still learning, but really enjoying the process of experimenting with dual extrusion and different print settings. Lately, I’ve been hearing a lot of buzz around AI in creative fields, especially something called generative AI.
I keep wondering…[what is generative AI] exactly? I get that it creates things (like images or music), but can it be applied in any way to 3D printing? Maybe for generating design ideas, optimizing models, or even suggesting print settings?
I figured this community might have some folks who are exploring the tech side of things beyond just the hardware. If anyone has used AI tools for 3D design or workflow improvement, I’d love to hear about it. Or if you just know a good beginner-friendly explanation of generative AI, feel free to share!
Thanks in advance—and excited to keep learning with you all.
I tried to do some AI 3d stuff month ago, but it was very experimental. I am also waiting for some nice tools, but i think the Data pool is quite small and complex. Much more then text or Images.
Generative AI is mostly used for generating images, videos, and sounds, and it isn’t very good at any of those things yet. There is at least one service, that I forgot the name of, that you can use to generate very basic 3d models using text prompts. I never tried it but I read about it somewhere and users said it was only useful for simple designs. I personally think that creating my own models is fun. Learning a CAD software is challenging but rewarding.
There are a couple of AI sites around that can generate STLs based off of your description, however you need to be able to describe what you want in a way that it can understand, so it’s ok for simple things, not so for complex things yet.
Yeah. Generative AI basically slurps up tons of existing examples of something to train it on that thing. Text is an obvious one, and organizations like OpenAI and Meta have sucked up basically the whole internet to train on the written word. (Often without openly asking consent to do so.)
The same can and has been done for solid models. However, that data is arguably much more complex, so it’s likely very difficult to train.
Regardless, Bambu Labs has already done it, as one example. They’ve had a few tools that let you generate very specific 3d models. Like little cartoon figures and stuff.
I wonder and fear Maker World and similar are really being used to accumulate a large data set for solid models to teach AI. Worrying only because it’s often done without explicitly saying that’s what’s being done with the data. And the AI then essentially reproduces the work in different forms, leading to a lot of questions about intellectual property.
At any rate, generative 3d models are indeed a thing. It’s hard to overstate the satisfaction of learning to craft models then bringing them to life though! Especially with 3d printing.
I think it’s not generative but there is a beta based on Orca which supports your slicing process with an Ai agent, it’s developed from the same dude which funded the spaghetti detective (later obico, Ai spaghetti detection), it’s called JusPrin
and it’s exciting!
A bit AI in the field right now but only little generative.
I find that some models are useful for creating OpenSCAD files to render STL files (i.e. to replace a handle on a Delta faucet in my kitchen among other things).
It actually produced rather nice code that rendered well in OpenSCAD and easily exported to a printable STL.
My prompt was quite simple and and the result was OpenScad code that rendered pretty close to what I wanted.
So, I think that using certain models that are good with writing code are a great way to “scaffold” a 3D print model to which you can tweak and modify from there.
Oh damn… looking at it again, I can see that it’s a template generated message. Or, it seems clearer to me that it is.
So many flags there… and because I’m reading about a hobby I enjoy, I overlook them. As a guy who works in the realm of building AI integrations for my company’s systems, I hang my head…
There are already AI engines which are able to create 3D models just from a text prompt. These are of course also usable for 3D printing. Using free 3D software like Blender or others it’s easy to convert that into i.e. a STL format or whatever you need - if the AI site not already outputs ready-to-use STL files.
