00:52 <kintel> J25K53 The "strange" triangles is just a shading artifact; we use per-face normals, and in this case a non-planar quad was generated which needed to be subdivided into two triangles, and the two triangles are not in the same plane

00:53 <kintel> Btw., I'd love to see that crash reproducible outside of Windows, but we've been unsuccessful so far..

00:55 <kintel> Btw. if the nonmanifold issue is really about double-precision coordinates being merged into the same when converted to floating point, it would be interesting to pass each model through a very strict mesh optimizer. I believe the Manifold folks were also talking about that at some point.

01:37 <tja> I think the docker image update mentioned in #5972 minorly broke the wasm build workflow -- I needed to throw in a reset "WORKDIR /src" at the end of the ephemeral Dockerfile in "scripts/wasm-base-docker-run.sh" to get a build (from earlier this year) working. I haven't yet pulled in upstream changes on my old branch, but as far as I can tell this isn't fixed in master's copy of "scripts/wasm-base-docker-run.sh"

01:43 <J25K53> kintel I do understand what is happening when using quads, just wonder why there is only this single region and not symmetrical on the other half. but there is only one area while all other 7 identical regions of the cube are fine.  Guess it is ok for higher resolutions.

02:03 <kintel> J25K53 hmm. I think some careful inspection may be warranted..

02:06 <kintel> Wait, where is that screenshot from? Still MeshMixer? ..and it looks good in OpenSCAD?

02:44 <gbruno> [github] kintel pushed 1 modifications (Fix GUI test resource copy) https://github.com/openscad/openscad/commit/34f612071fdf137e5b2c6ea581b9fbdba34db872

02:44 <gbruno> [github] kintel opened pull request #6119 (Fix GUI test resource copy) https://github.com/openscad/openscad/pull/6119

04:13 <J25K> kintel dont't got issues in openSCAD  (expect the 3mf export crash)

04:16 <J25K> I quite often have working 3mf files that show some errors on inspection with other 3d tools - but print quite fine.

08:02 <gbruno> [github] dizcza opened issue #6120 (Make 3D/2D rendering export button optional) https://github.com/openscad/openscad/issues/6120

14:41 <pca006132> sorry, I was on leave for 2 weeks with no computer (intentionally)

14:41 <pca006132> is there anything that I should check?

14:43 <kintel> pca006132 I don't think so, but I have one question :)

14:44 <kintel> We've had a few cases where external tools identify OpenSCAD exports as nonmanifold. I suspect it's a precision-related issue where double->float might merge some very close vertices making some triangles zero area

14:45 <kintel> Do you think using Manifold's Simplify() algorithms as a post-process, before exporting, would help mitigate that?

14:45 <kintel> I guess we'd have to choose some tolerance

14:45 <pca006132> not fully, manifold's simplify only works on things that are topologically close to each other

14:46 <pca006132> and does not guarantee the output to be degenerate free, we just try to make it better

14:47 <pca006132> I think discarding degenerate triangles is probably a bug in their tools if they really do so... can openscad import the mesh? is that still a manifold?

14:47 <kintel> In this case, I was hoping to simply remove very small triangles, where two vertices are so close together that they'd be coincident in 32-bit space

14:48 <kintel> Yeah, It seems to be perfectly manifold (i.e. epsilon-valid). I haven't managed to establish a pipeline yet as the only tools complaining so far are Windows-only : /

14:48 <pca006132> I thought people mentioned prusa slicer and meshlab

14:48 <pca006132> those are cross platform iirc

14:48 <kintel> meshlab using STL, but STL isn't great for this

14:48 <pca006132> I think it accepts obj

14:48 <kintel> I haven't yet managed to reproduce using Prosa

14:49 <kintel> *PrusaSlicer

14:49 <kintel> yeah, one report said that OBJ is OK, while 3mf is bad. More work needed to structure the investigation

14:49 <pca006132> ah, weird

14:50 <kintel> I was mostly curious what Simplify() could do for us if the issue is very close vertices

14:50 <pca006132> sounds like a bug either in our export or in the import of that external tool

14:50 <pca006132> I was discussing what simplify should do with our gsoc intern during my trip, I went to hangzhou and met him :)

14:50 <kintel> heh, I try to blame other tools too, but for some reason, people believe that OpenSCAD is better at fixing workflows than other tools :/

14:52 <pca006132> they probably say the same thing when talking to the other project maintainers :P

14:52 <kintel> haha :)

14:52 <pca006132> but yeah, ideally we should have a version of simplify that helps with this

14:52 <pca006132> ideally can also collapse thin gaps etc.

14:53 <kintel> J25K For the issue you had, could you open a GitHub ticket with very specific instructions on how to reproduce? ..but make sure you're using the latest version of all tools, so that we can try to reproduce

14:53 <pca006132> I should start a discussion in manifold's repo

14:53 <pca006132> (about "advanced" simplify)

14:55 <J25K> kintel for the crash?  or the different export result?

14:58 <pca006132> probably both, if you think both are bugs

15:00 <J25K> well the crash exporting 3mf is already documented but with color not some intersecting face resulting in it

15:02 <J25K>  https://imgur.com/a/TzdwKau   is the same object but meshmixer only show issues on one of them

15:04 <J25K> also the polyhedron has an intersecting face - so if you make wrong polyhedra you can't expect clean results - so i am not considering it as a bug

15:05 <kintel> pca006132 yeah, that's a better place to discuss the details!

15:35 <kintel> J25K I'm curious about the second issue. Could you open a GitHub issue for that, with exact info on how to reproduce?

15:36 <kintel> If you're able to also reproduce with a more coarsely tessellated object, that would make things easier

15:37 <kintel> ..and if any other tools than meshmixer indicates an issue, that would be good info too

18:06 <J25K> kintel hope i got it

18:07 <J25K> 6121

18:07 <gbruno> [github] UBaer21 opened issue #6121 (mirror() prevent CSG polyhedron face correction) https://github.com/openscad/openscad/issues/6121

20:32 <Guest75> Parametric Shoe Generator OpenSCAD

20:32 <Guest75> I need for test project

20:33 <teepee> what sort of test project would that be?

20:36 <Guest75> 3D printing test

20:36 <teepee> shoe generator sounds like a slightly strange choice :)

20:37 <Guest75> where can I find Parametric Shoe Generator OpenSCAD

20:37 <teepee> never seen such a thing before, but who knows, people create all sorts of stuff

20:38 <Guest75> have you heard of it?

20:40 <teepee> no

22:09 <gbruno> [github] Sparsh-N opened pull request #6122 (Fixes error with Hershey Draw call on windows) https://github.com/openscad/openscad/pull/6122

22:12 <gbruno> [github] Sparsh-N opened issue #6123 (GSoC 2025 - Windows Native Build) https://github.com/openscad/openscad/issues/6123

23:45 <sculptor> YES