(Not sure why there's a "vibecoding" tag though, it has nothing to do with vibecoding)
Happy to chat more about it!
The main question I tend to get from hacker-types is "what are the tech specs?". The demos run on a Mac with custom software I built for projection mapping/rendering/etc. (the hand tracking uses Apple's built-in framework, which is very good) & a small consumer 4k laser projector.
But this is primarily a research/design prototype - I think it's much more important to figure out what the user experience is, if the interactions are plausible & desirable, and then work backwards to the technology.
For these demos, is the laser projector actually located within the lightbulb, or is that still the speculative part? What's the tracking/projection FOV of the prototype shown in the demos?
this is great @guilaume
I can imagine, with some enhacements -- this could could become a primary way
a patient and doctor interact, may be even helpful in surgery.
for Folks with impairment may also find a concept of projection -- only not to walls, but to devices specialised to reduce effects of their specific impariment.
For every day hands-on work: car mechanics, house repair -- where LLM faciliated interaction right where the person is working (where the other part of the 'light bulb' device is a near-by camera working side-by-side to provide camera-enhanced feedback for the interactons.
I would say for car mechanics you end up at best with halfway-to-glasses form-factor: you need it to be a head-mounted lamp, and it might end up that glasses are a better choice in this specific case (constrained space with large relative differences of depth of physical elements inside it)
This is an amazing concept and one of the most original yet obviously useful tech I have seen lately. Smart glasses are an obvious competitor to this, but the lightbulb computer has a multiplayer, "slow tech" feel to it and does not bring as many privacy nightmares with it..
. I'd buy this!
It only doesn't have privacy nightmares until some company commercialises it while using server-side machine-learning processing for performance… regularly putting glasses into a padded Faraday cage sounds more practical than regularly removing lightbulbs on the ceiling.
Author here; I was definitely inspired by my visits to Dynamicland many years ago, which Folk Computer is also inspired by.
One big difference in direction is that Dynamicland + Folk Computer use projector+computer vision as substrate to try and reinvent/rethink the interactive/dynamic medium, and what end-user programming could be (which is very cool).
In that way I am much less ambitious; this exploration is more about how there is an entire class of existing interactions (the ones showed in the piece) that we perform today on screens, that could be much more interesting + collaborative if overlaid in the real world. And I am also terribly uninspired by the eyewear-based computing vision all major companies have pushed lately.
I have this bad feeling that if you put out ideas like these out there, Meta is going to steal it and call it their own a in a few years, and you're lucky to be hired in the process. (And getting an offer from Meta doesn't sound as nice as it did.)
I don't only want to comment doom though; this seems like both a great user experience, and a good use of LLMs that don't replace humans. (Unless you're thinking of that Black Mirror episode.)
Especially for the "turn on this light" moment, my initial reacting was "oh it's taking a second or so, is it faster than turning it on yourself" but then I was more like "okay this is a prototype, and finding the switch takes me longer than that and my back doesn't thank me for it afterwards".
If you're able to nail down the form factor then fair but every still, I think a reactive projector on 3-axis motors that does stuff like this is still awesome: bonus points if you can just unmount and carry it around with you and place it on tables like one of your lightbulb examples.
guillaume | a day ago
Oh hi, this is my project! Fun to see it here :)
(Not sure why there's a "vibecoding" tag though, it has nothing to do with vibecoding)
Happy to chat more about it!
The main question I tend to get from hacker-types is "what are the tech specs?". The demos run on a Mac with custom software I built for projection mapping/rendering/etc. (the hand tracking uses Apple's built-in framework, which is very good) & a small consumer 4k laser projector.
But this is primarily a research/design prototype - I think it's much more important to figure out what the user experience is, if the interactions are plausible & desirable, and then work backwards to the technology.
eyesinthefire | 23 hours ago
For these demos, is the laser projector actually located within the lightbulb, or is that still the speculative part? What's the tracking/projection FOV of the prototype shown in the demos?
vladislavp | 19 hours ago
this is great @guilaume I can imagine, with some enhacements -- this could could become a primary way a patient and doctor interact, may be even helpful in surgery.
for Folks with impairment may also find a concept of projection -- only not to walls, but to devices specialised to reduce effects of their specific impariment.
For every day hands-on work: car mechanics, house repair -- where LLM faciliated interaction right where the person is working (where the other part of the 'light bulb' device is a near-by camera working side-by-side to provide camera-enhanced feedback for the interactons.
Anyways, cool stuff , has good potential.
k749gtnc9l3w | 15 hours ago
I would say for car mechanics you end up at best with halfway-to-glasses form-factor: you need it to be a head-mounted lamp, and it might end up that glasses are a better choice in this specific case (constrained space with large relative differences of depth of physical elements inside it)
codekobold | a day ago
This is an amazing concept and one of the most original yet obviously useful tech I have seen lately. Smart glasses are an obvious competitor to this, but the lightbulb computer has a multiplayer, "slow tech" feel to it and does not bring as many privacy nightmares with it.. . I'd buy this!
k749gtnc9l3w | 15 hours ago
It only doesn't have privacy nightmares until some company commercialises it while using server-side machine-learning processing for performance… regularly putting glasses into a padded Faraday cage sounds more practical than regularly removing lightbulbs on the ceiling.
darth-cheney | a day ago
Dynamicland.org and folk.computer vibes!
guillaume | a day ago
Author here; I was definitely inspired by my visits to Dynamicland many years ago, which Folk Computer is also inspired by.
One big difference in direction is that Dynamicland + Folk Computer use projector+computer vision as substrate to try and reinvent/rethink the interactive/dynamic medium, and what end-user programming could be (which is very cool).
In that way I am much less ambitious; this exploration is more about how there is an entire class of existing interactions (the ones showed in the piece) that we perform today on screens, that could be much more interesting + collaborative if overlaid in the real world. And I am also terribly uninspired by the eyewear-based computing vision all major companies have pushed lately.
frontsideair | 16 hours ago
I have this bad feeling that if you put out ideas like these out there, Meta is going to steal it and call it their own a in a few years, and you're lucky to be hired in the process. (And getting an offer from Meta doesn't sound as nice as it did.)
I don't only want to comment doom though; this seems like both a great user experience, and a good use of LLMs that don't replace humans. (Unless you're thinking of that Black Mirror episode.)
Especially for the "turn on this light" moment, my initial reacting was "oh it's taking a second or so, is it faster than turning it on yourself" but then I was more like "okay this is a prototype, and finding the switch takes me longer than that and my back doesn't thank me for it afterwards".
tobz619 | 12 hours ago
I love it! Pretty sick!
If you're able to nail down the form factor then fair but every still, I think a reactive projector on 3-axis motors that does stuff like this is still awesome: bonus points if you can just unmount and carry it around with you and place it on tables like one of your lightbulb examples.
klardotsh | 4 hours ago
I hope this either comes to market, or comes to a BOM and a source code repository so I can make it myself. This is incredible.