surface quality

[cekuhnen]

This is not a very easy object to chew on

A circle (or half) blending into a two lines with a blend smooth curve (or g2 aling Alias)

The Fusion curvature graph kinda gives me some concern. It is not even even along left and right side.

Alias comes with its own bag of problems but in general seems much more cleaner

Building top and bottom cross sections as halve parts obviously makes it work smoother

But also here Fusion builds questionable results!!!

So questions

1: how usable is the surface actually when the graph is kinda crazy in Fusion

2: as you can see Fusion struggles here or is this my approach issue? is there a better way?

[Mike.Grau]

Hi @cekuhnen,

Thank you for sharing your pictures. I got touch with surface team and

we do work on that together. We´ll keep you updated.

Thanks,

[jakefowler]

Hi Claas,

I'll ask the development team to look into the two results you're seeing in Fusion.

The first one might be down to us merging tangent faces that might be better off as separate faces. The second one... I'm less sure about 🙂

But if you unsitich the inputs from the second example and build that face with a surface loft, you'll get a good surface out (attached):

 

I'd say it's always worth trying building a loft as individual surfaces if solid loft gives an unexpected result. In many cases they should give the same shape (underneath they are using the same algorithm), but not always.

Hope this helps!

Jake

[cekuhnen]

@jakefowler If Fusion would offer similar common surfacing tools like most NURBS modelers I think this would also be easier.

Another case where adv surfacing tools just really is more than simply important for fusion.

Loft and such cannot have G0 to G2 along all 4 edges when creating a 4 sided patch.

I will look into your file.

[cekuhnen]

@jakefowler 

There is another thing I noticed

When the loft is done between complete parts (not just half parts) the loft tool works as expected

But also here the curvature graph concern me.

They  are uneven while the have to be the same with mirrored solid parts and they are of crazy shapes.

I hate being a nagger here but I really strongly feel that the vision of Fusion ignores quality surfacing tools.

The curvature issue can be a really tiny math problem that in manufacturing might not show up.

I would need to do an injection molding and then measure to see if there are real visible differences.

But what concerns me a lot more is that one needs to explore different approaches to make the solid modeling tool work.

A designer should just s=choose a tool and it should work as expected instead.

[TrippyLighting]

---

@cekuhnen wrote:

@jakefowler 

But what concerns me a lot more is that one needs to explore different approaches to make the solid modeling tool work.

A designer should just s=choose a tool and it should work as expected instead.

---

+1 A buggy thus unpredictable tool is worse than having no tool at all!

T-Splines and surfacing are coming up a little short on the Roadmap. I find that surprising these tools and the way they can be combined with each other and with solid modeling are one of the distinct advantages of Fusion 360. Rather odd, really.

[cekuhnen]

@TrippyLighting Actually there are many apps that offer nurbs and SubD or solid and subd,

I am close to say that Fusion is still rather a good solid modeler with a very basic surface modeler as a bonus.

But I share your views - I mentioned it often in the the past that all those fancy things dont mean much when

the tools for creating work are limited.

[PhilProcarioJr]

@cekuhnen

This is why I am constantly using T-Splines, because the surfacing tools need a lot of attention. As I have said so many times before I'm not sure why the core tools are not getting the attention they need but other useless features are being added. They keep saying that other teams are working on the other feature and nothing is being taken away from, yet here we are with another update and no work being done on the core tools. This makes no sense...what good is things like simulation if you can't even model what you need to test, or branching designs if you can't model what you need to check in different design iterations, or advanced assemblies if you can't model what you need.....the problem here is the same it has been with all software companies, no one is looking down the road for oncoming traffic. They are only looking at the ground a foot in front of the car while doing 90 miles an hour....doing this always ends badly. They are doing the equivalent of texting and driving...

Just my 2 cents

[cekuhnen]

@PhilProcarioJr

plus T-Splines is slow for modeling complex designs and has its own issues.

curvature graph with TS is less usable as an example because of the TS BREP nature

[TrippyLighting]

---

@cekuhnen wrote:

  curvature graph with TS is less usable as an example because of the TS BREP nature

---

Could you elaborate on that ?

Surfacing, T-Splines etc. are an areas I am very interested in.

Also, what other application offer T-Spline combined with Solid Modling ? Emphasis is "combined" not just coexisting side by side.

