Creating a marble rail

[mroek]

I have been thinking about a new marble machine design, and today I thought I'd just play a bit with some basics as a learning exercise. I'm sure there are many ways to accomplish something like this, but I tried the following:

1) I created a 3D spline from scratch, along which I wanted my rail to follow

2) I then created a plane at the end of this spline, and sketched a circular U-profile for the rail itself

3) Then I used a neat trick which @PhilProcarioJr showed in another thread to create an offset 3D sketch of my spline using surfaces

4) Now I could sweep the profile and use the offset 3D sketch as a rail to prevent it from twisting uncontrollably

This turned out OK, but then I wanted to try adding some banking in one of the turns, after all the marble shouldn't fall off. To do that I tried a few things, but I ended up splitting the track with two sketches (should probably have used planes along path instead). Then I could use "Offset faces" on the outer/upper face in the turn, to "bank" the rail. Then I combined the bodies again, but I ended up with visible seams at the splits, even though I was just combining exactly where the split was. Which doesn't feel natural, combining what was previously split should be possible to heal?

I can think of other ways to create a rail like this, but suggestions for best practises are welcome.

I have attached my test project, in case anyone want to take a look.

[PhilProcarioJr]

@mroek

Take a look at the file I posted for what I think fixes your problem.

[mroek]

@PhilProcarioJr

I was thinking about the seams both on the inside an the outside of the track itself, which is left there after combining the bodies again. They may not represent a real problem per se, it's just that since at those seams, the bodies that was split hasn't been modified, so I don't quite understand why there are seams left. Take these, for instance:

The outside surface here is continuous and exactly as it was before the split, yet the combine operation leaves these seams where the split was previously.

[PhilProcarioJr]

@mroek

If you check zebra stripes that area is CC and you can see no seam other then model edges. But if you really want to get rid of that what you should have done was copy the model, cut the copy, make your changes and stitch the lip to the original. One min.....

[PhilProcarioJr]

@mroek

See if this is what you want?

[mroek]

Yes, and that's why I don't really understand why the seam is even there. But as I said, it isn't really a problem.

I also tried a different method for creating the lip, by sketching a profile on a plane along path, and then sweeping that profile, using the edges of the rail as the path and the guide. That seems to work OK, but the result is segmented, and has seams (but curvature seems OK): 

File attached.

[mroek]

@PhilProcarioJr

Yes, in your latest fix you removed the seams, but it took quite a few operations to get there. I think it might be better to create that lip in some other way, but thanks anyway, because I'm learning new stuff every time you post something.  🙂

[PhilProcarioJr]

@mroek

The reason the seam is there is because you have a sharp edge terminating into a patch NURBS surface...in short (without going into complexity) mathematically it has to have that after you changed the surface the way you did.

The reason it took so many steps is because Fusion Surfacing tools are VERY lacking.....yet another reason I ditch the history a lot of times.

When trying to do this the normal ways which would have been a lot less steps Fusion trashed the surfaces for no reason...

But again if you had just split the top surface at each end it would have solved a lot of this. I didn't do this because I was working with the method you started. 

[mroek]

@PhilProcarioJr

I actually started out splitting just the top surface, but then I wasn't able to offset it, so that's why I splitted the whole body instead.

[PhilProcarioJr]

@mroek

The problem is your diving into the weakest part of Fusion as far as modeling goes. So it becomes less important in the amount of steps and more about the quality of the surface you end up with. Fusion is capable of doing almost anything...can it do it as well as other more mature CAD apps no...so for now if you want quality surfaces...more steps.

[mroek]

@PhilProcarioJr

Ok, I get it. Did you see the other lip I created (with a sweep)? Is there a way to avoid those segments/seams, or is the method you showed using surfaces the only way to get a good result without all those visible seams?

[PhilProcarioJr]

@mroek

You can't easily get rid of the segmented areas using sweep because of the way you created the initial geometry.

What I mean by that is when you use that trick I showed you Fusion (for some unknown reason breaks the edge up with segments). Those breaks are what is causing segments in your loft.

Over building the geometry is the best way to get good surface quality. There is another way and if I get time I will show you.

[michallach81]

Hi Øyvind ( @mroek), the first mistake you've made is unnecessary Extend. Phils ( @PhilProcarioJr) trick was about 3d offset, in your case you don't need that. The relation between two edges of extruded surface is same as in your pick. Direction to control sweep was defined by extrusion. Segments did show up because the data for sweep command were "secondhand", degenerated.

Look at my way to achieve right results:

There is a small bug that I didn't notice. There was a problem with a profile for a sweep from the bottom. I've moved one line and now it's fine, the file is attached.

[mroek]

@michallach81

Thanks for looking! I'll study your screencast later, when I have the time. I guess there are many ways to do stuff, and I also did it in another way which worked better, where I swept to a full pipe, and then created a surface (from the initial 3D spline) to split said pipe and remove the top part (in other words, no guide rail in the sweep). After doing that, I could split just the top surface of the edge where I wanted a lip, and could offset the splitted face section correctly.

[mroek]

@michallach81

I've now looked at your sceencast, and you did something I also thought about, but didn't really test. It is a bit tedious if there are many such banked sections in a rail, but the result is good.

However, my latest method is easier, and seems to give an equally good result. I've attached a file showing that method too.