Sequenced subcomponents for dealing with complex designs

Sequenced subcomponents for dealing with complex designs

GRSnyder
Collaborator Collaborator
520 Views
2 Replies
Message 1 of 3

Sequenced subcomponents for dealing with complex designs

GRSnyder
Collaborator
Collaborator

This is a question about best practices and good model hygiene. Specifically, how to work with long chains of operations on a single body so that they're easy to review, navigate, and modify. Or more bluntly, how to avoid this:

 

Nightmare.png

 

There's a longer explanation and motivation below, but the basic idea is to use subcomponents to represent phases of construction:

 

Separation.png

Here, the subcomponents are design phases of the same body, not independent objects. My question is whether this is a reasonable approach or whether it's an abuse of the components system that's likely to create confusion and problems down the line.

 

More details: When creating complex bodies, Fusion 360's main organizational tool is to group features in the timeline so that they don't take up so much screen real estate. But those timeline groups are unfortunately not so nice to use. The only way to see a group's name is to hover the mouse over the group. Trying to find, e.g., "the place where the main body gets holes that correspond to the holes in the fan mount" is like hunting for socks in somebody else's chest of drawers. Are they in drawer #2? Let's see, no, that's "shirts and ties." #3 perhaps? No, still too early in the timeline...

 

Sketch names are listed in the browser, so you'd think they might be of help for navigation. Selecting a sketch normally highlights its timeline icon. BUT, if the sketch is inside a collapsed timeline group, there is no highlighting, not even for the group. Even editing the sketch does not roll back the timeline to the appropriate group. (I suspect this latter behavior may be a bug, but nevermind.)

 

So, fie on you, timeline groups. You're really only useful for hiding repetitive or mechanical changes that one will never want to revisit again.

 

The usual option: my current approach is to push complex modeling tasks into separate designs. For example, one design might create the general form of a part, and a second design that Insert-derives from it might add embellishments for mounting and molding. Yet a third design might derive and assemble multiple bodies into a final product.

 

I like this better than timeline groups, but it creates folders full of intermediate parts that will never be referenced outside of their role in this multi-stage build process. You also lose all direct representation of time and dependency relationships other than what you can encode into the filenames. Does "Cuff with mounting flanges" come before or after "Cuff with ventilation holes"?

 

This new option I've been experimenting with: sequenced subcomponents instead of separate designs. At each stage of construction, bodies are copied from the previous subcomponent into a new subcomponent, and then the original body or bodies are Removed.

 

Why remove? Mainly so that all subcomponents can remain visible all the time, obviating the need for manual visibility toggling when you return to an earlier stage. It also seems to help to lower the risk of unintentionally modifying the wrong subcomponent/body, which, if you're me, is an ever-present danger. There is only one body (or set of related bodies) extant at any point in the timeline, so... wherever you go, there you are.

 

Pros: this scheme keeps everything in one file, gives a nice sequenced display of construction steps with explanations of what they do, and makes it easy to focus in on any part of the sequence. Just activate the appropriate subcomponent and you see a dedicated timeline with exactly the relevant operations. For example, the section in which you add mounting features is listed right in the browser as "Add mounting features." You can also activate the top-level component and see a complete timeline for the whole shebang. And it's easy to hide all the construction details: just roll up the triangle on Cuff and treat Cuff as an atomic entity.

 

Cons: This scheme requires you to add joints and body copies just to get the sequence defined. You can't simply define Cuff as a rigid group early on because later-added subcomponents won't be grandfathered. You have to joint in the subcomponents as they're created. So that's at least four operations per subcomponent: create component, joint, duplicate body, remove body. And philosophically, it's just not the function components are intended to serve.

 

So far, though, I've been liking it. But what do you think?

0 Likes
521 Views
2 Replies
Replies (2)
Message 2 of 3

davebYYPCU
Consultant
Consultant

No context, Cant see what you are modelling

 

Yes all those cons are unnecessary and are self induced,

how does fusion manage efficient recomputes?

 

 

 

 

Message 3 of 3

GRSnyder
Collaborator
Collaborator

@davebYYPCU wrote: Yes all those cons are unnecessary and are self induced,

how does fusion manage efficient recomputes?

It's not free, that's certainly true. I added an example design above if anyone wants to take a look.

 

I'm not sure why recomputes would be affected. Duplicating bodies is a cheap operation and the face, edge, and vertex IDs seem to carry over. As far as I can tell, everything remains fully associative. Traveling through the timeline does not cause recomputes on its own.

0 Likes