<teepee>
I guess at some point I'd like to have the toolbars just configurable, but in the meantime having the option to change buttons probably makes sense
<teepee>
no change of keyboard shortcuts though, we really need to get that PR in :/
<Jack21>
so F7 is exporting always STL but we have a button for 3mf ( or svg?)
<teepee>
yes
<Jack21>
probably confusing if shortcuts change .. but a lot programs have customizable shortcuts
<teepee>
there's probably a need for some "smart export" action too which would just remember the export format or something like that
<teepee>
yeah, the shortcut editor shall solve that part
<Jack21>
"default export"
<Jack21>
i im happy with a button anyway
<teepee>
printing my traffic lights today I wanted to finally switch to 3mf, so I needed that button
<Jack21>
haha create demand
<Jack21>
so i have to wait until you want your traffic lights 3mf in color - Ü
<teepee>
the threads worked perfectly without any tweaking, but I'll try the stuff from your lib at some point too
<teepee>
hmm, printing in natural pla, using rgb led rings
<Jack21>
i mean when you upload this and want colors within the 3mf displayed
<teepee>
full color support is probably waaaay out unfortunately, unless someone finds heaps of spare time
<Jack21>
.. the Gewinde() module is now 4generation and will make a lot and has presets for M threads and some other 1/2 or soda bottle
<teepee>
but there might be some quicker wins, like tagging top level modules with color and exporting that
<lf94>
Nice. I skipped today's too, but once again, I'll go back to it
<InPhase>
I tried very hard to get rid of the magic numbers that are typically used to make these. I inserted only sqrt(5)-1, and 360/5, and did all of the rest with translate, rotate, and mirror, using nested module calls.
<lf94>
Nicely done
<InPhase>
There are no vertex tables. :)
<lf94>
I wouldve done vertex table :D.
<InPhase>
I felt like it shouldn't have to have a vertex table. Although it was pretty tricky to figure out the right pattern to do it without. :)
<lf94>
I made an isocahedron not using a vertex table but it may have been easier
<jack21zzzzz>
inPhase but you don't have the option to replace the spheres with little dodecahedrons - Ü
<InPhase>
jack21zzzzz: Oh yeah, I was going to ask about that. I saw that comment in the code, but then they didn't look like dodecahedrons in the image?
<jack21zzzzz>
in the customizer you can say spheres false
<InPhase>
Ah.
<InPhase>
jack21zzzzz: And what option do I select to put dodecahedrons at the vertices of the vertex dodecahedrons?
<jack21zzzzz>
but more usefull is the different orientations to print
<jack21zzzzz>
you want recursion? .. you can have recursion - Ü
<InPhase>
Dodecafractal.
noonien has joined #openscad
<jack21zzzzz>
just replace the dodecaeder (why did i used that name?) with the wiredodecahedron and it is recursive
noonien has quit [Client Quit]
<InPhase>
I'm going to attempt to recurse mind and see how far it can go before OpenSCAD blows up.
<lf94>
It's an interesting question because meshing these fractals should create tons of triangles. I personally think they meshed with a voxelizer and then triangulated.
noonien has joined #openscad
<lf94>
But do you think they're using an open source software or commercial? I think commercial.
<lf94>
(I think they did it that way because if you look at the 3D view, you can see the patterns reminiscent of being meshed using voxels)
<InPhase>
jack21zzzzz: lol
<InPhase>
jack21zzzzz: (At the first one.)
<jack21zzzzz>
this is a sphere .. now this are 1000 recursive dodecahedrons
<InPhase>
Got distracted by the kids, so you beat me to it. :)
<InPhase>
I will try anyway.
<jack21zzzzz>
hmm no without rotations it creates a dodecaheadron but 3 recursion already took some time .. not sure if 4 or 5 is possible
<InPhase>
lf94: I suppose you could do that in a lot of different ways. But at $141.01 a piece, I guess they don't have a good motivation to explain exactly how they did it.
<jack21zzzzz>
20736 dodecaheadrons are quite a lot
<InPhase>
lf94: I could certainly envision an OpenSCAD program which would generate that. It's not too far from what I did for this, just merged with an affine transform path traced out. https://www.thingiverse.com/thing:1693052
<othx>
InPhase linked to "Fabric Weave Library by rcolyer" on thingiverse => 5 IRC mentions
<jack21zzzzz>
lf94 maybe they just used the make pattern in meshmixer i
noonien has joined #openscad
<InPhase>
And of course a slightly different polyhedron generator that pops out a connected hexagon weave.
<lf94>
I'll have to look that up
<InPhase>
lf94: Let's put it this way. I wouldn't do that for fun. If someone contracted me to design it in OpenSCAD, I wouldn't take the contract unless it ranged up to at least 8 hours worth of my time. Because I expect it would take somewhere between 4 hours and 8 hours depending on things getting messy.
<InPhase>
lf94: But, doable.
<InPhase>
Some designs are just automatically more fun than others. That one looks a little painful. :)
<InPhase>
I wish it showed how many of them this person sold. That would be interesting information.
