The new perpetual 3D Printing Thread

Has anyone here used a SnapMaker 1.0? Sounds like a great idea, in theory. But does it actually deliver the goods?

And how about SnapMaker 2.0? Twice the awesome? Or twice the vapor?

I don't really have a need for any of the 3 possible machines that this thing could be, but having this in my company could spur some creativity.

At a minimum we could use the laser engraver to etch our logo and text onto 1,000 of these every month:



using a rotary attachment.

stevenkan said:

Has anyone here used a SnapMaker 1.0? Sounds like a great idea, in theory. But does it actually deliver the goods?

And how about SnapMaker 2.0? Twice the awesome? Or twice the vapor?[/url]

Click to expand...

Does this mean "No?"

Flipper35 said:

Just the ability to immediately problem solve like you said. My latest, the trucks for the caboose used M2 bolts and the existing were M3. Print M3 threaded rod and put in with some super glue and redrill for the M2. Done. Didn't have to get a wooden dowel or plastic rod to size, glue in, wait, file smooth then drill.

Click to expand...

Yeah, I bought a Bambu H2C for the office to have our 3D CAD summer intern experiment with a bunch of stuff. I was amazed at how well it does with threads.

How fast did you have to work when threading in your M3 rod?

I'm also pretty amazed at how well it does with hinges.

Flipper35 said:

I threaded by hand until it bottomed out like I would any other bolt. Once I was sure it threaded properly I took it out and put a drop of super glue on it and ran it back in with my fingers. By the time I got the package opened with the new trucks and the drill and bit out it was ready to drill and assemble.

Click to expand...

I was worried about the opposite problem, e.g. the glue hardening before you could get it screwed all the way in. It sounds like that's not a issue.

stevenkan said:

Yeah, I bought a Bambu H2C for the office to have our 3D CAD summer intern experiment with a bunch of stuff.

Click to expand...

Now that she's gone back to school I've had a chance to starting playing with the printer, and I'm hooked.

Regarding design software, I'm finding that Claude is pretty fucking amazing. Things I learned this summer that are at last tangentially on-topic for 3D printing:

1. Claude Code can do amazing things with FreeCAD and Blender, but:
   1. Tell it to run both headless and silent, so you don't have windows popping up all over your computer while you're doing real work.
   2. Tell it to record process ID, so it can kill stuck processes by PID, instead of the FreeCAD instance you're actually using yourself.
   3. FreeCAD has a long-standing, well-known units declaration bug that's you to work around as long as you remind Claude about it every fucking day.
2. Claude can edit STEP file by directly editing the text, and it's amazingly good at understanding the 3D geometry underlying the models, even without consulting a real CAD/3D kernel like FreeCAD and/or Blender.
   1. This is useful for working around other bugs/issues like FreeCAD dropping colors on STEP export or linearizing NAUO instances and blowing up file sizes.
   2. Fable is a massive improvement over Opus in understanding 3D geometry, and even Opus 5.5 is 10x better than Opus 4.x was when I first starting futzing with this 6 months ago.
   3. So I've instructed Claude to use FreeCAD as a geometry-solver only, and then to generate all final STEP output via text generation.
      1. This also allows Claude to script a lot of my STEP generation in python, which is much lighter weight than round-tripping through FreeCAD.
      2. I am working on a bunch of stuff that could be served through a web interface with python STEP generation in the background.
3. Claude has been very good at recommending settings to debug and improve my prints, including step-by-step instructions for where to find settings buried 3 layers deep in Bambu Studio.
4. Apparently there is a CLI for Bambu printers, but Claude says it's not fully implemented and/or supported, but I'm not using it yet, but that could be a game-changer for automating workflow.
5. Claude installed a go2rtc relay server on my BlueIris server (security camera server running $65 software package to manage IP cameras)), so now I have my H2C camera feed all the time, anywhere, without needing to launch the Bambu app, and I can replay video and do time-lapses without having any local storage on the printer.
6. Claude can generate/assemble STL from STEP to produce multi-color and multi-material builds from a parametric model, including sacrificial/release layers, etc.
7. Claude can reverse-engineer STEP from STL, which I used to reverse-engineer a molded plastic enclosure for a power supply that we bought from a Chinese contract manufacturer
   1. They sent me a 3D PDF with the model, from which Claude extracted an STL.
   2. The model was actually wrong in several places, but Claude and I worked out how to make the changes to match the delivered product. I can mate either 3D printed half to the opposite half from the factory, and it all fits perfectly.
   3. Now I have many different options for making one-off variants and/or presenting customization options to my customers.
   4. I also have protection against that mfr going out of business or jacking up the pricing, since I can second-source this any time I want.
   5. 
8. As a non-work-related project I'm also working with Claude to clone a custom silicone ice cube mold I bought for my brother and his company a few years ago:
   1. 
   2. 
   3. 
   4. 
   5. This is one of the projects that I might turn into a web service, e.g. upload logo and text; download STEP/STL files for printing, then mold your own molds!

