00:00
88
More
See all Show me
Using the three phase structured light technique. More context here: vimeo.com/3193063

Everything is sped up 3x (the system runs at 10fps, but this video is 30fps). There's a quick dance at the beginning, then experiments with different capture settings. The random Rubik's cubes were to help with calibration. That's me at 1:02.

The next issue I'm going to address is the sine distortion. The webcam-projector sync needs some more work, too.

Developed in Open Frameworks: openframeworks.cc/
Flickr set: flickr.com/photos/kylemcdonald/sets/72157613657773217/
Music: soundcloud.com/kyle-mcdonald
Me: kylemcdonald.net/

Credits

Likes

See all likes
  • Memo Akten plus 2 years ago
    hardcore
  •  
  • Memo Akten plus 2 years ago
    This is really super awesome. Do you get point cloud data (unordered, like the radiohead LIDAR data)?, or is it ordered like a heightmap? For the radiohead data I had to pre-process the data and create an animated height field to be able to do certain things with the data that involved persistence across frames (and I just noticed that you commented on the video a year ago! vimeo.com/1371683 )

    P.S. by ordered I mean on a grid (And by unordered I mean randomly placed, just a collection of points in space)
  • Kyle McDonald plus 2 years ago
    Hah hah, I think I commented on that when I was trying to figure out who everyone at of lab was going to be :)

    Yeah, it's ordered. The three phase decoder basically gives me a distance from the camera for most of the points in the 480x640 image. I say "most" because some pixels fall outside of the projection, and others are in weird boundary cases where three phase scanning doesn't really tell you their depth.
  •  
  • vargasz abolcs 2 years ago
    super awesome!! i think so toooo
  •  
  • toneburst plus 2 years ago
    Very very impressive!

    a|x
  •  
  • Rob McDougall plus 2 years ago
    great stuff.
  •  
  • Inklen 2 years ago
    very cool indeed
  •  
  • Lawrie Cape 2 years ago
    Amazing! I love it.
  •  
  • Eoghan Cunneen 2 years ago
    Epic!
  •  
  • Friedrich Kirschner plus 2 years ago
    dope!
  •  
  • Tj Nelson 2 years ago
    what in the... so dope!
  •  
  • Hexstatic 2 years ago
    I like this!
  •  
  • Kurt Rummel 2 years ago
    Tight!
  •  
  • blick 2 years ago
    that's great !
  •  
  • Todd Jakubisin 2 years ago
    i like! i like! over and over, i wish i could like!
  •  
  • AntiVJ / Joanie plus 2 years ago
    very impressive
  •  
  • Accent Creative 2 years ago
    awesome!
  •  
  • Nicolas Guyon 2 years ago
    sick
  •  
  • jaysinspace 2 years ago
    Wow! Really nice results.
  •  
  • Randy Jones 2 years ago
    Brilliant way of doing more with less!
  •  
  • Steve Pomeroy 2 years ago
    You could swap out the visible light projector with something like this:
    web.media.mit.edu/~raskar/LumiNetra/

    Which would make it go quicker, reduce costs and you could use IR LEDs :-)
  • Kyle McDonald plus 2 years ago
    The way I understand that system is that it requires one sensor per point to be scanned (i.e., it's motion capture, not surface capture).

    If you wanted to use it in this setup, you would need to couple it with a high-speed camera. Totally doable, but not as DIY :)
  • Steve Pomeroy 2 years ago
    I was just referring to the static projection mask greycode projector. It's just a greycode projection mask, a big lens and a bunch of LEDs. The rest of the system can be ignored for the purpose of 3D capture :-)
  • Kyle McDonald plus 2 years ago
    Ah, I see! I'm actually working on creating a generic IR projector right now. I'm looking at two versions: one with a colorwheel that projects at 3x the refresh rate (e.g., 180 Hz) and one without that is a modification to a pocket projector.
  • Jesse Stiles 1 year ago
    @kyle this is superfantastic (i love the music too)!
    @steve good to bump into you on the interweb!
  • Kyle McDonald plus 1 year ago
    Thanks Jesse :) The music was composed under funny circumstances. There was a Dan Deacon show in downtown Troy in 2005. After the show, probably about midnight, I wandered over to the Hudson where there was a floating dock that used to be gated, but unlocked (now it's locked). You could just walk down there. It was low tide, and I walked down and sat there with my laptop, headphones, and trigger finger. I don't really have your talent for making infinitely catchy things, but this is one of the rare circumstances I was left with a song I really liked :)
  • Gabriel Shalom plus 1 year ago
    yeah, the music is really making it AND its great to see you advancing with the medium. this material has the most poetic effect for me until now. so, remind me again when you're coming to Berlin? ;)
  •  
  • Steve Pomeroy 1 year ago
    @kyle By color wheel, do you mean greycode projection wheel? Because if you do an IR-only camera + greycode projection wheel (sync'd to an oscillating 3W IR LED) + lens you could manage to get a really cheap, DIYable greycode projector. You wouldn't even need the wacky bar lens that Ramesh uses (which I seem to recall as being one of the main expenses).

    By my rough calculations, you could get a 6 inch disc spinning at 1500 RPM with ~0.5 inch greycode patterns on it (you could probably just print this on a laser printer or laser-cut it for crisper blacks) to about 625hz.
  • Steve Pomeroy 1 year ago
    Ehem, that was meant for the previous thread.
  • Kyle McDonald plus 1 year ago
    You could make a color wheel like that, I wasn't think about that at all... it might be tough to make something that is precisely printed and constructed that can spin accurately at that speed. You would probably put it at a different point in the optics chain than a normal color wheel, too.

    I was thinking more about just removing the color wheel from a normal DLP projector, as it then turns into a single-channel projector that does 3x the framerate (assuming it's an oldschool 3-part color wheel). This technique is the basis for Song Zhang's higher speed scanning research.
  •  
This conversation is missing your voice. Take five seconds to join Vimeo or log in.

Advertisement

Statistics

Date Plays Comments
Totals 15.1K 247 30
Feb 14th 4 0 0
Feb 13th 9 0 0
Feb 12th 4 0 0
Feb 11th 3 0 0
Feb 10th 10 0 0
Feb 9th 4 0 0
Feb 8th 6 0 0

Related lessons from Vimeo Video School

Check out these lessons to learn more about how you can make videos like this one!