05:01 <J24k97> hmm importing 3mf (that has been exported before) causes CGAL error

05:10 <J24k97> haha the space navigator moves the view even if not active window

05:11 <J24k97> but only translation not rotation

07:34 <gbruno> [github] ochafik synchronize pull request #5185 (Simple color support in 3D rendering (includes OFF color import / export)) https://github.com/openscad/openscad/pull/5185

12:27 <TylerTork>  what's the preferred way, these days, to create a round-cornered cube? I found subroutines for it that use minkowski on a sphere and a cube, but this doesn't produce an object of exactly the specified dimensions because spheres are composed of facets and their bounding box is smaller than the specified diameter. Also it's pretty slow. I wrote

12:27 <TylerTork> a version using hull that solves both of these but I wondered whether there might be a better way still

12:35 <InPhase> TylerTork: Extractable examples that can be copy/pasted for this are here: https://github.com/rcolyer/smooth-prim

12:40 <InPhase> It is true that there are very small dimensional deviations for the normal hull, but with $fa=1; $fa=0.4; set, these are like 10 micron deviations.

12:42 <InPhase> The example there handles a few extra cases like compressing the box beyond the dimensions of the rounding specified but still keeping its dimensional accuracy adequately. Also there are a few other implementations for the other cases like hollowing a cubical wall, and rounding in only one plane.

12:44 <TylerTork> Thanks. This code is doing what I did except I also added three cubes that push the edges out to the exact dimensions. The code could use some for loops and parameter rationalization also

12:47 <InPhase> TylerTork: One useful trick for exact sizing is to not use $fa and $fs, but instead specify $fn as an integer multiple of 4, and then adjust your dimensionality by dividing it by cos(180/$fn)

12:49 <InPhase> TylerTork: If you measure outputs you'll find that's roughly float precision levels of exact.

12:49 <InPhase> Although this is usually not important for printing purposes.

12:51 <InPhase> Note that to be precise, this achieves exactness in an x, y, z axial manner. There will be some diagonal points where the sphere has a larger cross-section than the radial specification.

12:51 <InPhase> s/than the radial/than the original/

18:09 <Virindi> I do rounding without minkowski or hull...in this example, I would take a large cube and subtract the edges, corners, and face cubes from the inside using cylinder, sphere, and cube. then subtract that from the original cube. allows more control over exactly which parts are rounded, because often you don't want to round everything

22:52 <nomike> Hi

22:53 <nomike> InPhase: Now that I've published the initial version of my project, I can show you what I used your funcutils library for which you recommended a few days ago:

22:53 <nomike> https://www.printables.com/model/924920-parametric-pride-flag-rainbow-pins

23:31 <InPhase> Ah, to specify the variations in geometry as a string?

23:31 <InPhase> Looks nice, but let's see this code. :)

23:36 <Virindi> seems like a good use for the now more cheaply available full color soldermask PCBs :)

23:40 <Virindi> I wonder what happens if you try to oven reflow a steel pin assembly to a PCB footprint

23:41 <InPhase> nomike: See the edits to line 46, 50, and 72, which remove the preview artifacts for this model. This is described in the manual as the overlap rule. https://bpa.st/UM5Q

23:41 <InPhase> nomike: It looks like similar would be required for some of the other selectable options, but I demonstrated for you in the default displayed rainbow. :)

23:43 <InPhase> They can sometimes lead to render failure, so it's more than just a preview artifact to be worried about addressing on those.

23:44 <nomike> Ahh....nice. Good to know. I will try to revise my design accordingly. Thanks a lot for the tip.

23:46 <InPhase> nomike: Soooo, is there some sort of international standardizations organization for all those pride flag color patterns? :) I don't think I've ever seen so many varieties specified at once before.