This is fun!

stevenkan said:

Yeah, I bought a Bambu H2C for the office to have our 3D CAD summer intern experiment with a bunch of stuff:

Click to expand...

Now upgraded with dual AMS and a 3D printed, PETG dual AMS riser/stand assembly:



As useful as modern printers are, I'm also seeing that we're still in the early stages of their development. I don't think we're in the DOS/8088 days like we are with AI, but we're probably in the WfW 3.11 era. Good enough to be useful to lay people like me, without needing to learn a bunch of the technical bits, but also with a huge future runway of massive improvements.

How far away are we from arbitrary color mixing like CMYK toner printing, along with multiple independent print heads with linear motors?

Drizzt321 said:

you can't put it into CLAUDE.md or something?

From what I hear, Opus 5.5 is a huge leap in object/spacial understanding.

Click to expand...

Yes, you need to put into your .md file AND remind him every day. And then it will only bite you half as often.

BigLan said:

Color mixing seems like a really tough problem to solve - mixing filament together would likely need a bunch of purging, and you'd likely be locked into a specific brand of filament to make sure the colors match well.

Adding an ink pigment to white filament could be interesting, but would I don't know how easy you could get it to mix in a hot end. Having a separate brush head to basically spray paint while printing might be an option too.

Click to expand...

Yeah, I'm sure there are people noodling over this, but it's hard to find any research, because all searches end up discussing layer-based dithering.

What's the 3D printing analog for a pixel? Not quite a voxel. How about an extrudel? xStrudel?

Someone will come up with a full-spectrum color approach that comes out of left field. Maybe not filament at all.

DB63 said:

There were attempts to real-time colours mixers, feeding in four stock filaments each with their extruder gear running at different rates into a single nozzle. But they didn't mix properly and came out more like candy-cane stripes than smooth blends. Also very difficult to make transitions between pure colours when needed.

The Full Spectrum stuff is working pretty well though. I did a short test (very wasteful on X1C), and the results were pretty convincing in real life. Combined with thinner layers and 8+ colour inputs (RGBW+CYMK) it would be very close.

Click to expand...

The problem with layer-based dithering is in-plane planar surfaces. Unless someone comes up with a solution for that.

stevenkan said:

Yes, you need to put into your .md file AND remind him every day. And then it will only bite you half as often.

Click to expand...

and

gregatron5 said:

I'm sure there's a large gradient in between.

Click to expand...

The goal is no gradient. Badum bum.

DB63 said:

Yeah. I've been monitoring the poop from my printer this year and it's working out to about 20g/month (=~50c/month). Even if you allow that I might do more multi-filament prints with the right gear, efficient filament swaps are never going to be a big thing for me.

Click to expand...

Yeah, given the cost of filament, and the amount of waste I generate through prototypes, filament waste isn't of any concern to me. But time definitely is.

DB63 said:

I am interested in using multiple nozzle widths in a print though, mixing speed and resolution. I'm a bit surprised that none of the tool/nozzle swappers have made that mainstream yet.

(Edit). And separate nozzle for support, that would be nice.

Click to expand...

