<InPhase>
rue_mohr: STL files are not in general fixable. They are sometimes fixable. It depends on how messed up they are, and exactly what the nature of the problem is. There have been discussions about adding some capability for this, but it is not yet in place.
<InPhase>
rue_mohr: The fundamental problem is that many people out there are designing STL files with programs that are really not built for making 3D printing models, and they sometimes make STL files that are quite substantially invalid. For objects designed for display, rather than for printing, only the surface matters. But for printing, everything that goes on inside matters too (some objects are
<InPhase>
intentionally hollow or structured inside), which is a very different scenario.
<InPhase>
rue_mohr: For example, when OpenSCAD merges a cylinder going through a cube, it makes a new object representing the outer surfaces of this combined structure. But some display-oriented programs still have the cylinder having an outer surface inside the inner surface of the cube, and it's not mathematically obvious in an automated way what that is supposed to mean, and what's really the outside or the
<InPhase>
inside.
<myosotis>
I use the 3dprint tools in blender, but that's a lot like gambling
<InPhase>
And then for every one scenario you invent a solution for, there are more. :)
<rue_mohr>
InPhase, I do not think I have seen a model yet from openscad that slic3r didn't report having to make a few hundred fixes to
<rue_mohr>
I get ya tho
<InPhase>
rue_mohr: I've not tried slic3r, but I've never seen an OpenSCAD model cura had any complaints about, except a few cases where critical features were below the minimum print dimensions, which were design flaws.
<InPhase>
Does slic3r have some special expectation for stl content that we don't satisfy by default?
<InPhase>
Also, have you ever tried another OpenSCAD output format like 3mf in slic3r?
<InPhase>
I guess I vaguely remember a few designs from the old days that didn't load up in Cura. In some much older releases sometimes invalid outputs with design flaws didn't quite trigger the necessary error messages and responses. But that's pretty much all resolved now at the render step.
Junxter has joined #openscad
ur5us__ has joined #openscad
ur5us_ has quit [Read error: Connection reset by peer]
<rue_mohr>
the other one isn't taking to the repaired STL
LordOfBikes has quit [Ping timeout: 256 seconds]
myosotis has quit [Ping timeout: 272 seconds]
LordOfBikes has joined #openscad
arebil has joined #openscad
<InPhase>
Might not actually be repaired. It is notoriously difficult to do.
KimK has joined #openscad
<InPhase>
I suspect at this point the global success rate for people setting out to repair randomly found stl files actually succeeding is somewhere less than 50%.
<InPhase>
It might even be less than a third.
ur5us__ has quit [Remote host closed the connection]
ur5us has joined #openscad
myosotis has joined #openscad
J2227 has joined #openscad
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
J22 has quit [Ping timeout: 256 seconds]
ur5us has quit [Ping timeout: 240 seconds]
<rue_mohr>
arg, I'm trying some other programs to modify it and no success so far
<rue_mohr>
well
<rue_mohr>
I created this
<rue_mohr>
but
<rue_mohr>
with a few programs, regenerating it is much harder than just cutting the hole I need in it
<rue_mohr>
or, it should be
<rue_mohr>
the last program that modified it was openscad
<rue_mohr>
I'm not sure what kinda error I'm looking for
myosotis has quit [Quit: myosotis]
<InPhase>
Do you have an image somewhere of the model you've been trying to import?
<rue_mohr>
I have a hole in a hole that causes the render anomoly
<rue_mohr>
but if I remove that it still cant do it
<rue_mohr>
my best idea is to try to go back to the two stls I made it from, precut and remerge the parts
<rue_mohr>
I cant redo it from scratch, I dont have a way of re-radiusing the corners of it like that
<InPhase>
That can definitely be done. It's not a 15 minute design, but you started working on this 6 hours ago (not sure if you were working the whole time). It should definitely be redoable from scratch in a lot less time than that.
