Patch Layout

Discussion created by Kevin De Smet on Mar 23, 2010
Latest reply on Mar 30, 2010 by Kevin De Smet

Like • Show 0 Likes0
Comment • 8

I want to start a discussion about an important aspect of surfacing: patch layouts.

This seems to always raise varied opinions, as it should, and it is a topic that I personally feel fuzzy about. Fuzzy being a nice way of saying... absolutely stumped.

When are patch layouts beneficial, and why?

When do patch layouts become a nuisance?

I think it is heavily dependant on the model at hand, and the construction techniques for that particular model to get from start to finish.

Somehow though, when I see a big surface with not a lot of topology, it just feels...wrong. I want the rounded off bits to be separate surfaces curvature continuos, I want the straight bits to be the simplest geometrically as they can be to represent the shape it needs to be. I do not want this big mental surface pulling all the weight! What is wrong with me though? Because by the same token, all I feel I am doing is getting more boundaries and in turn more discontinuities to worry about!

To sum up a few pros and cons, the way I see it (do not know if I am right about these) on patch layouts:

Pros

Simpler, but multiple surfaces mean less tension on the surfaces resulting in less pinches, bumps, inflections or dimples.
Makes editing the model easier and more cordoned off, making a G2 round bigger will not disturb the rest of the model.
Aids the construction of a model by providing more framework to work with, during curve and especially surface creation.

Cons

More surfaces means more boundary edges, which in turn means harder to maintain continuity.
Seems like it only really starts to become highly beneficial in high end class-a surfacing for auto or aero with higher order surfaces.

Anybody else feel the dilemma sometimes?

We could get some insightful dialogue going

Outcomes

Charles Culp Mar 23, 2010 5:13 PM

If I can do it in one surface, I will.

I think my big caveat for that is if the surface is made of open or closed loops. That will make a big difference, as building off of those it is easier if they are all open or all closed. So if you are modeling a tube (like a handle, or an aero wing, etc), then it is important to decide if each rib is a closed loop, or 2+ open edges.

The only time I can think of when I will purposefully add new patches is when the surface changes direction over 90° from the normal of the existing surface. Once that happens there are bound to be issues somewhere, and it is best to make it multiple surfaces. Also, the obvious abrupt changes in curvature or non-curvy shapes.

Actions

Kevin De Smet Mar 25, 2010 2:33 AM

Here I have an example of an interior of a mould cavity I did some time ago now, and I did it like is shown on the left, with the design intent being lines and arcs. So there really was no way around it, strictly technically speaking.

However doing a simple fit spline, and lofting that, I get the result on the right. Which has significantly less topology.

Both results sure look similar enough to each other, but the one on the left is unquestionably more work.

This is kind of a linear "loft" in one axis so perhaps here, the big surfaces are fine?

Still my intuition tells me that the layout on the left is just... better, period.

Any insight appreciated!

Attachments

patch_layouts_mouldcavity.x_t.zip193.2 KB

Actions

Charles Culp @ Kevin De Smet on Mar 24, 2010 9:45 AM

Your "straight" line edges will not be perfectly straight with a fit spline. Within manufaturing tolerance of straight? maybe... But if that was a hand created spline, and not a fit spline, it would be tough to edit it and keep the straight edges perfectly straight.

I've always been a fan of "if the model tolerances are tighter than the manufacturing tolerances than it is good enough", but there are plenty of perfectionists out there who would disagree with me. Then again I work mostly with aluminum castings.

Actions

Kevin De Smet @ Charles Culp on Mar 28, 2010 2:31 PM

Good point.

There is also the modeling strategy of making a "theoretical model" first, just main slabs without any transitions. And then model the transitions. That will indirectly create a comprehensive patch layout.

But when is this a good way to model, and when is it better to build with smoothness right into your primary sweeps, lofts and boundaries?

No doubt that building it in directly is faster, but can it run you into a wall later on?

Actions

Anna Wood @ Kevin De Smet on Mar 28, 2010 10:38 PM

When I learned surfacing back in the day, manual drafting boards and early wireframe and surface CAD systems, we always developed surfaces using a primary, secondary, tertiary, etc techniques.

I actually have a hard time envisioning surfacing with the newer techniques of the CAD systems. Something that would be easier for me if I had more practice beyond the basics with the SolidWorks surfacing tools.

FWIW,

Anna

Actions

Matt Lombard Mar 28, 2010 9:34 PM

Yeah, I always think it is appropriate to think about this kind of stuff. My rule of thumb is to make as much surface area as a feature is comfortable with, but if you gotta break it up you gotta break it up. All other things being equal, it's better to have fewer edges in a model. Still, sometimes I'll make a single feature, then blast a hole in it and use Fill to patch it over because the single feature doesn't do a particular area very well. I agree its very model specific.

