Creating a Recessed Snug-Fit Cavity for a Complex STL

[johnswork15]

Hi all!

So, I'm currently working on a design for a set of board game boxes for the board game Villainous. I have these complex STLs that I have generated from scanning the board game villain movers which have fairly complex geometries. For example, the Captain Hook mover looks like so:

For obvious reasons, I can't attach the STL models but if a simplified replica of the model would be helpful in discussing this, let me know and I can try to make something. I'm not the fastest with fusion though so it'll take me a minute.

I want to create snug-fit recessed cavities in the box that can hold a mover (feel free to correct me on the terminology used here, I'm not an engineer and don't work in this space so there may be a better term I can use). Ideally, I want the mover to

• Sit snugly in the recess
• Not have any part of the mover be above the top of the box
• Have the mover be hugged on all sides by the box excluding the front that allows it to be pulled out of the box

For each mover, I have a high triangle count version, and a 10k triangle version. I then use the organic mode convert to solid body to convert each mover into a solid body. This works pretty well and comparing it to the high triangle count version, it's fairly accurate if a little inaccurate in a couple of spots that aren't that important for my purposes. So for each of the attempts I'm going to describe, I do have a solid body to work with and not just a mesh.

Here's what I have tried so far that I can remember

1. I attempted to create silhouettes of the model (using the project function in the sketch mode) from the top perspective, left perspective, and bottom perspective to attempt to create a punch. I did this by extruding the top silhouette to create the base punch and then cutting out the correct shape on the bottom of the punch by extruding the side silhouette with a line that maximizes it's length while still being coincident with the mover's left perspective silhouette. This nearly works except it's an extremely tedious process and I ideally didn't want to have to do this for 12 different movers again and again. It also does not hug the bottom of the mover and instead only supports it at the widest points. This isn't the worst but doesn't really do what I want it to.
2. I have tried using the boundary fill but I don't really understand how to use that correctly to do something like this so it didn't go anywhere
3. I tried creating a silhouette and then projecting that to the surface of the solid body version of the model to create a 3D sketch that I could theoretically then use to extrude a body that would slot into the top of the model and allow me to join the that body together with the solid body model to produce a punch. 

I really want to do this and I will write a script if that's the only way to do it (I just don't know where to start there) but I was really hoping there was something built-in-ish.

[laughingcreek · Accepted solution]

it sounds like yo might benefit from using a solid sweep (as oppose to a profile sweep) which was very recently added to fusion.

but I suspect you will have a difficult time doing this in fusion.  fusion doesn't really handle high count meshes, or meshes in general very well.  If you want to work with mesh files with any regularity it would benefit you to learn a software that is specifically for manipulating mesh files.  my current choice is blender, but there are other possibilities out there.

attaching a sample model would be helpful.  It's unclear what the "obvious reason" for not being able to attach scan data you created is.  (trying to sell your scan data of copyrighted items would be a different story).

[TrippyLighting]

It doesn't make sense to reduce the high tri-count mesh to a 10k version and then convert that.

The organic conversion first re-meshes the hires mesh into a quad mesh and then converts that into a T-spline and then a solid (or surface). A higher triangle count provides the re-mesher with more data.

There is no obvious reason why you could not share one of those model scans.

[johnswork15]

@laughingcreek Thanks so much for telling me about the solid sweep! I didn't know that existed. Like I said I have managed to convert it into a solid so I can try with that. I also don't typically work with meshes so I didn't want to make the jump to another software suite (that also doesn't run on my mac) unless I absolutely had to. I also couldn't find any resources on doing this task in any other software either so I was near giving up and doing some vacuum forming and sculpting.

As for the reason that I can't share the models @TrippyLighting and @laughingcreek, it is the fact that the boxed game materials are copyrighted and I tend to avoid the risk with anything involving that even if it is unlikely that someone will make a copyright claim on some random Autodesk forums post. Especially since I am not a lawyer and I don't understand the IP laws well enough to know the level of trouble I would or would not get into if I did post copyrighted materials. Because as far as I can tell, even just making the scan can technically be considered breaking the copyright but I'm not sure.

Finally, @TrippyLighting, I tried with a high triangle count and the conversion would never complete and instead my computer would just freeze. If there's something that I am missing there, I'd be glad to know otherwise.

[TrippyLighting]

---

@johnswork15 wrote:

... I also don't typically work with meshes so I didn't want to make the jump to another software suite (that also doesn't run on my mac) 

---

Blender runs perfectly fine on a mac, both on Intel and Apple silicon.

---

@johnswork15 wrote:

Finally, @TrippyLighting, I tried with a high triangle count and the conversion would never complete and instead my computer would just freeze. If there's something that I am missing there, I'd be glad to know otherwise.

---

The conversion code in Fusion is pretty slow and does not provide a lot of options. I can recommend another external software for re-meshing into a quad mesh. I've done this a lot with InstantMeshes.

I created a tutorial for converting meshes with it a few years ago:

[johnswork15]

It may run on mac in general but it does not run on my Mac. I'd probably
need to download an older version and that sounds like more trouble than
it's worth.

[TrippyLighting]

Please make @laughingcreek suggestion of the solid sweep as the answer if it works.

[johnswork15]

Ok, @laughingcreek, I tried the solid sweep and it may have some constraints on the solid body it can sweep because when I tried to sweep the simple cube, it worked fine but when I tried to sweep the mover, it failed. I'm going to create an example to demo the issue.

[johnswork15]

Here's example with a different STL under a public domain license that has a similar type of curves and seems to be made up of splines like the solid bodies that I have.

[TrippyLighting]

---

@johnswork15 wrote:

... that has a similar type of curves and seems to be made up of splines like the solid bodies that I have....

---

the object doesn't "have curves". The Organic conversion remeshes the scanned results into a quad-mesh. The quad mesh is then converted into a T-Spline and then the T-Spline into a patchwork of NURBS surfaces that are stitched together into a BRep. 

The end result is very complex geometry, and I am not surprised that it fails.
I've had some success using the solid-sweep with a much simpler hand-modeled T-Spline object.

[johnswork15]

@TrippyLighting, I'm not familiar with what is going on behind the scenes so I was not speaking about that. I was saying that they are made up of similar curvatures and thus hopefully would be on similar levels of complexity for the purposes of illustrating the problem I was having.

It does not surprise me that the level of complexity is likely the source of the issues that I am having. It is why I am here asking for possible workflows to perform this kind of task.

I am fairly comfortable programming and could possibly implement an algorithm to do this myself but if I was to do this, I want to verify that this isn't something I could easily do already. I'm also not averse to simpler solutions that will get close to the solution I am looking for. I'm already attempting to just build up clay and using a vacuum former to approximate a decent fitting solution so anything that can do approximately as well as that solution but built into my model is an improvement.

[johnswork15]

Also, side note, I think that in similar contexts the solution provided by laughingcreek would be the solution so should I accept that as the solution even if it doesn't work for me?