Machining mesh versus solid - does Fusion care?

[steveBKDRW]

I don't know if this question is for someone at Fusion behind the coding curtains or people who have fought battles in the war of mesh.  

For machining an existing organic mesh shape in Fusion is there any benefit to machining time if a mesh is converted to a form of solid first? 

If no benefit, is a tri-mesh different from a quad mesh for machining? 

If a benefit, which solid type is better; brep, tspline, step...?

I have convinced myself that adaptive 3D clearing and Morph Spiral are best, are any other tool paths options people consider more mesh efficient?

Steve.

[johnswetz1982]

I think as long as you are using 3D operations it does not matter. You would only need a solid if you need "edges" for something like swarf where you need a set of control edges. Otherwise you could use a sketch for just having containment areas.

[steveBKDRW]

While appreciating a reply, I had hoped for something more definitive than "I think".  I'm going to wait a while.

[JaumeMartinez]

Hi @steveBKDRW , in my experience it is always better to work from solids.

Solids are the most accurate representation of our design, lot better than a mesh.

As they say in 3D operations it would not be any problem, but in industrial design parts (automotive, aerospace, ...), we work with different geometric and dimensional tolerances.

Speaking clearly, we need a line to be straight, a flat wall to be a single surface, a circle to be unsegmented, and the edges to exactly represent the "real" shape.

Working with meshes as you suppose is not compatible with the above criteria.

I usually use step and parasolid.

It´s allways better a quad mesh than a tri mesh.

I hate STL.

Kind regards.

[johnswetz1982]

The posters said "organic" not geometric or prismatic. I dont know what exactly he wants to machine but in something like topography or something generated from a program like Z brush you are not going to get a solid model and there may be too much data for fusion's limitation of 10,000 face to convert to a solid. So if you have something organic like a lions head or a Pickle Rick,

even if you can get it defined as a solid does it really matter because fusion is going to convert it internally to a mesh for tool path calculation?

[johnswetz1982 · Accepted solution]

I say "I think" because I am not the guy who programmed it. If you are waiting an official reply you are going to be waiting a long time because this is a user forum. You will occasionally see some Autodesk employees on here but they are generally tracking bugs and examples of said bugs.

I have some experience with machining and programing organic surfaces whether it is a Lions head, a storm troopers helmet or a Pickle Rick. There are only a couple people on here with higher rating for solutions found. You can believe whatever you want but you might be waiting awhile for a better answer/ I can only offer this as an example of organic mesh machining using the 3D toolpaths,

 

[seth.madore]

@johnswetz1982 is right, the Autodesk employees that frequent the forums are typically chasing down bugs and helping out users who have gotten in a bind with their educational journey.

I can ask this question to the devs who would know the most about this, but, as has been said, it may be a while. Pretty much everyone is on holiday until next week, and at which point, they are only going to be a couple thousand emails behind..... 😩

[jeff.pek]

For the 3d strategies, it really won't make a difference. The 3d strategies work off of a mesh, which is generated from a solid if that's what you have, or the mesh directly if not.

The advantage of converting to solids would be when working with 2d strategies, for drilling (via cylindrical faces), etc.

Jeff

[JaumeMartinez]

yes, @johnswetz1982 , 

That is why we have indicated that for 3d or surface operations there will be no difference.

Only explaining why we prefer solids.

Anyway I think that even if it is something organic if it is in solid better. I personally like to keep things simple.

Kind regards.

[JaumeMartinez]

@jeff.pek  That’s the real deal.

[cody]

Steve,

John, Seth and Jeff are all correct in what they are telling you. I will add to what they said, as I have a lot of experience with reverse engineering and taking the probed or 3d scanned data and turning it into usable solid/prismatic geometry or just machining the organic mesh model. Don't waste your time converting meshes

to solids. If you absolutely need a solid for your design or to machine, then bring the mesh into fusion and build the solid model from the mesh data with an actual prismatic tree. There are plenty of tools in Fusion to accomplish this. If you don't need the solid model, then just leverage the design side of Fusion with the CAM side to get the part made. you can use any toolpath in Fusion to machine a mesh with a little work on the design side. Study using the project tools in sketch mode. A very handy one is "Intersect". Just create the geometry you need to either contain or drive any toolpath in Fusion from mesh geometry. I make a lot of molds for various industries in this manner, many 

of them were hand made originally and I make new replacement or modify existing molds from scanned data. It's all

there in Fusion, no excuse you can't use meshes. Hope this helps.

Thanks,

Cody C

[JaumeMartinez]

Hi @cody ,

All you are right , don’t waste time converting your mesh into solid if you received the model as a mesh.

