Designing on Complex Surface Geomoetry

[Anonymous]

I need some general workflow advice for modeling and designing in F360!

@TrippyLighting has been a great help and has provided invaluable advice - but it is time to reach out in a broader sense. 

How would one go about modeling + designing an EEG net onto the surface of a head model in F360 where:

1. Dimensions and parameters are realistic.

2. Cylinders (electrode pedestals) are normal to the head surface. 

3. A net is connected to each cylinder (imagine rubber bands between each cylinder).

4. Can such a design be modular/scaleable? 

Originally my intended workflow was to create the net as a 2D body, assemble the electrodes to the net using joints, and then wrap the net around a solid body head model, but this specific workflow does not seem proper (or possible) in Fusion. 

After reaching out to @TrippyLighting , my next attempt was to take the 2D net design and Project to Surface the design onto the surface of the head model. I believed from there I could split faces and press/pull the projected net geometry as a 3D net on the head model. While I did make some progress here, it ultimately proved to be a very laborious workflow that did not have the desired outcome I believed it would. Although I was able to get portions of the projected geometry as a 3D net on the head - it was not as intuitive as I had imagined and I ran into plenty of modeling errors as I tried to use features and tools in undesired ways. 

Can anyone imagine a workflow in F360 to achieve this? Is this something possible to achieve in F360 or am I asking for the impossible?

From fruitful discussions with @TrippyLighting I understand now that other CAD softwares might be better suited to tackling this type of problem - however if I only have a license for F360 at the moment 😄

[TrippyLighting]

A question that I've forgotten to ask in our previous conversations, is this for manufacturing, or for visualization ?

[Anonymous]

A very important question!

Two major goals of this design + concept is to:

1. Render the design to use as figures.

2. Manufacture a prototype net based on this design. 

Ideally this design would cover both visualization and manufacturing. Realistically the design will be used for visualization. The current proposal is to create a manufacturing prototype design based off of this concept - so it looks like I will be re-designing the final version specifically for manufacturing either way. 

[TrippyLighting]

---

@Anonymous wrote:

2. Manufacture a prototype net based on this design. 

---

How, or by which means & methods ?

[Anonymous]

I have found a company that specializes in prototyping solutions and works with elastomers/plastics/polymers. 

They are going to help me manufacture the prototype designs for the net. Strangely enough they desire to have Gerber files to manufacture from.

Once we finalize the net design, the design will be translated into Eagle, exported as Gerber files and off to the manufacturer to create some prototypes!

I find it funny to create a 3D net this manufacturer wants to start with a 2D design - but that is entirely as a result of the manufacturer's chosen method for fabrication. 

[TrippyLighting]

Sooo.... we've established already that a 2D structure will naturally deform when mapped on a curved surface.

That is a requirement governed by the laws of nature/math.

So if you create a 3D structure around a head and map that onto a 2D plane, how is that not going to distort ?

And once that 2D structure is manufactured, how is that going to actually fit in a head ?

My question about manufacturing was actually really aimed at figuring out why you actually would need precise CAD data as opposed to say a mesh model in Blender. That is answered, but I am suspicious of what the manufacturer is asking for ( raised eyebrow here!).

[kb9ydn]

---

@TrippyLighting wrote:

Sooo.... we've established already that a 2D structure will naturally deform when mapped on a curved surface.

That is a requirement governed by the laws of nature/math.

So if you create a 3D structure around a head and map that onto a 2D plane, how is that not going to distort ?

And once that 2D structure is manufactured, how is that going to actually fit in a head ?

My question about manufacturing was actually really aimed at figuring out why you actually would need precise CAD data as opposed to say a mesh model in Blender. That is answered, but I am suspicious of what the manufacturer is asking for ( raised eyebrow here!).

---

I would have to agree; wanting gerber files for something like this makes no sense.  This type of fabrication is outside of my area of expertise but it's interesting enough I thought it worth posting an idea.

