<J2234>
ok this is annoying .. i have a (complex) polyhedron with a radius, that is copied 3× around (120°) two are fine the third failed with cgal error - if the radius is 0.05 .. it is fine with 0.049 and 0.051
<J2234>
probably some rounding binary floating fuck .. how can a loop rotate() cause a polyhedron to cause "Expr: e_below != SHalfedge_handle()"
arebil has joined #openscad
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
J2234 has quit [Quit: Client closed]
J2234 has joined #openscad
<InPhase>
J2234: testcase?
TheCoffeMaker has quit [Ping timeout: 272 seconds]
TheCoffeMaker has joined #openscad
arebil has joined #openscad
arebil has quit [Client Quit]
arebil has joined #openscad
J2234 has quit [Quit: Client closed]
J2234 has joined #openscad
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
ur5us has quit [Ping timeout: 250 seconds]
arebil has joined #openscad
marcus has quit [Read error: Connection reset by peer]
marcus has joined #openscad
marcus has quit [Ping timeout: 240 seconds]
marcus has joined #openscad
marcus has quit [Read error: Connection reset by peer]
marcus has joined #openscad
marcus has quit [Remote host closed the connection]
marcus has joined #openscad
Joel has quit [Ping timeout: 240 seconds]
Joel has joined #openscad
drfff has joined #openscad
drkow has quit [Ping timeout: 256 seconds]
qeed_ has joined #openscad
qeed has quit [Read error: Connection reset by peer]
linext_ has joined #openscad
EkpyroticFrood8 has joined #openscad
toulene8 has joined #openscad
rawgreaze_ has joined #openscad
<Scopeuk>
J2234 my guess would be our old friend degenerate triangles and grid snapping (rotation causing alignment. a test case for it as InPhase says would be awsome though
EkpyroticFrood has quit [Quit: Ping timeout (120 seconds)]
toulene has quit [Quit: Ping timeout (120 seconds)]
linext has quit [Ping timeout: 256 seconds]
rawgreaze has quit [Remote host closed the connection]
EkpyroticFrood8 is now known as EkpyroticFrood
markasoftware has quit [Quit: No Ping reply in 180 seconds.]
toulene8 is now known as toulene
<J2234>
i am working on one .. ( as always it is some weird tinkered module in my lib)
la1yv_a has joined #openscad
markasoftware has joined #openscad
<J2234>
InPhase Scopeuk at the moment the problem doesn't came up with a normal loop but only with my "Polar(3)" which is a loop internal but with more options - and there it seems something strange happens
<J2234>
if you remove the ! .. the front faces are gone and you have mesh not closed - in my local version the faces are there but the third object is gone
ur5us has joined #openscad
<J2234>
when exporting the 2 objects from (2022.03.21) they look fine and meshmixer can't finde any issue when analysing.
<J2234>
one of those again .. if you render both they are fine .. if you remove the top one .. the bottom one is only 2
<J2234>
(top one has working values while the bottom one causes issues - but if the working one is present - the issues are gone)
lastrodamo has joined #openscad
othx has quit [Ping timeout: 248 seconds]
othx has joined #openscad
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
arebil has joined #openscad
lastrodamo has quit [Quit: Leaving]
J2234 has quit [Quit: Client closed]
J2234 has joined #openscad
Guest9159 has joined #openscad
<Guest9159>
Hey there!
<Guest9159>
Is there a way to animate with View -> Wireframe?
<Guest9159>
Every time I pick Wireframe and then set Steps for animation, View sets to Preview before animation starts.
ur5us has quit [Ping timeout: 250 seconds]
TheAssassin has quit [Remote host closed the connection]
TheAssassin has joined #openscad
Guest9159 has quit [Quit: Client closed]
TheAssassin has quit [Remote host closed the connection]
TheAssassin has joined #openscad
<InPhase>
J2234: Well the implementation of Gewinde is very difficult to follow from the classic lack of whitespace. But when I try the example with or without the !, I see some massive z-fighting going on with the top of that cylinder. Do you have an overlap issue hiding within Gewinde?
<InPhase>
J2234: Or wait, there's not a cylinder in the one online...
<InPhase>
Ok. I updated ub.scad from the git repository, which no longer seems to work under the last release, so I switched to the master branch. Now there is no cylinder.
<InPhase>
J2234: Ok, I take out your for loop, take out the !, and just make it i=0; and then I add a cube(1); and I get the same render issue. So the problem exists with only a single instance.
<InPhase>
I'm not really getting your exception that you previously reported, but the models are defective in some manner if they're hollowing out like that, so later exceptions are not surprising. But I cannot discern where it would be as it is hiding somewhere in Gewinde.
<InPhase>
J2234: In Polar I notice you are doing a rotate in a for loop of the children.
<InPhase>
J2234: So is this doing an arrangement of Gewinde4's such that it violates the overlap rule?
<InPhase>
Or wait, g went to e, and e was then 1, which meant only one instance?
