1 Like
The simple truth is that certain type of shapes are easier to create with a certain type of modeling technique, this is why it’s not possible to achieve everything with just one type of geometry. Depending on the goal, one can choose to use free-form NURBS, parametric NURBS, traditional polygonal mesh modeling (with or without subdivision), voxels, deformation-based polygonal modeling such like the one used in ZBrush etc. All of those methods have they strengths and weaknesses, so none of them is totally flawless and universal for every type of geometry and shape.
Rhino is extremely flexible in terms of providing a wide range of tools to deal with the major type of geometry, though I must agree that it still has some important tools missing or underdeveloped. It’s perfectly capable to handle 90% of the needs for car body modeling (still can’t complete with Alias, Catia and Icem), chassis, suspension, reinforcement panels and basically everything used on a car, except that it lacks an integrated mating tool with inverse kinematics (IK), complete with start and end limits to create and simulate hinges, suspension mounting points etc.
“I’ve spent 2 weeks looking at this and have all the answers (except I have no actual actionable results,) but I now know more than everyone else in this industry.”
Not what I wrote or implied. Like the painter who changes the landscape to match their painting.
Yeah, you’re right, the technology today is abysmal, what we need are Star Trek replicators, holodecks, etc
I’d prefer their no-money advanced society [1] where hunger and poverty are completely eliminated.
In the meantime, as you stated, “all today’s software is too primitive to be used efficiently”…
Since in all the industry up to now nobody has got it right yet, it might be that it’s just a little more complicated than you think.
Or we’re still in a primitive profit-driven era [1] (in contrast to Star Trek above), when we haven’t even figured out how to make highly-efficient software.
And since I’ve found solutions, it shouldn’t be more complicated than I think, think about it.
Some like parametric, some like polygons…
By “like” they actually mean they prefer the features and methodologies they have invested thousands of hours to master, because there is a high threshold in mastering any different ones.
Which proves my point: the vast majority of software is mediocre in terms of overall efficiency and usability.
There is a difference between “it’s not possible today” and “it’s not feasible at all”.
So what I’m saying, is that it’s feasible to make everything, exclusively with parametric modeling, assuming it has been developed to be flexible and efficient in making whatever you want.
Parametric modeling is closer to math and by definition less restricted than any other method, and since math can describe anything, so can(should) that software too.
You can’t make the majority of free-form car bodies and many other shapes with parametric modeling. A car body requires tons of manual adjustment of curve and surface control points based on visual analysis. Many areas of the model need to include trimmed surfaces, which breaks the history used by parametric software. The same goes to human body and creature modeling, which is best done via polygonal mesh (be it with traditional subdivision technique or mesh deformation in Zbrush).
1 Like
Let’s remember to keep it civil folks…
Plus… @gustojunk is the only one who can use a tool for 2 weeks and know everything wrong with it and the industry that created it…
I keeed, I keed… 
(i’ll see myself out)
HAAAAAAAAAA!! nice-
1 Like
What is your actual experience with actual parametric software? 'Cause it sure sounds like something you’ve only read about. Your premises are completely wrong, there’s geometry that users of such software–and Rhino–create every day for which no actual precise mathematical description exists and the system just bodges out an ugly little approximation.
Here is an example of a car designed with SubD. The SubD was then converted to NURBS, in order to split the body into separate panels and add extra details. Post #16 explains the technique used to do so:
There are two classes of CAD people: inventors/designers and engineers/technicians. The first class makes conceptual models. Rhino SubD is suitable for this work. The second class designs parts which are manufactured. The essence of their work is finite element analysis and tolerancing. These two classes of CAD people have nothing in common. Airlines divide them into four classes: inventors = first class, designers = business class, engineers = premium economy, technicians = economy.
That was definitely true in past decades. Today in only true for people working with very arcane workflows, in teams with a high level of specialization/narrow division or labor, or for people without enough time/exposure/mentorship to develop at all levels.
I always recommend people to develop at all levels, then you can be the pilot and fly your own plane. Even start an airline.
G
4 Likes
But what if it’s the same person? Are you telling me that I suffer from multiple personality disorder… wait… it all makes sense now! 
-Jakob
1 Like
I had the same “issue” so many times while talking to customers asking me if I know people who can handle 2d conceptual sketches, NURBS surfacing, engineering, g-code programing and CNC-milling. 
Many insist that those things must be done by different persons.
3 Likes
Can you share your outcome? I’m very interested in a direct comparison Rhino-Blender
I’m surprised you find it buggy, I hardly ever run into bugs, specially modelling related ones. Can you give some examples?
I’m really struggling to understand what this sentence means, I find it hard to imagine how one could get more math driven than a NURBS modeller and more parametric than Grasshopper
2 Likes
What is your actual experience with actual parametric software? 'Cause it sure sounds like something you’ve only read about. Your premises are completely wrong, there’s geometry that users of such software–and Rhino–create every day for which no actual precise mathematical description exists and the system just bodges out an ugly little approximation.
