@Tom_P fyi, I just added your second sample to the YT thread that pascal made.
1 Like
Well, if fillets arenāt designed in parametrically, and donāt succeed 100% of the time during design iterations, even if done parametrically, then they just get in the way of development.
On the common occasion that the fillets fail, even parametrically, their failure will inhibit the design process.
Hence, a sound workflow will usually have the filleting stage near the end, if not the last step of design.
Therefore, fillets just get in the way.
For example, design a 0.0625"-0.125" thick sheet metal part and put a 0.02"-0.03125" radius along every single edge and corner, then send it off to a manufacturer, and see if they bother deburring all the edges accordingly. 9 times outa 10 theyāll leave all the edges and corners sharp. So, why bother modeling round edges and corners at all? 
Rhino has lost its bearings since version 7.
It spent resources on trendy features but betrayed its original mission of being able to do everything by building its own workflow and of being a Quick tool for aesthetic surfaces for designers!
Today it is not able to make complex radiuses as solid modeler (because it requires strong strategic work on the historical sequence) and today it is not able to offer handmade aesthetic radiuses because it has not developed tools to manage curvatures such as Icem or Alias !!!
To make truly aesthetic and controlled radiuses the user must spend an enormous number of clicks when it would be enough to guarantee a tool that can make you choose a management based for example on perpendicularity, on the choice of final and initial points, and so on, in a fillet surface toolā¦
in Rhino, however, for years you have had to cut the edge to define the endpoints, you have to build lines to define the directions, you have to project them sometimes onto the normal of the surface or onto the projection in viewā¦ and the hours pass.
And I donāt want to talk about how complicated it is to manage the CV of a surface that must guarantee curvature on 4 sides! Ok that Rhino is a NURBS modeler but after years of waiting I expected that a tool capable of managing the explicit value of CVs would appearā¦ currently after two rounding loops the risk of being faced with a huge number of CVs is close 100% while Alias allows me to choose how many CVs it must have as a minimum number to guarantee continuity of curvatureā¦ for at least 12 years ago!!!
Iām not asking for the latest Icem or Alias āātools but at least the basic ones to speed up the workflow which was Rhinoās original approach up to release 6. if, as someone wrote, ārhinoās mission is not to be a solid modeler but rather a of surfacesā that at least rhino proves to be consistent with its original mission to do QUICKLY aesthetic radii for designers.
9 Likes
I can only speak to jewelry purposes, so I havenāt had to suffer fillet-frustration nearly as much as machine-shop/injection molding designers. All edges get rounded over pleasantly in mere seconds during final polishing after casting. For machining purposes though I can understand why one would need fillets to work well, since the cutters have limits, usually with at least a tiny radius.
But for rendering purposes, tiny filleted edges all over the place do please customersā eyeballs, and that has failed nearly every time I try it, even at 0.01mm (though I havenāt lost a sale because of it).
Iām not a coder at all, but Iām wondering if thereās some alternate mode that would enable fillets easily, like when we QuadRemesh to a SubD? Then have it able to toggle back and forth between NURBS and āFilletModeā or something? Like simultaneously existing in 3 forms in the same spaceā¦
1 Like
Maybe you know this, if not, consider Edge Softening.
ā¦ obviously this is just a cheap trick and good only sometimes for aesthetical previewā¦ OT.
ā¦if you put it that way it seems Rhino started doing thing worse than in past. Itās not, but the progress in this matter is small to none.
Rhino 8 today canāt do fillets as good as other software did in 2010ā¦ and even surfacing tools are somewhat stagnant.
In my field Iām lucky I work with SubDs so the few times I need good fillets, I almost have fun doing them manuallyā¦
3 Likes
Ah yes, the Edge Softening even at 0.01mm has failed many times. Which for my purposes isnāt that big of a deal. But I do rather like to play with the SubD soft fillets 
1 Like
Rhino hasnāt gotten worse but has made overly commercial choices, losing speed, losing depth, losing purity.
If it has to be a tool for designers it must give me the guarantee of reaching production on complex objects and it must do so in a reasonable time.
Depth is the possibility of using each type of entity with the same paradigm and with the understanding of how they work.
Purity is being able to do practically everything with the right balance with the automation.
To date Rhino cannot go into production in certain areas of design because it got distracted and spent resources on features that have distorted it.
Where I work you must be able to control the exact value of a cordal radius while preserving the curvature and relating it to mold axes and draft angleā¦ with Y-shaped radii which are the norm and with curvature on surfaces on all sides with organic shapes and where the number of CV must be as low as possible to be able to calibrate the zebras on time.
In fact forms light years away from prisms or extrusions and please not mention sub-d that are totally usless when i have to meet this constrains togeder.
For anyone who has ever brought a steering wheel into production knows what Iām talking about, and itās about design and Rhino sells itself as a tool for designers.
2 Likes
That is good to know and really gives something to look forward to. As someone who is relatively new to the community I find it a bit unsettling how these types of requests are handled. The OP made a clear and informative post and shows a gap and a problem in user experience. But it seems like a few old timers want to insist that it is the way things should be. As a UX designer one thing I have learnt is that you have to constantly evolve to the changing user needs. Just because something works does not mean it can not be improved. Working on better default behaviours while still giving full control to the experts is a nice middle ground to satisfy both newcomers and long time users. I am really liking integrating Rhino and Grasshopper in my workflow over the last year and I hope to see this community to grow and make better tools so that Rhino can be even more enjoyable to use by all types of creators.
