How to work with irregular and organic .stl files.

[TrippyLighting]

I created this tutorial based on a conversation in a recent thread:

[lichtzeichenanlage]

Thx for sharing this. Instant meshes is new to me and I guess I'll use it in the near future. 

[TrippyLighting]

Just to be clear, InstantMeshes is not the only software that allows to re-mesh. A number of Sculpting and Sub-D modeling tools also do this. ZBrush and 3Dcoat both have builtin remeshers.

SolidThinking Evolve has Polynurbs that can also do this and has shown to be very fast when I tested it a while ago.

Then there is the technology developed by Cyborg3D that is also used in Solid Works Powersurfaces and available as a plugin to MoDo.

However, for many people these tools are out of reach for occasional use and InstantMeshes is the only free and open source solution I am aware of.

[JDMather]

This thread should be pinned.

[lucasliraAU]

Congratulations on the tutorial! 

Thank you for sharing your knowledge!

I'm always working with projects that are involving knits in Fusion 360 this tutorial will help a lot in coming work!

Greetings!

[Anonymous]

Once you have the .stl into the final format; is it possible to then cut it on a cnc?  For example; I would be interested in cutting that Head Sculpture on a 4th rotational axis CNC.

[TrippyLighting]

.stl files can be machined in Fusion 360 directly without using the workflow above. That workflow only makes sense if you want to modify the geometry with solid modeling tools.

But, if you do follow that workflow above, then you can also machine this in Fusion 360 like any other solid model.

[Anonymous]

---

@TrippyLighting wrote:

.stl files can be machined in Fusion 360 directly without using the workflow above. That workflow only makes sense if you want to modify the geometry with solid modeling tools.

But, if you do follow that workflow above, then you can also machine this in Fusion 360 like any other solid model.

---

I feel like I am very behind on this topic...

when last I tried, admittedly a couple of years ago, you could not generate a CAM toolpath from a .stl.

is this now possible to do?

Are rotary toolpaths from .stl supported as well?

I have been creating artifacts (see: https://www.malcolmstanley.com/physicalworks/) using XZA machining using Deskproto as I thought Fusion 360 would not support toolpath for XZA or XYZA directly from a .stl input.... 

if Fusion will now do this that would be pretty exciting...

[TrippyLighting]

@Anonymous  I "know" that .stl files can be machined directly but that statement exhausts my knowledge about the CAM topic 😉

Let me see if I can get some of the Expert Elite's form the CAM forum @HughesTooling @LibertyMachine to chime in.

[seth.madore]

Thanks Trippy for tagging me in this!

Yes, there are some toolpaths that WILL recognize .stl files. As far as I know, all the 3D Toolpaths can be used for programming from stl files.

You CANNOT use any of the .stl model for alignment purposes during Setup. You will need to construct sketches or primitives to represent Z&X orientation of the model

You CANNOT use any of the 2D toolpaths with an stl model. So, 2D Contour to cut around the part is out. However, any sketches you can create are fair game.

[HughesTooling]

The quick answer is it should work as long as the STL is closed, last time I tried it would not recognize an open STL. I did try a very complex STL from a discussion on the Rhino forum and Fusion did not recognize it as a good closed model, seem to remember it had some internal voids. I don't have a lot of experience working with meshes so gave up on that one but as long as you have a clean closed mesh it should work.

As for rotary, I think the biggest problem will be positioning in Fusion depending on the shape. If you can position accurately in another program so the centre of rotation is what you need in Fusion setup will be a lot easier.

Mark

[Anonymous]

it sounds like Deskproto is still much more plug and play on this.

Thanks guys for the perspective!

[StephenCim-001]

Thanks Peter, exactly what I needed, I can now go back and redo some designs that did not come out the way I wanted.:)

[Anonymous]

Thank you for this video. Very informative. I have a question about triangles vs quads. When I tried inserting a mesh into F360 it complain about the number of triangles when I tried to convert. My question is was F360 more irritated with the number or the shape. In essence would I have had more success converting them into quads in a package like Instant Meshes rather than just trying to reduce the number of triangles

[TrippyLighting]

The difference between a quad mesh and a triangulated mesh is not so easily explained, but let me try ...

It depends how that quad mesh is interpreted. You cannot convert a quad mesh into a BRep directly in Fusion 360, because a quad polygon is not necessarily flat.

A triangle on the other hand is always flat. One will need to understand what the difference is between CAD data and a  mesh. CAD data works with analytical and NURBS surfaces in form of a BRep. These are mathematically precise definitions of geometry that are theoretically free of resolution. CAD data also has topology in a mathematical sense.

A triangulated mesh has a finite resolution and no topology. When it is converted into a BRep every triangular facet is converted into a flat mathematically precise surface. That has a few implications. Firstly it takes more data to store than a mesh. Secondly most of the operations in the Model (Solid) work space have a lot more data to process.

Imagine you create a cube and then model a hole in that cube. Fusion 360 only has to calculate the intersection of a few surfaces. However when you "drill"a hole into one of these highly faceted BRep bodies it has to potentially intersect with hundreds of surfaces. That takes a lot longer to process and is the reason why the polygon limit exists Fusion 360. There are other issues that can result from this.

Near tangency and near coincidence are geometry problems that can easily prevent a modeling operation from completing. A split body operation or a combine/join for example can easily fail because the split face is near coincident with one of the facets edges and would create geometry that the geometric modeling kernel cannot digest mathematically.

A quad mesh on the other hand can be interpreted similarly to a NURBS surface. If you follow the link to NURBS I provided above you'll find that the the shape of a NURBS surface is controlled by a mesh (yep, a quad mesh!) of control points.

