<teepee>
altough it needs a small extension to allow parameter extraction + design generation in one go, no sure how ciritical that would be for WASM use case
<linext_>
so can i feed in openscad code and get back HTML code?
<linext_>
and vice veras
<linext_>
versa
<teepee>
no, JSON with the extracted parameter types and meta info
<teepee>
but you don't need to parse the scad code
<linext_>
is there something in the docs that shows an example? is it available via command-line?
<linext_>
if it's accessible via wasm, i'd rather use that
<teepee>
it's command line so it should work via wasm too
<linext_>
so to import those json vars, i would use -p or -P
<linext_>
customizer parameter file and customizer parameter set
<linext_>
is there an example of using those args
<teepee>
that's just the normal customizer files that are saved automatically
teepee_ has joined #openscad
teepee has quit [Ping timeout: 258 seconds]
teepee_ is now known as teepee
califax has quit [Ping timeout: 258 seconds]
TheAssassin has quit [Remote host closed the connection]
TheAssassin has joined #openscad
califax has joined #openscad
<JordanBrown>
J1A84 "i wonder when someone fixes the convexity issues with render()"
<JordanBrown>
AFAIK render() has the same convexity issues as polyhedron(), import(), linear_extrude, surface(), and maybe others that I am forgetting.
<JordanBrown>
Pretty much anything that creates a 3D object that isn't a cube, cylinder, or sphere.
<JordanBrown>
I don't know why the previewer needs that convexity value, but it does.
<JordanBrown>
I tried to push for increasing the default to 10 or so, so that it would almost always be adequate, but the microscopic-for-most-cases performance costs had some people pushing against it.
<teepee>
that's 2 separate issues
<JordanBrown>
In what way?
<teepee>
whatever convexity is set, render seems to lose that value
<JordanBrown>
And of course right now I can't get convexity to misbehave...
<JordanBrown>
I thought a linear extrude of a concave object would demonstrate the problem, and it's not.
<teepee>
I think there's an issue on github but I can't find it right now
<JordanBrown>
I can't even get the standard convexity misbehavior. Examples that I thought were enough aren't demonstrating it.
<JordanBrown>
I mean, I would be happy if it went away, but I don't think it has.
<JordanBrown>
I grabbed one of my old models that I knew demonstrated the problem, deleted the convexity=5, and indeed it started demonstrating it. (I'll have to track down exactly what makes it complicated enough to occur.)
<JordanBrown>
But when I slap a render() in front of it, no convexity artifacts, even with no convexity specification anywhere.
<JordanBrown>
(What I was forgetting is that you have to have a difference() in the picture, or maybe something else similar.)
<JordanBrown>
render() does seem to discard (or not pick up) the convexity of its children. But if I explicitly include the convexity in the render call, it seems to be OK.
<JordanBrown>
I don't know enough about the use and implications of the convexity parameter to know whether it would be desirable to pick it up from the children. It wouldn't surprise me if children accumulate large effective convexity values, and maybe when you render() you want to start over in that tally.
noonien has joined #openscad
<J1A8431>
just tried some simple forms and the convexity value works .. but i have a complex module where it doesn't
<JordanBrown>
My first suggestion would be to start with the complex module and chop it down to isolate what makes it misbehave.
<JordanBrown>
That's what I get with no convexity parameter to the render.
<JordanBrown>
I've never used fast-csg, so don't know how to turn it on. I checked the checkbox in Features, but that didn't seem to make any difference.
<J1A8431>
yes that is normal .. i get this with convexity = whatever .. in fast-csg
<J1A8431>
flush cache
<JordanBrown>
Indeed.
<J1A8431>
Ü
<JordanBrown>
I think I would characterize that as a problem with fast-csg rather than with render().
<J1A8431>
.. just had someone using a script of mine with a mac and it crashed oscad on his system .. but works fine on my pc
<J1A8431>
yeah looks like fast-csg ignore render or treat differntly
<J1A8431>
when your design renders in 5min with fast-csg .. you will not turn it off again
<J1A8431>
JordanBrown lazy union let you make multi material or multi object 3mf files .. that might be a side effect ..but it does and is very useful
<JordanBrown>
Kind of.
<JordanBrown>
See where I said "except in trivial cases".
<JordanBrown>
Triviality in this case is associated with your material divisions being at the top level.
<J1A8431>
yes i design them that way so i can use the feature that way so i get what i need
<JordanBrown>
I just wrapped a shell script around it.
<J1A8431>
that doesn't produce 1 file does it?
<JordanBrown>
Nope.
<JordanBrown>
Oh, did you mean uploading one 3D object file?
<JordanBrown>
I was thinking of source files.
