Fascinating video. I watched almost the whole thing without planning to, I got sucked in.
This is one of those examples of software that reminds me of my struggle to understand how LLMs are passing code evaluations that culminate with people declaring that they are now better than even the best human coders. I have tried to get LLMs (specifically, Claude and ChatGPT, trying various models) to assist with niche problems and it's been a terrible experience. Fantastic with CRUD or common algorithms, terrible when it's something novel or unusual.
The author creates his own version of a "FLIP simulation". I'm going to go out on a limb and posit that even ChatGPT's unreleased o3 model would not be up to the task of writing the software that powers this pendant. Is this incorrect? I realize perhaps that my comment is a little off-topic given that this is not an AI project. However, this project seems like an excellent example of the sort of thing that I am quite skeptical the supposedly "world-class" artificial software engineers could pull off.
I've implemented fluid mechanics using Claude (through Cursor) and it had no problem writing the logic and integrating it using my custom physics engine and custom renderer.
So no, I don't think your assessment is correct. LLMs shine when they get to implement something from scratch on a blank slate with clear API boundaries, whether it's a CRUD app or a physics simulation. Where I think they struggle the most is in big legacy codebases on tasks spanning multiple modules with lots of red herrings.
I'm a computational fluid dynamics (CFD) engineer. Like others have said, I'm pretty confident basic CFD algorithms are in the training data of many LLMs. I would say a bigger problem is the accuracy of the generated simulator. A LLM would not be able to generate good tests. You need both tests for the math ("verification") and tests for the physics ("validation"), and LLMs can't do either at the moment.
Gold-standard verification tests are constructed using the "method of manufactured solutions" (MMS), which can be largely automated with computer algebra software, but are still quite tedious. I know from experience. I don't believe LLMs can handle the algebraic manipulation here particularly well.
Worse, LLMs won't be able to produce actual experimental data to do the validation test with. You'll need to track down one or more experiments from the literature or do your own experiment. LLMs might in the future be able to point you to appropriate experiments in the literature, but they don't seem able to do that at present. I think LLMs might provide useful advice when a simulation ends up not matching the experimental data. LLMs seem to know a thing or two about turbulence modeling, though I would question their knowledge of the most recent advances.
(If you're only interested in fluid simulation for games or computer graphics then physical accuracy is not a priority. But you probably should still use MMS to make sure you've implemented the math correctly. MMS is an interesting technique that has no parallel in software testing in general. Abstractly, the idea is to make a minimal modification to the software so that you have an oracle, and the nature of the modification is such that the modified software passing the test implies that the unmodified software would also pass the test. This idea probably can be applied in other areas.)
The "issue" with saying an LLM can't do this is that CFD simulations are not actually that niche. Many university courses ask their students to write these types of algorithms for their course project. All this knowledge is present freely on the internet (as is evident by the Youtube videos that the author mentioned), and as such can be learned by an LLM. The article is of course still very impressive.
Great point. Niche to me, but not to thee. I was unaware. This is actually one of the frustrating things about the LLMs - they don’t tell you when what you asked for is outside their training data!
A teacher once told me their (middle? high?) school was the last in their state with lathes. Something which couldn't be created there today, and couldn't exist now in some other states, but which had been repeatedly grandfathered, since its creation with many others, just after World War 2. And that it would disappear upon the first serious accident, leaving none at all.
I mentor at a public high school that has a dozen metal lathes, a half dozen wood lathes. Plasma cutters. CNC everything. A giant planer. My favourite thing is that half of it is super modern and half of it is indestructible WWII surplus, which looks really cool mixed together.
I cannot imagine the concept of lathes being taken out of schools because of accidents. What a self destructive cultural attitude.
People are good, maybe a little too good, at assessing the risks of acute instances like an injury from some equipment. We seem to be awful at assessing systemic risks like growing up without knowing to make things for yourself, and what that does to a society.