<othx>
teepee linked to YouTube video "I have a brain injury." => 1 IRC mentions
LordOfBikes has quit [Ping timeout: 250 seconds]
<jack21zzzzz>
as said if you have the borromean surface as object not just surface you can do this with 3 clicks in meshmixer (probably also making a surface into an object)
<lf94>
InPhase: I would think this is pretty easy to define as a surface using mathematical formula.
<lf94>
And then intersection with hexagon ?
<lf94>
But this looks like a shape which was deformed.
<lf94>
shape >> intersect hexagon >> deform
<lf94>
Not good for openscad. But Curv could do this pretty easily
<lf94>
Because deformation is natural in SDF universe
<InPhase>
It's possible it could go easier there.
<jack21zzzzz>
most of these objects are very easy to make if you know how - and as they look complicated people pay for them
<InPhase>
Just from intuitive experience it looks like one of those things that gets messy in a way you can't see in advance as soon as you try to actually do it.
<jack21zzzzz>
bit like modern art can cost 10000$ for just someone squirting paint onto a carnevas
<InPhase>
teepee: I don't think I've seen Strange Parts before, but that reminds me of what Simone Giertz went through.
<ali1234>
and those pendants are in sintered metal at the prices shown
<jack21zzzzz>
strange parts had some nice iphone mods and insight into shenzen markets and manufacturing
<jack21zzzzz>
but a tumor and concussion are very different
<teepee>
It says on one of the designs: "I used Mathematica and Kenneth Brakke's Surface Evolver to compute this section of it. My own creative contribution: the holes."
LordOfBikes has joined #openscad
<jack21zzzzz>
every mesh has triangles - so making hexagons is not that difficult ..
<InPhase>
Oh wait, that's the guy I watched with the "made my own iphone in China" video years ago.
<InPhase>
Yeah.
<teepee>
yep, that's likely him
<InPhase>
Looks like that was his first video. :)
<InPhase>
Then he did a whole bunch more stuff in that style.
<InPhase>
Guest30: Are you designing out buildings or something with it?
submariner has joined #openscad
Jack21 has joined #openscad
<Jack21>
just want to thank you for the correction ( polyhedrons ↦ polyhedra) - i really appreciate those. Although seems both can be used https://www.dict.cc/?s=polyhedrons (for sure not my silly version with ea )
<Jack21>
dTal:
<Jack21>
inphase: i am still baffled how you did that dodecahedron .. that moving the cylinder via minkowsky is mind-bending however as you turned this into a contest - here is my version https://bpa.st/KS4A
PovilasCNC has quit [Read error: Connection reset by peer]
Guest82 has joined #openscad
Guest82 has quit [Client Quit]
<InPhase>
Jack21: Yeah. We're used to minkowski working on objects where they are. So if you make one object a really tiny dot, then it's basically just a translate-to-point operation with a few faces added that are too small to worry about.
<InPhase>
Jack21: I did this just to make the math easier, since you can't rotate a cylinder into position without disorienting it, and I was trying to avoid hardcoding the position.
<Jack21>
i think you could use a translation with the vector of the rotation (sin,cos) so the object is not rotated itself
<InPhase>
Jack21: Your new one is very visually appealing. I like those smoothed connectors. :)
<Jack21>
Thanks, these take 50% of the code - Ü ..
<InPhase>
Jack21: Yeah, that was my first plan. But then for style I decided to try to minimize precalculating.
<InPhase>
I actually wrote up the precalculated coordinate by rotations first, then deleted it later. :)
<InPhase>
I wanted the language to do the work on that one.
<InPhase>
Looks like you have overlapped spheres on the vertices. I had that issue for a little bit.
<InPhase>
There was a very specific pattern I had to generate to avoid overlapping and vertex spheres.
<InPhase>
The overlapping was really slowing down the render enormously when I tried a render.
<Jack21>
i had to dissect the code to find out what you have done
<InPhase>
That really subtle if (z) was key to getting rid of the last one, so like a fencepost problem I had three bars and two spheres.
<Jack21>
and yeah it does but you can not have these smoothe bones without (you could cut the ends accordingly so 3 form an egde)
<InPhase>
I also had to flip one part around to make the if (z) work out as just a subtle if (z). I originally had the ordering in the other direction. :)
lastrodamo has quit [Remote host closed the connection]
bedroller has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
<InPhase>
Jack21: Your smooth one renders really slow. I fixed the overlap some time ago, but I've been running the benchmark comparison before and after for a while now. :)
<InPhase>
It's one of those pretty-at-a-price designs.
<InPhase>
Jack21: https://bpa.st/TOGQ It's not perfect, but that's the minimal change I could find to make that resolves the sphere overlap issue. It turns the render time from 28 minutes down to 8 minutes. I added the i to Strut so I could switch off of it, and made a circle sliver to offset to.
<InPhase>
It seems some symmetry is disrupted though.
<InPhase>
This would probably work better with icosahedral spheres at the junctions. :)
<Jack21>
can't see the symmetry is off .. maybe if removing a ball from each of the struts ends and keeping the balls separate could help
<Jack21>
i shouldn't remove spheres i should use rotate_extrude circles so they fit .. the problem is that on odd numbers of fragments rotate extrude doesn't fit the position but on a sphere there is always something not fitting..
