Interesting tethering of camera and Kinect

[Petr Dvorak]

I really love how people think...

https://vimeo.com/39505902#

http://www.jamesgeorge.org/works/clouds.html

 

[Brian Iannone]

Wow, now that is pretty cool.

 

[Erich Ocean]

You can also capture depth from stereo capture with an Epic, and pair that with the RGB data from one fo the sides (or both, but most people just use one). Works quite well.

 

[Gavin Greenwalt]

You can also capture depth from stereo capture with an Epic, and pair that with the RGB data from one fo the sides (or both, but most people just use one). Works quite well.

Quite well but far less accurate than a kinect--but does have the advantage of working outside and at a range > 20'.

 

[Felix K.]

Quite well but far less accurate than a kinect--but does have the advantage of working outside and at a range > 20'.

But with a higher resolution and framerate... they should pair a stereo-epic-set with kinect (or two)...

 

[Tony Rudenko]

What would be really interesting is to have multiple cameras (say 8 or so) shooting form different angles & heights simultaneously and merge the all data/video into one virtual camera... or virtual stereo 3D camera for that matter... That would be neat to see...

 

[Petr Dvorak]

What would be really interesting is to have multiple cameras (say 8 or so) shooting form different angles & heights simultaneously and merge the all data/video into one virtual camera... or virtual stereo 3D camera for that matter... That would be neat to see...

Exactly, this is just the begining. Multicam or multimiror octopus with kinect will be next ...

 

[Gavin Greenwalt]

But with a higher resolution and framerate... they should pair a stereo-epic-set with kinect (or two)...

Higher resolution is only useful though if the resolution is accurate. In my depth-extraction tests I'm getting less z resolution than with a kinect from 4k RED. There just isn't enough detail in most scenes/objects to find depth. If you shoot a white-wall for instance the stereo rig will be useless since there is nothing to triangulate. A Kinect is clever in that it projects an invisible pattern into the scene and then captures that--so it doesn't matter how much texture there is in your scene.

 

[Erich Ocean]

Higher resolution is only useful though if the resolution is accurate. In my depth-extraction tests I'm getting less z resolution than with a kinect from 4k RED. There just isn't enough detail in most scenes/objects to find depth. If you shoot a white-wall for instance the stereo rig will be useless since there is nothing to triangulate. A Kinect is clever in that it projects an invisible pattern into the scene and then captures that--so it doesn't matter how much texture there is in your scene.

Well, project a texture then, that's what we do. The resolution with the Epic is killer, and we're doing stereo capture with a single Epic (not two).

I'll see if I can post some of our stuff.