<kintel>
teepee The 3mf export looks reasonable, didn't do a deep review, but ran it casually on macOS and it worked. Didn't exercise all the options though. Let me know if you want a proper review :)
<teepee>
I'm thinking of adding support for commandline too. lets see if that happens shortly, otherwise it would make sense to not let it bitrot
<teepee>
kintel: ahem, "playground repo" :-)
<teepee>
= WASM
<teepee>
there's multiple repo notifications
<kintel>
oh, I keep forgetting to read the URL :)
<kintel>
Even nophead is moving to the web version, perhaps time to consider elevating that to an official surface?
<teepee>
yes, moving slowly into that direction
<teepee>
missing: a build pipeline for it
<kintel>
It will take a while before we support previews though, but what's there is a great start
<kintel>
It makes me wonder if it's feasible to merge the entire implementation of native + web, including (most of) the UI. That might become a massive project though.
<teepee>
merge how?
<kintel>
..but apparently, Qt6 supports building for the web (perhaps not with the classic widget set though)
<teepee>
yeah, I don't think a simple 1:1 GUI would work on a web page
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
snaked has quit [Remote host closed the connection]
<Scopeuk>
There is some debate if with the manifold speed boost there is as much value remaining in preview as their once was
guso78k has joined #openscad
<guso78k>
maybe somebody here can aid me with little 3d maths. if i have two vectors v1=[1,0,0]
<guso78k>
.. and v2=[0,0.7070, 0.707), how can a gradually change v1 into v2 direction in e.g. 5 linear/angular steps ?
<guso78k>
of course i can calculate the total angle between v1 and v2 and i can calculate the delta angle of all steps, but how do i get the actual 3d directions ?
germ has quit [Read error: Connection reset by peer]
<teepee>
gradual how?
<teepee>
like just linear of the end point?
<teepee>
or rotating somehow and scaling length so it's an arc?
<guso78k>
teepee actually rotatating between v1 and v2 while maintaining the unity length
lastrodamo has quit [Quit: Leaving]
<guso78k>
i believe, that angle_axis_degrees function can help here
germ has joined #openscad
mmu_man has quit [Ping timeout: 252 seconds]
<teepee>
I don't know that one
<teepee>
plus I'm very distracted by c++ templates :)
mmu_man has joined #openscad
<guso78k>
this is the very function which is used when you rotate(v=vector, a=angle) object in openscad
<InPhase>
guso78k: I have a routine for this.
<InPhase>
Let me see what happens if I do it in steps...
<guso78k>
InPhase, I am loooking for a effective funcion, as it might be used a lot ;)
<teepee>
omg! it did compile
<InPhase>
guso78k: https://bpa.st/5CJA So part of the ambiguity comes down to which you consider the correct rotation between vectors, and also how you choose to define the in between path.
<InPhase>
guso78k: But this is one definition. There exist others.
<guso78k>
InPhase thx! will look into your past now!
<InPhase>
guso78k: I plotted the path with a loop so you can see how that rotation sweeps out.
<InPhase>
Uh oh, will you hire a PI?
<guso78k>
PI ? hire ?
<InPhase>
To look into my past. ;)
<InPhase>
PI == Private Investigator
<guso78k>
i justy know the rasperry thing , ahh
<InPhase>
(You wrote post as past. I was just riffing off of the typo.)
<guso78k>
i am just a private programmer without the ability to hire specialists ...
<guso78k>
haha,
<guso78k>
InPhase , i just tried your example and the result looks like great art. I feel you should sell it
<guso78k>
actually i need the stuff in c++, i need to check if i can use the idea behind your code
<InPhase>
guso78k: I guess the straight-arc version is not hard to do either. This does weighted vectors between the two, so you get a linear combination of the directions. It requires normalizing the inputs first: https://bpa.st/repaste/N6CQ
<InPhase>
That arc one is basically the "geodesic" path on a sphere. Although I think the arc steps are not even along that path with that weighting method.
<InPhase>
There exists another update one can do the the weightings to make the arc steps even. It escapes from the top of my heaad, but it could be figured out. But I'll let that simmer as some examples.
<InPhase>
s/the the/to the/
<guso78k>
InPhase, thanks, definitely helpful
<InPhase>
And yeah, that should convert to C++ trivially. Rotations are done easily with matrix operations. You could reduce that all down to a single set of matrix operations with some linear algebra.
<InPhase>
A good symbolic math engine might reduce and simplify it even further.
<InPhase>
As the comment notes, this is the set of operations chosen for xy plane orientation preservation.
<InPhase>
Because I figure this is most often what is desired, and at the very least, one can provide transformations when wanted to make use of a well defined orientation preservation.
<InPhase>
I no longer remember how I derived it, but just thought it out carefully and recorded the answer.
<guso78k>
Yes, i have derive6 in my toolbox
<guso78k>
xy plane preservation is not intended here. i want to stay flexiable in all direction
<guso78k>
https://imgur.com/a/AWWHnHW i need it to get rounded corners, i dont like to follow the minkowski approach
<guso78k>
https://imgur.com/a/pdba8zp after accepting, that the angle_axis_degrees takes degrees instead of radians
<guso78k>
reading often helps O:3
<guso78k>
https://imgur.com/a/tfzDsgu the corners are quite round, but there are still some optical artefacts from trying to mesh the connection from 4 faces with triangles. 4 != 3 ...
guso78k has quit [Quit: Ping timeout (120 seconds)]
guso78k has joined #openscad
guso78k has quit [Client Quit]
guso78k has joined #openscad
sinned69158 has joined #openscad
<InPhase>
guso78k: Well there's always a favored orientation. There's no one vector to vector definition, but an infinite set with every resulting axial orientation.
<InPhase>
guso78k: This is best exemplified with [1,0,0] to [-1,0,0] where it's painfully obvious that all transformations are equally favored. And actually my Arc example breaks for this, but the Steps example chooses a favored path and works.