Sometimes it feels like people think: "Yeah what's the point of learning how to make things when you can buy cheap things made in a factory in china by some low skill people. If they can make that stuff so cheap it must not be worth that much"
It only takes one student to lose an arm, face, life and their parents will sue the school into oblivion and start PAPT (parents against power tools) or something "for the kids!"
I'm on your side that I'd rather see the tools at the school. At the same time, I took an auto-shop class in high school and the majority of students in the class were dipshits and were lucky not to get more hurt. The teacher managed to stop them just in time from trying to turn over an engine on a rack that would have crushed them if the teach had been 5 second later.
We actually did have an unprecedented accident a couple years back and as a result some new safety requirements were put in place. But no tools or access to tools were removed.
I wonder if the culture here is different than the U.S.? Dipshit students are removed from these classes before they can get anywhere near tools. It’s definitely a privilege to be in a shop class.
I don't want to generalize it but as soon as you said "if the culture here is different than the U.S" I thought "Okay yep that's why."
I don't even know if it counts as culture, but the US inevitably seems to have such a refined sense of litigation and, insurance mindset for lack of a better term?
Anecdotally, schools are indeed a prime example, we had cooking classes in high school that were stopped because of the costs to insure for it.
> I wonder if the culture here is different than the U.S.?
There's definitely going to be cultural differences between the US and wherever you're located, but be careful not to generalize too much from one comment you read on the Internet.
Access to a CNC mill and a CNC lathe should be a fundamental right.
Being CNC with an enclosure mitigates 99+% of all the safety issues and makes them 1000% more useful. With CNC, the most harmful injury is likely to be a broken bone because you're only in the envelope with servos active but not the spindle. With manual, the spindle or chuck can grab you and kill you or send shrapnel into you.
And I really wish we had a good, open source pick and place. Those still can't handle 0201 components and higher density BGAs even though those PCBs are really cheap nowadays.
I've seen a Massachusetts public library with an associated hacker space (sewing machines; 3D printers), and another with a "tool lending library" (including a battery drill, IIRC?). And have read of European libraries with a "be a general community space" vibe. So that seems plausible at least.
I picked some random public high school in my area (twin cities, MN). Here's some classes they offer:
> Machine Technology 1 - Increase your knowledge and skills in the use of common hand tools, engine lathes, shapers, drill presses, milling machines, and
grinders. In this class students will have hands-on experiences in elementary mechanical fundamentals and principles used in
precision manufacturing industries.
> Welding 1 - Interested in becoming a welder? Learn how to use the art of welding in artistic and industrial settings. Develop basic skills related to
safety, equipment set up, metal transfer, gas shielding and welding various types of metals. This class focuses on oxyacetylene and
gas tungsten arc welding.
> Construction Trades 1 - Interested in a career in construction or learning valuable home maintenance skills? In this course you will develop basic carpentry, metals, electricity, and pipefitting skills along with safe usage of hand and power tools applicable to each skill set.
I think there was a period (90s-2000s?) where trade skills were mostly not taught in public schools, but they're thankfully working their way back in.
That is to come! Once I get it working the way I want it haha.
Wifi is tricky and I seem to have lost the code that made that video work and I reflashed with old code that doesn't work :(
Yeap, Tim is some serious maker/tinkerer/creator. Besides the very good hardware things he makes, there are also some other software stuff: https://mitxela.com/other/
This project is an incredible mix of artistry and engineering. the attention to detail in both the fluid simulation and the hardware design is mindblowing(especially the clever use of charlieplexing to optimize LED placement
congrats!!
this is so incredibly impressive. I find projects and the people who do this so inspiring and also deflating because while I can appreciate the work ethic and polymath skills needed for this I couldn't recreate this. I'll just have to hang back and appreciate.
Any other similar creators with this level of polish (pun intended) and interests to follow?
