topic Re: What's the "new" mesh->BRep polygon limit ? in Fusion Design, Validate & Document Forum

What's the "new" mesh->BRep polygon limit ?

TrippyLighting — Fri, 29 Jul 2016 23:15:30 GMT

I had hoped with the new mesh work space the silly 10k polygon limit would be raised to something higher.

Is there a new higher limit ?

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Fri, 29 Jul 2016 23:21:19 GMT

@TrippyLighting

It is higher but now you have to triangulate the mesh to convert it to a B-rep......grrrrrr

I just converted a mesh that's 14260 but be warned performance with meshes and B-reps is terrible now...

Re: What's the "new" mesh->BRep polygon limit ?

cekuhnen — Fri, 29 Jul 2016 23:35:17 GMT

@PhilProcarioJr @TrippyLighting

To me it seems they introduced a big amount of issues with the new updates

however besides that what would be the point to BREP such a dense mesh?

In Alias I would use it to snap new and clean surfaces over it only.

Since rotated meshes are being split into tri angles anyway the MESH to BREP process in Fusion is dead for me.

It worked fantastic with furniture design 2 years ago.

According to AD developers when rotating a cube the faces are not planar anymore ... ah well ...

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Fri, 29 Jul 2016 23:45:39 GMT

@cekuhnen

Say you wanted to build head gear that perfectly fit someones face or better yet I design prosthetics and they have to fit perfectly....

or stuff like this:

https://knowledge.autodesk.com/support/fusion-360/getting-started/caas/screencast/Main/Details/c0b1c787-53d6-45dd-a028-c6d371fbf7cf.html

Re: What's the "new" mesh->BRep polygon limit ?

daniel_lyall — Sat, 30 Jul 2016 02:21:59 GMT

I have a lot that are just under 11000 so I can't do a lot with them just that extra 1000 would be good

Re: What's the "new" mesh->BRep polygon limit ?

TrippyLighting — Sat, 30 Jul 2016 12:09:03 GMT

The point @cekuhnen is this!

The curved arms that hold the PCB with the LEDs were designed in Blender, using the array and curve modifier on a relatively simple base mesh. The head holding the LEDs was added manually by good ol' sub-D modeling. The control mesh, for the object has 19k+ faces.

This was imported into Fusion 360 and converted into a T-Spline, with the aim actually not editing the T-Spline but simply adding the mechanical features, the tabs to snap the LED PCB in and the interface that connects it to the center piece.

Adding the mechanical features explaines the reason for getting this into Fusion 360.
Fusions 360 lack of more advanced T-Spline modeling tools such as a modifier stack is the reason I did not actually model it in Fusion as it simply cannot be done in Fusion. The other reason is that this T-Spline brings Fusion to a grinding halt, not a crash, which is truly admirable, just a halt. Editing the T-Spline is impossible (on my machine).

Once in Fusion 360 I have two avenues to get me to a smooth surface.

As indicated above, a T-Spline. T-Splines are mathematically perfect surfaces without a finite resolution, which would be the limitation of a Sub-D mesh.
Mesh to BRep. I really don't need to edit the T-Spine and as explained above cannot anyway due to the slow T-Spline performance. However, in oder to get to a smooth surface, then I need to vastly increase the level of subdivision before I import such a mesh into Fusion 360. I've increased the resolution for the .stl mesh that I sent to shape ways to their limit of almost 1 Million, triangles and if held right in the right light I can still see the faceting as present in the .stl file. So an increase from 10,000 to even 100,000 is still not enough.

These limitations above are actually limiting me from exploring my next idea at least as it involves Fusion 360. I'd like to explore direct metal 3D printing a similar curved LED arm, roughly 2 times the size with more intricate fractal like patterns in stainless steel, or titanium with the optics designed using high contrast caustics design (video). I have contacted the researchers.

I think I have a pretty good Idea of what a strategy could look like that allows us to get around these limitations, but I'l post this in another thread 😉

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Sat, 30 Jul 2016 12:22:41 GMT

@TrippyLighting

I could retopo this for you and solve your problem....