What is the conceptual hurdle that makes nozzle-width mixing difficult? From my human brain it doesn't seem like it should be difficult.

Excellent. This is exactly what I hoping to find out about.

Aurich said:

No, there is no melt zone, it is as @feeverte just said, closer to an inkjet printer, but mixed with a UV resin printer.

So it's spraying fine droplets of colored resin, and hardening them with UV light. Which is a very standard thing, and been around for ages now. And layering the resin up isn't new either. They're basically taking it to a bit of an extreme to build up the layers into a true 3D print, instead of say just a thicker gloss texture like you might find on a more typical commercial UV printer.

Click to expand...

Very interesting! And in the kickstarter video they even mention that they're "powered by an Epson printhead," which adds credibility in my book. And if this is fundamentally just inkjet printing with new materials, throw an HP Page Wide printhead at it, and you're looking at a new class of device.

Aurich said:

I don't think any of this will ever compete on price, ease of use, speed etc with FDM. But it does open up new use cases, and that's super fun.

Click to expand...

I might take that bet, especially on points 1 and 2. There's a bunch of physical infrastructure that just goes away when you do away with filament. Piping mixable fluids from a tank is just so much simpler than handling N discrete filament rolls.

From the limited amount that I'm understanding of the fundamental technology behind UV printing, the major stumbling blocks are material science issues. New materials could flip the script on that.

Aurich said:

This tech is not new, commercial UV printing has been around for years and years. What's new is trying to break into the consumer market. And that will hopefully push for more consumer friendly solutions. But it's still a somewhat messy tech, just by the very nature of things.

Click to expand...

Is printing upward with printable, water-soluble support material new? I'm not any head-to-head competitors in this space, but I also don't know the vocabulary, so I'm not searching for the right things.

I understand why the support material is required for . . . support, but I don't understand why they print the bubble over the finished model. Is that for protection some post-UV curing stage?

Yes, it's messier than filament printing, but it looks way better than resin-bath printing.

feeverte said:

stevenkan said:

Yes, it's messier than filament printing, but it looks way better than resin-bath printing.

Click to expand...

I have to disagree with you here.

Click to expand...

LOL. I meant "looks better" from a messiness point of view, not print quality!

VirtualWolf said:

I was initially looking at Bambu's H2D because I've seen a lot of good things about how plug-and-play their stuff is, and I'd rather spend a bit more up-front for something more prosumer versus just dipping my toes in at the bottom, but the cloud stuff is a bit offputting (though I gather there's a LAN-only mode? I have zero need to be able to monitor my prints remotely because I work from home).

Click to expand...

We've had an H2C here at work since July, and we've been very happy with it. As most other reviewers note, it just works.

I've been playing with the dual nozzles to print text on things, and 0.2mm prints quite legibly at reasonable font sizes:


with some visible artifacting up close:


That was printed text-side down on a smooth plate. I'm going to try it text-side up and see if it's any different.

I haven't played with more than 2 colors/materials yet, despite buying two AMS units to feed both sides, but once I get a full set of six 0.4mm nozzles for the right-side printhead, I'm going to try printing a full, 7 color (6 faces plus mechanicals), 100% print-in-place Rubik's Cube, just for the sake of doing it.

One things I've been pleasantly surprised at by Bambu Lab is their reasonable pricing on accessories. For example feeding more than one AMS into one nozzle requires a 4:1 PTFE adapter that I would have expected to be priced at $10 - $15, and instead it's $4.99:

Similarly, if you want to feed either of 2 AMS units into either of 2 nozzles, you can buy their Filament Track Switch, which works with the printer to route either to either:


I was expecting a price of$99 or more, and instead it's a perfectly reasonable $59.99. If HP were ever to enter this market I'm sure they'd be charging 3-5x what Bambu is charging for these types of accessories.

That doesn't absolve Bambu of all their sins on the IP front, but at least they're doing a lot of other things right.

stevenkan said:

That was printed text-side down on a smooth plate. I'm going to try it text-side up and see if it's any different.

Click to expand...

Here's text-side up, no ironing on the text: