G1 or g2 doesn't work on loft

[kahjun1991]

Hello guys

I'm a industrial designer and now I've been considering to change the 3D tool to Fusion 360

So I've practiced Fusion 360 and i have an issue with loft modeling

[TrippyLighting]

@cekuhnen did you by any chance read any of the previous forum posts ?

Yes, we did try that. How else would you do that anyway ?

There are two main issues with this model :

1.

In my initial attempts the 2nd from left strip I was only able to apply a G1 constraint to the upper/lower boundary in order to even loft this. The models that exhibit that behavior are all attached to some posts so the behavior should not be difficult to observe.
Now that I've changed my workflow (more on that in the 2nd point) the surface quality of the 2 left strips is completely unacceptable and that is a behavior I have observed many times in other models!

It is also observable that there is a horizontal "streak" going through the middle of some of the left most strips, right in the area the middle rail was applied.

2.

There is a workflow issue here for people that come from other CAD applications and that does NOT exist in other CAD applications! The one I am going to discuss first I have explained many times. Fortunately it does not appear in this model, but I have observed this many times in many other models.

The OP created  Sketch20 and projected the lines in that sketch onto the curved surface resulting in Sketch21. The goal was to use the projected curves as loft rails. There are two problems with that.

2.1

In many cases and many posts I have demonstrated that in Fusion 360 projected curves , intersection curves and offset splines exhibit horrible curvature problems. In other applications it is very common to use such projected curves to create further geometry and naturally people try workflows they are used to first. In Fusion 360 surfaces created from such curves exhibit the same curvature problems as the curves they are created with. Not a surprise 😉
But, again, in this case that is no a problem. That, of course presents another challenge, because let's assume a user does actually check the curvature of these curves. They are fine in this case. Why should he assume they aren't in the next project ?

2.2

In this model these curves can only be used as loft profiles but not as rails. That limits a users ability to control U and V directions resulting in the undesirable behavior of the Iso curve analysyis tool  you observed and described on the EE Slack Channel.

But it does allow you to create a loft with Gx constaints ... for the first strip from the right. In that case you can actually still select G1 or G2 for the rails in the loft dialogue ... for the first rib only, however, When you try to loft the 2nd rib you don't get the same selection:

So what might the user try then:
Lets say that user comes from a software that does not require for the boundary surfaces to be stitched together but they can be individual surface bodies such as this:

He then uses the split faces command 10+ times (not overly efficient either if one might say so) and splits the center and right surface to get his loft rails/profiles. He also cannot freely change what to use for rails and profiles.

But things work out just fine for the fits rib loft, but then for the second loft he also does not get the continuity selection he got for the 1st loft:

I have analyzed hundreds and hundreds of Fusion 360 files created by other users. Sometimes you just work with what's in the model and this was where I ended up. A place any other user could easily end up!

Then with that model as I have it right now, you can stitch everything together and lofting works as one would expect, but in one of the posts above is a model where that stitching removed the entire upper and lower horizontal boundary edge and only left the vertical split lines. Complete bummer!

Yes, you might get lucky as you and Jamie did and take the "right" Fusion 360 "enlightened" path the first time, but if you don't you'll end up pretty frustrated, definitely one you hit the 2nd left rail.

 My point her and in the EE slack channel is that there is a lot more work to be done in the surfacing front before Fusion 360 is adapted more by people that come from other applications. 

[richard.hallT2EK9]

Hi to both @kahjun1991  and @TrippyLighting from development, and apologies for the problems you have had to go through here, but we are grateful for you mentioning this issue. I've investigated the matter, and found that the problem seems to have been within Fusion for a long time. There is a significantly tighter tolerance that the functionality initially uses to decide whether the  take-off dropdown for a rail curve should appear in the dialog or not, than compared with the actual construction of the loft surface using rails. Hence, it is very dependent on how accurately the user has modelled the rail data in the vicinity of the profiles, as to whether this option appears in the interface. I have made the code consistent at the coarser tolerance now, and submitted the fix for it. The dropdown now appears for this data (as it should have done). Many thanks again to both of you.

[TrippyLighting]

=Hi @richard.hallT2EK9 thanks for taking a look at this.

Yes, this has been in Fusion 360 for a long time and I am glad that an improvement is on the horizon!

I can see that finetuning some internal tolerances can improve Fusion 360 behavior in certain situation. However, I fail to see in this case how we would model this more accurately. Can you elaborate on that please ? 

[TrippyLighting]

---

@cekuhnen wrote:

Hi there

one practice is to work with target surface

in your case I slice the target surfaces so I can easily loft all

those duplicate / target surfaces can later be removed before stitching all together

did you try that ?

 

the tricky part will later be filleting the sharp edges

---

@cekuhnen thanks for the helpful tip!

Including the upper and lower surfaces in the splitting did indeed not occur to me.

[cekuhnen]

Anyway I still run into surfacing issues

The second last loft to the left is always crap

[cekuhnen]

Yes I would be interested to learn more here too

I have some other very basic cases where loft fails - in many areas actually

I shared the file with Jeff - it is a simple bend handle where I try to custom round the edge a via a loft blend surface

[richard.hallT2EK9]

Hi @TrippyLighting I am not suggesting that you should have to model this example more accurately. You should have been able to obtain the g1/g2 constraints on the loft rails from the word go, but unfortunately there has always been an inconsistency in the software with regards that. You will be able to set the constraints via the aforementioned workflow and selection, once you get your hands on the release version with the fix in it. Thanks.

[cekuhnen]

@richard.hallT2EK9 

Hi Richard instead of hard coding - it might be a good idea to like Alias offer a preference panel where the user can set the tolerance values.

[richard.hallT2EK9]

Hi @cekuhnen, yes, I think that there are probably quite a few areas of Fusion modelling that would benefit from the user having some control of the geometric tolerance(s), but there are of course some fundamental limitations on the flexibility of such inputs, imposed by the current state of the modelling kernel.

However, an overarching tolerance management system would be a good step forwards in that respect.

Just out of interest, which pieces of the Fusion modelling functionality would you like to see the user being able to decide at what tolerance the operation should be performed at, which is not available today ?

JFYI, I worked on Delcam PowerSHAPE for nearly twenty years, and there was an options page where linear, angular tolerances for modelling operations could be set. There was some, but marginal overlap with the topological tolerances. The main geometric tolerance could be interactively changed from the main UI, so effectively set per operation if that was what the user required.