<rue_mohr>
it was made over the last 6 years tho
<rue_mohr>
there are countless modifications over the process
<rue_mohr>
my origional 3d design guy had 3 strokes and cant move his hands anymore
<rue_mohr>
I think he used solidworks
<rue_mohr>
results from meshlab are completely corrupt going into slic3r
<InPhase>
I'd use some combination of minkowski with spheres, slapping together different rotate_extrude segments to get the smooth edge and corner pieces, and maybe a bit of this library, like to make the holes easier: https://github.com/rcolyer/smooth-prim
<rue_mohr>
its frustrating that it can preview but not render
<InPhase>
Preview just stops when it finds a displayable surface.
<InPhase>
It doesn't care what the volume is.
<rue_mohr>
hmm
<InPhase>
But the plastic goes into the volume. :)
<rue_mohr>
huh
<rue_mohr>
if I do a difference with an object that doesn't even touch it, it still faults...
<InPhase>
Yes.
<rue_mohr>
so it cant break it down?
<rue_mohr>
e_below != SHalfedge_handle()
<InPhase>
Because any union or difference operation will pass it into the library which will attempt to determine whether or not the difference or union intersects, and it will give up when it realizes there is no defined inside or outside for a non-manifold object.
<rue_mohr>
non manifold means its two volumes, right?
<rue_mohr>
or its just not closed?
<InPhase>
It means it's not closed, or self-intersects, or some other similar problem.
<rue_mohr>
in meshlab I saw no anomolies, slicer can load that stl with "no repairs"
<InPhase>
Basically, the inside and outside are no longer well defined.
<rue_mohr>
I suppose the anomoly could be infentisimally small
<InPhase>
Yep, it's often small.
<rue_mohr>
oh wait
<rue_mohr>
I need to zoom out here
<rue_mohr>
this is all ok
<rue_mohr>
I dont need to modify this model actually
<rue_mohr>
the screw that we put in is almost cosmetic
<rue_mohr>
right...
<rue_mohr>
the goal is to change it to fit a nut so we can use a machine screw
<rue_mohr>
casue the wood screws keep splitting the plastic
<rue_mohr>
ok, well, I'll remember I have to backtrack or restart if I ever want to modify that model
aiyion1 has quit [Remote host closed the connection]
aiyion1 has joined #openscad
<rue_mohr>
thankyou
arebil has joined #openscad
use-value has quit [Quit: use-value]
arebil has quit [Read error: Connection reset by peer]
<InPhase>
rue_mohr: If you don't need a lot of them, never underestimate what you can accomplish with epoxy.
<InPhase>
I'd be happy to give tips later on a redesign, but easiest path to the critical solution is wisest. :)
ferdna has quit [Quit: Leaving]
toulene has quit [Quit: Ping timeout (120 seconds)]
<InPhase>
rue_mohr: This sort of filler task is what epoxy pastes were invented for.
<InPhase>
rue_mohr: It's not as precise as printed work, and not as reproducible. It's a quick hack. But much progress in my life has been made with quick hacks done selectively where they are appropriate. :)
oldlaptop has quit [Ping timeout: 250 seconds]
oldlaptop has joined #openscad
teepee has quit [Remote host closed the connection]
<Scopeuk>
linext I wonder if you could do a sort of "rod projection" and have hollow tubes that run all the way from the back of the shape to the immage surface
<Scopeuk>
would give a path out for the resin and could look cool (no guarentees)
<J2227>
teepee is it much hassle to have a $lpr for the lightsource position? At the moment it is nearly impossible to get true colors or only with distorted geometry
<teepee>
I don't think that would work with the current lighting
<J2227>
or just an option to have light from the viewport
<J2227>
so a color is that color in an exported image
<J2227>
at least for a flat surface
<J2227>
(when viewed orthogonal.. so like a square from z top view)
<teepee>
there's a ticket for better lighting and maybe making it configurable
<teepee>
I'm not sure we should extend the inflation of $ variables :)
<teepee>
there's maybe other way to get the same feature, the variable assignment is a bit messy
<J2227>
as long we have no shadows there is not much need for changing the light .. except to get 2D colors .. so maybe 2D objects could just have flat color without lighting
<teepee>
IIRC the ticket asks for multiple lights
<J2227>
sounds like overexposure
ur5us has quit [Ping timeout: 250 seconds]
<J2227>
would be nice so you don't need to use processing to get a proper image for CGI maps
<J2227>
(or to tilt the object for a scene background color )