Assuming I'm understanding the end result correctly, the part being manufactured is essentially a bunch of rings (that the sensors fit into) connected together by a web, correct?  For manufacturing something like this I think I would try making it from some sort of cast-able silicone, using a 3D printed mold.  I can imagine having a negative mold in two halves (left and right) with tracks for the material to be molded.  Then use a balloon to fill the space where the head would be leaving only the tracks for the casting material.  You'll have to figure out sprues for the material to flow in through.

As far as designing it goes, the only thing I can come up with is this:

1) model a "head" from a tspline form so that you can use the vertices to define the positions of the sensors

2) construct a plane parallel to the "head" surface at each of the sensor points

3) use the torus feature to cut the rings into the "head" surface.

4) Use the pipe feature to cut the tracks to for the web connecting the rings.

5) Repeat for all sensors.

6) Use symmetry to you only have to model one side of the "head"

This is method will be tedious and has some issues to work out but I think it could actually get the job done.

C|

[Anonymous]

To my understanding, this vendor specializes in working with elastomers and specifically thermoplastic polyurethane (TPU) foil. It seems they have created a novel fabrication method for circuits with significant elongation properties, and have turned it into a proprietary process. It seems their fabrication process for projects involving high elongation starts with a 2D planar foil. Somehow the net structure is etched or punched out from a large planar sheet. Circuitry is laid onto this planar sheet and laminated via another layer of the foil. The final product allegedly has a significant amount of elongation and has the ability to conform to complex geometries. 

As far as @kb9ydn suggestions goes - can you provide a screencast or elaborate a bit further? This seems like a very viable workflow for what I am trying to achieve, but I do not have the Fusion knowledge to understand it completely from the provided bullet points. 

To address your suggestions:

1. I currently have a solid body head model I have been using to reference and design from/on. Can I easily convert this to a tspline body or are you suggesting to create my own head model as a tspine? Using verticies to define sensor positions sounds like a great idea. 

2. How do I construct a plane parallel to a point on the curved head surface? Is this something I have missed in general parametric modeling or is this in respect to working with tspines (which is the sculpt workspace - not mesh?). 

3. I am not specifically familiar with this feature however this should be simple enough to Google and research on my own behalf. I assume it is to cut normal to a curved surface. 

4. I am also not familiar with the pipe feature - is this in the sculpt workspace? Again, a feature I can most likely research on my own 😄 

5. YAY! @TrippyLighting knows how much I enjoy hours of repetitive workflows (still working on efficiency + patterning). 

6. Patterning when possible saves a lot of time... so I have heard. Time to use it. 

I think the main objective is to create a head model with the net design located on the scalp for the purpose of visualization. You both have expressed a very valid point that it is not useful to spend hours accurately designing geometry - only to have to it lost from going 2D --> 3D, vice versa, or the like. As far as manufacturing a stretchy 2D net for a 3D net application - it is a proof of concept prototype and it is ironic how self-contradictory the concept is. 

tl;dr: I have extremely limited CAD experience + knowledge - but my current task is to tackle this project. It is people like @TrippyLighting and @kb9ydn that keep me sane in this trying times. As a chemical engineer by trade, stepping out of my professional engineering realm and attempting to be a Mech/CAD engineer is not easy! One day I will be mixing the polymers to make the net - but until then!

[Anonymous]

I attempted to attach a .step of a head model on my most recent reply - is this not possbile?

I can attach the file to this reply message. Upon pressing 'Post' it reloads the page and my attachment is removed.

Am I doing something wrong? 

[davebYYPCU]

1) model a "head" from a tspline form so that you can use the vertices to define the positions of the sensors

Not sure in my limited sculpting that this is available, (first thing I thought of too,)

when finishing the tspline, it gets converted to nice curve continuous breps,

vertices are not available to surface or modelling area,

I think the brep has to be saved out as STL or OBJ to obtain edges and vertices, (low Poly ) Trippy will know the how of this.

Following with interest....