If you're into the custom PCB design, I highly recommend Carl Bugeja: https://youtube.com/@carlbugeja
He has lots of super interesting projects using micromagnetics, motors, flappers, and the like.
If you liked the part where he machined the housing, check out Clickspring. I think it's fair to say he is the greatest machinist currently active on YouTube, both in videography and storytelling, and in actual skill.
https://youtube.com/@clickspring
I of course don't know your specific circumstances, but nonetheless I'd like to recommend you the author's own take on their work described in a rant titled "Spare Time and Hard Work".
It's available here: https://mitxela.com/rants
(unfortunately It's not possible to link directly to it so you'll need to scroll a little bit down to find it)
I personally like coming back to it from time to time, since it's inspiring in it's own way for me.
I second this. I bet the creator of this would be bummed to read this - it's not like their skills came from out of nowhere, just because they're humble about it.
I do my own little hacking project that I enjoy - and definitely where I sink my spare time when I can carve it out! But this level of dedication and final result can be very humbling.
Weirdly I am drawn to his previous idea of using an actual liquid (mercury) to trigger the LEDs. So much simpler to lay out in KiCad, no 4-layers required....
I would love to see a version with wireless charging, though that might be tough with the all metal enclosure, perhaps it could be done through the front/screen section.
Also, adopting the screen off function Apple watches use could be a cool addition, for that it would need a light sensor. On an Apple Watch, you can switch the screen off by simply covering the screen with your hand.
Fascinating that with such a simple simulation you can see grain boundaries like those in metals forming. Look at the second image, those are almost identical to what they looked like in our material science courses.
I am tempted to make a slightly sillier version of this w/ a circular lcd rather than the uhh retro-cool smt leds honestly just to avoid the multiplexing design
I'd be tempted to fill it with water with something dissolved in it to make it more conductive (and probably more viscous), and include LEDs whose brightness and/or color reflects the conductivity at their location. The rationale being, why should I do all that math when Nature will do it for me?
Isn't that precisely his first idea? A device containing a conductive liquid (mercury) that takes care of closing the circuits and the turning on the LEDs.
The HN guidelines say not to complain about the formatting or presentation of the article, but often the page design is so insanely terrible that someone can't help but post a complaint about how they couldn't read it because X, or how it didn't work on browser Y, or how the information density was so annoyingly low it wore out their mouse or pointing finger, or...
This isn't one of those posts. Good job to the author. This article is not only nicely written but also a pleasure to read.
Fascinating video. I watched almost the whole thing without planning to, I got sucked in.
This is one of those examples of software that reminds me of my struggle to understand how LLMs are passing code evaluations that culminate with people declaring that they are now better than even the best human coders. I have tried to get LLMs (specifically, Claude and ChatGPT, trying various models) to assist with niche problems and it's been a terrible experience. Fantastic with CRUD or common algorithms, terrible when it's something novel or unusual.
The author creates his own version of a "FLIP simulation". I'm going to go out on a limb and posit that even ChatGPT's unreleased o3 model would not be up to the task of writing the software that powers this pendant. Is this incorrect? I realize perhaps that my comment is a little off-topic given that this is not an AI project. However, this project seems like an excellent example of the sort of thing that I am quite skeptical the supposedly "world-class" artificial software engineers could pull off.
I've implemented fluid mechanics using Claude (through Cursor) and it had no problem writing the logic and integrating it using my custom physics engine and custom renderer.
So no, I don't think your assessment is correct. LLMs shine when they get to implement something from scratch on a blank slate with clear API boundaries, whether it's a CRUD app or a physics simulation. Where I think they struggle the most is in big legacy codebases on tasks spanning multiple modules with lots of red herrings.
I'm a computational fluid dynamics (CFD) engineer. Like others have said, I'm pretty confident basic CFD algorithms are in the training data of many LLMs. I would say a bigger problem is the accuracy of the generated simulator. A LLM would not be able to generate good tests. You need both tests for the math ("verification") and tests for the physics ("validation"), and LLMs can't do either at the moment.