In Sub-D modeling a quad control mesh is used to control the shape of a surface subdivided with Catmul-Clark or Open SubDiv . Sub-D math and NURBS math is different but similar enough so that methods such as T-Splines exist to convert a Sub-D control (quad) mesh into a NURBS surface. There are other technologies that do this as well, e.g. PolyNurbs.

Thus for these irregular meshes the idea is to semi automatically re-mesh them for example with Instant Meshes or re-topologize them for example with Topogun or RetopoFLow2.  That is an intermediate step to get back that smooth, mathematically  precise and resolution free surface and in the case of manual re-topology the topology.

Based on the above it is easy to see that when a model is converted from a BRep or NURBS surface model into a .stl mesh a substantial amount of the information contained in the original geometry is lost. 

[jeff_strater]

That's a very nice explanation, @TrippyLighting .  Thanks for writing that up and adding it to this already very useful thread.

We should submit another AU class in 2019 together on this whole topic of mesh, BRep, Sub-D, etc in Fusion and other tools.  A sure winner!

[Anonymous]

Wow, even though it will take me a couple more readings (and then a few more after that) to digest all that information, I can almost grasp the herculean task that managing a native model F360 is confronted with, never mind the issues that  a conversion presents. I thank use for providing that context. This forum rocks.

[Anonymous]

Trippy, that is an awesome explanation of a very complicated subject and I didn't understand a word you said. 

One of the interesting things about this forum is the wide variety of knowledge of the participants here. Some people here are well versed in the various kernels used by CAD/CAM systems, the mathematical and representational models they use, and the limits those models place upon the actions the software may easily perform. Other people here... aren't.

Being in the latter category, non-expert in the basics underlying the CAD/CAM system I am using (well, actually, not using so much right now, long story) I fall back on my training as a complex system product manager and start to look at it from a use case / agile user story perspective. This thread started, i think, in an attempt to answer a user story with a tutorial. 

The user story I might phrase this way: 

As a user of Fusion 360 I want to import an arbitrary .stl file, 
work with it using the standard Fusion 360 workspace tools (i.e. drill holes, divide bodies, etc) 
and generate 3d printing jobs or CAM toolpaths from the resulting geometry.

The tutorial in the original post showed a number of techniques that make a ton of sense if you understand the aforementioned mathematical models, differences in types of surfaces and solids, 

but make no sense at all if you do not, i.e. to a user not expert in the mathematical underpinnings, it looks like a lot of arbitrary actions to achieve an uncertain goal of poorly understood value.

In response, I might edit the user story this way: 

As a user of Fusion 360 I want to import an arbitrary .stl file, 
work with it using the standard Fusion 360 workspace tools (i.e. drill holes, divide bodies, etc) 
and generate 3d printing jobs or CAM toolpaths from the resulting geometry,
without loss of resolution or fidelity from the imported model,
being guided though any transformations of the underlying geometry required to enable the actions I wish to undertake, 
utilizing software and tools workflows that are wholly encompassed by the Fusion platform.

In other words: 

 - I need help to understand the steps required to do what i want to do

 - My goal is to identify a geometry, manipulate it, and render it physically

 - complicated workflows using third party tools are enabling, but confusing, so please encapsulate them: if I need to use a 3rd party tool to convert 'something' to 'something but different under the hood' when I try to take an action, please give me a button that says "to (drill that hole, split that surface, _____) you need to convert 'something' to 'something but different under the hood', press OK to proceed... 

 - if when I press OK Fusion then proceeds to spin up a 3rd party package, do the thing, and quit out with the result, I am fine with that. I'm sure it is not that simple, but the ask/desire is to make these processes look more like an integrated workflow and not a concession to the limits of the product where we have to surrender the Fusion application context in order to accomplish some enabling task we do not understand and could not explain.

 - power users will hate all of this, so make a switch somewhere that allows us to turn it off and on: I'll bet hardly anyone turns it off if the user experience is properly designed. 

The user story above is poorly written because I am in a hurry, and I would not submit it to an engineering team without additional work, but you get the idea. The tutorial at the top of the thread is a good tutorial, and clear, and defines a poor user experience, a work-around workflow to enable a popular use case apparently unsupported by product management and engineering. The subsequent explanation makes the complexities clear, and maybe that is why the use case is unsupported? Maybe rather than writing code in the core to address this, some 3rd party integrations and macros would be sufficient to enable a better customer experience: perhaps this could be explored? Until it is, as a non-expert, this all screams to me that if I want to do the things I say i want to do, until the user experience has been given more thought, another tool might be a better option.

just 2c, offered with best intentions.  As I say, I am a product manager with complex systems experience, and currently unemployed: if there were a way to join you, even on a temp basis, to look at issues like this, would be thrilled to have the opportunity...

[Anonymous]

Hello,

I have a mesh done in a plane design software that I need to edit in Fusion360. But there seems to be some problems with the mesh, it is broken or with holes. At first sight the structure seams fine, but if you try to Brep, quad transform, repair it or remesh it, it distorts the model.

I've seen a lot of videos, I tried, Fusion, Netfabb, Meshmixer, Recap, Instant Meshes, Blender, but nothing solves the problem without a lot of distortion.

Need some help please, (file is attached).

[TrippyLighting]

Thin thread discusses irregular and organic meshes. Your mesh in neither of those, in fact it is exactly the opposite in that it is  highly regular. Please create a separate thread. 

You can tag me in that new thread with @TrippyLighting 

The trick here is also to convert this into a quad mesh and will involve a bit of manual work and a rather different workflow.. But it is quite doable in Blender. That's what I have so far  but its still work in progress.