6 Likes
That clear and informative complaint about Rhino fillets has been made many times by many users in the past 25 years. I donāt think anyone has stated that it is the way things should be.
3 Likes
hmm, thatās actually something Rhino is not constrained to.
How much does āIcem or Aliasā cost annually?
I think thatās probably a matter of technique, workflow, and talent or etc.
Yes the hrs will pass indeed.
Yeah itās not something thatās easy, otherwise everyone would be doing it. The world would probably be a better place if it was easy though. 
How much does 12 yrs of Alias cost?
I think Rhino is the best CAD, itās merely in itās infancy. The next 100 yrs will be amazing I bet. 
Sounds parametric. That would be awesome. I wish I could afford Alias or CATIA, etc.
2 Likes
The use of BĆ©ziers surfaces through the explicit control of the CVs is not a question of parametric but of a recursive check on tolerances.
Obviously the preparation of the starting surfaces and the work of perpendicularity or radiality of the sections etcā¦ but here it is not a question of costs as with many more clicks even Rhino could come close ti Alias but only of implementation of an algorithm that would already be implementable using current routines.
Personally Iām talented enough 
to be able to create rather high quality surfaces on objects much more complex than the exterior of a car but the software shows limitations that make certain workflows frustrating that would be within reach.
And therefore due to these extra loops it is not physically possible to get close to the quality of Alias without spending too much time.
I repeat: itās not a question of costs, the routines are already available but not harmonized or organized to make this leap in quality.
PS. There are many videos online that show the creation of complex fillets with Alias and you will be able to see what explicit control is and how much it allows you to focus on the quality of the zebras.
2 Likes
K, so 12 yrs of Alias would cost at least $64k just for the concept version, if you want the surface version well thatās going to cost $133,440 
So, Iād rather buy a few seats of Rhino and hire a few people full time to model surfaces for me, than to buy 12 yrs worth of one seat of Alias with endless fees.
I could hire a few newbs and pay them to run Rhino and have them follow paths of discovery that will lead them to be more productive than any singular speed modeler that tries really hard to be super fast with Alias.
I disagree. I could hire a CAD modeler, or even train multiple newbs to become CAD modelers with Rhino and accomplish more in a yr than any singular CAD modeler that uses only Alias for 12 yrs. Especially for the following 12 yrs after that, cause that could be another $133k ā think about the compound interest you could earn off that instead of giving it to autodesk 
I could care less about zebras. Itās funny how much ppl get distracted by things that arenāt actually requisites for creating a better world.
Rhino has zebras anyways. But seriously, the cost of Alias makes the comparison pretty irrelevant.
Instead of giving autodesk $100k, I could pay a newb to pretend to be a CAD guy, do a bad job and make terrible surfaces, and then pay a newb to pretend to use sand paper to smooth out the surfaces, then pay a newb to paint it. And the revolutionary design would still look better than the tesla cyber truck, so 
4 Likes
Just curiousā¦ Where do you work and what do you create in Rhino?
2 Likes
How many people do you think you could hire for $133,440 over 12 years? Thatās $11,120 a year. Maybe one part-time student. 
3 Likes
I could hire a few newbs and pay them to run Rhino and have them follow paths of discovery that will lead them to be more productive than any singular speed modeler that tries really hard to be super fast with Alias.
For most of the design I do Rhino wouldnāt work. I need a parametric design space. Right now Iām making changes to the size of a complex housing and because itās built parametrically almost the entire model has updated properly, just need to spend a few minutes fixing things.
With Rhino I could never make such drastic changes so quickly - I canāt scale the entire housing whilst maintaining wall thickness and relationships to PCB components etc. I donāt have the time in my day to rebuild surfaces whenever I want to make a change to something, so in my case the cost of parametric software and the speed with which it allows me to create and iterate is worth the cost.
2 Likes
Cost, money, prices, etcā¦
If we want to go that way, ok, Rhino is not parametric, but included in the price you also have grasshopper.
Iām doing paneling works for an office that have 10+ Inventor licences with same amount of enginers working on those.
Why? Because they are unable to do so. What they ask require literal programming, not āparametricā stuff.
All this is OT. Rhino have its ups and downs.
Letās all stick to the āfilletā part maybe? (it really is interesting)
Rhino really need some love for handling more fillets cases (and also more mature tools for surfacing).
8 Likes
Cheap software + cheap labor = entrepreneursā favorite around the world. 
Of course, the provided wisdom is pricelessā¦
To not make it offtopic:
Fortunately, I rarely use fillets, but in the past, I learned some basics and from time to time I do some simple surfacing work.
In one task I was saved by the TestSetbackFilletEdge command but itās not working anymore. Was it merged into another command already?
1 Like
Yeah.
Beeing able to make better use of what already exists would already be a big step forward.
Something like what you posted here :
2 Likes
Try TestSetbackBlendEdge
see 
1 Like
Setback corners are now default for BlendEdge (and blend edge only); the test command - as far as I can see - gives no different results.
4 Likes