<pca006132>
for manifold, because we work with floating points, we don't really care about if the number is rational or not
<pca006132>
hard to say if the epsilon is an absolute value or not, because it changes according to the model size and operations on it
<InPhase>
pca006132: Yeah, I went with just generalized high dynamic range floating point abuse attempts.
<InPhase>
But at the very least, nothing seemed to end up mis-ordered or value tweaked except for that translate bug I identified at the largest values. Something was somewhat amiss there.
<InPhase>
Although it was intriguing that the shape stayed preserved.
<InPhase>
It seemed like maybe someone was doing a differential calculation for translate. Like rather than subtract all points, perhaps manifold translates a reference point for an object, and recomputes the object with respect to that reference point to preserve relative shape information?
snaked has quit [Quit: Leaving]
<pca006132>
InPhase not really
<pca006132>
I did thought about it, but it probably is not worth it to do so
<pca006132>
but we do apply transformations lazily, so the matrices are combined first
<pca006132>
that may reduce some numerical error
<InPhase>
Hmm, that might have been sufficient for the result I got.
<InPhase>
I will have to percolate in my head to see if there's a different abuse of the system that would trip it up with the lazy transformation approach.
<InPhase>
Although still something resulted in a net shift from moving and moving back.
<InPhase>
Some value mutation must have occurred.
<pca006132>
you can force transformation by unioning with something disjoint
<pca006132>
because so far we don't track per vertex transform :P
<pca006132>
it is possible to make things more precise, but the issue is whether it is worth it
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<InPhase>
pca006132: Ah, well I had in fact done a difference operation before translating back.
<InPhase>
So that probably blocked the lazy matrix operations.
<bambukah>
I was in the MakerFaire Lille yesterday, and I speak with some Prusa team member about OpenSCAD :D
cart_ has quit [Ping timeout: 264 seconds]
<buZz>
f prusa
<buZz>
they'd just fork it to PrusaSCAD and never push back changes
<bambukah>
oh shit... I don't know that, sorry :'(
<bambukah>
I talked about OpenSCAD to other people like PolyFab3d (a French 3dPrint seller) and FrancoFil (Filament creator & seller)
<buZz>
hehe its fine
mmu_man has quit [Ping timeout: 252 seconds]
<InPhase>
Prusa is one of the rare cases of someone who used to do and release open source designs in OpenSCAD, endorsing the merits of both the approach and the open source benefits, and then walked away from that converting their approach to proprietary gui-based programs.
cart_ has joined #openscad
<InPhase>
It coincided with the heavier commercialization of the Prusa printers
<InPhase>
And of course it's completely legitimate and helpful even to attempt to commercialize printer production. But I did not and do not understand the rationale for converting their approach to model design.
<InPhase>
Especially considering that constructing designs for printer parts is one of those things OpenSCAD truly excels at, as it aligns exceptionally well with everything that OpenSCAD is particularly good at.
<bambukah>
So, for my projects (which is made with OpenSCAD, what is the best 3d printer brand for a commercial partner ?
mmu_man has joined #openscad
mmu_man has quit [Ping timeout: 265 seconds]
mmu_man has joined #openscad
GNUmoon has quit [Ping timeout: 260 seconds]
juri_ has quit [Ping timeout: 265 seconds]
mmu_man has quit [Ping timeout: 255 seconds]
bambukah has quit [Quit: Client closed]
GNUmoon has joined #openscad
juri_ has joined #openscad
juri_ has quit [Read error: Connection reset by peer]
mmu_man has joined #openscad
juri_ has joined #openscad
cart_ has quit [Ping timeout: 276 seconds]
L29Ah has left #openscad [#openscad]
califax has quit [Remote host closed the connection]
califax has joined #openscad
krushia has quit [Ping timeout: 264 seconds]
krushia has joined #openscad
crazy_imp has quit [Ping timeout: 246 seconds]
crazy_imp has joined #openscad
L29Ah has joined #openscad
<Scopeuk>
InPhase my best guess is it is much easier to find mechanical design engineers who are happy with that design flow. probably adding in some of the analysis tools that go with it.
<Scopeuk>
I'm not sure it's necessary but I can see their might be reasons to do so
mmu_man has quit [Ping timeout: 252 seconds]
<JordanBrown>
pca006132 I think I came up with a high-performance (O(N)) scheme for deduplicating "close" points. Send me email if you're interested.