Gold-standard verification tests are constructed using the "method of manufactured solutions" (MMS), which can be largely automated with computer algebra software, but are still quite tedious. I know from experience. I don't believe LLMs can handle the algebraic manipulation here particularly well.
Worse, LLMs won't be able to produce actual experimental data to do the validation test with. You'll need to track down one or more experiments from the literature or do your own experiment. LLMs might in the future be able to point you to appropriate experiments in the literature, but they don't seem able to do that at present. I think LLMs might provide useful advice when a simulation ends up not matching the experimental data. LLMs seem to know a thing or two about turbulence modeling, though I would question their knowledge of the most recent advances.
(If you're only interested in fluid simulation for games or computer graphics then physical accuracy is not a priority. But you probably should still use MMS to make sure you've implemented the math correctly. MMS is an interesting technique that has no parallel in software testing in general. Abstractly, the idea is to make a minimal modification to the software so that you have an oracle, and the nature of the modification is such that the modified software passing the test implies that the unmodified software would also pass the test. This idea probably can be applied in other areas.)
You could go and try it out. You'd be surprised how capable the LLMs, specifically Claude are now at creative projects such as this
The "issue" with saying an LLM can't do this is that CFD simulations are not actually that niche. Many university courses ask their students to write these types of algorithms for their course project. All this knowledge is present freely on the internet (as is evident by the Youtube videos that the author mentioned), and as such can be learned by an LLM. The article is of course still very impressive.
Great point. Niche to me, but not to thee. I was unaware. This is actually one of the frustrating things about the LLMs - they don’t tell you when what you asked for is outside their training data!
Claude is pretty good at writing Python programs.
> access to a lathe is a fundamental human right
A teacher once told me their (middle? high?) school was the last in their state with lathes. Something which couldn't be created there today, and couldn't exist now in some other states, but which had been repeatedly grandfathered, since its creation with many others, just after World War 2. And that it would disappear upon the first serious accident, leaving none at all.
I mentor at a public high school that has a dozen metal lathes, a half dozen wood lathes. Plasma cutters. CNC everything. A giant planer. My favourite thing is that half of it is super modern and half of it is indestructible WWII surplus, which looks really cool mixed together.
I cannot imagine the concept of lathes being taken out of schools because of accidents. What a self destructive cultural attitude.
People are good, maybe a little too good, at assessing the risks of acute instances like an injury from some equipment. We seem to be awful at assessing systemic risks like growing up without knowing to make things for yourself, and what that does to a society.
Sometimes it feels like people think: "Yeah what's the point of learning how to make things when you can buy cheap things made in a factory in china by some low skill people. If they can make that stuff so cheap it must not be worth that much"
It only takes one student to lose an arm, face, life and their parents will sue the school into oblivion and start PAPT (parents against power tools) or something "for the kids!"
I'm on your side that I'd rather see the tools at the school. At the same time, I took an auto-shop class in high school and the majority of students in the class were dipshits and were lucky not to get more hurt. The teacher managed to stop them just in time from trying to turn over an engine on a rack that would have crushed them if the teach had been 5 second later.
We actually did have an unprecedented accident a couple years back and as a result some new safety requirements were put in place. But no tools or access to tools were removed.
I wonder if the culture here is different than the U.S.? Dipshit students are removed from these classes before they can get anywhere near tools. It’s definitely a privilege to be in a shop class.
I don't want to generalize it but as soon as you said "if the culture here is different than the U.S" I thought "Okay yep that's why."
I don't even know if it counts as culture, but the US inevitably seems to have such a refined sense of litigation and, insurance mindset for lack of a better term?
Anecdotally, schools are indeed a prime example, we had cooking classes in high school that were stopped because of the costs to insure for it.
> I wonder if the culture here is different than the U.S.?
There's definitely going to be cultural differences between the US and wherever you're located, but be careful not to generalize too much from one comment you read on the Internet.