[cekuhnen]

@TrippyLighting

because TS is G1 curvature graphs will not be smooth they have kinks in them like around span points in Alias.

so this renders the CG cool but less functional for me as a designer serious with surfacing perfection

in Fusion because AD has not yet included

3D sketching

some serious update to surfacing tools so it is even with what a standard is

T-S can replace NURBS patches

So you can in TS build your patches and in timeline work with the patches like in Alias for trimming etc

essencially if you move CVs of a NURBS patch in Alias or a TS patch in Fusion is workflow wise the same

TS patches exported as STEP however have a heavy geometry so they are not ideal

but I was told that inside Fusion it is treated differently and thus the geometry is lighter

A TS model exported into Alias gives every surfacing specialist a heart attack 😉

[cekuhnen]

@jakefowler 

There is another thing I noticed

When the loft is done between complete parts (not just half parts) the loft tool works as expected

But also here the curvature graph concern me becuase also there the graph shows a lot of noise where there

should honestly be no noise at all. Does that translate into bad surface aka not perfectly smooth?

Any news from the developers? It is somewhat important for us to know how to read the comb in Fusion when

teaching it.

[jakefowler]

@cekuhnen

No update yet, I'm following up with the geometry team, will let you know when they've analysed this model. Apologies for the wait!

Is this an issue you see with regularity? Or is this model a particularly bad example?

Thanks,

Jake

[cekuhnen]

@jakefowler Hi Jake, from time to time I had few discussions with various AD people about the data quality.

For example what Fusion creates internally and what it produces when it exports as STEP.

I noticed over the years those curvature graph noises but assumed most times it was linked to it being a new

product.

But since Fusion 360 seems to be more accepted in industries around me and we (I) switched from SW/Alias to it

for teaching it became more of a focus of interest now comparing surface quality created in Alias vs Fusion.

This might be only a fine detail but without machines I cannot verify the impact well but still for engineers coming

down the road this should be explainable which is why I prefer some clear answers so I can pass this to the students.

Claas

[cekuhnen]

@Mike.Grau and @jakefowler

It is a month ago that I asked about how usable the surface quality in Fusion is considering how crazy the curvature combs often are.

Did you get any answer from the developer team or does AD not want to answer here?

[jakefowler]

Hi @cekuhnen

Sorry for not getting back to you. We've had some discussion around this, but there wasn't a straightforward answer yet I'm afraid.

In regards to the general question: in most cases clean input curves with give you a good quality output surface; the surface quality is strongly dictated by the curve quality. That being said, this particular case doesn't have any obvious problem with the inputs - this one appears to be an edge case with unexpectedly difficult underlying math.

So the problem here is the surface, not the comb reading; I think the curvature comb is reflecting real issues with the resultant shape.

If you're able to attach/point towards other cases where the comb is not as expected, we can look at these collectively and see if there's a common underlying issue.

Re future enhancements: CV curves are on the roadmap; that should offer much better control for surface creation. As for full surface control, it has been discussed but no firm plans yet AFAIK.

Sorry I can't be of more help for this particular model, other than the previous recommendation of trying separate surface lofts if the combined solid loft behaves unexpectedly. 

Many thanks,

Jake

[cekuhnen]

@jakefowler 

So with a very very basic form the Curvature Combs are fine

Both loft targets are single closed sketches

But then when the sketches / forms get more complex there is not only a noise in the Curvature Comb, along the mirror axis they are also NOT the same!

Now this could be because of how Fusion projects a curve onto the surface. Fusion still fails to offer this as a tool.

When I instead of using projected curves use split edges then the combs are much cleaner 

[jakefowler]

Good catch @cekuhnen - this is an unexpected discrepancy, I'll follow up and see if I can get you an explanation of this.

Thanks,

Jake

[cekuhnen]

@jakefowler 

Yep I think one can forget the projected curve for curvature graphs at the moment.

But this also means 3D curves (projected) are of questionable quality for surfacing tasks then as well!

Fusion really needs a dedicated surface curvature comb tool like SW or Alias has. The current options and workflows are complicated, limited and now obviously also flawed.

With the last screenshot one can at least see that the surface the loft created is somewhat good.

There are some kinks in the comb specifically where the flat top flows into the arced sides.

The combs all show breaks - drastic direction changes.

[cekuhnen]

@jakefowler

Here is another test - a cylinder blends into a solid that is extruded out of 4 curves that are G2.

Works nice and flawless the loft tool and the combs are sweet.