Fusion 360 Mesh Modeling on 2016 Roadmap

[TrippyLighting]

This was an interesting revelation as it was briefly discussed in this thread.

My qyestions woud be:

1. will that ablity will go along with a more realistic limit then the <= 10k faces import limit ?
2.  will this include catmul-clark subdivision surfaces ?

With the mesh object in discussed in the linked thread I ran into these limitations. It is the comttrol mesh of a subdivision surface I created in Blender with 19k+ quad faces with the goal of 3D printing it, which went fine.

In order to have a smooth surface it needed either to be converted into a T-Spline or a subdivision surface modifier neded to be applied.

The liitations were:

The T-Sline conversion took 5 minures and etiting the T-Spline stalled Fusion 360 to a screeching halt and made it unusable. This also semed to have cause the serious pload and download times of 5 minutes.

As Fusion does not have catmul-clark subdivision the only other option is to import an even higher polygon mesh but the 10k import limitation already makes it if not impossible than at least impractical to even do this with the lower polygon base mesh.

Lifting both of these limitations would really be a boost for many things!

[PhilProcarioJr]

@TrippyLighting

If they don't figure out how to make meshes with A LOT more polygons perform fast then there is little point to adding the tools. The part I don't understand is the current performance issues with polygon meshes. Almost all other 3d modeling apps can handle millions of polygons fine. I know someone is going to say but those apps are not CAD apps.....the only real difference here is accuracy, history and if your using a polygon mesh in your design accuracy and history to a CAD level is not needed for the polygon mesh. So again I don't get it. Also I use Maya all the time and converting from Polygon to Sub-D to NURBS is a very fast and painless process. In the past this was how I got editable solids into Solidworks all day long and if I use Solidworks to export these I have the same capability in Fusion. It sucks jumping through all these hoops all the time. I really hope the dev team gets the tools and implementation right for these tools. 

I am working on a huge project right now where I have to combine the two worlds...I spend more time jumping through apps doing cleanup and conversion then I do getting work done.

Just my 2 cents.

[colin.smith]

Hi @TrippyLighting

The term "Mesh Modeling" may be a bit misleading.  The tool set we are referring to are those that are found in Mesh Mixer.  The tools will be included in a Fusion workspace in order to allow slicing and healing of mesh bodies. This won't have any effect on the amount of time it takes to convert a dense mesh to a T-Spline, and it won't change how Fusion works with a very dense mesh.  We hope to introduce some adaptive quadrification that will help lower the poly count to something more manageable for a T-Spline conversion (quad size is determined by the curvture of the surface rather than a uniform number of quads based on highest rate of curvature). This is a long term project, not all the issues will be solved in one release.  

Hope that is clearer.

Colin

[schneik-adsk]

For a preview of the mesh work that ha been ongoing you can watch this dev video we shared last year. My how time flys!

Performance will be improved over the current mesh support. https://www.youtube.com/watch?v=nQr20fTQS5M

As to triangle to quad conversion we are working on it. We have chosen to work on an improved conversion that does not need to uniformly subdivide the t-spline surface such as other conversion techniques do. The power of tsplines is that it is different from other sub-d like surfaces in that it only requires denser patches where you need smaller detail. Traditional sub-ds require uniform division, so you end up with the entire model as dense as you need to capture your smallest detail.

[TrippyLighting]

@colin.smith well it may be a little misleading as what I had imagined with mesh modeling is more inline of what one would usually do with 3DS max or Blender, but looking at the video Kevin has posted that still is mesh modeling of sorts 😉

It will certainly help a lot of people in the realm fo 3D printing.

@schneik-adsk if the video and the reference to Meshmixer is any indication of Fusions future Mesh handling capability it should be able to easily handle the object I designed in Blender even if I apply one level of subdivision in Blender. That should be enough to get a smooth surface for a 3D print. Provided of course that translates into the same performane when converted into a solid.

I just tested the object in question by exporting three .obj files from Blender each with a differnt level of subdivision, then imported the .obj into Meshmixer and exported it as a .stl.

The base mesh without subdivision has 19,444 quad surfaces

1 level of subdivision 77,756 quad surfaces. File size of the .obj 9.2 MB

2 levels of subdivision 311,024 quad surfaces. "" 38.7 MB

3 levels of subdivision 1,244,096 quad surfaces 159.4 MB. Eve

4 levels of subdivision 4,976,438 quad surfaces 664MB

Import of even the largest file was under 20 seconds al others were from instantaneous to under 10 seconds. Similarly with export to .stl. 

The point with T-Splines is understood, but for my particular object I only have two ways to get a smooth surface.

1. I convert it into a T-Spline, but with 19k faces the T-Spline perfomance made it uneditable as a T-Spline. Editing the T-Spline was not necessary. It was editable just fine as a solid model. 

2. The other and in this case preferred avenue would have been to import it into Fusion 360 with one, or better two levels of subdivision and then convert it into a breb. At the moment this is not possible due to the 10k limit. Also importing a high poly mesh with 300k facets into Fusion brings viewport performance to it's knees. For comparison In Meshmixer viewport navigation of the 1,2Mio qud surfaces object was still smooth and manageable in the almost 5Mio facet version.

This stuff needs real consideration if generative design is ever going to make it into Fusion.

[PhilProcarioJr]

@colin.smith

I'm sorry to hear that. I was definitely hoping for a different approach to dealing with polygon modeling and CAD combining in Fusion. 

@schneik-adsk

Some of the stuff in the video is neat but of little use to my needs, given the current info I will still need to use a bunch of apps to get my work done.

You said:

"The power of T-Splines is that it is different from other sub-d like surfaces in that it only requires denser patches where you need smaller detail. Traditional sub-ds require uniform division, so you end up with the entire model as dense as you need to capture your smallest detail."

You must have used different Sub-D tools then I do because Maya and Lightwave and Modo allow Non-Uniform divisions in Sub-D. Lightwave's and Modo Sub-Ds are arguably the best in the business and honestly I'm sorry to say but more powerful then T-Splines too, at least when compared to Fusions implementation. Anyways thanks for clearing that up on where the dev team is headed.

[colin.smith]

@PhilProcarioJr - not saying there will never be poly modeling tools in Fusion, but that is not on the short term roadmap.