Surface modeling workflow for complex organic shape

[brianvonlehe]

Hi,

Maybe it's not "complex" but it's complex to me 😉 I'm looking for help in choosing the right surfacing workflow for this organic model. Unfortunately I can't share the whole thing, so I trimmed it down and removed the history. However, beyond where the loft is trimmed it just continues in a relatively straightforward series of G2 profiles, so this model includes all the relevant parts.

This part is made from a sweep, loft, patch, and mirror. (plus an extrude & trim on the centerline to get the curvature continuity across the mirror). Since the sweep only makes up a portion of the first profile for the loft, I'm only able to get direction continuity in that corner of the loft. I would like to keep using the sweep as that's a particular area where I need particularly clean and repeatable geometry.

There's some lumpy curvature in the middle of the patch but that's not a problem because in the actual model I can add more G2 rails that cross that area. 

Honestly, it's probably OK for my purposes as is, as my primary need is clean surface geometry for the downstream operations of creating molds for the part. I'm pretty happy with the outcome of the Thicken operation. But I thought I'd throw it out to the surfacing geniuses out there and see if there's a better strategy for producing this shape before I spend multiple days dialing in this method.

(FWIW, to other readers: I already use T-splines ("Form") for prototyping and development of the product, but they're just not accurate and clean enough for building tooling so I need to do this with proper NURBS surfaces)

Cheers & thanks

[brianvonlehe]

Ok after stepping back from this it does seem like I need to break the loft into multiple parts. I think I'm trying to do too much with that one surface. A loft that matches the end of the sweep and continues along the edge, and then another one or two lofts to get over to the centerline. Still would deeply appreciate it if the experts can make recommendations. 

[TrippyLighting · Accepted solution]

---

@brianvonlehe wrote:

... it does seem like I need to break the loft into multiple parts. I think I'm trying to do too much with that one surface. ...

---

Correct. But it is hard to say without having a timeline and being able to take a look at how this geometry was actually created.

[brianvonlehe]

So the loft has one profile that is bent and one that is planar. On the bent profile which includes the end of the sweep, I'm using curvature continuity to connect the curve to the corner of the sweep.

I just tried to use both the sweep end edge and the sketch curve to create a single loft profile, but the loft won't recognize that as one profile, so therein lies the problem. If I use projected geometry from the end of the sweep for the loft profile, the loft will work because the bent profile is all from one sketch curve, a combination of projected and normal geometry. But, this also makes it impossible to keep continuity between the sweep surface and the loft since it's just connecting to a curve. 

So I guess I do have to break the loft into sections that align better. This gets more fun the better I understand it. Thanks!

[TrippyLighting]

IF you can share the complete model of the saddle incl timeline I can perhaps provide some guidelines.

Something to keep in mind when surfacing is that trying to combine areas with high curvature such as the ridges in the middle with low curvature (the rest o the top surface) is usually not going to be very successful.

[brianvonlehe]

Patents for this kind of thing are probably, um, patently ridiculous, but I still can't publish. Thanks though!

Am I understanding correctly that I should try to keep each individual surface limited to roughly one curvature? Like run an individual loft or sweep along the sharper areas that match with a separate surface for the gentle areas?

I've been trying put off modeling it this way for as long as possible, but alas the time has caught up 😉

[TrippyLighting]

Before going any further it would probably help you to familiarize yourself with a few basics.

Here is a link to the Autodesk Alias Theory Builders. While Alias is a different software the general theory applies to any work with spline-based surfaces.

I see a number of areas of he surfaces in your design, which do not maintain good curvature continuity across the boundary between surfaces. Not only is this visible in the zebra stripe analysis (which should always set to "high), but also in a render with high gloss material such as the black glossy paint.

While you did use CV Splines many of your curves are multi-span curves because of the number of control points exceed the number of degrees+1 of the curve. The points are also pretty unequally spaced in some of the curves.

When you encounter the need to go to multi-span and to very unequally spaced control points it's time to think about breaking the problem into smaller patches.

This youtube channel shows many techniques.

One thing you'll notice is that he talks about "slabs". A slab is a large main surface. So the workflow is to creat these large surfaces and then add smaller surfaces such as fillets and blends.

While Fusion 360 does not nearly provide you with the same control over surfaces as Alias, you can build pretty nice surfaces in Fusion 360 with a bit of experimentation and patience.

[brianvonlehe]

Great, thank you for the pointers on directions to go. Control points = curve degree +1 is super helpful and alone clarifies a lot of workflow decisions about when to break up into smaller surfaces. Maybe I've been stuck in too many fusion-specific tutorials that are pretty light on surfacing. 

I'll take another stab and try to shoot for a "B+" surface. If I get stuck maybe I'll create another model that has similar curvature but not the shape or the dimension of the actual product and post that. 

[Anonymous]

Hi there,

can you please re-upload youtube channel link for my preference?

Thank you so much!

[TrippyLighting]

The link in my post works just fine.