This topic remind me the time when I machines dental prosthesis.

We have to kind of clients, some with CAD technicians that send the design as a solid and others that send you the model as a mesh.

 

Work with solids was better because you didn’t have to do nothing, but with meshes you have to adjust the model because you have some tolerance in the implant connection.

Is a lot better to work with a solid if you can, but as all of you say, if the model don’t have any 2D shapes there is no difference if you work with meshes.

 
If it is not my design I always try to ask for it in solid, it is clear that if you receive directly what comes out of a scanner you will have a mesh, which is not a problem for 3D operations.

Just personal opinion.

Regards.

[steveBKDRW]

John, I'm not doing anything as detailed as you are, but along those lines.  I'm getting requests for wood appliques for custom cabinetry.  Most of the "art" comes as tri-mesh STL.  I've been trying Carveco, Meshmixer, Blender, Photo recap and some others to convert the incoming STL to a solid (brep/step..).  Results are not consistent and detail usually gets degraded. 

It could be the degraded detail, but it has been my experience that a solid conversion machines in appreciably less time than the original STL and ends up needing less cleanup.  Even converting to quad-mesh doesn't seem to make much difference. 

The answers given leads me to see what the request market can afford or walk away from that.

Steve.

[hfcandrew]

I mill organic shapes in the O&P industry, I use Meshmixer daily. I use Deskproto for my CAM.

In my experience all that matters is that you use a high enough .stl mesh density, then use a multi-directional tight toolpath.

[hfcandrew]

I mill organic shapes in the O&P industry, I use Meshmixer daily. I use Deskproto for my CAM.

In my experience all that matters is that you use a high enough .stl mesh density, then use a multi-directional tight toolpath with a high sampling rate To make sure every vertex is accounted for.

If you are milling wood, the wood itself will be the limiting factor in accuracy, not your CAM.

[johnswetz1982 · Accepted solution]

Here is a simple experiment you can do then. Next time you get something like this or pick a model to try out on. Record how long it takes. And my suggested work flow.

-STL first,

-Do an Adaptive or 3D contour toolpath(s) with .015 stock leave (I think that would OK for wood)

-Follow up with your finish strategy whether that is a Parallel or Morphed spiral. A .015 step over since you would probably be finish sanding by hand?

-Save your operations as templates

-Record the actual programing and machining time.

-Solid. Convert your STL to a solid

-Use the previous toolpaths that were saved as templates and apply them that way you have the same settings for your solid model

-Record the actual programing and machining time including the conversion time. This will even be a little bit shorter because you have the templates from the first STL model.

I have a feeling it is just your perception that a solid behaves any differently. Like was stated by Jeff Pek your solid model gets converted to a mesh behind the scenes anyways. The only point where you would need a solid is something like Swarf where you are using the solid features (2 edge lines) to drive the toolpath by picking control lines, or ?Pencil? where your tool is always going to be normal to a surface. But those are the multi axis toolpaths and not the 3D toolpaths.

[steveBKDRW]

"I have a feeling it is just your perception that a solid behaves any differently. Like was stated by Jeff Pek your solid model gets converted to a mesh behind the scenes anyways." 

Maybe I didn't read this properly earlier, but I have to agree.  Another thing that I've figure out is that my g-code with STL (WinCNC PP) is 95%+ G0 lines due to the short moves.  My  experience with solids was always better due to longer moves, more 2.5D.

Thanks everyone for their comments and insight. 

Steve.   

[craig.chester]

There are some 3D strategies which will not work on a mesh - Examples: Flow, Blend, Swarf etc or any strategies whereby you want to contain the toolpath between feature curves (Such as fillets where there is a big curvature change either side of it)..  Other strategies where the quality will not be quite as good on a mesh - Example: Steep and Shallow (It uses surface data where available to more accurately determine the dividing curve between steep and shallow regions).

Computation time may also be better on a solid, because, although most strategies will convert to a mesh behind the scenes, there is a good chance that they will optimize the mesh specifically for machining. For example: Minimum facets on a flat region.

The other obvious advantage of a solid/surface model, is that the automatic mesh generation will ensure that the resultant toolpath respects the desired toolpath tolerance.... I'll try explain better... If you have a meshed sphere, meshed at a tolerance of 0.1mm.... Then you can't really machine it better than that. However, if you converted it to surface/solid, you could then machine the new model accurately to a smaller, more desireable tolerance.

In saying all of this.... As many others have said. If this was me, and I didn't NEED to convert it to a surface/solid, then I'd prefer to machine the mesh directly.