[TrippyLighting]

The T-Spline idea proposed above is neat and likely based on the idea that a net and a mesh are similar enough for this to work, but the topologies for a mesh that creates proper head geometry and the topology of the net are different so there's no 1:1 match.

Just look at commonly available mesh models of human heads and compare this with the structure of a mesh that would place all these sensors. It might be an interesting study for someone with too much time on their hands to see whether or not this can be made to work.

BTW a T-Spline can be converted in a way that maintains the T-Spline mesh edges.

Another approach would be to subdivide the T-Spline fine enough so it works like a fine enough grid to place things on. But that'll create an incredible amount of incredibly dense NURBS surfaces.

The next task after that is to figure out how to place 128 sensors aligned with the h face normal. That can be done but is tedious.

[kb9ydn]

---

@TrippyLighting wrote:

The T-Spline idea proposed above is neat and likely based on the idea that a net and a mesh are similar enough for this to work, but the topologies for a mesh that creates proper head geometry and the topology of the net are different so there's no 1:1 match.

Just look at commonly available mesh models of human heads and compare this with the structure of a mesh that would place all these sensors. It might be an interesting study for someone with too much time on their hands to see whether or not this can be made to work.

BTW a T-Spline can be converted in a way that maintains the T-Spline mesh edges.

Another approach would be to subdivide the T-Spline fine enough so it works like a fine enough grid to place things on. But that'll create an incredible amount of incredibly dense NURBS surfaces.

The next task after that is to figure out how to place 128 sensors aligned with the h face normal. That can be done but is tedious.

---

So up until this point I've had literally minutes of experience with Tsplines, so my idea suggested above is guaranteed to be half baked.  With some messing around I was able to create a kinda sorta proof of concept to demonstrate the idea.  Now that I've actually tried it I think using Tsplines is maybe not really necessary.  What you need is a way to create paths in the shape of a head (roughly spherical) that you can use for patterning torus and pipe bodies.  I think you could probably do this with 2D sketches intersecting the head form. (I didn't have time to try that but it should work in theory)

I could try to explain further but it's probably better to just look at the file.

[Anonymous]

I also have very limited experience with tsplines. 

@kb9ydn I took a look at your provided design - wow what a proof of concept! Placing a tangent plane to a face at a point is something I need to use more. Using the pipe method to create traces/pathways for the net seems like a good idea - now I understand what you mean! I have never used a torus before, quite an interesting tool. Now I can place a torus to define a sensor location - from there I can place a point in the center of a torus, place a plane tangent to that point and create geometry normal to the head surface by sketching on that plane. This is incredibly useful! I seriously appreciate the attached .f3d file. 

I am unsure what you mean @kb9ydn when you say this can probably be done with 2D sketches intersecting the head form. Maybe with some elaboration or clarification I will understand it . 

I may not have time today to explore this concept on my head model - but I will report back when I do. 

An additional question, should I look towards remodeling a head model using a quadball/tsplines? Or is this type of workflow (placing pipes + torus) possible on a solid body? For example, I could use foresee myself using split body/face to parcellate the solid body head model to create paths/edges to lay pipes on. 

[kb9ydn · Accepted solution]

---

@Anonymous wrote:

I also have very limited experience with tsplines. 

@kb9ydn I took a look at your provided design - wow what a proof of concept! Placing a tangent plane to a face at a point is something I need to use more. Using the pipe method to create traces/pathways for the net seems like a good idea - now I understand what you mean! I have never used a torus before, quite an interesting tool. Now I can place a torus to define a sensor location - from there I can place a point in the center of a torus, place a plane tangent to that point and create geometry normal to the head surface by sketching on that plane. This is incredibly useful! I seriously appreciate the attached .f3d file. 

I am unsure what you mean @kb9ydn when you say this can probably be done with 2D sketches intersecting the head form. Maybe with some elaboration or clarification I will understand it . 

I may not have time today to explore this concept on my head model - but I will report back when I do. 

An additional question, should I look towards remodeling a head model using a quadball/tsplines? Or is this type of workflow (placing pipes + torus) possible on a solid body? For example, I could use foresee myself using split body/face to parcellate the solid body head model to create paths/edges to lay pipes on. 

