Mine was entirely mechanical (driven by punch cards and a hand-crank), and changed all of the pixels in parallel, but a lot of the mechanism development looked extremely familiar to me.
This is incredible! I can appreciate how much work it took to make this happen. Well done!
I was recently in the presence of some linotype machines from the 1800s and it's so good to be humbled by the achievements of people who came before us. That machine was so complex, I could barely begin to figure out how to manufacture one. Your discussion of looms reminds me of that!
If you enjoy linotype machines, I'll suggest you watch 'Farewell ETAOIN SHRDLU', a documentary on the last night the New York Times ran its hot press system
Really cool! I just watched it finish "cat saying 'hi'". It doesn't look like any new posts have shown up on @kilopx.com on Bluesky for the last 9 days though.
A few suggestions for improvements:
- After completing a submission, move the "pen" out of the way as much as possible to get a clean photo of the completed art before moving onto the next submission.
- On the website, show attribution for the currently in-progress submission.
- On the website, have a "history" gallery for completed submissions. It looks like pending submissions have permalinks that say "Timelapse will be available after this is drawn", but there's no way to discover permalinks for completed submissions (or the in-progress one).
I think the issue is that I'm streaming to disk with ffmpeg and recording at 5 fps to save space. OBS must be locked to the same frame rate since it's sharing the webcam?
My original concept included two webcams, one for OBS, one for ffmpeg. Guess I should have gone with that!
This has to be the the most expensive cost per pixel display I've ever seen. And I've never loved a display more. This is absurd in the best possible way
For what it's worth, dollar stores typically sell wooden cubes for arts & crafts purposes (board game designers also like them for prototyping) in bags that work out to a few cents per piece. I guess they're quite a bit smaller than what you ended up using, though. And of course that doesn't account for the frame or the control mechanism. (And now you have me trying to think of more robust ways to turn the pixels...)
When I was a kid, my school had several "1 litre" tubs of 1cm³ wooden cubes so that we could stack them 10x10x10. This would have been very early 80s UK.
Just googled and I found some on Shein with 200 cubes for £2.50. They also have 2cm sized ones at £1.31 for 20 cubes and 4cm ones for £1.88 for 4 cubes.
You'd still have to drill holes in them all, but I wonder if a different solution might be possible - for instance holes in the wooden strips between the rows of cubes that are slightly wider than screws that hold the cubes suspended from the strips. If they weren't too tight, the cube could rotate freely. But maybe just drilling holes using a CNC would easier (and potentially you could drill all the holes on a flat plane of wood before cutting up into cubes).
I came to post about Rosin's work as well. I personally love that he uses clever lighting and angles to create the shading for his pixels instead of just painting one side. It makes it feel like a mirror, all one material like a magic wallhanging.
That said the one I experienced was an earlier work had was fully driven by hobby servos (or something that sounded very much like them) and when you get even one of those going it's loud as hell. I didn't get to look at the construction too closely and this was many years ago. I expect that he did some kind of sound dampening because it wasn't as.. deafening as I expected. But it still kinda 'took me out of it' a bit.
Another idea: have the cubes point an edge straight forward (instead of a face). Then if each cube has two adjacent dark sides and two adjacent light sides, one could setup two ‘simultaneous’ images: one viewed from the left at 45° and another viewed from the right. (Each pixel would have four possibilities.)
I guess the patents are long expired now and don't really apply to pixels, but that concept exists already for non-pixelated images and sadly these are replaced mostly by LEDs now in the wild:
For this to work, you'd want two adjacent faces painted, rather than opposite faces being painted, which seems to be how they're currently done (unless they only have one face painted?). Then the four possible rotations would allow for each possible pixel-pair. (The cubes could perhaps instead be squat rectangular prisms, to correct the aspect ratio, too.)
... But that's as far as you could take it, since 16-gons would show at least 7 faces while only having an encoding for 4.
I also thought of using hexagonal prisms, showing two faces at a time in paired colours but using three colours. These would also need much less clearance in order to rotate freely, compared to face-on cubes.
