<ran>
which combination of these options yields the best performance?
<InPhase>
Unclear, and it might vary, although the first 2 are strongly recommended.
<InPhase>
"exact" almost certainly slows everything down.
<InPhase>
The remesh one is unclear.
<InPhase>
remesh predictibly is probably a slowdown and definitely spelled wrong.
<InPhase>
There were at some points a very small number of testcases where fast-csg actually made the whole render slower. But I think all of the known cases of that were resolved. But because it is fundamentally different, it's possible some other cases exist where this could happen. But it is overwhelmingly between a major speed up and close to even on almost all cases.
<InPhase>
Common values one can expect are 5-10 times faster, but sometimes more than that, sometimes a bit less.
<ran>
thanks
<InPhase>
fast-csg-remesh probably depends on precisely what you're doing. For example, minkowski of minkowski of minkowski probably benefits a lot from that. But... it also fundamentally changes designs from their specification, which could be a significant issue in some cases. You might lose edge integrity that you were intending to preserve. So that makes it a deviation from normal expected operation.
<InPhase>
I know that for certain mechanical designs it mattered a great deal to the outcome I achieved to align the edges of the tessellation with the mechanical motion to get a smoother result. Even features that one might otherwise think of as below print resolution can have some moire rippling effects that bleed into the friction response of a part if this is not attended to.
<InPhase>
A remesh blows that sort of property out of the water.
<InPhase>
Scopeuk: Ah, thanks. I tried to watch that the first time.
<teepee>
probably, trying to spoil the fun others ahve
<teepee>
*have
<teepee>
that was like 2 seconds before the time where it would have to abort
<guso78[m]>
what type of printer to you need to print a rocket ...
<teepee>
big one :)
<teepee>
i think they said it's 7m diameter?
<InPhase>
Presumably a metal printer...
<JordanBrown[m]>
Or lots of superglue.
<guso78[m]>
any 3d printed part is not as tight as a mold created plastic
<teepee>
they should have quite a number of videos online of the printer
<guso78[m]>
wondering it that is suitable for airospace applications ...
<teepee>
it's probably very challenging but with high reward in regard of being able to create complicated structures
<teepee>
like printed rocket motor
<InPhase>
Back on.
<teepee>
yep, 2 minutes
<teepee>
oh, no
<teepee>
nothing happened
<InPhase>
Early rocketry is full of anticlimactic moments. :) (Interspersed with kabooms. But no kabooms today.)
<J2356>
who was using the far away zoom behind trees cam for the launch moment?
<teepee>
it's still a hold at T-0
<InPhase>
J2356: Yeah, that was a bad choice. Maybe they'll fix it for the real launch.
<InPhase>
teepee: That's probably a scrub for the day.
<teepee>
some palm tree nut? :)
<teepee>
yeah, likely, but they still call it hold - what they earlier said is not possible
<InPhase>
I think that's the default state transition, and then a human needs to process the state transition to scrub. Although typically once you go past "go/no-go" their protocol doesn't allow any other choices.
<teepee>
yep, they said after T-70 it's all automatic abort
<teepee>
now lets see if the'll give it another go today
<J2356>
now it turns into a popsicle
<teepee>
where's the people frantically typing on their keyboards?
<teepee>
the control room looked bored
<InPhase>
Hmm, ok. Perhaps they think they can still go.
<teepee>
looks like launch window is another 1,5h or so (if I heard that right)
<teepee>
so now they need to find the person saying "yeah, yeah, all good, I fixed that sensor by setting it to ignore"
<J2356>
45min
popejohnpaul has joined #openscad
<popejohnpaul>
is there a way to rotate an object with respect to the center of the object instead of with respect to the origin?
<JordanBrown[m]>
OpenSCAD doesn't know where the center of the object is.
<JordanBrown[m]>
If you know where the center of the object is, translate it to the origin, rotate, and translate it back out. Writing a module that does that is pretty easy.
<popejohnpaul>
the nvm I'm just stupid
<popejohnpaul>
my problem was that I was rotating before translating instead of translating before rotating
<JordanBrown[m]>
That would be bad.
<popejohnpaul>
I'm kinda a newbie I'm just testing out some primitives
<JordanBrown[m]>
Newbies welcome!
<InPhase>
popejohnpaul: Yeah, it's generally easiest to maximize the work done at the origin, and then place things.
<InPhase>
popejohnpaul: As designs get more complicated, that principle holds, and you keep doing more on each part at the origin.
<popejohnpaul>
yeah that's generally what I've been doing
<InPhase>
If you have subparts of parts of a larger piece, build up each at the origin in a nested manner.
pah has joined #openscad
<InPhase>
What you'll find is that this makes the parametric qualities stronger, because the whole design will adjust better to parameters that define the total part shape, but without those parameters being spread throughout the design.
