Can anyone please fillet the following surfaces with G2 continuity?

I wish I could be one of them… Just playing around with it from time to time after work and wonder how much distinctive and original in many ways this unique software is :slight_smile:

I don’t understand the point of this. Try something really challenging. This is like trying to become a concert pianist while only playing chopsticks.

Well, as Alex_Wright says he wish to learn how to use VSR tools to solve this puzzle and I agree that it does make sense to start with the basic examples to understand the principles of how it works and then move further.

1 Like

That’s not a setback corner, as the OP asked for.

That corner is a standard 4-boundary corner and there is a standard NURBS method to do it correctly (align by project for the shortest side, here shown on top). A setback is something very different, also in terms of surface tension and visual effect, meaning what the product’s owner sees.

SolidWorks, Alias and ICEM can handle this and I am more than puzzled why after years and years Rhino does not offer an align by project option, as the algorithms are no longer patented. Instead, rendering modes seem to receive most effort, not modelling and analysis tools.

2 Likes

In a little while someone will tell you that the programs you just mentioned cost a lot of money, and they can not be compared to Rhino …
Rhino for certain operations is a little “rough”, not very sophisticated, focuses his attention in other areas…

1 Like

I really missed that thing, indeed it’s a whole another story and as far as I know only Catia GSD is able to handle ball corner with the setback and within a single splitted patch (face)

Catia ISD makes the same producing three pairs of patches

VSR (ADS) the same story with its own pluses and minuses…

I assume there’s a similar tool in Alias which may produce more controllable result but I’m not quite sure that it is possible to achieve curvature continuity in this case and especially when the shapes aren’t just flat planes.

1 Like

What is the name of software you are using? (Blue one)

Its quite simple: McNeel neither understands the topic nor the workflow and thinks that surface quality is an “Automotive” thing and therefore neglects that complete field, otherwise we would have seen appropriate tools years ago.

The filleting tools in Rhino are still a complete joke, the matching tool is stoneage technology, and the way Rhino handles manipulating surface CPs is useless if you want quality. Additionally the lack of surface evaluation tools shows McNeels comprehension of surface quality.

McNeel simply had no idea about the value of the knowledge of Peter Salzmann and Michael Wieczorek, and thus saw no need to work with those guys. Autodesk did, and in that process improved their own tools to ICEM level. But thats just an “Automotive Thing” you know…

It’s an IceM Surf owned by Dassault Systemes. Check the wiki for a short info about it

1 Like

Btw, I were searching for some discussions about native IceM technologies applied in AD software and couldn’t find anything specific… What are the tools which Salzmann And Wieczorek had improved, Alias I assume?

Yes, it was Alias

The Catia solution usually is not clean if you analyse it. It is easily done in ICEM and Alias.

You are likely right, but, again, since many algorithms or UI approaches concerning interactive controls are not or no longer patented, why not just add them to Rhino’s functionality? It’s not that there is black coding magic necessary, or having to hire Fields Medal winners ; )

If you do industrial or consumer products for top-tier and even good second-tier brands, the design quality in terms of formal aesthetics is so very important these days. You even see it in some Danish, Italian and Swedish furniture brands.

1 Like

actually,

I’m a big fan of “baby steps”. Sure it is the basic corner blend, but I always wonder why so many “professionals” don’t even know how to do this one. So its not that bad starting with this one.

@Alex_Wright

Next challenge could be doing a corner blend when 3 fillets are flowing a bit different.
Here 2 other types of corner blends that might be useful:

initial shape:

type one -> works but isn’t the best choice here due to the fillet flow

type two is better in this situation:

blend2vars.igs (735.8 KB)

well I try to reverse some Icem algorithms for years now and it turned out to be extremely difficult.
Sure I’m not a mathematician so maybe this is the problem I have. But under the hood these algorithms are extremely powerful and complex. You can get close, but the accuracy and reliability are superb. I actually think that McNeel will have problems as well. In addition to that there are no papers around on how these things work. I actually believe its rocket science.

I also know both people mentioned and they have made an impressive career. Furthermore Icem was developed very close with its specialized customers needs for decades. Whats good for some, may not the best for broader mass. This is the dilemma of Icem Surf. And the question remains, those couple of hundreds of surface experts are simply no market compared to the millions of cad users. And that’s okay. Because although there is an increasing demand for class A, I actually doubt that there is often a real need for class A. Its just “I want the best, yesterday and as cheap as possible”. Let us not forget that there already has been ASM/VSR, but hardly anyone bought it.