Access to a CNC mill and a CNC lathe should be a fundamental right.
Being CNC with an enclosure mitigates 99+% of all the safety issues and makes them 1000% more useful. With CNC, the most harmful injury is likely to be a broken bone because you're only in the envelope with servos active but not the spindle. With manual, the spindle or chuck can grab you and kill you or send shrapnel into you.
And I really wish we had a good, open source pick and place. Those still can't handle 0201 components and higher density BGAs even though those PCBs are really cheap nowadays.
I really wish they'd repurpose public libraries for this.
No we need public libraries
I've seen a Massachusetts public library with an associated hacker space (sewing machines; 3D printers), and another with a "tool lending library" (including a battery drill, IIRC?). And have read of European libraries with a "be a general community space" vibe. So that seems plausible at least.
Rumor has it some libraries have added makerspaces
I picked some random public high school in my area (twin cities, MN). Here's some classes they offer:
> Machine Technology 1 - Increase your knowledge and skills in the use of common hand tools, engine lathes, shapers, drill presses, milling machines, and grinders. In this class students will have hands-on experiences in elementary mechanical fundamentals and principles used in precision manufacturing industries.
> Welding 1 - Interested in becoming a welder? Learn how to use the art of welding in artistic and industrial settings. Develop basic skills related to safety, equipment set up, metal transfer, gas shielding and welding various types of metals. This class focuses on oxyacetylene and gas tungsten arc welding.
> Construction Trades 1 - Interested in a career in construction or learning valuable home maintenance skills? In this course you will develop basic carpentry, metals, electricity, and pipefitting skills along with safe usage of hand and power tools applicable to each skill set.
I think there was a period (90s-2000s?) where trade skills were mostly not taught in public schools, but they're thankfully working their way back in.
I used a lathe by myself at school when I was 12 or younger. But we live in different times I suppose
So primary school? That seems another possibility for what made the teacher's school singular.
Id take it with a grain of salt.. I used to work with Dream It Do It in western New York. There were many schools with machine and carpentry shops.
Charlieplexing:https://en.m.wikipedia.org/wiki/Charlieplexing
I learned recently that Charlieplexing is kind of like the discrete version of the Radon Transform: https://en.wikipedia.org/wiki/Radon_transform
This is super cool! I love the simulation aspect of it.
On the same note, I've been working on a bike POV display with lots of LEDs and just got "live-ish" video working over wifi with it:
https://youtu.be/hxAHBvuyqpY?si=8XraFuG_Fi54Bs7T
Awesome! I hope you'll post some write up or something on how you did it!
That is to come! Once I get it working the way I want it haha. Wifi is tricky and I seem to have lost the code that made that video work and I reflashed with old code that doesn't work :(
Back to the drawing board
Ooo nice! Looks sweet
I really love mitxela's project ideas and process videos, I recommend checking out his other projects as well
Yeap, Tim is some serious maker/tinkerer/creator. Besides the very good hardware things he makes, there are also some other software stuff: https://mitxela.com/other/
This project is an incredible mix of artistry and engineering. the attention to detail in both the fluid simulation and the hardware design is mindblowing(especially the clever use of charlieplexing to optimize LED placement congrats!!
this is so incredibly impressive. I find projects and the people who do this so inspiring and also deflating because while I can appreciate the work ethic and polymath skills needed for this I couldn't recreate this. I'll just have to hang back and appreciate.
Any other similar creators with this level of polish (pun intended) and interests to follow?
If you're into the custom PCB design, I highly recommend Carl Bugeja: https://youtube.com/@carlbugeja He has lots of super interesting projects using micromagnetics, motors, flappers, and the like.
If you liked the part where he machined the housing, check out Clickspring. I think it's fair to say he is the greatest machinist currently active on YouTube, both in videography and storytelling, and in actual skill. https://youtube.com/@clickspring
I of course don't know your specific circumstances, but nonetheless I'd like to recommend you the author's own take on their work described in a rant titled "Spare Time and Hard Work".