As an experiment, I just spray painted 66 magnet spheres in half, to make a physical display in 5 minutes. I manually rotated the sphere into position and it holds the image.
There is hope of physically moving them if you put each sphere into a 3d-printed countersink hole over some metal sheet (so that the magnet is hold in place against the plastic), moving a electro-magnet head over you can rotate the magnet, like a scaled-up version of a 2d magnetic tape.
You may even create a Ising model if you put magnets too close to each other.
Very cool. I loved reading your write-up. It reminded me of something I'd read in a steampunk novel once. I had to Google it to get the details. It's the kinotrope from Gibson & Sterling's Difference Engine.
I found a blog post about it and someone who made one with a servo for each pixel. Now that would be expensive!
It’s great to see someone doing something just for the love of doing it. We so often get wrapped up in reaching a goal we forget the journey is what matters most. The curiosity and will to learn new things. I think this project reflects this quite well, and bravo on this amazing achievement. It’s seriously badass.
There's a full colour one of these in the entrance hallway inside the Berlin Technikmuseum, near the Zuse exhibit - there are 12800 cubes, each with 4 colours: https://www.youtube.com/watch?v=BfWFLnsy6QA
Speaking of "alternatives to e-ink for a zero-power-use-when-not-updating dot-matrix display"...
Has there ever been designed a "display" that is just a thermal printer hidden in one end of a box, and a take-up spool + tensioning spring hidden on the other end, such that the "display" is then a continuous thermal paper "scroll" stretched across the box behind [UV-protective!] glass, that can be "refreshed" by printing a new full-width image to the thermal printer?
I like that idea. The printing process should probably be inverted: cool the paper as a whole to darken the whole sheet amd use a small heating coil to erase.
Ooh, I like this idea. You could also use the box structure to stretch the display so it has 4 sides if you build the mechanism correctly, which means as you refresh the image on the "primary" display it moves the other images to the secondary, tertiary, and quaternary displays before it gets taken up. You can use tensioned rollers at each corner hidden by the frame if you plan for a gap for "bezel".
I was wondering about the algorithm to drive the plotter and update pixels, which ties into this.
Given the current image being shown and the next image, you (presumably) want to plot the pixels of the next image as quickly as possible. I believe the optimal algorithm is:
1. Calculate the set of pixels that are changed between the current and next image.
2. Find the shortest path from the plotter's current position through each of those pixels. I believe breadth-first search (O(n)) is sufficient here.
Running this on all potential upcoming images and choosing the one with the lowest total path cost would do what you propose under "better".
What about some system to shoot wooden spheres into a tube or channel for each scan line, selectively feeding different color spheres. Some combination of gravity or pneumatics to drive it. So a scan line would flush out one end and refill from the other. Then scale it up to a stadium size unit with bowling ball pixels.
I guess a challenging part would be proper timing to recycling the colors back into their appropriate supply channels. And also introducing some kind of damping to quiet it down and reduce the wear and tear on the pixels.
On the other extreme, you could go active matrix and have blocks that simply rotate in place to show different face colors based on some solenoid/servo action.
There was a fish project on here a few days ago that also had to deal with uh... adverserial images and it was (mostly?) solved by training a neural net to detect those.
This is pretty cool as-is, but I can't help but try to think up ways to increase the speed. (Not the point, I know.) I feel like it should be able to do a whole column pretty quickly with some optimizations. If the device that turns a block could do so without needing x-axis alignment to change, then you could do a whole column pretty quickly. Or perhaps it'd be better to do rows instead of columns, since the y-axis alignment shouldn't need to change with the current device. As for the block-turning device itself, I think some sort of thing that rotates would speed things up since you wouldn't need to reset, I think. I bet a manufacturing automation specialist could get this thing cruising...
BTW I love that you initially went with a very direct e-ink analog with the balls!