StudentWeis has joined #openscad
StudentWeis has quit [Client Quit]
<InPhase>
J2234: Well that about ends that exploration. GewindeV4 (which I think is the one it's using) is pretty indecipherable in its complexity without a ton of study. I confess my own thread implementation has some complexity issues like that (although with a bit more code layout alignment). I can share that if I got myself into a non-manifold problem like that with it and set out to fix it, I'd probably
<InPhase>
attempt a radical refactor to try to break it into much smaller bite-sized pieces with clearer defined well-labeled input/output boundaries. This is the strategy that tends to insert decipherability into complexity.
<InPhase>
Generally you run some working examples when doing the refactoring, and then go back to the non-working one to see what's going on once things are clearer.
StudentWeis has joined #openscad
StudentWeis has quit [Client Quit]
<J2234>
inPhase thanks for your effort - the think i wonder is how i can render a single instance fine without error but 3 of them causing an error but only in one of them - and when adding a slightly changed version - the ones that didn't render suddenly render fine .. this is very unexpected behavior
myosotis has joined #openscad
<J2234>
i found a way to get this rendering https://imgur.com/a/6kcZaJS - i assume the problem ist that in the end / start region points are getting very close to each other which causing these problems.
<J2234>
(normaly the loop is not needed if you use g=3 .. but - - well complicate to explain you will get warnings from the module as this is not a viable thread geometry if you do full turns )
<InPhase>
Well somebody's work slowed down the clean build of master a bit.
<InPhase>
From 2.5 minutes up to 4 minutes.
<InPhase>
Well hmm, now it works.
<InPhase>
Maybe that was a fast-csg issue. I forgot I had left it turned on.
<InPhase>
Okay, but I also now have no issues with the full design.
<InPhase>
The problems with this design are completely gone with 2022.04.14 on my system.
<J2234>
well sounds like i should update Ü thanks
<InPhase>
If that doesn't work, then there's something weird going on. If it does, then that makes sense. You were perhaps just poking at the edge of what fast-csg was doing correctly during that Feb/Mar work.
<J2234>
äh wait .. did you try rad1=0.049 ?
<InPhase>
Pastebin a version currently broken for you with that.
<InPhase>
If I start editing I might end up with something slightly different.
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
lastrodamo has joined #openscad
<joseph_>
teepee InPhase: I have made good progress on the proposal. It's on GitHub, but I made it a private repo because it contains info like my email. To confirm, which GitHub usernames should I share it with?
<teepee>
or could you maybe just split into public / private?
<teepee>
it would be nice if other people could have a look and comment if they want to. I don't expect that many will do so, but some people here in the channel might have some additional insight
califax has quit [Remote host closed the connection]
califax has joined #openscad
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
ToAruShiroiNeko has quit [Ping timeout: 256 seconds]
ur5us has joined #openscad
<InPhase>
joseph_: Yeah, the rcolyer one is me. thehans is peepsalot on here.
lastrodamo has quit [Quit: Leaving]
<joseph_>
teepee InPhase: I just sent the invites. I've only so far described #55 and haven't added in issue #56 yet. I wanted to hear your thoughts about the merits of the medium 175 hour option vs long 350 hour project. At this point my schedule could potentially accomodate either
<joseph_>
Once we've discussed your initial thoughts, I'll copy the bottom section of my proposal (excluding my personal section) and put it in a public repository so the whole channel can offer feedback
<teepee>
joseph_: we discussed yesterday that the wording of #55 is a bit misleading as it talks about Preview and that's a bit special
<teepee>
the clearer path is probably targeting the display of the F6 result, often called "render"
<teepee>
as in that case the mesh fully exists and there's no trickery going on as with the F5 "preview"
<teepee>
just having dinner, will read after that :)
<joseph_>
teepee: Okay, sounds good. So to confirm, even the Preview.vert and Preview.frag shaders are actually influencing the F6 "render" only? If this is the case then I will disambiguate my proposal to always refer to it as the "3D viewport render output"
<teepee>
joseph_: I think they are only used in a special preview case with "view edge" enabled
<teepee>
I suppose render output should cover things, or maybe just mention that there are different display modes and we target one specifc case
<teepee>
so you don't have to repeat that all the time
ToAruShiroiNeko has joined #openscad
teepee has quit [Remote host closed the connection]
myosotis has quit [Quit: myosotis]
teepee has joined #openscad
peepsalot has joined #openscad
<peepsalot>
joseph_: did you see any of the messages from last time I addressed you (on 4/12)? not sure because you had no response
<peepsalot>
eg: shaders are currently tied to only working for F5 preview with "show edges" enabled
<peepsalot>
and there is an issue with some unsightly artifacts for objects under CSG (difference or intersection) much worse if you use perspective view as opposed to orthogonal
<peepsalot>
snippet to reproduce under F5, with show edges enabled, and orbit the camera around: difference() { sphere(10); cube(10); }
<peepsalot>
also, there's the whole VBO vs non-VBO code, so display is effectively implemented twice
<peepsalot>
sorry if I seem overly pessimistic, but trying to give a accurate representation of the state of things. which is unfortunately pretty complicated at the moment for GL code
<teepee>
right, maybe it's time to switch to VBO?
<peepsalot>
yeah, i'd be fine with that
<teepee>
it did get promoted from time to time on the mailing list and I have not heard any issues
<teepee>
I guess that still does not mean there's no issues to come but probably the best that can be done upfront