---

So the attached file is what I was thinking about with using 2D sketches intersecting a (non t-spline) head form to create paths.  This is much more like how I normally model things (since I mostly make machine parts and such).  It's much more parametric and tries to take advantage of patterns to define the locations of things.  I originally thought of t-splines because it allows you to create a roughly spherical form with an evenly spaced grid on it very quickly.  Or so I thought anyway.  Subdivision modeling (basically what t-splines are) is mostly foreign to me because I just don't ever need to do that kind of thing.  It's possible it could be the easier way to go but I'm just not familiar enough with it to make that determination.

Anyway, take a look at this version and see what you think.  It's still fairly chaotic but I think you'll get the idea.

C|

[TrippyLighting]

@kb9ydn that approach makes much more sense and is also how I would do this in Fusion 360.

Edit: At first sight this looked OK, but I've looked a little closer at the model and and the problem with the pattern on path is that it does not properly honor the surface normal, meaning that further along in the pattern the rings dip to one side and why I aborted previous attempts ( in other models) with the pattern along path approach.

This should work, but is a limitation. of the current tool.

[Anonymous]

---

@kb9ydn wrote:

Anyway, take a look at this version and see what you think.  It's still fairly chaotic but I think you'll get the idea.

C|

---

@kb9ydn Thank you for the provided design file - extremely helpful! It is powerful to see more experienced F360 users tackle the same problem in different ways. I now understand your method in this regard. The way you setup the model for designing is eye-opening, I would have never thought to create a sketch in the middle of the head model to use as anchor/reference points for creating a plane from three points. This is quick and painless compared to how I would normally tackle the same step (thanks!). I also liked how you projected geometries on the head surface, this is another basic technique I never picked up on! 

This thread has become very informative - thank you to @kb9ydn and @TrippyLighting I appreciate your efforts! As of now, the most viable solution has been provided by @kb9ydn (should I accept solution to this response?). From my understanding, using pipes to create the net geometry will not be directly useful to manufacturing/designing the net for prototyping; however for this is a good solution for rendering/creating visuals from the F360 Design. 

If anyone has additional suggestions/solutions, or wants to continue discussion on this topic let me know - I will be here . 

Seriously hats off to everyone involved. Not only am i getting viable solutions, I am understanding and learning aspects of F360 that I could not seem to learn on my own so far. 

[TrippyLighting]

---

@Anonymous wrote

---

From my understanding, using pipes to create the net geometry will not be directly useful to manufacturing/designing the net for prototyping;

---

Why ? Would you need a different cross section ?

[TrippyLighting]

---

@Anonymous wrote:

---

As of now, the most viable solution has been provided by @kb9ydn (should I accept solution to this response?).

---

Yep!

[Anonymous]

---

@TrippyLighting wrote:

At first sight this looked OK, but I've looked a little closer at the model and and the problem with the pattern on path is that it does not properly honor the surface normal, meaning that further along in the pattern the rings dip to one side and why I aborted previous attempts ( in other models) with the pattern along path approach.

This should work, but is a limitation. of the current tool.

---

@TrippyLighting does this still hold true when the 'orientation' field of the Pattern tool is set to "path direction"? I originally was under the impression that patterning would not adhere to the surface normal - however setting the field to path direction seemed to have F360 attempt to make each patterned body normal to the surface. I can definitely understand if this is not a true surface normal, meaning it appears to be normal to the surface even though it is not. If this is the case, for complex paths I can see patterned geometry being skewed significantly from surface normal. Clarification might help substitute for my lack of understanding here

[Anonymous]

If I use the method suggested by @kb9ydn , meaning I project the head geometry at specified locations to create the desired net traces --> then create pipes on these paths. What are my biggest limitations using this workflow? Can I combine all the pipes at the end to create a single solid body that represents the net? Can i specify width of the net, or thickness/height of the net or will I be limited to pipe diameter?