Here’s the latest. Not sure if it’s the actual first 3D digital camera or not. There may have been other before it:
http://www.fujifilm.com/products/3d/camera/finepix_real3dw1/

http://www.stereoscopy.com/jasper/

I met the owner. He does all his own machining and has been quietly making them for about 25 years. I own the panorama head and the single vertical mount. The construction is top-notch.

They bolted the cameras on whatever pits of ply-board they had handy… and it took them a couple of hours. It looks like crap, and the two cameras probably don’t even point in the same direction.

Photoshop makes anaglyphs a breeze. Paste the right eye view over the left as a new layer. Select Layer->Layer Style->Blending Option and click the Red button for red/cyan or the green button for green/magenta. Use the cursor keys to set the convergence plane while viewing through glasses. I have an Action that does the layering and blending, so all is left is using the cursor keys for setting the easiest and most natural viewing. Very slick, and works with either colour or monochrome images. I have one version of the Action that de-saturates the images prior to layering which produces less cross-talk between eyes.

I’ll just answer a few of the points brought up in the comments here.

Yes, the lenses are 35mm AFS 1.8.
Focusing was done by having the subjects stand on a mark, flip to live view, find manual focus for both cameras, switch off live view, engage preload triggers.. Everything was shot at 2.8, this caused issues with both the subject moving off their mark, and asymmetrical focus when something was bumped or nudged accidentally..

As for the oversize inter-occular, it was a byproduct of the construction technique. We have no metal working experience so the rig was bolted together using available materials and skills. Its backyard, but it works.
The impact on the stereo is to exaggerate the depth, but the effects are negligible at the kind of display sizes we are viewing at, life is also made easier by having a blank background.

As for the rig’s lack of ability to converge the cameras, we subscribe to the parallel school when it comes to camera setup. In a perfect world we would have purpose designed camera bodies with mechanically shiftable sensors, or be shooting with tilt/shift lenses to offset the sensor.. But we don’t have those things, so we shift in post. It means you lose some resolution and field of view, but its fast, easy and avoids all the keystoning and vertical disparities issues inherent in shooting converged.

The Anaglyph conversion was done in Nuke. I implemented the conversion fomulas found here:
http://www.3dtv.at/knowhow/anaglyphcomparison_en.aspx
Specifically the Optimized Anaglyph formula, it dramatically reduces luminance asymmetry by knocking out colours that only appear in one eye or the other. You lose colour range but reduce the visual buzzing. Its responsible for the heavy green cast, but it feels better in 3d relative to traditional anaglyph.
If anyone is interested they can download my Nuke gizmo below:
http://dl.dropbox.com/u/3656251/OptimizedAnaglyph.gizmo.zip
We dint use Ocular but most likely will in the future. We aren’t using it in our day jobs because Guardians is all CG and doesn’t suffer from the issues that Ocular is designed to address (mainly repairing problems with mismatches in live action stereo footage).

@Larry N.Bolch
Are you the guy who takes all the great 3d Lambo and Ferrari stuff on Flickr?

As for the W1.
I have one now, its great and lives in my bag everywhere I go. Takes seconds to fire up unlike the 30 minutes of frigging around with the D90s. I love it.
Its worth noting that the W1 is also a parallel camera, doing the convergence as a post process..

@Rob Bannister
Yes, reconstructing 3d point clouds from stereo pairs is an area that interests us.
Its non-trivial but we have tried some stuff with projecting into a deepimage.. Its cool, but we need more hours in the day..

As for Nikon making dedicated stereo cameras in the future?
I think its going to have to evolve out of wherever they go with mirrorless cameras, smaller lenses for the same size sensor will make it easier to position the sensors close together..
See similar ideas with people getting excited about stereo rigs built on RED’s small sensor scarlet.

it looks like somthi´ng has gone wrong

bad rumor/not interesting anyway