Determine mesh rotations to minimise overhangs for 3D printing | Hacker News Reader

Determine mesh rotations to minimise overhangs for 3D printing

github.com

63 pointsalonfnt32 comments

Very nice, but often minimum overhang is not the only constraint. Sometimes you want to take (directional) strength into account. And sometimes it simply looks better if an object is sliced in a certain way.

It's very familiar to anyone who's done some woodwork. Keeping the grain direction in the same direction as you require strength is 90% of joinery.

This might be good for slicing random decorative models that weren't designed with printing in mind.

Because when I care about strength and printability I do so from beginning and base the whole design around it. There is no need for a tool to pick print orientation for me because I know before the first CAD extrude how it's going to be printed.

How would you design a screw thread adapter that needs to resist shear?

I certainly wouldn't 3d print it, that is a job for off the shelf metal part.

Nobody has any clue what the thing is. :D

It claims to be BSP, but it’s 1 inch (1.1?) with 19 threads per inch (18?), which matches no standard I’ve ever found.

Print a mold, cast it in wax, put that in casting sand, then do a lost-wax casting in aluminum or something more exotic if necessary and you're able?

I think at that point I'd have to ask someone to make it for me.

Lathe would be the tool to use in that case.

You're absolutely right. But I have a 3D printer, casting sand, and a torch... I don't have a lathe, sadly.

A 3D printed plastic part could be fine depending on the application. Certainly worth considering/testing. Printing it flat on the bed (with a small flat side) so layers are lengthwise may work.

If it needs any amount of strength you'll want to find someone with a lathe. BSP tells me it may be tapered(?) so find someone with a taper attachment. Even as a hobbyist I could probably make one of those on my lathe faster than drawing it and printing it on my 3d printer.

See also threaded inserts for the other half of that fastener.

At some point you have so many competing compromises that you're sacrificing one aspect to improve another. When this happens, I pick what's easiest to print and make up the strength with a lot of extra wall loops, slightly higher print temperature (for better layer adhesion, as print quality allows), and stronger material selection.

If this is a critical or load bearing part, I'd skip 3D printing entirely and get it machined.

45 degrees to the bed, 3+ walls, gyroid infill 40%

Although true, you can optimise this considerably by using materials with very strong layer adhesion like PET (which I've never had break along layer lines) or by annealing or even remelting. So for me at least who prints entirely in PET or PET-CF (note PETG to be clear) this is super helpful.

Tangental:

Most consumer printers I’ve come across have differing XY and Z axis spatial resolutions

I’ve been wanting a decent algorithm to determine the optimal angle to print models such that the result is aesthetically as high fidelity as possible

Take for example a rigid model crane set to a non orthogonal angle. Is it pleasing to print with respect to one vertical column or the other? What if they are not equally sized? What about which is most likely to be viewed up close?

Generalising that problem to more organic shapes is incomprehensible but I can’t even answer the aforementioned toy example

I’m sure there are dozens of excellent potential approaches but I can’t easily find it for arbitrary XY and Z (or other combinations thereof) spatial resolutions

If you need detail in more than one axis, you should use a smaller nozzle and lower speed. Along z-axis, the feature size limit is layer height, and along XY it's linewidth. Layer height can be very small with any nozzle (I've seen as small as 50 microns), but linewidth will typically need a smaller nozzle hole to work well, particularly to get below 100 microns.

I think XY actually has better resolution depending on your perspective. I agree that the minimum extrusion is approximately a line width circle. But that circle can move less than the line width for features on larger surfaces. For example, if you emboss something on the side of an object with 0.3mm depth using a 0.4mm nozzle / line-width, it will work fine.

Yeah, I suppose it depends heavily on the kind of detail. I was mostly thinking along the lines of printing lithophanes or hueforge prints.

With lithophones it's typically recommended to print upright so the image is formed by stacking layers, as that allows for "pixels" to be thinner. Hueforge instead is typically printed flat because color stacking/blending needs more resolution than the shapes.

My cheeky observation to that would be: if you need detail in any axis use a resin printer. ;)

Yeah, if you're doing detailed prints often, especially decorative stuff, resin can't be beat by standard fdm.

Great stuff, but working with STLs isn't a good option for me. Doing this in STEP would be amazing. But I suppose I could use STL to guide me on orientations I can then set up manually with my STEP files?

This tool makes total sense to me as-is, but I don't quite understand why it would be more useful to work with STEP files.

As-is, this is a tool I could see myself using when I've already got an STL which I maybe don't have input in (downloaded from printables/thingiverse/etc, or extracted and converted from a source like game files). But if you've got the STEP file, you have far more ability to make modifications to it to make it more compatible with 3d printing, like reducing overhanging bits or adding structures to support the overhangs.

I directly print STEP files. Faster, quieter and better finish than STL for curved surfaces. I print at fairly high speeds (300-400mm/s, >20k accels, 25 SCV on quality) so this might not be a universal consideration but for me it's a deal breaker usually unless I'm doing something silly

you could convert the STEP file to STL just for the purposes of this tool, it looks like you have to apply the resulting rotation manually in your slicer anyway using the rotate tool.

Well I'm quite curious what you're using for a printer now!

Nothing especially unusual I think. Mostly use it for fixtures (for woodwork, DIY), improving/fixing random things and bike mods. Currently designing a derailleur for a folding bike.

Reason why I've optimised it to print fast is that there's only so many hours between getting back from work and then going to bed (I don't trust my wiring enough to leave it overnight)... So pretty much everything has to be printable in max 5 hours... which is especially hard to achieve if I'm printing at high infills and 0.1mm layer heights.

I think they meant what printer do you use

Thanks. I can't read it seems. It's a custom VzBot

That's pretty impressive performance! Was just looking over at their github repo and documentation and I'm tempted... because I definitely need yet another project :)

Custom VzBot

Reply on news.ycombinator.com