Looking through past hacks is a great source of inspiration. This week, we saw [Russ Maschmeyer] re-visiting a classic hack by [Jonny Lee] that made use of a Wiimote’s IR camera to fake 3D, or at least provide a compelling parallax effect that’ll fool your brain, without any expensive custom hardware.
[Lee]’s original demo was stunning, and that alone is reason to revisit it. Using the Wiimote as the webcam was inspired back in 2007, because it meant that there was no hard computer vision work to be done in estimating the viewer’s position – the camera only sees IR LEDs anyway. The tradeoff is that you had to wear two IR LEDs on your head, calibrate it just right, and that only the person with the headset on gets the illusion just right.
This is why re-visiting the past can be fruitful. As [Russ] discovered, computing power is so plentiful these days that you could do face/eye position estimation with a normal webcam easier than you could source an old Wiimote. Indeed, he’s getting the positioning so accurate that he’s worried about to which eye he’s projecting the illusion. Clearly, it’s time for a revamp.
So here’s the formula: find a brilliant old hack, and notice if it was hampered by the state of technology back when it was done. Update this using modern conveniences, and voila! You might just find that you can take the idea further, simply because you have more tools in your toolbox. Nothing wrong with standing on the shoulders of giants.
But beware! Time isn’t sitting still for you either. As soon as you make your killer 3D vision hack, VR goggles will become cheap and ubiquitous. So get it done today, before your hack becomes inspiration for the future.
Not sure how to feel about this … I still use a note2 (12? Years old now, android 12l though:p) as my daily driver. My desktop is a motherboard with a 790 and chipset, I think 13 years on that. CPU is newer as u upgraded it a year ago to an fx from the original phenom2, so 10-11 years there I guess. My macbook (running arch) is a retina 2015, so 8 on that.
Still for the most part feels alright, not sure how much faster new stuff is.
So comparatively this hack is only 4 years old to me :) While I remember actually trying this out in 07 and was stunned, I think the real problem was lack of applications, other than the demo.
I think computationally it may have been possible, but software has advanced tremendously, which is probably where the big win is.
Actually I was trying to source a usb camera 6 or 7 years for $work, and one of the options was a little module, running something embedded on some arm, _maybe_ an FPGA in it, dong multiple facial recognition, gender recognition and age recognition, all in realtime. So probably was not a big issue back then. But software hen, do we have better applications for this tech this time around? Or just a very cool demo …
Slightly OT, but *love* the artwork. Jane Austen meets Robert Zemeckis.
Yeah! Joe Kim made this one for an article ages ago, and I’ve just been dying to find the opportunity for a re-run.
I was blown away with Me. Lee’s work back in the day. I copied a couple of his projects. Used a projector and a Wii mote as a light pen. Worked great for strategy games like warcraft and command & conquer. Hard to believe that was that long ago.
Mr. Lee* hate using my phone to leave commits.
I’ve already tried bringing this back, but the powers that be would rather do VR headsets than IR glasses
That, and it’s an existing FOSS project (Freetrack) a good decade old. And opentrack if you want to do it with just head tracking without needing a fiducial.
I’ve honestly been looking for this for the past few years. I started on this rabbit hole again and I find it novel that I ended up back where I started when it was first announced. I’d like to know what the forks are that use facial rec but at the same time its really hard to force a 4k resolution desktop even with a spinner top cap dawned with a vive tracker… I am less initiated than most people here so I need the Man tutorial to make it work =_=