## The All New FractalForums is now in Public Beta Testing! Visit FractalForums.org and check it out!

 Pages: 1 [2] 3 4 ... 12   Go Down
 Author Topic: Shapeways for 3D printed fractals  (Read 40930 times) Description: 0 Members and 2 Guests are viewing this topic.
taurus
Fractal Supremo

Posts: 1175

 « Reply #15 on: November 29, 2012, 11:55:40 PM »

Great stuff for the 21st century - impressing!

The problem is to iteratively find what is the maximum precision they can accept while staying in the 1 million triangle count limit.
Dunno if you own a major 3d modeller. Some have pretty good polygon reduction modules. Depending on the model, you can get rid of up to 90% of triangles without a visible loss. Even a complex shape like the mandelbulb might benefit from such a tool (while I didn't try yet).
 Logged

when life offers you a lemon, get yourself some salt and tequila!
bib
Global Moderator
Fractal Senior

Posts: 2070

At the borders...

 « Reply #16 on: November 29, 2012, 11:59:02 PM »

These are just Sierpinski tetrahedrons stitched together. Did you use a sphere to make your Mandelbulb hollow?
 Logged

Between order and disorder reigns a delicious moment. (Paul Valéry)
fractower
Iterator

Posts: 173

 « Reply #17 on: November 30, 2012, 12:14:16 AM »

I used a sphere fold on C to mirror the exterior on the inside.

if(|C| > C_cutoff) {
C_new = C*(C_cutoff/|C|)^2;
}

Use C_new in place of C in the bulb calculation.

 Logged
bib
Global Moderator
Fractal Senior

Posts: 2070

At the borders...

 « Reply #18 on: November 30, 2012, 12:41:33 AM »

Thanks, I am trying that. I want to make a half bulb so we can see the inside too.

@taurus, for polygon count reduction I used Meshlab and followed this procedure but I'm not convinced by the result. http://www.shapeways.com/tutorials/polygon_reduction_with_meshlab

I think I can improve a lot the first mesh if I understand better how to use Fiji.
 « Last Edit: November 30, 2012, 12:43:05 AM by bib » Logged

Between order and disorder reigns a delicious moment. (Paul Valéry)
cKleinhuis
Fractal Senior

Posts: 7044

formerly known as 'Trifox'

 « Reply #19 on: November 30, 2012, 03:40:16 AM »

so, people, this is why we use raymarcher instead of marching cubes for visualising

what the guy basically is talking about is that voxel representation is far more efficient than using vector/triangle representation

to understand it, just compare a raymarched mandelbulb with a marching cubes mandelbulb at the same iteration level

the triangles ad an amount of uncertainity to the result, thus, false results lead to wrong connected triangles, this
can not happen when working with raw voxel data, although vector data use more memory
this is what he is talking about in the memory footprint section

awesome, dudes, we just have right now another example of that the art or indefinitely beautiful objects serve a purpose
and he is using just the mandelbulb for analysing such data sets for 3d printing, lols, if he would have hand on a real
3d mandelbrot it would just show more complexity in the differencies between voxel and vector representation due to the
far more detailed surface a true mandelbrot would exhibit

awesome, this is great because i always need examples why i do the calendar, and what art has to do with science,
and the mandelbulb served a purpose in this case, to demonstrate the lack of triangle representation for 3d printers

nice!
 « Last Edit: November 30, 2012, 03:42:04 AM by cKleinhuis » Logged

---

divide and conquer - iterate and rule - chaos is No random!
cKleinhuis
Fractal Senior

Posts: 7044

formerly known as 'Trifox'

 « Reply #20 on: November 30, 2012, 03:44:41 AM »

@taurus its true that polygon reduction would provide an optimised mesh for 3d printing ....
we have a problem here, this would work on a low iteration bulb like bib used in his shop, the more detailed
is added (due to iterations) the more complex the surface gets, and a really good algorithm would not have anything
to optimize, comparable to a jpeg image compressed via zip ....

and this is basically the answer to why voxel representation for 3d printers is far more efficient for such detailed objects
because the voxel representation is directly mapable to the dpi or dpsi ( dots per square inch ) ah no we are in 3d dpci (dots per cube inch )
and the vector representation just explodes in memory footprint because for one dot you need 5 points to define a pyramid... ( how is the triangle based pyramid called?)