It's available here: https://mitxela.com/rants (unfortunately It's not possible to link directly to it so you'll need to scroll a little bit down to find it)
I personally like coming back to it from time to time, since it's inspiring in it's own way for me.
Direct link: https://mitxela.com/rants#:~:text=Spare%20Time%20and%20Hard%...
That's a great read - thanks for sharing!
Don't be deflated, get started.
I second this. I bet the creator of this would be bummed to read this - it's not like their skills came from out of nowhere, just because they're humble about it.
I do my own little hacking project that I enjoy - and definitely where I sink my spare time when I can carve it out! But this level of dedication and final result can be very humbling.
Weirdly I am drawn to his previous idea of using an actual liquid (mercury) to trigger the LEDs. So much simpler to lay out in KiCad, no 4-layers required....
So, so beautiful. I was shocked at the price. I was expecting something about 10x what he's charging.
Seems like it’s sold out now except for a special one at £1,000. Did anyone catch what the original price was?
I believe they were £500 when I first checked (approximately an hour after the video was uploaded).
I would not use "prototype" to describe this as he did.
One sense of the word is “a first full-scale and usually functional form of a new type or design of a construction” (Webster)
I would love to see a version with wireless charging, though that might be tough with the all metal enclosure, perhaps it could be done through the front/screen section.
Also, adopting the screen off function Apple watches use could be a cool addition, for that it would need a light sensor. On an Apple Watch, you can switch the screen off by simply covering the screen with your hand.
I think a plastic housing might look really nice TBH.
This is the only piece of jewelry I've ever seen that I immediately crave. Wish it were still on sale!
Those LED's could have been packed much closer together if one was willing to have adjacent LED's share a single pad.
Since the LED's are all the same component and therefore heat at the same speed in the oven, I don't think that would cause tombstoning.
Fascinating that with such a simple simulation you can see grain boundaries like those in metals forming. Look at the second image, those are almost identical to what they looked like in our material science courses.
wait, this is amazing !
I am tempted to make a slightly sillier version of this w/ a circular lcd rather than the uhh retro-cool smt leds honestly just to avoid the multiplexing design
great work!! super cool
dang, great video and nice write up
I can see a good video as far as show/don't tell my stuff is dry/way too much info. Guess it depends verbatim tutorial vs. just showcasing.
Such a cool project. The only thing - I wish LEDs were blue
Is this for sale? Its so cool
£1k, on his website he has a link.
the whole mitxela store is t-i-g-h-t the metal shop close-ups are zomg :salivating-face: ...
Does it send MIDI?
The led screen is disappointing
put diamonds on top of those LEDs and you can sell it to rappers.
Or we could just fill the same container with....water
I'd be tempted to fill it with water with something dissolved in it to make it more conductive (and probably more viscous), and include LEDs whose brightness and/or color reflects the conductivity at their location. The rationale being, why should I do all that math when Nature will do it for me?
Isn't that precisely his first idea? A device containing a conductive liquid (mercury) that takes care of closing the circuits and the turning on the LEDs.
Water is a lot less luminescent than LEDs. I think the effect might be reduced if changed to a water based display.
What about a fluorescent dye with one (or few) UV LEDs on the edge facing inward to illuminate it?
Maybe not “better” than OP’s design, but could be interesting in its own way.
At such a small scale it will not slosh around satisfyingly.
i was wondering whether this could be hardware accelerated.
The HN guidelines say not to complain about the formatting or presentation of the article, but often the page design is so insanely terrible that someone can't help but post a complaint about how they couldn't read it because X, or how it didn't work on browser Y, or how the information density was so annoyingly low it wore out their mouse or pointing finger, or...
This isn't one of those posts. Good job to the author. This article is not only nicely written but also a pleasure to read.