Re: What's the "new" mesh->BRep polygon limit ?

cekuhnen — Sat, 30 Jul 2016 14:06:36 GMT

Can you share the obj file with me? I am curious how shark fx deals with this.

it also offer subd-do to NURBS but others their on technology.

if this works I can send the brep back to you 😉

MOI 3D also has a subd-do to NURBS module.

Re: What's the "new" mesh->BRep polygon limit ?

cekuhnen — Sat, 30 Jul 2016 14:09:03 GMT

@TrippyLighting This are the same reasons why o feed data done in blender to fusion. Currently I work on a brain spatula with the bio mechanical engineering department and neithe alias or fusion can help here exploring ideas that incredible fast when using bend curve modifiers and many others

Re: What's the "new" mesh->BRep polygon limit ?

jeff_strater — Sat, 30 Jul 2016 14:42:39 GMT

Mesh to BRep, as it is in Fusion today, is really intended to be a temporary solution. This method creates a BRep face for every mesh facet. You really don't want, as @cekuhnen says, to directly convert a mesh to a BRep for any significant sized mesh using this method. The performance of such a BRep would be horrible. That's why we added the 10k limit. The number is arbitrary, but we wanted to protect users from doing something that a) would take a very long time, and b) would produce bad results.

The long term plan is to use some quadrification technology that will produce quad meshes from triangle meshes. Fusion can already do quad mesh->TSpline body. So, the flow would be: tri mesh -> quad mesh -> TSpline -> BRep. This will work for "organic" shapes (like you would build in TSplines). I did a quick search in the Fusion Idea Station, and did not see this particular idea, but I admit I did not look very long. Anyway, please feel free to create a "tri mesh to quad mesh" idea, or a "tri mesh to BRep" idea. If enough folks are interested, it helps us make our case to get if funded.

The "exact mesh" -> BRep path is less clear. This case happens when you have a mesh that was generated by another CAD program (most likely from the display mesh), or by a tool like Sketchup that is a mesh modeler, but allows you to draw planes, cylinders, etc. The above path is not likely to be effective in these cases, as what you want out are BRep planes that you can sketch on, or cylinders you can create a work axis from. The path via TSplines will not produce analytic geometry. So, there is another whole project needed for "precise mesh->BRep"... There are some techniques, but today, most are not very good.

Jeff

Re: What's the "new" mesh->BRep polygon limit ?

cekuhnen — Sat, 30 Jul 2016 15:38:13 GMT

@jeff_strater

Till today I never found a software that can take what ever STL file and very closely including sharp edges rebuild the surface with quads.

Thats why I never put such an idea down.

A small software that is very popular is: Instant Meshes - https://github.com/wjakob/instant-meshes

Because if you want to use quads then you need to do it in a way that makes sure that the topology is fine so that TS will be happy.

I personally still feel like besides all the high tech we have some workflows cannot be made that much better today sadly.

LOL imagine a laser scanner that pops up the NURBS surfaces 😉

EDIT:

@jeff_strater I just did a simple test:

U used a ReMake scan to retopo in Instant Mesh resulting into roughly 27000 faces - that part worked very fine and fast - could in crease the resolution

T-Splines loaded the model and converted it but the UI response gets problematic and Fusion cannot build a Brep from it

MOI3D loads the mesh easily and quiet quickly gives you the mesh2brep result.

The UI starts to slow down (this is an X Display Server app for OS X) but you can still work with it fine and export the STEP file.

Theses meshes get so dense I am not sure if this is really the ideal idea to work with compared to well importing a scan data and rebuilding it cleanly with NURBS coming from a practical standpoint.

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Fri, 05 Aug 2016 21:25:21 GMT

@TrippyLighting

I didn't forget about you...sorry it's taking so long I have been really busy.

Anyways I worked on this for a few hours testing which methods work best and I think I am on the right path.

