3d printing a GH gyroid

The attached Gyroid picture is made up of 1518 closed meshes. Produced by the attached script. If I were to print this Gyroid ?

1. What's missing to make a 3d printed object? (besides a 3d printer) :)

2. Do I have to clean up the 1518 closed meshes and make them 1 mesh ?

3. What to do about thickness ? etc.

Attachments:

gyriod_FX_dual graph.gh, 10 KB

Replies to This Discussion

Permalink Reply by Nenov Ivo on March 13, 2016 at 9:38pm

http://www.shapeways.com/

Permalink Reply by martyn hogg on March 15, 2016 at 1:23pm

Materialise

I prefer these guys

Permalink Reply by Hyungsoo Kim on March 14, 2016 at 12:18am

Hi. Kim.

The output of "Dual Graph" are 1518 polyline curves, so your mesh pipe component create bunch of duplicate mesh geometries.

Also MeshPipe can't create nodes.

If you have "T-Splines" installed, tspipe is another option. Or you can use WB's Picture Frame+Mesh Thicken.

Attachments:

gyriod_FX_dualgraph_re.gh, 14 KB

Permalink Reply by peter fotiadis on March 14, 2016 at 12:51am

TSpipe eh? well ... yes ... and no (too many bugs AND very very slow for that amount of data [takes ages to finish - if at all]).

IMHO the WB is rather the most realistic approach.

Permalink Reply by Kim hauer on March 14, 2016 at 12:39pm

Thanks Hyungsoo, I don't have T-splines, I'll stay with the WB stuff

Permalink Reply by Kim hauer on March 14, 2016 at 3:55pm

Hi Hyungsoo:

Using "T-Splines" might be irrelevant for my particular example, I did some more WB experimenting, and added another WB comp. to your rewrite.

Permalink Reply by Nik Willmore on March 14, 2016 at 5:28pm

T-Splines pipe does work quite well, radius 0.01, but the entire network takes a few minutes to form, then way too many minutes more for smooth mode, as in a half hour, and the resulting *mesh* is nothing special in terms of size, I imagine.

Rhino is too slow in smooth mode to do much now.

When I do convert smooth mode to a mesh, I lose smooth mode:

If I Grasshopper subdivide that quad mesh and triangulate I get a 160MB STL file, so that's not much help versus Cocoon and Meshmixer:

It seems like one could use lofting between little prism faces assembled from any penta/hexa mesh like this, no? Thats would give the same result as T-Splines, in a way. Since things are individual and then in pairs for the lofts, Python could run those steps in parallel too. Curious. Maybe a future project. It would work on triangular meshes too, using mostly hexagonal prisms.

3D printers use triangulated STL files though, so how efficient this would be I don't know.

Permalink Reply by Nik Willmore on March 14, 2016 at 4:08pm

As long as closed meshes fully overlap instead of kiss, most 3D printing services are fine with it. They run it though Materialise Magics to union it much better than Rhino 5 can do. However, your mesh tubes are redundant so surfaces overlap instead of interpenetrate, so it is not a good system.

Cocoon is the best answer these days unless you can get Exowire/Exoskelton to work. If you want more control over shape, feed your uncapped tubes into Cocoon as meta-surfaces and delete any and all of the inner meshes to just keep the outer single closed one, but this is just duplicate-culled lines used as meta-lines:

Turn down the CS input to 0.005 for this result, from 0.02 used for faster preview. In fact bake the lines and only test Cocoon on a few of them in order to get the result you want before doing the whole thing.

Whole thing at 0.005 cell size takes 5 minutes for Cocoon and 2 minutes for refinement to a smooth and even mesh.

Actually, seems like 0.005 is way too fine, giving a 600MB STL file.

So, 0.01 cell size at less than a minute total:

159MB STL which is still a bit too big for places like Shapeways. Wow. OK then 0.02 cell size, but I have to increase diameter or my two smoothing steps in refine collapse things too much, an in fact I set it to no smoothing, getting more volume and a reasonable 46MB STL file:

Alas, now it's more frail and overly organic rather than mechanical. Increasing diameter just merges it into perforated plates too much. File size is simply an issue with this complexity level, so different 3D printing services will have different file size limits.

Exowire/Exoskeleton would work but your original mesh hasn't been MeshMachine remeshed to be regular, so short segments ruin it. Here is just a corner:

I think that's why more wires fails, at least. Pretty temperamental component.

Switching to MeshMachine is needed, I guess, instead of Cocoon refine, to remesh away so many small triangles along the boring tubes. Crucial for good remeshing was to set Flip to 0 or I failed to get a rough enough mesh.

It's an adaptive mesh so I can retain good detail while roughing out the tubes.

MeshMachine is terribly slow for this whole thing, like 6 minutes, and blows up for this overly rough setting, 20 steps, so less rough, ugh, I'm out of time. I think free Autocad Meshmixer is the way to make a better smaller mesh, after a refined output from Cocoon. MeshMachine is just too slow to tweak and when it blows up, creating massive triangles jutting out, it hangs too when you change settings.

Starting with a Cocoon refined mesh certainly helped Meshmixer. Using triangle budget lets me have full control. Here is 150K triangles instead of 200K:

STL file size down to 40MB. I think Shapeways is 70 or 100MB limit? So it can be even finer. Here is the Cocoon output versus the Meshmixer reduction:

To use Meshmixer, turn on View > Show Wireframe, Command-S to select all and use Edit > Reduce from the palette that appears.

Cocoon can end up making a few inner meshes where things get weird in your uneven original mesh with small holes so fish out the main mesh by adding a List Item node.

The best strategy for Cocoon is indeed to make an overly fine STL so you avoid any need to tweak forever in Grasshopper, but then you can achieve a smaller mesh file size while preserving shape instead of things turning all smearly organic in Grasshopper.

Attachments:

gyriod_FX_dualgraphB.gh, 20 KB

Permalink Reply by Kim hauer on March 14, 2016 at 6:39pm

Thanks Nik, for addressing the most important issue, FILE SIZE... related to 3d printing

FYI: I managed a ~93mb stl file Using Meshman, No further reduction with Meshmixer yet!

Permalink Reply by Hyungsoo Kim on March 14, 2016 at 10:03pm

When I convert my T-Spline generated mesh into STL, file size = 25.7MB

Permalink Reply by Nik Willmore on March 14, 2016 at 8:03pm

Running your initial isosurface mesh through MeshMachine while fixing the joined loop naked edge curves, gives a much better behaved dual that should be easier to pipe with various methods: