Should I build all fillets with the Fillet command?

This question is about general approach of modeling high-end products.
Let’s say it’s a real model of kitchen sink, and should be made with the best possible quality.
Should I use tangency or curvature for radii?
Since there is a Fillet command, simply all radii should be made with this command?
So all fillets are tangency, is it acceptable?
What if the radii are big enough like 2 and 3?

I numbered radii (some of them aren’t there) in a case of detailed feedback.

Radii 1: 2 – 4 mm
Radii 2: 12 – 30 mm
Radii 3: variable radius 2 – 30 mm
Radii 4: 1 mm

I would be happy to hear some industry standard approach.
Thank you!

Hello - if you’re using multiple radii, you can combine them in one filleting operation or work from large radius to small - that is, make the largest radius fillets then as a separate operation make the next size down, given what you now have from the first set, etc.
I’d say for a kitchen sink or any consumer product, really, tangent fillets will not be good enough, you’ll probably want to go in and create these transition surfaces more carefully, depending of course on the design etc. But my guess would be you’d end up making ‘hand made’ transitions to fully control how light and reflections fall.



Are you saying we have to do this “by eye”?! Seems unworkable. What could be more accurate than a curve coming off a flat surface as a tangent? I’m seriously confused now.

Hello - I’d say in general for surfaces that are on consumer products for which the design is important and control of light and reflections is important, a tangent, arc section fillet on transition surfaces is probably not clean enough - there will probably be requirements for controlling curvature continuity and flow between surfaces that are more complex than that. I should think a kitchen sink would fall into that category but if fillets are good enough, use them by all means,


1 Like

The rate of change of curvature abruptly changes between lines and tangent arcs, so although they will be tangent, reflections will have sharp edges. Have a play with the options in the BlendSrf command between two perpendicular surfaces, and compare results using the Zebra command.

update: added an image demonstrating differences in smoothness with different types of continuity. increasing smoothness from left to right. I’ve never used G3 or G4 because i’ve never been able to see the difference in smoothness beyond G2, only the increased curvature in the middle.

1 Like

You may already know this, but i’ll point is out for the ot :
this is a bit misleading example, because you point out a funny special case. Actually the G1 blend looks better than the g2 blend here. Because even if there is a little kink in the zebra (G1) if the flow of the highlight is better, G1 can be better. How strong the kink is can be adjusted by the blendfactor. one thing you always should do when working with g2 is having curvature on all surfaces. Otherwise you get this „overhang“ shown in your g2+ highlights. G3 matching (not blending!) is useful on extremely stretched and symmetrical surfaces. When working with symmetry you use g1 or g3, since g1 is defacto g2 on symmetry.


Formal aesthetics are in the eye of the beholder. In this case, the client (or in-house design manager). What looks “good” for one, looks “bad” for another. Per se an never ending topic.

In general, it is practical to work with degree 5 or 7 curves whenever possible, so you always have G1, G2, etc. options, depending what your client (or in-house design manager) requires.

In case you work in this industry (I did courses for companies in this sector), a very helpful idea is to take some time during the Salone del Mobile to see high-end products and concepts in the sector one is interested in and take good pictures and measurements. Such material is also very helpful in discussions with clients who want to cheap out on formal aesthetics ; )


Lagom, thank you for detailed answer, it helps a lot! I just didnt get about

can you explain this again please?
Thank you

Hi Tom, thank you for such a practical advice, the only thing i cant get is

would you be so kind to illustrate this concept?
tnak you

Hello - if you mirror a curve or surface across a plane that defines the tangency, then you get curvature continuity ‘for free’ since the two are exactly the same -

The curvature graph has no gap or step where the graphs meet - that is G2. The lowest example has the graph smooth where they meet in the center - there, ‘flow’ is continuous, or the rate of change of curvature. (G3)


1 Like

Although this is for Alias, see these well illustrated explanations related to NURBS curves, best practices, etc. Since Rhino is NURBS based, you can then apply these principles in Rhino, too.


Very nice link Lagom. I have bookmarked that page for occasional reference as I feel required. I like it.


1 Like

I have a following basic question. How do you decide between G2 and G3? Is there some rule of the thumb? Or some scale dependency?

There is a simple rule. Look at your highlights and decide which looks better. In the end its about the look. We usually say: If the eye stops because of some irregularity (bumps, twisted surfaces, discontinuities…) , then you need to fix it. Usually G3 makes sense at your biggest main surfaces. The smaller and more curved it gets, the degree of continuity becomes less important. Nevertheless, G3 is used rarely, never used G4 and don’t know where this ever will be relevant. We constantly use G3 on a cars front/rear bumpers, bonnets and roofs for symmetry

Edit: One important aspect which is not obvious in Rhino, since you cannot really measure deviation of continuity: There is no 100% G2 or 100% G3, sometimes you have G2 but your controlpoints are well distributed so its close to G3. Nobody can tell if its G3 or not. Same for G1 and G2. A close-to-G2 can look better than a true G2. Cannot stop saying this.


Since I don’t design cars and related products, G2 is all I need. It very much depends on client demands, industry standards, habits, etc.

1 Like

Also remember to provide a good “lead-in” on your primary curves and surfaces if you want good highlights.

Hi Andre. The detail on your object depends on the end need and fabrication process. Industries have multiple modelling passes most of the time (as far as I saw at least) and rhino is the first in scale so you will need to leave your surfaces workable for the next guy (developable surfs etc, depending on needs). If you will make a 3d for an industry or for ready to go you’ll need higher gcon most of the time (again depends on what you will do). If that’s the case you should start at g5, that’s more than enough in most cases. By the time you’ll be at g7 you’ll have all figured out anyways so no worries. I used to make injection molded toys and all the next guy needed was intersections most of the time. It really depends on what is needed. If you are going for renders I’d say use as less meshes as possible for speed reasons. Cut mercilessly wherever possible. It isn’t worth it most of the time. 3d printing is also another story, depends on the machine…
But if you want to make career in surface engineering you should start learning how stuff is made, that way you’ll have a better understanding on how surfaces should flow. Good luck

1 Like

what do you mean with g5 and g7?

I also asked what this means, then accidently found the answer in Alias tutorials: so he means degree, g5 means degree 5 and 6 control points, g7 - degree 7 and 8 points, as far as i understand, so G2 and G3 respectively

that’s what I was thinking as well and what would make sense. So a Bezier surface of order 6 and 8. Degree 5 doesn’t necessary means having 6 control points when using Nurbs. Maybe “g” is the German term for “Grad” -> translation of degree. Never heard that notation, though. Thats why I’m asking.