 « Last Edit: November 30, 2012, 03:54:36 AM by cKleinhuis » Logged

---

divide and conquer - iterate and rule - chaos is No random!
cKleinhuis
Fractal Senior

Posts: 7044

formerly known as 'Trifox'

 « Reply #21 on: November 30, 2012, 03:51:34 AM »

the text about voxels vs polygons is just perfect! and should be made sticky somewhere

fractals provide an abstract superficial representation of what should be possible, and funny enough the 3d printers
can provide much higher resolution using voxel data ( because this is what they actually print ) than with vector representation
(because a single dot in 3d would need at least 4 points connected to a pyramid )

and this can be very good demonstrated using fractals because they provide an (sic) perfect (no real 3d mandelbrot yet ) way of what is
basically doable with such a printer!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!eleven!!!1111!thousandonehundretandeleven!!!!1!!!!
 Logged

---

divide and conquer - iterate and rule - chaos is No random!
cKleinhuis
Fractal Senior

Posts: 7044

formerly known as 'Trifox'

 « Reply #22 on: November 30, 2012, 04:08:37 AM »

i just love the article about voxels vs polygons!!!!!!!!!!!!!!!
 Logged

---

divide and conquer - iterate and rule - chaos is No random!
fractower
Iterator

Posts: 173

 « Reply #23 on: November 30, 2012, 05:17:43 AM »

Triangles rule!!!
 Logged
KRAFTWERK
Global Moderator
Fractal Senior

Posts: 1439

Virtual Surreality

 « Reply #24 on: November 30, 2012, 09:18:46 AM »

Cool stuff bib! (even if they are a bit small and low resolution)
The inside mandelbox is really beautiful!

Did you order any of them for yourself?

 Logged

taurus
Fractal Supremo

Posts: 1175

 « Reply #25 on: November 30, 2012, 09:23:39 AM »

we have a problem here, this would work on a low iteration bulb like bib used in his shop, the more detailed
is added (due to iterations) the more complex the surface gets, and a really good algorithm would not have anything
to optimize

I know in principle, but I thought of the rather plain, whipped cream parts of a bulb. I would think, a good algorithm should spot those and leave the detailed parts untouched.
But obviously bib tried and wasn't convinced - so as long as I don't try, I simply believe him.
 Logged

when life offers you a lemon, get yourself some salt and tequila!
bib
Global Moderator
Fractal Senior

Posts: 2070

At the borders...

 « Reply #26 on: November 30, 2012, 02:31:56 PM »

@CK a triangle based pyramid is called a tetrahedron!

@Johan, I have ordered the low precision bulb. How many triangles were used in your model? At first sight, I would say between 5 and 10 millions, and Shapeways limit is 1 million :-(

And regarding the size, small is beautiful! And also much cheaper, because the price is proportional to the quantity of matter, so when you divide the dimensions by a factor 2, you reduce the quantity of matter by a factor 8 (in the simple case of a plain cube)
 « Last Edit: November 30, 2012, 02:42:05 PM by bib » Logged

Between order and disorder reigns a delicious moment. (Paul Valéry)
bib
Global Moderator
Fractal Senior

Posts: 2070

At the borders...

 « Reply #27 on: November 30, 2012, 02:37:38 PM »

I know in principle, but I thought of the rather plain, whipped cream parts of a bulb. I would think, a good algorithm should spot those and leave the detailed parts untouched.
But obviously bib tried and wasn't convinced - so as long as I don't try, I simply believe him.

I could post images, but I confirm that the whipped cream areas are still full of small triangles even after polygon count reduction using the Quadratic Edge Collapse Detection in Meshlab. Ideally I would like to be able to define what specific areas of the mesh need to be simplified. I still need to test the myriad of options that exist in Meshlab and Fiji, when time permits...
 Logged

Between order and disorder reigns a delicious moment. (Paul Valéry)
KRAFTWERK
Global Moderator
Fractal Senior

Posts: 1439

Virtual Surreality

 « Reply #28 on: November 30, 2012, 03:14:15 PM »

@Johan, I have ordered the low precision bulb. How many triangles were used in your model? At first sight, I would say between 5 and 10 millions, and Shapeways limit is 1 million :-(

Actually I have no idea, it was done using voxel layers from Daniel White converted by the guy who made the prints.
But the details of the mini mini bulbs are really small <1mm...
 Logged

bib
Global Moderator
Fractal Senior

Posts: 2070

At the borders...

 « Reply #29 on: November 30, 2012, 04:06:54 PM »

What material has been used? I guess the fine details are very fragile?
 Logged

Between order and disorder reigns a delicious moment. (Paul Valéry)
 Pages: 1 [2] 3 4 ... 12   Go Down