1 Like

Well, there are many excellent research journals in this field, just like there are research journals in the field of medicine, consumer psychology, branding, etc. Professionals usually read those publications and attend relevant conferences and working groups. I have been to a few, as a non-coder, just for professional interest. I would be very surprised if those doing the coding at McNeel are cut off from what’s going on in the relevant fields and what their peers are up to. Anyone can subscribe to the Journal of Topology or attend the International Conference on Isogeometric Analysis to learn about new NURBS or manifold handling concepts and discuss topics with peers, etc.

1 Like

My impression over the years is that as a relatively smaller company that frequently offers “lack of resources” and “priorities” as an explanation for why things are done a certain way and other things are not done at all, Mcneel’s activities center on development, training, sales and support.
I think the traveling that developers do is related to those activities and there’s probably not a lot of budget for “professional development” as those at Intel, Nvidia, Boeing, Dassault or IBM would think of it. McNeel is not NASA or the University of Washington. I’m sure they do the best they can, though.

Maybe I’m wrong.

@Lagom @AlW

when I was talking to the Icem devs and looking at our own in-house developers the main issue with software developers and especially research on universities is simply the lack on knowledge about surface modelling. I believe Icem became so good because they went nearly weekly to the companies doing surface modelling, during their main development phase. Our experienced coworkers told me about this. There was one example they always tell me about which is about offsetting. At first they were always trying to create exact offsets. However our engineers don‘t need exact offsets, they need clean offsets with little deviation allowed. So what they did they implemented three different versions. And this what you can do in nearly every command now( construction mode for technical exactness, standart for most situations, and design mode for early surfacing, where controlpoints are strongly optimized but also have strong deviation.
That they became so good regarding matching, is due to the fact that modellers told them how they manually match in difficult situations with already included deviation analysis. And i remember talking to one the vsr guys about research papers. They said, some are good but at best they give you 10 % of whats needed for a new command. Its not just about the math, its also about realibility, prioritising, the fact that the result is workable with. Thats the same issue when people say there has to be always g2 or higher, which is simply not true. If you are mathematical correct your result can look much worse compared to non exact, but prioritiesed result.
The asumption of always having g2 was leading to t-splines, which is very good technology. But at first they thought this technology will even replace class A . At least I remember advertising like this during Rhino times. At some point, probally when migrating to Autodesk they realised that this technology has very good application in concept and organic modelling, but its pretty useless on class A. So I believe CAD technology should be developed very close to the targeting customers. I think McNeel even does that, I just believe that class A is not they target audience. Although Rhino is not that far away. :slightly_smiling_face:

2 Likes

Excellent and informative comments.

Every technology intended to be practically useful should be developed close to the targeting customers. :wink:

I personally know of one such case from a completely different field (transport logistics). From the role of a business consultant on a transport company I ended up designing (and implementing) an entirely new logistics system from scratch (actually a complete ERP system). The world was already “filled with logistics systems” at the time, but none of them really solved the problems and customer demands which a transport planner in Scandinavia had to deal with (today this has changed, but anyway).

But since I was talking directly with the “domain experts”, on a daily basis, including the top managers in charge of the business strategies, I could distill exactly the kind of info which you gave examples of (regarding filleting in the context of CAD), namely the “real world needs” and in the logistics case, the exact “business driving” aspects of the rules in the system.

One interesting thing we noticed when evaluating some optimization strategies for the route planning functionality was that the best algorithms on the market had proven to achieve +11% better results in a comparable business. But, my end customer had a workflow concept that, “by implication”, gave nearly 30% more efficient routes than their competitors in the market segment. In plain language; If I had applied an “academic approach” to the the design of the new system (for example introducing automated route optimization) it would most likely have destroyed the company’s competitive advantage.

The company’s competitive advantage, and the best rules for achieving the most efficient end result, was inherent in the culture and in the minds of the transport planners at the company. Adapting the design of the tool to serve the way these guys were doing their job (which they werte doing better than most others in the field) was key.

So if something similar was also the strategy by Icem Surf, then I can imagine why they are doing things better than most others in their field.

// Rolf

1 Like