<Jack21>
so if you take the rotate(180) out you get the rings residue
<Jack21>
problem wenn using $fs - you never know the number of fragments currently used
<InPhase>
Jack21: When I worked with peeps[win]'s icosahedral spheres module, I noticed that the cubic spherical subdivision one had nice properties for lining up with cylinders on the axes. I think the only thing that will line up perfectly with cylinders for dodecahedron axes is an icosahedral subdivision sphere.
<InPhase>
The dodecahedron and icosahedron have that special relationship of faces and edges that I think makes this work out.
<InPhase>
I suppose it's also possible a dodecahedron spherical subdivision would do this.
<InPhase>
For that to work you'd need to go back to a tabulated vertex approach though, so that you can spherically subdivide them. (Which is by finding appropriate intermediate points and scaling them back out by the radius to preserve the symmetry as you subdivide.)
<InPhase>
An example is on github.com/thehans
<InPhase>
One of those almost appropriately named repositories. (Back to my errands.)
<Jack21>
only that the offset and rotate_extrude will have a different symmetry
GNUmoon has quit [Ping timeout: 276 seconds]
<teepee>
Jack21: sorry, lucky accident, at least i don't remember otherwise. or maybe you did?
<Jack21>
hmm however it is just perfect
<Jack21>
you did the reordering 12.11 - Ü but it was different in the initial order
<teepee>
sometimes random + luck does produce interesting results :)
<teepee>
totally different questions... what's the simplest way to create screw posts with fillets? I'm using Strebe(single = true); so I have to put the hole in myself.
<peepsalot>
Jack21: yeah if you need to ensure that a certain multiple of fragments are used, then the only option is to specify $fn directly. i was recently thinking that maybe we could use another special var like $fd (d for divisor) to specify that the result should be a multiple of $fd fragments, but i feel like the current situation is already too complicated for many to understand well
GNUmoon has joined #openscad
<Jack21>
teepee what kind of fillets? .. the post schould have an internal thread ? use Gewinde (innen=true) and either Kehle(dia=10) or VariaFill(dia=10);
<teepee>
it does not matter too much, just to look fancy ;-)
<teepee>
well, and a bit of mechanical stability of course
<teepee>
using old GewindeV2!! ?
<Jack21>
use<ub.scad>
<Jack21>
Gewinde (innen=true,new=true,h=20);
<Jack21>
VarioFill(dia=8, l=[4,10]);
<Jack21>
if not with the template Gewinde will not automatically use the new version due to backward compatibility .. assume this will change soon
<Jack21>
peepsalot thanks - yeah it is not easy and when rotate_extrude switches between odd and even fragmentnumbers you just need to check or define the case working
<teepee>
ah,a it just overlays
<Jack21>
teepee the old Gewinde also can work with a child and cut (the new is cutting space and insert itself iirc) but this is only usefull for one central thread (i had variables to move but for multiple threads this is getting confusing )
<peepsalot>
teepee: what do you think about the $fd idea above?
<teepee>
I've not seen that yet. just back from a birthday so my day so far was mostly eating :)
<peepsalot>
or maybe "$fm" for "multiple"
<Scopeuk>
peepsalot to put a personal opinion in it regularPolygon(side=nnn) might be nicer than another global ish variable on cylinders/circles, they are a bit of a complexity best avoided
<Jack21>
there are not so many "good" values for $fn .. so 24 36 72 144
<peepsalot>
Scopeuk: i'm not sure you are discussing the same issue. a regular polygon module is trivial: module regularPolygon(r, sides) circle(r=r, $fn=sides);
<Scopeuk>
peepsalot in this case side would be the side length not the count
<Scopeuk>
a bad choice of parameter on my part
<peepsalot>
i'm talking about a way to ensure that fragments which are scaled via $fa and $fs would be some multiple of divisions
<peepsalot>
while still having the benefit of choosing an appropriate number fragments based on radius
<Scopeuk>
ok then I have misunderstood the issue at hand
<Jack21>
but a multiple wouldn't really help, you need the same number .. sure with multiple you get even distributed residue .. and with rotated spheres (even with only 90°) you never get a match
<Jack21>
not sure is isopsheres work better .. maybe a subdivided truncated octahedron could work
<peepsalot>
are you wanting to match up spheres of varying radii? if the fragments can be guaranteed divisible by 4 (360/90), then rotating by 90 over the z axis should always match up
<peepsalot>
Jack21: did you try the alternative_spheres script? subdivided octahedron is indeed an ideal option for corners of rounded cubes etc.
GNUmoon2 has joined #openscad
GNUmoon has quit [Remote host closed the connection]
<Jack21>
peepsalot maybe the sphere() could get a variable to choose different symmetry sets - this polar design is only useful for some cases.
TheAssassin has quit [Remote host closed the connection]
TheAssass1n has joined #openscad
<Jack21>
it is a bit anoying that spheres with even fn never have a point in the z center and uneven fits but can't be rotated without headache