Here is what I have so far, keep in mind this is still rough by my standards but it is a solid in Fusion... Anyways I just wanted to give you a progress update.

And the current curvature map...

Zebra...

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Fri, 05 Aug 2016 21:42:47 GMT

@TrippyLighting

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Fri, 05 Aug 2016 21:43:45 GMT

@TrippyLighting

Re: What's the "new" mesh->BRep polygon limit ?

TrippyLighting — Fri, 05 Aug 2016 22:08:04 GMT

@PhilProcarioJr that looks indeed very promising. I am very interested to see what the poly could could be before/after and if I can use the retopologised structure to guide me in reducing polygons when creating this.

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Fri, 05 Aug 2016 22:19:06 GMT

@TrippyLighting

I took your model and increased the polycount to 3 billion. That's what I am working off of for the final model. What is your target polygon count so I know what I'm working towards?

Re: What's the "new" mesh->BRep polygon limit ?

TrippyLighting — Fri, 05 Aug 2016 22:33:53 GMT

After popping my eyes back into their sockets I am wondering if there is a misunderstanding. 3 Billion ? in Blender on my 6 year old iMac viewport pivoting and panning gets choppy past 9 Million.

The original control mesh had 19,000+ faces.

Thats what I created the .obj file from that I sent you.

That's what I imported into Fusion 360 and converted into a T-SPline.

I don't have a set target, but my goal would be a control mesh with fewer than 19,000 faces.

Re: What's the "new" mesh->BRep polygon limit ?

PhilProcarioJr — Fri, 05 Aug 2016 22:42:51 GMT

@TrippyLighting

LOL, no you misunderstand what I am doing...I have to increase the polygon count for the Highres mesh to retopologize on. That's the only way to get rid of the surface tension in the LOWres mesh...lol I'm not going to give you a mesh with 3 billion polygons...The mesh I give you will have the least amount of polygons needed for T-Spline/solid model.

Come on Peter you know me better then that...anyways you made me smile when I read that.

If I use your lowres mesh for retopo your surface curvature would be absolute crap, thats why I need the 3 billion polygon mesh.

Re: What's the "new" mesh->BRep polygon limit ?

cekuhnen — Fri, 05 Aug 2016 22:53:56 GMT

@PhilProcarioJr@TrippyLighting

I am lost here right now - what do you two try to achieve? Retopo this in a different app?

Re: What's the "new" mesh->BRep polygon limit ?

TrippyLighting — Sat, 06 Aug 2016 00:19:33 GMT

Phew...

Now that I've had some dinner my brain cells are back to normal function (no snide remarks please 😉 and looking at the Zebra analysis and remembering posts you made in other thread about surface tension I realize what you are doing.

Perfect smoothness is not a requirement for this model as it is 3D powder printed in Alumide, which is fairly coarse.

I don't know how large the model is that is generated in different apps from the .obj but the bounding box of the .stl hat goes to Shapeways is roughly 243 x 131 x 22 mm.

The history behind the design is explained to a degree here. When investigating the design requirements for SLS 3D printing it became clear that a solid tube would not be very attractive and also not economically printable and also, of course that a more complex structure would not be more expensive. Simple holes in a tube also turned out to be ugly so I experimented in Blender with some shapes and forms.

What I arrived at is what we're discussing here. The solid spiral I wanted to be smooth with the grid part more hard edged to look like almost carved out of a solid tube. So don't take it to far with the smoothness. I am less interested in a perfectly smooth model for this piece as I am interested how you approach re-topologizing the mesh to reduce face count.

While I was aware that retopolgy is a very normal workflow e.g. for game models and rendering, I have already learned something that only occurred to me very recently after the posts were you mentioned the surface tension.

It will likely improve the area I was am not too happy with, which is the transition between the gridded tube and the head.

After following many of the T-Spline related threads, including this one I think I fully understand conceptually what you are talking about with CAD accurate T-Splines and how you achieve that.

Indeed this methodology makes T-Splines a much more powerful tool.