, it looks like there is a lot more in the user data table on your file, for one thing - you must have some plug-ins that I do not.
-Pascal
Pascal, Wim, this is the thing, I have rhino 6 and 5 with several third party plugins installed but none in rhino 7. Well just recently I allowed substance importer but never used it… don’t have substance subscription…
Running audit3dm returns something strange about materials. It shows 9 different materials, while in reality, reality as I understand, there is only one plastic. So where are the remaining 8…?
Corner_Audit3dm.txt (6.8 KB)
4 Likes
Yes, so many good ideas and interesting comments, I’m glad I asked. People with experience sharing their knowledge, it can’t get any better.
I see you came out with yet another very creative workflow. Thank you.
2 Likes
awesome. I think there is a test comand to make those spheres on fillet corners in a simpler way but I can’t rememner the name. I saw it here in the forum. maybe @pascal remember the name.
1 Like
SphereTangentToThreeSurfaces is what comes to mind.
-Pascal
2 Likes
I mean this one. maybe now has that new name
I think that is indeed the same, just made official now.
-Pascal
1 Like
@Pascal, I tried that hidden command, however, it requires a manually set radius, which is basically impossible to use in cases like my sample model, because its sphere has a radius of 7,3137085 millimeters. Quite a random number if you ask me. 
Also I noticed that this hidden command only does a 3-sided patch between the selected edges, but in reality after using ! _Untrim i realized that it’s just a halved square and not an actual sphere. This makes it unable to fill in the 3-sided hole on my model unless I use the ! _ExtendSrf command multiple times.