attracting a plane to a bowl

[SpiritOfMadMesh]

Hi everyone,

I am trying to attract a plane created from a honeycomb, to fit on the outside of a bowl. 

My work flow to get to this point was the same as if I were to get text onto the bowl. Selecting the surface of the honeycomb, inverse, delete inverse, set bowl as target, select all of plane and then attract to target. The process begins, but doesn´t seem to make any progress at all, it stays in the computing stage and the progress bar stays gray. I´ve let it compute for an hour and nothing happened. I needed to kill MM in process explorer to get out of the program.  The diameter of the bowl is larger than the diameter of the honeycomb, so it is not a to small a target.

Is what I am trying possible or am going about this all wrong? 

[MagWeb]

This should work (although I'm not sure about computation time with such a big number of open boundaries).

One reason which might cause the long computation:

AttractToTarget does not remember the last used setting for PreserveBoundaries (which needs to be disabled for this task). It always starts with PreserveBoundaries enabled. So IF the honeycomb only owns vertices at the boundaries there's nothing to project as there are only vertices which are preserved in their positions. This might cause some unhandled exception....

You might try to use Remesh at RemeshMode = LinearSubdivision and a Density to get additional vertices inside the surface in this case.

Now AttractToTarget can move these new vertices (which will be garbage) but there's the chance to disable PreserveBoundaries now...

If this doesn't solve it: Please post the honeycomb pattern.

[SpiritOfMadMesh]

Hi Gunther,

Thanks for helping again! 

See attached screenshots for the honeycomb.

The total of vertices is only 65085 so I wouldn't think it that hard on my machine to compute. 

[MagWeb · Accepted solution]

There are triangles spanning from an open boundary to another boundary.

I would try to do Remesh with BoundaryMode=FixedBoundary and a higher Density to make sure that those faces get split.  After that you may check the result in Inspector (without doing some repair) to make sure that there are only blue markers (open boundaries which aren't "connected" via some single triangle).

Maybe it's possible to post the honeycomb mesh instead of an image (zip a STL or OBJ to attach it here)?

[SpiritOfMadMesh]

I did as you suggested and did a re-mesh with Boundary Mode = Fixed Boundary and a higher Density and it worked. Right now the final computations after turning off Preserve Boundaries are in progress. 

I'll need to remember this for next time! 

When it is done I'll post a screenshot of the end result

Thank you very much for your help, again!!   :)) 

[SpiritOfMadMesh]

Computation took a long time and in when I tried to change my viewpoint with my 3dconnexion mouse meshmixer crashed. On restart of MM there was no auto-safe restore, so it was all for nothing.. I was a bit too soon with accepting the solution as I can't seem to replicate the previous results. I started cleaning up the mesh and started the attract to target and now it gives me a red honeycomb. I can't get past this point.

I included the mesh of the honeycomb and a screenshot of the red mesh.

[MagWeb]

Here it took 25sec to run AttractToTarget with fixed boundaries to the default sphere and after disabling PreserveBoundaries another 19sec to get this without any modification on your file. Are you sure that the backfaces of the honeycomb are pointing towards the target in your scene? 

I attach the input scene and the output after Attract.

Note: I moved the honeycomb to an average position in BeforeAttract.mix . This way you'll get less distortion on the pattern.

EDIT: Just noticed in your screenshot that there are unneeded options enabled. Disable all checkboxes.

[SpiritOfMadMesh]

Thank you very much for the files!!

I am positive  I had the right face facing the target. I have no idea why it didn't work an hour ago and now it does the whole thing without any problem, or at least it looks like it.  The big difference is my attract job is still computing.. It might be a hardware issue on my end then..

I have one core of my cpu at around 25%, my video card isn't even being used for these computations..

Do you know of anything I might change settings wise, in my laptop?

I run windows 10 with a 7th gen core i5, 16gb memory and an Nvidia GTX 950m with latest functional drivers. Not the fastest machine perhaps, but most other stuff in MM goes pretty quick.

meanwhile my machine is still busy trying to reproduce your results.. 

[SpiritOfMadMesh]

On your edit,

I am still not even at a point I can check or uncheck those settings. With the red mesh that window pops right up when it shows that it failed, so unchecking wouldn't have made a difference yet.

But I indeed had the assumption I only needed to uncheck the preserve boundaries checkbox, since that is how I learned to do it with attracting text to a target. This was my first attempt to do it with anything other than text. Still have a lot to learn I keep discovering 🙂 

[MagWeb · Accepted solution]

The other checkboxes of AttractToTarget ARE remembered.

So do some simple attraction task first to disable these options and Cancel AttractToTarget. Now do the attraction with the honeycomb. 

[SpiritOfMadMesh]

Gunter, you are a lifesaver, thank you so much for all the help!

Unchecking those boxes reduced computation time to about 5 minutes, still not your 25 seconds, but so much better than the previous endless waiting.

I do have a question though about reduced sharpness towards the outside perimeter of the attracted mesh. If I extrude now, the middle looks perfect, but towards the perimeter it gets somewhat distorted.

Is there any way to get this as crisp and sharp as the middle part other than stitching smaller section together?

If you would stitch smaller sections together, how would you do this?

Would you stitch smaller sections before the extruding or stitch the extruded section? 

[hfcandrew]

If the processing time is taking a while you can also reduce the triangle count. For a simple 2D mesh place like that 160K triangles is crazy high. A 90% reduction would make Meshmixer happier while still maintaining the exact the same shape of the honeycomb.

[MagWeb]

Well, computation time depends much on the machine you're working on... I'm on a MacPro 2011 with 8cores processor.

About distortion: The only way to minimise distortion is to run the projection from an average plane (as I did it in the BeforeAttract.mix above). This pulls the center region while pushing the exterior. You can not avoid distortions wrapping a plane on a sphere (same as in real world). Think stitching without seams braking the pattern is a hard, almost impossible job. The problem of regular octagons is that they always build a plane...