Is this curving tapering fillet best done with sweep or fillet?

Fillets ask for radii and this has a constantly changing radii,
Not sure anyway how to tackle a curved fillet if the radii are constantly diminishing.
How would such be done ?

I am still struggling with the concept of FilletSrf where the result of the first pick loses one the surfaces required for the 2nd pick.

Furthermore the fillet starts at one edge and reaches out across the other surface so perhaps a sweep with profiles ?

What method is best ?

Others also to be done.

curving diminishing fillet2
IMG_002405 Fillets stbd



1 Like

Has been in Rhino since at least V5.
For variable non-circular blends use VariableBlendSrf
For variable chamfers use ‘VariableChamferSrf’

1 Like

Having a quick go with first the VariableFilletSrf, its giving an arced fillet and expecting me to establish the radii at start and finish, however doing it visually when one zooms in I see I am ‘out’ a bit.

Then the result doesnt follow the edge.

can it not establish the max rad needed for the selected surfaces, and if the radii changes asd it does here, how does one get it to follow the edges ?

I then try for VariableBlendSrf and again same issue, establishing the fit, and it doesnt go to both edges.

its a standard radii it seems.

again no auto follow edge, and zoom in and see I am over edge.

fillet issues.3dm (762.1 KB)

I need some decent vids on these tools, and how a varying edge is followed and how as here the fillet can fill the entire area.
curving diminishing fillet2


1 Like

Are you sure it is a variable radius?
Your picture is not very clear, but it looks to me that it is a constant radius that was already there before the slots were cut.
Can you measure the radius?

1 Like

Not sure what you mean by slots were cut, the image is a photoshop doodle using a paintbrush.
I refer to radii changing in the fact that the distance from the purple wall to the edge of the light blue shelf is getting greater as one goes Clockwise round the shape.
If the radii are determined by that distance they must be constantly changing.

I cant get the fillet do go to the top of the purple surface as it would then stick out beyond the lt blue shelf.

A Cobweb fillet would fit that space but there is no command as such.

Wish there was, especially if an animated spider was to be seen doing it :smile:

Not sure what you want me to measure.
attached is the file.
fillet issues.3dm (762.1 KB)


1 Like

The picture that I was referring to was this:

The edge indicated by the red arrow looks to be at a different level then the edge that the green arrow points at. Same with the magenta and cyan.
You are trying to make those edges at the same level.

Yes how does that not make the radius constant?
Do you have access to the object in the picture or is the picture the only reference you have?

1 Like

something like this can be done with V8 / WIP
but the command hassles with Arcs…
EDIT bug reported here

rebuild the curve and it works

see result as V8 and V5

fillet issues_tp_rh5.3dm (779.5 KB)
fillet issues_tp_rh8.3dm (4.6 MB)

hope this helps - kind regards -tom

1 Like

Hi Jim,
optical illusion.
the item is of 5 levels.
red and green are both level 4 and cyan magenta are both level 5.
Pig of a thing to measure and understand.
My CAD is best stab and I would challenge anyone to better it, as its a few days and x10 glasses and a vernier !

Just need to fillet the steps from level 2 down to level 4, and 4 down to 5.

as per that image, like a cobweb fillet as opposed to an arc.

Tom your V8 FilletSrfToRail is a good attempt, but its an arc, only goes a bit up that cliff face.
see image of the oily item.

Is fillet the wrong tool as if its designed to do a radius , then it wont do a curving inward cobweb type fillet. maybe fillet is wrong word. It wont fill the area from edge of ledge up to top of cliff there.

It can be that a radius arc can go from ledge edge to cliff top, .

What tool can do that though ?

If I were to do sweep2 using ledge edge and cliff top (hope my terms make sense !) and establish an arc at the largest ‘web’ then sweep to point, would that do it ? (is there sweep to point ?)

note how the ‘cobweb’ ends at prof1


Doesn’t look that way to me.
This is what it looks like:
fillet issuesx.3dm (816.9 KB)

I’m lost in translation :thinking:

Not sure but this doesn’t look like intended:

you show ‘prof 1’ twice with different shapes. and ‘rail’ twice. Do you mean prof 1, prof 2, rail 1, rail 2?

Is this what you’re after?

Filleting or cobweb fill exercise_emod.3dm (4.2 MB)

guess I have to give you ancient V5 format brb

Filleting or cobweb fill exercise_emod_ancient V5 version.3dm (3.2 MB)

1 Like

Fillet is the right tool when you need precise circular / arc based shape. If you just need “fillet like” shape, you can use blending and basic surface modeling to sculpt out the surface needed.

1 Like

I used sweep2, using history in this case to adjust one of the profiles, and then rebuilt the srf after:

Looks like it made cleaner isocurves than my previous netwrksrf version, so… kinda interesting. It would be cool if sweep2 would allow the user to select more than 2 profiles or something, to add more control. Idk, maybe it does, but I doubt it. There should be a ‘sweep4ormore_profw/2rails’ command. Or ‘cobweb_sweep’ :joy:

Hi, I have asked for a sweep 3 or sweep 4 in the past .
That looks nice.
Its better than fillet as fillet wants an arc, i.e curve of one radius per moment of fillet.
It will need the end of that ‘cobweb’ filling in and the junction with the existing ‘cobweb’ filleting.

