many problems with the shell command

[cekuhnen]

in most cases shelling fails when we try rather organic shapes. here is one example.

also in DM mode the surface offset fails with 2 mm and this is pretty bad.

[subpartner]

Thanks for the files. I will download and take a look.

Cheers.

Jon.

[Jon.Dean]

I have tested on the Mac and the maximum shell is 1.5 mm, any thicker and you will get the error.

I tested also on the PC version and I get the same error.

I noticed that on the end of the model there is a circular face, which gets split when you remove the top half.

I think this is the cause of the problem.

Jon.

[cekuhnen]

if I go into DM I cannot even offset a single shell to 2 mm which is quite unfortunate for Fusion because Alias and Rhino can do it - even when the surface might need then further adjustment to fix self intersecting problems.

[Jon.Dean]

If you look to the inside of the model, on the circular face. You will see that the surface is distorted.

This is why the shell is limited to 1.5 mm.

The outer surface needs to be smoother before invoking the shell command.

Jon.

[cekuhnen]

Fusion cannot even offset in DM the normal shell without the cap.

Alias does just fine.

In addition the Alias surface is pretty simple. The rounded cap is done with the Fusion patch tool.

if the surface is bad then it seems the patch tool code is faulty here.

From my experience the shelling in general in this application is quite hard - simple objects work well,

more organic shapes fail. this harms then the usability of T-Splines when the solid modeler bugs out.

[Jon.Dean]

Take a look at my Video, link shown below.

Fusion 360 Shell

You will see that I have gone back to the orininal solid and deleted the circular patch.

I then re-defined it and stiched the body, now I can create the Shell command with a thickness that you wanted and beyond.

Jon.

[cekuhnen]

Hi Jon,

thank you for testing this, but there are two key differences here.
First the tip is patched flat
Second you shell through the end cap

Now if I offset in DM just the shell -2mm it fails

if I cap the front flat the offset works -2mm

so thats why I think the shell or offset command has to be improved. Alias and Rhino can simply offset the shell just fine without the trick to cap it G0 and stitch it first.

in other files I also noticed that shelling a full object does not work splitting it works or visa verse if a full works the split half do not work with shelling.
if the 3d input data is the same it should not matter which way it is shelled.

[schneik-adsk]

What are you picking to create the shell? The face to open, or the body?

[schneik-adsk]

How was this model created?

[cekuhnen]

modeled in Alias

to make the open shell I just picked the open shell.

[schneik-adsk]

Not sure if this will make a difference but lets start trying to figure out what is going wrong...

In alias can you look at the preferences. I recommend if you are going to send data to Fusion 360 to use the Inventor Construction preset.

• Set Preferences > Construction Options - Construction Presets to Inventor.

The issue is completely localized to the end cap face. Note the discontinuity in curvature:

I created a new cap face cleaning up this end and shell works up to a thickness of  2+ mm.

At some point the radius of curvature at the tip causes the offset to self intersect. This will cause shelling problems.

[cekuhnen]

so here is a Wire file.

you see in Alias all works fine. the offset works fine.

in fusion already the offset not solid shelling fails already for the shell.

now my students data is not the best (they are beginners) but only Fusion bugs out here.

If at least the offset would work we could build the shell by hand.

[deyop · Accepted solution]

Claas,

One distinction between the Alias/Rhino offset and the shelling operation is the accuracy that is held.  Unless the Explicit Control option is unchecked in Alias the offset is very approximate.  In the example you provided the distance varies between 1.42 mm and 1.98.  The offset also doesn't recognize self intersections so it can be successful but cause problems later on. There are times that the tradeoff in accuracy for control layout simplicity can be benificial especially when additional direct modeling is anticipated.   If the Explicit Control is uncheck the results will be much closer but then the controls will have much greater density to ensure the tolerance.

This is not to suggest that the shell shouldn't perform well but it has a goal of accuracy that a surface offset is not trying to achieve.

This design also seems like an excellent opportunity to utilize the Sculpt tools rather than suffer with the continuity issues of the NURBS surface.

Thanks

[cekuhnen]

Suffer from NURBS continuity lol I have to pass that to my students (they do).

I force them through the pain because otherwise they will go with easy and do everything with TS and full stop trying to learn NURBS.

The model here in this thread is also a test surface, the student rebuild it in Alias with a better topology, it is also a closed surface.

His new model is build out of two sides and in Fusion everything worked perfect. (if we know could get his file open as Fusion refuses that).

Your point is right, technically speaking the offset in Alias they way I did was not spot on perfect.

Which is why I can understand why the shell command with this form bugs out.

And I really appreciate the technical aspect of Fusion. For shelling, ribs, and such things, the solid modeling approach is incredible useful

and furthermore a pleasure to work with. The timeline really is a great helper. 

I guess at least the offset command I wish would be able like in Alias to have both a closer and an approximate result.

This way in case the surface is hard to shell, we can build the B surface by hand.

[lindsay.fowler]

I have no idea what you are communicating when you do a quick succession of laps around objects with your mouse. Can you guys please try talking in your videos? 

[chrisplyler]

I'm just curious. How do the real world, non-educational licenses for F360, Rhino and Alias compare in terms of pricing?

[TheCADWhisperer]

Alias Concept

Rhino 6

[TrippyLighting]

@TheCADWhisperer already provided links so I felt I have to provide an opinionated follow-up.

Rhino, for what it does and for its maturity is pretty inexpensive. It has a vast array of tools and the generative/procedural modeling in Grasshopper makes it a very tempting choice.

There are plenty of tutorials online for learning and a good online community.

Alias is a high-end tool for Class-A vehicular and product surfacing. That goes along with a staggering price!

It also has by far the highest learning curve (Class A surfacing is very technical). This youtube channel is very educational and the guy is an industry insider. Not the most polished tutorials, but you'll quickly see that he knows what he is doing.