Actions

Jerry Steiger Mar 29, 2010 1:24 PM

Kevin,

The only rule I have about this is that there are no rules. The unsatisfcatory answer is "it all depends". I wish I had some meta-rule that would help me figure out when I should make something with one surface or three, but I usually end up learning which way is best by trying all of the possibilities I can think of. And then someone else usually comes up with a better way of doing it. The closest I can come to a rule is that when the "character" of the surface changes, you should probably make it a separate surface. But sometimes you can do a single surface that has regions with multiple characters that looks better, or at least as good, as multiple surfaces.

Jerry Steiger

Actions

Kevin De Smet Mar 30, 2010 4:21 AM

Looking at my attached parasolid example file again, I do notice that the one with less edges still keeps 4 sided-ness.

Maybe that's at least one good factor to gauge when you can, and when you can't, make everything just one big surface.

Actions

8 Replies

Charles Culp Mar 23, 2010 5:13 PM
If I can do it in one surface, I will.

I think my big caveat for that is if the surface is made of open or closed loops. That will make a big difference, as building off of those it is easier if they are all open or all closed. So if you are modeling a tube (like a handle, or an aero wing, etc), then it is important to decide if each rib is a closed loop, or 2+ open edges.

The only time I can think of when I will purposefully add new patches is when the surface changes direction over 90° from the normal of the existing surface. Once that happens there are bound to be issues somewhere, and it is best to make it multiple surfaces. Also, the obvious abrupt changes in curvature or non-curvy shapes.
Like • Show 0 Likes0
Actions
Kevin De Smet Mar 25, 2010 2:33 AM
Here I have an example of an interior of a mould cavity I did some time ago now, and I did it like is shown on the left, with the design intent being lines and arcs. So there really was no way around it, strictly technically speaking.

However doing a simple fit spline, and lofting that, I get the result on the right. Which has significantly less topology.

Both results sure look similar enough to each other, but the one on the left is unquestionably more work.

This is kind of a linear "loft" in one axis so perhaps here, the big surfaces are fine?
Still my intuition tells me that the layout on the left is just... better, period.

Any insight appreciated!
Attachments
- patch_layouts_mouldcavity.x_t.zip193.2 KB
Like • Show 0 Likes0
Actions
- Charles Culp @ Kevin De Smet on Mar 24, 2010 9:45 AM
  Your "straight" line edges will not be perfectly straight with a fit spline. Within manufaturing tolerance of straight? maybe... But if that was a hand created spline, and not a fit spline, it would be tough to edit it and keep the straight edges perfectly straight.
  
  I've always been a fan of "if the model tolerances are tighter than the manufacturing tolerances than it is good enough", but there are plenty of perfectionists out there who would disagree with me. Then again I work mostly with aluminum castings.
  Like • Show 0 Likes0
  Actions
  - Kevin De Smet @ Charles Culp on Mar 28, 2010 2:31 PM
    Good point.
    
    There is also the modeling strategy of making a "theoretical model" first, just main slabs without any transitions. And then model the transitions. That will indirectly create a comprehensive patch layout.
    
    But when is this a good way to model, and when is it better to build with smoothness right into your primary sweeps, lofts and boundaries?
    No doubt that building it in directly is faster, but can it run you into a wall later on?
    Like • Show 0 Likes0
    Actions
    - Anna Wood @ Kevin De Smet on Mar 28, 2010 10:38 PM
      When I learned surfacing back in the day, manual drafting boards and early wireframe and surface CAD systems, we always developed surfaces using a primary, secondary, tertiary, etc techniques.
      
      I actually have a hard time envisioning surfacing with the newer techniques of the CAD systems. Something that would be easier for me if I had more practice beyond the basics with the SolidWorks surfacing tools.
      
      FWIW,
      
      Anna
      Like • Show 0 Likes0
      Actions
Matt Lombard Mar 28, 2010 9:34 PM
Yeah, I always think it is appropriate to think about this kind of stuff. My rule of thumb is to make as much surface area as a feature is comfortable with, but if you gotta break it up you gotta break it up. All other things being equal, it's better to have fewer edges in a model. Still, sometimes I'll make a single feature, then blast a hole in it and use Fill to patch it over because the single feature doesn't do a particular area very well. I agree its very model specific.
Like • Show 0 Likes0
Actions
Jerry Steiger Mar 29, 2010 1:24 PM
Kevin,

The only rule I have about this is that there are no rules. The unsatisfcatory answer is "it all depends". I wish I had some meta-rule that would help me figure out when I should make something with one surface or three, but I usually end up learning which way is best by trying all of the possibilities I can think of. And then someone else usually comes up with a better way of doing it. The closest I can come to a rule is that when the "character" of the surface changes, you should probably make it a separate surface. But sometimes you can do a single surface that has regions with multiple characters that looks better, or at least as good, as multiple surfaces.

Jerry Steiger
Like • Show 0 Likes0
Actions
Kevin De Smet Mar 30, 2010 4:21 AM
Looking at my attached parasolid example file again, I do notice that the one with less edges still keeps 4 sided-ness.

Maybe that's at least one good factor to gauge when you can, and when you can't, make everything just one big surface.
Like • Show 0 Likes0
Actions

Patch Layout

Outcomes

Attachments

Related Content

Recommended Content