So fillet was wrong for that area, as my rethink away from fillet was steering to now, so it would seem.

In the other post even the long bar with radiused ends that you say needs removing from the boolean should be given a fillet at its base.



I guess I’m interpreting this more clearer now. It looks like you’re reverse engineering something, and I see the shapes you’ve been mentioning. But it’s hard to tell what the design intent is here, and almost looks like a damaged or worn out component you’d like to repair or replace idk.

Also, if said component is basically 2.5D and you’d like to add some ‘organic curvature’ to it, then it’s going to boil down to some freeform surface modeling. So, good thing you’re using Rhino :smiley:

But in RE, capturing the worn out component is one thing, and recreating the way it was when new, is another thing. :sweat_smile:

I’m seeing things like burrs, gouges, worn edges etc., in that image. Not sure what you’d like to recreate from that.

And, the spacing here initially doesn’t look very accurate, so it’s probably a matter of design intent. If you want a perfect duplicate of the original, then you probably need the spacing to be more accurate before ever thinking about the rest of the surfaces you’re after, like the ‘fillet’ or ‘cobweb’ etc:

and looks like the original doesn’t have that slot all the way through, so idk, it’s hard to interpret what I’m looking at from one image, and not knowing the GD&T’s, design intent, etc…

That area is in fact not undergoing any trauma, was always like that from the start.

Yes the area circled by you in red differs, and I knew it would end up spoted, and after one week of taking loads of .001 accuracy measurements and trying to recreate it all, its not worth anymore effort fathoming out that small area, both levels are circles at ends, the gap between them at top is accurate, somehow their milled and my cad see a different area, and I have measured thrice !,

I am leaning over the item with my back aching badly, I have 5 fractured vertebrae, and there is a limit to what I must put myself through, if this one area of a few .001 inches wipes out another week, it will be my undoing, so that red circled area whilst gnawing at my perfection brain must be as it comes out in CAD and we must fillet and cobweb MY CAD… forget the original image from now on. It was to show the cobweb areas effect only.

we must not get hung up on comparing them. 5 weeks when 1 was charged for.

all the very accurate dims I made should have all matched, but theirs is done in an era of no CAD, and what a milling machine does at the time can wander it seems when in a human hand.

lets just focus on my CAD and the ledges and cliffs that need 'landslips. or cobwebs !

We now have one cobweb you did using sweep2.

even the long bar in purple should in fact have its base filleted with a smal radi, perhaps 0.02 or 0.01, so that it looks the part.

This is for visual effect, its a non functioning 3D printed insertion into a larger item.

So put to one side that photo as such. My fillets or webs go from edge to edge as per Pshop doodles.

I just need to know how to do the filets and cobwebs and how to tidy up the boolean.

I had to make it in layers, nightmare to get to see best way, and the layers of solids formed with cutting shapes was the key into getting it this far.

I presume the boolean union was the correct thing to do, then MergaeAllFaces then explode it and tidy up.

To see how its done is good for me and others.

To have a Rhino file exported as V5 I can study is good.

any video making of methods is VERY GOOD.

Now to fillets and cobwebs, cobwebs with a curve type face.



1 Like

I’m not the one trying to reverse engineer this part. Steve is. I’m just explaining why the tops of those round slots don’t look round. The reason is that the fillets were there before the slots were machined. The shape you see at the top of each slot is the shape of the intersection of the slot and and the constant radius fillet that was there before the slot was cut.

Your geometry may well be drawn accurately. Your question is - How did they make the fillets? The answer may be very simple. They made the fillets first and then cut the slots. The fillets were already there when the slots were milled.

You can model the part the same as it was machined. Just make the constant radius fillets first then cut the slots. That’s a far simpler way to model than to make edges and try to fill the spaces between edges with cobwebs.

fillet issuesx2.3dm (1.2 MB)

Also, if you would invest in a Radius Gauge you could measure those fillets and confirm that they are in fact constant radius fillets.

1 Like

Hi Jim,
well thats hit the nail on the head :tada: :grinning:

the straight sides above your fillets could be such in the item, or as near as.
Its a pig to get to, only lens in corner of mobile phone (hence lousy images, give me my macro canon anyday !) can manage it, whilst trying to tip the heavy item at same time and vernier, and a struggle and backache, and a lamp I cant angle enough.

Thanks for the curve gauge links, I have with it my two sets, but they wont get into it and angle of sight is tricky on them. It wasnt worth ultra analysis, just must look sort of like it.

I will run with what you have managed.

However I can’t select the dark grey part of it and yet the layer is not locked. :frowning_face: How do I select that part ?

I have to also do another which is the far side of it, so will try to follow what you did.

If there is chance of a video, it would be welcome, so I can see the steps involved, else a storyboard.

but to have the model is MOST WELCOME.

are those fillets of constant radii ?

sweep2 by CAD CAM CNC Lander was also showing promise.



1 Like

Indeed, I merely intended to reference that image.

very interesting :beers:

The object might be locked, maybe use ‘unlock’ command?


Unlock worked.


1 Like