If you had a rotating mechanism which allows slip then you could have rotor shafts which rotate all the blocks in a column while braking mechanisms prevent all the blocks in a given row from moving. Or you could have both rotor rows and rotor columns if you implement a rough mechanical equivalent of the hysteresis systems of ferrite-core memory. Or (I think GistNoesis suggested something similar https://news.ycombinator.com/item?id=44794092 ) if you hide a neodymium permanent magnet just under one corner of each of the blocks you could use a pair of electromagnets behind that block to pull the block from either orientation into the other, almost a solid-state solution apart from the axle the block would rotate on and potentially one which could set the entire display at once.
Thanks for thinking through it! I've found that moving left-right is a little noisier and has a little vibration - up-down is smoother. However, it's not that noisy and it'd be fun to experiment with different algorithms for finding the next pixel.
I'm working on something similar, but 3D and faster motion: PinThing / TAP (Tangible Actuated Pins) [1]
"Motorized pin art display" is what i'm going for...
The problem with passion projects: Progress is never as much or as fast as I want, though! Hard to find the people who want to throw money at things like this and/or buy them. And anything mechanical gets complicated and expensive very quickly. But it's so much fun and a great way to learn and apply so many new skills: laser cutting, 3D printing, CNC milling, circuit design, embedded programming, etc.
Opening the site and immediately seeing it drawing a frame from Bad Apple was amusing, I suspect someone will attempt to automate submissions of frames from that video at some point.
This is very cool. I might have the skills to replicate this but would never be able to accomplish it. My brain is warped .. to get motivated, I need the end result to be of some tangible value. Pure fun doesn't work for me, sadly. I am also unable to go and learn something just for the sake of learning.
Just an aside, an I realize efficiency is NOT a metric for your fabulous display .. here is another interesting mechanism. Maybe one could build a 10K or 100K display with this?:
Absolutely wild this came up today. Just last night I was fiddling with my kids 'magnetic ball board', bought at SF MoMA, and thinking that it could be turned into a cool display by loading a small magnet on the 3-d printer gantry.
Yes! I have an RGB sensor that could handle that, but it's more bulky than the simple IR on/off sensor I went with. Could be four colors, or four shades of a color.
y'know, I've been excited / feared that Bad Apple would show up. The good news is a lot of frames would probably just be a few pixels to change from the previous frame, so some might draw really quickly.
Basically you want to avoid keyframes on this thing, they'll kill you
Some of the ports of Bad Apple have had to deal with this and they narrowed it down to the few changes needed for each frame. When there were too many pixels to change all at once, they would make fewer changes in exchange for a loss of quality.
Very neat to see! I've been wanting to make one for ~15 years, one approach to making one would be inspired by a wooden binary counter where each 'pixel' has a gravity latch that rolls out when on one side and triggers a flip for the next adjacent pixel: https://youtu.be/zELAfmp3fXY
Mumbo Jumbo built a similar concept on Hermitcraft using trap doors. There is a revolution going on called "buildstone", using redstone to create aesthetics
I think I might put this in my friend's coffee shop but I'll restrict access to people in the coffee shop. Not going to let the internet get a hold of that.
In addition to the user-controlled modes I also have ambient modes. My favorite is a clock that struggles to draw the current time because it takes too long
Hah, cool, I had an idea for a similar project (although I'm not crazy enough to make 1000 pixels, or a robot to turn them for me). But I got as far as making a JavaScript simulation and realised I couldn't be bothered manually turning the beads https://incoherency.co.uk/beadboard/
In Christopher Nolan's "Interstellar," the "Interstellar library" refers to the tesseract, a 5-dimensional space within a black hole, visualized as a bookshelf. This structure is not a physical library but rather a construct created by future humans, allowing Cooper to interact with the past and relay gravity data to his daughter, Murph.
This is cool. I wonder, as you were iterating on the design and development, why didn't you start with a very small grid (10x10) to validate or test different options for their practicality and operation before scaling up to the 1000 pixel versions? It might have saved a lot of time and money, but maybe small scale tests aren't sufficient to work out the kinks?