<J1A8431>
what imbezile wants 30+ files if it fit into one
<J1A8431>
when you use multi material you want 1 file with all components
<JordanBrown>
Want? No. But do I consider it a problem worth spending any energy on? No.
<JordanBrown>
Yes, absolutely. Multi-material you want one file with all the components.
<JordanBrown>
And if I can't have a module that yields a multi-material object, the multi-material problem is not solved.
<J1A8431>
or in my case if a make some mechanics with 10 parts .. i want them in 1 file so when printing the infill is centered on each part not on all parts
<J1A8431>
well you need to adapt the workflow to the tools available
<J1A8431>
i mean i would like to have NLP but it just is not there for 3D design .. but for art and 2D it is .. so lets wait and see Ü
<JordanBrown>
Except that natural language is simply awful if you want to specify something precisely.
<JordanBrown>
I mean, English isn't even clear on what red and blue balls are.
<J1A8431>
well stable diffusion was trained with 600 000$
<J1A8431>
so when you follow the artists there are very specific prompts that lead to the wanted result
<J1A8431>
In the end it would be more iterative like star trek computer .. give me 20 hexagons .. now round the corner .. no the other corners
pa has quit [Ping timeout: 252 seconds]
<JordanBrown>
Yeah, maybe. I suppose that it depends a lot on your design process.
pah has joined #openscad
<JordanBrown>
Anyhow, I was reacting to the statement that having a module yield a multi-material object was a design flaw. I would say that it doesn't *work*, but that it's a perfectly sensible design. The flaw is that the tool doesn't support it.
<JordanBrown>
"You really ought to be able to do that, but can't" is very different from "You really shouldn't do it that way".
<J1A8431>
no it was that it is a design flaw if you put everything into one module without discriminator when you want to separate it later.
<J1A8431>
how can something be the "correct" way if it doesn't work .. sure the workflow will change when the tools change but you can't choose a workflow when the tools are not ready for it ..
<J1A8431>
the correct design is a module that can select every part that should be separated later - if you design that way you can loop through the selection to assign color or material or id or whatever.
<J1A8431>
or you have only one part per module .. then you can use them all as children with another module
<J1A8431>
In the end you need to be able to access the "atom" of your design
<J1A8431>
everything else is flawed
<J1A8431>
» Ancient Greek philosophers called these hypothetical ultimate particles of matter atomos, a word which meant "uncut".«
abff_ has joined #openscad
<J1A8431>
So if you design your model in a way to use "hull separation" to get your parts .. this is working as long you have clearances (and will help to spot errors) however for multi color prints this is not the approach you want.
<InPhase>
J1A8431: But we can give access to the material parts.
<teepee>
maybe we should default without -P if there's only one, but right now I think it requires both
<linext_>
fine by me
abff has quit [Read error: Connection reset by peer]
<teepee>
yes, that should work
<linext_>
generating a lot of parameter sets into STL files is a good example of where other users can parallelize at the same time
<teepee>
the "problem" with the parameter info extraction is that if you give both, input values via -p/-P and export the parameter info, you get the values from the parameter file, not the defaults from the scad
<linext_>
it doesn't get the pen_thickness, pen_inner_diameter
<linext_>
i guess the * symbol is not allowed
<Friithian>
wow I'm surprised makerbot has that, must be an entirely different section of the company to those who made cloud print
N4buc0_ has quit [Read error: Connection reset by peer]
N4buc0 has joined #openscad
<teepee>
yes, that is by design, only literals are picked up, not assignments of more complex expressions
<teepee>
I don't think the code is online though, there's some initial stuff from tbuser who maintained that for years, but I think he left makerbot quite some time ago
<Friithian>
oooh this was made by 1 guy who cared
<teepee>
no idea, I suspect it did have some support in the beginning, but now with stratabot, who knows what's going on
<teepee>
we talked a while ago with one of the admins who took over the stuff, but he never showed up again
<teepee>
seen moos*
<othx>
moos3 was last seen in #openscad 3 years, 302 days, 23 hours, 38 minutes, 51 seconds ago saying 'you can make a thing app :)'.
<othx>
moose was last seen in 7 years, 22 days, 3 hours, 23 minutes, 28 seconds ago saying 'I know this Demod guy so if anyone has an answer to his issue, I will pass it along.'.
<teepee>
looks like he's also not at makerbot (or I'm mixing up people which is totally possible too) https://github.com/moos3
<teepee>
theory: customizer maintained by a student coming by every 3 month for 2 days
<Friithian>
better results than what makerbot normally puts out
N4buc0 has quit [Read error: Connection reset by peer]
N4buc0 has joined #openscad
teepee_ has joined #openscad
teepee has quit [Ping timeout: 258 seconds]
teepee_ is now known as teepee
N4buc0 has quit [Remote host closed the connection]