Page 1 of 1

Box

Posted: Wed Oct 02, 2013 11:05 am
by CTZn
Box explores the synthesis of real and digital space through projection-mapping on moving surfaces. The short film documents a live performance, captured entirely in camera.
http://www.youtube.com/watch?v=lX6JcybgDFo

Re: Box

Posted: Wed Oct 02, 2013 11:21 pm
by StompinTom
Those are the guys over at Bot & Dolly, they beat us to the chase! Amazing stuff.

My favorite still is this one: http://vimeo.com/50197298

Re: Box

Posted: Thu Oct 03, 2013 7:02 am
by CTZn
hey thanks for the headsup Tom, I'm discovering this all. Art !

Re: Box

Posted: Thu Oct 03, 2013 7:32 am
by galinette
They are quite different, the second is 90% art and 10% technique, while the first is an impressive combination of well mastered technologies, including real-time 3d, projection, viewpoint capture, robotics...

Seeing the video, it seems quite clear that the POV of the camera is taken into account in the projected 3D. Which means, as the camera is handheld, that the 3D is real-time. They used only diffuse surfaces, so the illuminations are likely precomputed. Awesome stuff, thanks for the links!

Re: Box

Posted: Thu Oct 03, 2013 10:13 am
by PureSpider
galinette wrote:Which means, as the camera is handheld, that the 3D is real-time.
The cam is on a robot aswell, everything is precomputed: http://www.youtube.com/watch?v=y4ajXJ3nj1Q
Still impressive!

Re: Box

Posted: Thu Oct 03, 2013 12:17 pm
by CTZn
Yep there is absolutely zero lag involved.

On a slightly different note there is a lot of fuzz going on around Rapid Eyes Movements (REM) recently (Google Glasses, controversial - lies or disorders detection, neuro-marketing etc).

Re: Box

Posted: Thu Oct 03, 2013 8:20 pm
by galinette
CTZn wrote:Yep there is absolutely zero lag involved.
POV tracking with realtime 3D is a common thing in "Cave" systems, and there is definitely no perceptible lag in those (otherwise you would be sick quickly...). I've tried some myself. So this is existing tech!

http://en.wikipedia.org/wiki/Cave_autom ... nvironment

Re: Box

Posted: Thu Oct 03, 2013 8:22 pm
by galinette
PureSpider wrote:
galinette wrote:Which means, as the camera is handheld, that the 3D is real-time.
The cam is on a robot aswell, everything is precomputed: http://www.youtube.com/watch?v=y4ajXJ3nj1Q
Still impressive!
Are you sure? In your video at 4:02 I see a shoulder camera with tracking markers (LEDs). Realtime POV tracking (head or camera) is actually existing technology.

EDIT : OK, this is explained just after. They prerecord the camera track and move it with a robot during the performance. It could have been done realtime, and this would have allowed having a human watching instead of the camera (even with stereo glasses). Maybe for the next generation...
I'm a little disappointed but not too much :)

Re: Box

Posted: Fri Oct 04, 2013 6:38 am
by CTZn
It could have been done realtime, and this would have allowed having a human watching instead of the camera (even with stereo glasses). Maybe for the next generation...
It could indeed, and I would find the technology even more appealing than the holographic ads as seen in Back to the Future II movie.

But lag, can we really fool a brain yet instead of making it feel sick ? And stereoscopy (stereopsis) would require us to equip devices.

It is an amazing show as it stands already !

Re: Box

Posted: Sat Oct 05, 2013 11:20 pm
by StompinTom
galinette wrote:
PureSpider wrote:
galinette wrote:Which means, as the camera is handheld, that the 3D is real-time.
The cam is on a robot aswell, everything is precomputed: http://www.youtube.com/watch?v=y4ajXJ3nj1Q
Still impressive!
Are you sure? In your video at 4:02 I see a shoulder camera with tracking markers (LEDs). Realtime POV tracking (head or camera) is actually existing technology.

EDIT : OK, this is explained just after. They prerecord the camera track and move it with a robot during the performance. It could have been done realtime, and this would have allowed having a human watching instead of the camera (even with stereo glasses). Maybe for the next generation...
I'm a little disappointed but not too much :)
For part of my graduate project I did exactly that: I tracked a viewer's skeleton with a Kinect and corrected the perspective of a projected model using the viewer's head position, so you could 'look around corners' in a digital model that was an extension of the room. It's exciting stuff, and robotics + sensing are the perfect way to bring all this stuff to super-precise real-time performance.

The Bot & Dolly video is technically supreme - undoubtedly - while the dandypunk video is great because it combines projection with damn good human choreography and a bit more of an interesting story. It's cool to see humans that can be almost as coordinated and precise as machines. :)