Parametric just means building with a higher level of control, instead of low level using polygons, and thus with no artifacts and limitations in scale, flexibility, etc.
I was modeling with nurbs surfaces in Rhino 5.0 full time for a few months a decade ago - but today I had forgotten almost all and I just recalled some of it the last few weeks. A few years before that (early 2000) I was researching a method of my own to completely replace the polygon-based modeling and also the graphic cards of that era with parametric ones that would run many times faster and I regret I didn’t continue that research. 20 years later most of the (dinosaur) industry is still stuck in polygons…
You can’t make the majority of free-form car bodies and many other shapes with parametric modeling. A car body requires tons of manual adjustment of curve and surface control points based on visual analysis. Many areas of the model need to include trimmed surfaces, which breaks the history used by parametric software. The same goes to human body and creature modeling, which is best done via polygonal mesh (be it with traditional subdivision technique or mesh deformation in Zbrush).
You confuse theory with today’s specific implementations, and thus today’s limitations.
In theory you can make anything you want with parametric modeling, including human face and body, and it should be far easier, not more difficult, because the actual visual information has nothing to do with triangles and quads, that’s just a human artifact.
The reason that today it’s more difficult and time consuming, is because those tools are poorly made i.e not designed from the right perspective / philosophy.
Visual information can be represented by a smaller set of parameters that a human can easily create and modify, rather than masochistically manipulating low-level artifacts, losing the big picture and inspiration in the process.
Here is an example of a car designed with SubD. The SubD was then converted to NURBS, in order to split the body into separate panels and add extra details.
When I converted it to nurbs, I got a set of surfaces arbitrarily cut and having lots of lines each, and not equally spaced… I thought it would be a lot of labor to rebuild each of them, adjust them, blend them, cut them… Also, when I imported an obj mesh and converted it to nurbs, I got tens of thousands of little surfaces with no option to join many of them, except every two of them…
@guido
“very few have realized that, and that should be parametric, not polygon-based nonsense.”
I’m really struggling to understand what this sentence means, I find it hard to imagine how one could get more math driven than a NURBS modeller and more parametric than Grasshopper
Just think that science and technology will evolve, and all of today’s tools will become completely obsolete sooner or later. From this de-facto fact, you can conclude that there is definitely a ton of things that can be done to improve what we’ve got today to bring it to amazing levels.
In short, the problem is more algorithmic than mathematical. Today’s parametric modeling algorithms are at a very low level vs what is possible, hence the low efficiency and the vast room for improvements.
see also my reply to JimCarruthers.
I’m surprised you find it buggy, I hardly ever run into bugs, specially modelling related ones. Can you give some examples?
If you’ve found and are following the same optimum path(s) of doing things, obviously you’re skipping the bugs, or have got used to them.
The last 2-3 weeks I have stumbled on hundreds of bugs to the point that at one time I was thinking that it’s 80% bugs vs 20% clean code, and version 7.2 seemed to me far buggier than version 5.0.
I wasn’t motivated to record them, so here is just a couple of examples:
I got this annoying one too many times for different views:
Here is a crash when you can see that neither memory or CPU is too high:
and a minor one (visual misaligned feedback):
Can you share your outcome? I’m very interested in a direct comparison Rhino-Blender
Me too, but life is not always so generous… 
First, an objective comparison would take a significant time that I don’t have.
Second, I have spent the last 2+ weeks full time (all day long) trying different workflows, so I’m sorry, but I don’t feel comfortable to share what I discovered just yet - (and thus empowering the competition)…
I can only say that this is possible: I’ve found a workflow (that I’ve not seen anywhere) similar to Rhino’s SubD but far quicker and far more flexible, that allows me to make any basic unique vehicle form (incl. exotic ones) not in hours or minutes, but in seconds(!) with a very direct and intuitive control and feedback, while having real-time booleans eg cutting the bottom of the vehicle and the arcs of the wheels - things that cannot be done in Rhino.
That gives me very good polished surfaces already, and then I can add details with a number of different tools, including smoothly affecting the surface with another mesh using Shrinkwrap projection along with some more real time modifiers.
You look at the command line, there’s a FLIP option there.
I think you just don’t know how to work on Rhino.
2 Likes
I don’t confuse anything, my “theory” is based on my real-life experience designing actual cars for living.
As for the messy NURBS surface patches that come from your SubD models, those are fully dependent on the complexity of your geometry and whether you use SubD properly by creating good topology of the polygons considering the limitations of the Catmull-Clark subdivision.
SubD has two option for the conversion to NURBS: either transform each 3-sided or 4-sided polygon into a NURBS patch, or try to combine multiple co-linear polygons (like 4x4 quad polygons parallel to each other) that together form a large, single common NURBS patch.