I have to confess I only skimmed it, but it seems that if the choice is to rotate an object, then using a simple flap on an axis would be both cheaper and likely faster (less mass to move). I realize that efficiency was not a goal, but it does align with the pricing issue.
This is really exciting. The world of non-electronic computer interfaces has a ton of potential, so I love seeing people build them and write up their experiences.
Pretty quiet! I spent some time figuring out how to make sure the stepper motors don't whine (the answer is microstepping and decent motor controllers). The pixel turning is very quiet unless it misses slightly, then it makes a clunk.
Mine was entirely mechanical (driven by punch cards and a hand-crank), and changed all of the pixels in parallel, but a lot of the mechanism development looked extremely familiar to me.
I was recently in the presence of some linotype machines from the 1800s and it's so good to be humbled by the achievements of people who came before us. That machine was so complex, I could barely begin to figure out how to manufacture one. Your discussion of looms reminds me of that!
A few suggestions for improvements:
- After completing a submission, move the "pen" out of the way as much as possible to get a clean photo of the completed art before moving onto the next submission.
- On the website, show attribution for the currently in-progress submission.
- On the website, have a "history" gallery for completed submissions. It looks like pending submissions have permalinks that say "Timelapse will be available after this is drawn", but there's no way to discover permalinks for completed submissions (or the in-progress one).
https://en.wikipedia.org/wiki/Calm_technology
My original concept included two webcams, one for OBS, one for ffmpeg. Guess I should have gone with that!
Only drawback is having to hire a C130 if I want to take it on a trip with me :')
Just googled and I found some on Shein with 200 cubes for £2.50. They also have 2cm sized ones at £1.31 for 20 cubes and 4cm ones for £1.88 for 4 cubes.
You'd still have to drill holes in them all, but I wonder if a different solution might be possible - for instance holes in the wooden strips between the rows of cubes that are slightly wider than screws that hold the cubes suspended from the strips. If they weren't too tight, the cube could rotate freely. But maybe just drilling holes using a CNC would easier (and potentially you could drill all the holes on a flat plane of wood before cutting up into cubes).
With the most cost effective and creative "wear item" ever.
https://youtu.be/0o_9CHYeRvI
That said the one I experienced was an earlier work had was fully driven by hobby servos (or something that sounded very much like them) and when you get even one of those going it's loud as hell. I didn't get to look at the construction too closely and this was many years ago. I expect that he did some kind of sound dampening because it wasn't as.. deafening as I expected. But it still kinda 'took me out of it' a bit.
https://excalidraw.com/#json=driyv7dR-eODBzuh_hdrk,93QQvkYae...
https://www.rotapanel.com/trivision-mechanism-and-prism-type...
I also thought of using hexagonal prisms, showing two faces at a time in paired colours but using three colours. These would also need much less clearance in order to rotate freely, compared to face-on cubes.
https://gist.github.com/unrealwill/b8f585758880009113805bd95...
Small spherical magnets are quite cheap.
There is hope of physically moving them if you put each sphere into a 3d-printed countersink hole over some metal sheet (so that the magnet is hold in place against the plastic), moving a electro-magnet head over you can rotate the magnet, like a scaled-up version of a 2d magnetic tape.
You may even create a Ising model if you put magnets too close to each other.
I found a blog post about it and someone who made one with a servo for each pixel. Now that would be expensive!
https://differencing.blogspot.com/2010/04/kinotrope-clackers...
Has there ever been designed a "display" that is just a thermal printer hidden in one end of a box, and a take-up spool + tensioning spring hidden on the other end, such that the "display" is then a continuous thermal paper "scroll" stretched across the box behind [UV-protective!] glass, that can be "refreshed" by printing a new full-width image to the thermal printer?
This ink is interesting in that it fades when heated (60 C), but darkens when cooled (-10 C). In between those temperatures it is stable.
