Shapeways for 3D printed fractals

[vb2012]

Hello,
I am Vladimir Bulatov, that Shapeways developer. I am not rock star :-)

I would like to clarify the problem. Shapeways develops open source software http://abfab3d.com/
intended to facilitate process of 3D printing. It is extensive library of 3D voxel grid representation and manipulation
written in java. In particularly it provides the process to convert voxel grid into usable triangle mesh with appropriate
reduction of face count. the package is in early stages of development, but I think it is already useful
enough for experience programmer.

I am fractal lover myself and used that library to create printable fractal shapes.
Photos of my first real piece are attached.
The 3D model can be viewed and downloaded from
http://www.shapeways.com/model/730555/limit-set-i.html

I generate voxel grid and convert it into triangle mesh directly in my program using abfab3d API.

As I myself feel that abfab3d is capable to produce high quality printable 3D mesh for fractals
I though it may be useful for other fractal builders.
However it is just a library and it needs some user interface to be usable by non programmers.
If there is a person who wish to use the library to write program which
can accept some more useful input (like stack of images) , I would be happy to provide
pointers in the code where to start from.

This is open source software with generous LGPL license.
So the program will be yours.
At the same time Shapeways will continue the abfab3d development and improvements.


Vladimir Bulatov
3D graphics researcher
http://shapeways.com
personal site
http://bulatov.org

[cKleinhuis]

hi there, great to hear from such an experienced dude, shapeways is awesome, you can explain why the introduce the 1 million mesh limit, i mean is it just arbitrary set up or really given through technical limitations ?

[vb2012]

1M limit is due to current limited hardware/software capabilities of different printers Shapeway has.
1M really pushes hardware to the limit. We are working on improving that.

It is perfectly possible make printable fractal models with reasonable polygon count.
Abfab3d helps to do it.
Here is video of that fractal piece <a href="http://www.youtube.com/v/HsVBm2vUgX8&rel=1&fs=1&hd=1" target="_blank">http://www.youtube.com/v/HsVBm2vUgX8&rel=1&fs=1&hd=1</a>
It is about 20 cm in diameter and has 600,000 triangles

Current advise - the smaller triangle count you can get without compromising model's visual quality
the more smooth and faster the printing process will go and this will make everybody happier.

Keep in mind, that bad/unprintable models waste computing resource and make printers crash
and make printer's operators sad. :-(



Vladimir Bulatov
3D graphics researcher
http://shapeways.com
personal site
http://bulatov.org

[Tglad]

Hmmm, that makes this fractal a nice one to try on a 3d printer
http://www.fractalforums.com/new-theories-and-research/new-fractal-needs-a-name/msg54859/#msg54859
because it is already built directly out of a triangle mesh.
I might give that a go.

[DarkBeam]

Hey there Vladimir!

Your Mobius based 3d fractals are fantastic. We are trying to understand how you do those...
Sharing some (or just one? ) source codes would help the fractal community and make you a big rockstar. pleeease! Even just the ones you make with Fragmentarium...

[bib]

Welcome to fractalforums Vladimir, I was sure it would be clearer in your own words Thanks a lot for joining and taking the time to present the project.

BTW, in an e-mail you said

Remember stair steps are 0.1 mm and fractal details are around 1mm.

Assuming we have a 500x500x500 grid and 5cm object, then yes about the stair steps (marching cubes artifacts) but I don't get it about the 1mm fractal details. I thought there was a clear disctinction between the smallest printable details (around 0.1-0.2mm in WSF) and the minimum walls/wires thickness (Around 0.7-0.9 mm in WSF). As a matter of fact, some fine details on my 3D printed mandelbulb are visibly smaller than 1mm, but these are not walls or wires, just simple relief. Anyway, thanks for pointing that out. The devil is in the details, but if the details are invisible, no big deal!

@Tglad, your idea to use your new triangle based finding sounds great. If you can directly produce a mesh, then your "steps" (no "real" steps as I assume you wouldn't have to use marching cubes) and your "fractal details" would be at the same scale (?...)

[Tglad]

Yes that's right bib, in fact if anyone else want to try it out the mesh files are available, as .aoi file (art of illusion)
   https://docs.google.com/open?id=0B8S7Si-yu3DoM2VPWTJIc1BuV2s
and exported to a .obj file:
   https://docs.google.com/open?id=0B8S7Si-yu3DoOURIMXV6WWJKSXM
They're just a surface not a closed mesh, but would just add 2 triangles at the base to make it closed, easy!

Great work Vladimir!

[bib]

More 3D prints received! Cool (see models here: http://www.shapeways.com/shops/3Dfractals)

The bulb is bigger than the first one and much more detailed.

However on the "beamed octahedron" (the small version with 5 iterations, whereas the big one has got 6 iterations) the smallest wires are so thin (0.7 mm) that some were broken probably during transport.

This time they came totally covered with dust, like if the operator didn't dare cleaning them being afraid to break them, and the large beamed octahedron still has got loads of dust stuck in the middle.

Still no candidate to help Vladimir optimize code for 3D fractal printing?

[taurus]

one can forebode, that they look pretty cool, but I can't believe, that you make all those stunning fractal images and at the same time, you seem not to be able to make an acceptable snapshot of your 3d prints.

Or was this intension, to make us buy your pieces, to get a glace on them?

[bib]

I can't believe, that you make all those stunning fractal images and at the same time, you seem not to be able to make an acceptable snapshot of your 3d prints.

lol

Here are some more decent pics:

[Madman]

Better. Now try adjusting the white balance a little 

[taurus]

yeahzzze! Great work. 
On the bulb one can see the grainieness of the Fiji created stack-mesh. I hope some better solution than Fiji can be found soon.

[bib]

The grainy aspect comes from the material, the printing process, and the fact it is not polished. Fiji OBJ conversion and faces decimation is another topic. In fact, the mesh faces cannot be seen. The problem is how to maximize the level of detail (i.e. the lowest DE Stop) under constraints: min wall thickness (0.7mm for this material, see http://www.shapeways.com/materials/material-options) and less than 1M faces, and last but not least: minimize the amount of matter to minimize the price!

[taurus]

The grainy aspect comes from the material, the printing process, and the fact it is not polished. Fiji OBJ conversion and faces decimation is another topic. In fact, the mesh faces cannot be seen.

really? ok, it's only a photo, but those stairs look the same as the ones I had in my fiji exports. Anyhow great proof of concept. I need to try myself soon...

[bib]

Well spotted, I should have increased smoothing in Meshlab.