<popejohnpaul>
that's what I've been trying to do
<InPhase>
Then great start. :) Also, welcome.
<popejohnpaul>
when I was rotating I was testing how two physically separate pieces fit together
<popejohnpaul>
probably a better way to do it but the polyhedron doesnt play nice with center
<InPhase>
polygon I hope?
<InPhase>
Manual polyhedron is a lot of work.
<InPhase>
I have written myself a bunch of polyhedron interface libraries just because I don't like doing it manually. :)
<InPhase>
But polygon with linear_extrude or rotate_extrude remains pretty elegant to do manually.
<popejohnpaul>
actual polyhedron
<InPhase>
Is it not symmetric across z?
<popejohnpaul>
I just stole the cube example from the tutorial, change some parameters, and union'd it with some cylinders
<popejohnpaul>
then I took that module and union'd it with a rotated and translated version of itself, then I differenced that fat cylinder with the doublesided thing
<popejohnpaul>
that was the only time I'd need a polyhedron for the this project so I just did it
<popejohnpaul>
the rest is primitives
<InPhase>
popejohnpaul: An illustration of the value of working in 2D: $fa=1; $fs=0.4; linear_extrude(8) hull() { for (i=[0:3]) rotate(i*90) translate([(i%2)*15+7,0]) circle((i%2)*0+2); }
<popejohnpaul>
wow that is way simpler than what I did
<InPhase>
popejohnpaul: Now that modulo trick is extra that just happens to fall out of your particular shape. But even without it the 2D approach will easily come out simpler.
<InPhase>
Ah, and mine might be a bit incorrect if you actually needed those notches.
peeps[work] has quit [*.net *.split]
cbmuser has quit [*.net *.split]
juri_ has quit [*.net *.split]
Ckat has quit [*.net *.split]
mohnish has quit [*.net *.split]
escherial has quit [*.net *.split]
fardog has quit [*.net *.split]
n1essa has quit [*.net *.split]
juri_ has joined #openscad
mohnish has joined #openscad
cbmuser has joined #openscad
Ckat has joined #openscad
escherial has joined #openscad
fardog has joined #openscad
little_blossom has quit [Quit: little_blossom]
fling has quit [Remote host closed the connection]
n1essa has joined #openscad
little_blossom has joined #openscad
fling has joined #openscad
fling has quit [Remote host closed the connection]
fling has joined #openscad
<popejohnpaul>
is my computer just bad or is there a reason the viewport gets super laggy when I do this in my file?
<J2356>
popejohnpaul depends on the $fa value how many fragments generated
<J2356>
so i assume you have $fn defined with a high value
<popejohnpaul>
I didn't set $fn
<popejohnpaul>
but I removed the piece of code that set $fa to 1
<popejohnpaul>
it's still laggy though
<popejohnpaul>
but usable now
<JordanBrown[m]>
I agree that it's a bit laggy, and I don't immediately see why.
<popejohnpaul>
like the shape seems simple enough
<JordanBrown[m]>
I'm sure that it would be faster if wedge directly generated the wedge, rather than doing a difference(), but there are no trivial ways to do that.
<JordanBrown[m]>
rotate_extrude() and linear_extrude()polygon(... trig to make an arc...) are the two obvious techniques.
<JordanBrown[m]>
(Sorry, when I say "obvious", I don't really mean "obvious"... I mean obvious to somebody who is totally familiar with the program.)
<popejohnpaul>
yeah this is for a print so it doesn't matter at all to me
<InPhase>
Someday soon I hope we will figure out how to wedge color processing into rendering, but there is not an existing plan for this yet.
<InPhase>
(For multicolor/multimaterial prints.)
<J2356>
popejohnpaul $fa will not have any effect if you use $fn. $fa is limiting fraqments by a max angle .. set to 1 limits to 360 fragments
<J2356>
but if you set $fn then that number is used .. better is working with $fs as this will determine the fragment size not number
<InPhase>
General strategy for printing: $fa=1; $fs=0.4; if 0.4 is the extruder nozzle diameter. Then don't use $fn at all except in special cases like making a hexagon out of a circle/cylinder, or some case where you want to align a circle to some integer multiple of 4.
<J2356>
and if you use $fa=$preview?5:1; you will have fast preview and only for render the full resolution
<InPhase>
Also an option if things are getting laggy.
<J2356>
as you have 200 unit big parts a $fs of .4 will give you high fragment count and a slow preview
<InPhase>
Although aesthetically I really like seeing the smoothness early on to get a sense for what I'm going to get.
<J2356>
just set $preview=true; to check Ü
<J2356>
s/ false
<InPhase>
lol
<popejohnpaul>
I don't suppose any of you know the standard dimensions of a servo motor do you?
<popejohnpaul>
all the servos I'm looking at have these little wings on the end to hold them in place but I'm not sure whether that's counted in the length or not