Thus you could have one loop that is continuously reused. Not sure how many cycles you can get before the ink degrades.
https://en.wikipedia.org/wiki/Thermochromism
- naively: Levenshtein
- better: real world edit time based on a model of the display : probably dominated by XY travel distance
Given the current image being shown and the next image, you (presumably) want to plot the pixels of the next image as quickly as possible. I believe the optimal algorithm is:
1. Calculate the set of pixels that are changed between the current and next image.
2. Find the shortest path from the plotter's current position through each of those pixels. I believe breadth-first search (O(n)) is sufficient here.
Running this on all potential upcoming images and choosing the one with the lowest total path cost would do what you propose under "better".
[0]: https://knowyourmeme.com/memes/bad-apple
What about some system to shoot wooden spheres into a tube or channel for each scan line, selectively feeding different color spheres. Some combination of gravity or pneumatics to drive it. So a scan line would flush out one end and refill from the other. Then scale it up to a stadium size unit with bowling ball pixels.
I guess a challenging part would be proper timing to recycling the colors back into their appropriate supply channels. And also introducing some kind of damping to quiet it down and reduce the wear and tear on the pixels.
On the other extreme, you could go active matrix and have blocks that simply rotate in place to show different face colors based on some solenoid/servo action.
He's optimizing for some very different things though.
BTW I love that you initially went with a very direct e-ink analog with the balls!
"Motorized pin art display" is what i'm going for...
The problem with passion projects: Progress is never as much or as fast as I want, though! Hard to find the people who want to throw money at things like this and/or buy them. And anything mechanical gets complicated and expensive very quickly. But it's so much fun and a great way to learn and apply so many new skills: laser cutting, 3D printing, CNC milling, circuit design, embedded programming, etc.
[1]: https://youtu.be/tx4W3ZDA_Vg
Just an aside, an I realize efficiency is NOT a metric for your fabulous display .. here is another interesting mechanism. Maybe one could build a 10K or 100K display with this?:
https://en.wikipedia.org/wiki/Split-flap_display
Legends say this is how MicroLEDs are made, one pixel at a time. That's why they are so expensive.
Could turn this into a 4 color display at the cost of drawing speed?
Basically you want to avoid keyframes on this thing, they'll kill you
https://trixter.oldskool.org/2014/06/19/8088-domination-post...
https://trixter.oldskool.org/2014/06/20/8088-domination-post...
> I have a mechanism to quickly delete problem submissions.
Did you build a male genitalia swastika classifier like the fish guy? (What a sentence)
https://news.ycombinator.com/item?id=44719222
Also worth noting this art project: http://breakfastny.com/dot-screen
Mumbot 2.0 https://www.youtube.com/watch?v=uUzU8HnjBI4&t=1108s
Grian made an animated waterfall with dispensers, snow particles, and potioned arrows: https://www.youtube.com/watch?v=gGPS8hURZks
Also, I'm surprised "All your base are belong to us!" hasn't been submitted yet!
In addition to the user-controlled modes I also have ambient modes. My favorite is a clock that struggles to draw the current time because it takes too long
In Christopher Nolan's "Interstellar," the "Interstellar library" refers to the tesseract, a 5-dimensional space within a black hole, visualized as a bookshelf. This structure is not a physical library but rather a construct created by future humans, allowing Cooper to interact with the past and relay gravity data to his daughter, Murph.
I tested a 1×10 grid of the wooden pixels to try out some different variations as well.
https://en.wikipedia.org/wiki/Project_Hail_Mary
I need to go find some corgi art to upload next!
https://everyware.kr/home/portfolio/r/
Congrats OP!
Unfortunately I can't find the video. Will edit if I do (or anybody else finds it first).
https://www.smoothware.com/danny/woodenmirror.html
It had cubes in different colors so from further away it would look like an image.
Super cool project.
How is it volume wise while it's working? Manageable or painful?
https://digitalartarchive.siggraph.org/artwork/daniel-rozin-...
Edit: Oops there is, it was an issue on my side.