3D Cameras vs. 360-degree Cameras

All the rage in the news this week has been the introduction of some high-end “360-degree” cameras for virtual reality. Are these 3D cameras?

A simple 3D camera.
A simple 3D camera.

The elephant in the room is that these “360-degree” cameras do not produce stereoscopic output. Stereoscopy is the bedrock of creating the illusion of three dimensions.

What is “stereoscopy”? From Wikipedia: “Stereoscopy is the production of the illusion of depth in a photograph, movie, or other two-dimensional image by the presentation of a slightly different image to each eye, which adds the first of these cues (stereopsis). The two images are then combined in the brain to give the perception of depth.

Anyone who has used a ViewMaster is familiar with this trick. A camera that can be used to create stereoscopic images has two lenses, one to create the imagery for each eye. Like this:

3D camera from Panasonic

Source: http://uncrate.com/stuff/panasonic-pro-twin-lens-3d-camera/

Heck, the front end of this even looks like the back end of a ViewMaster.

Now, let’s compare and contrast this baby with these 360-degree cameras that have all the VR enthusiasts shrieking “game changer!” all over the web.

360-degree camera

Source: http://ir.net/news/virtual-reality-headsets/124125/orah-4i-live-360/

The idea behind this little guy is that it will capture “spherical video” – video footage will exist in front of you, behind you, above you and below you – all the way around.



But what it does not do – what it cannot do – is prepare different images for each eye, as the 3D camera above does.

Let’s say I used a 3D camera to take a picture of my cat sitting on the deck. Using a VR headset, the two different images the camera produces would get sent to each eye, and while my brain was stitching them together, the illusion that the cat is much closer to me would be rock-solid. Like how subjects in ViewMaster disc photos absolutely looked closer to the viewer than the background.

Anyone who has seen a movie with 3D glasses on knows all about this illusion.

But with the 360-degree camera – there will be no such illusion. The camera is only capturing one “spherical image”. The cat would appear flush with the “spherical” background. No matter how close it was to me when I captured the footage.

3D Cameras that produce 360-degree output?

In conclusion, I’d like to get you to take a moment to think about the challenges of designing a camera that can do both.

Could you simply use two 360-degree cameras, side-by-side? No, you could not. Because the two lenses have to be side-by-side. In “front” of the cameras you could do this. But “to the left” of the cameras – they would not be side-by-side. One would simply be closer to the subject than the other. Illusion lost.

A more likely possibility would be to use a bunch of 3D cameras on a bigger ball. That way, on each side of the ball, there would be side-by-side lenses, to capture the two required images, all the way around.

Then there’s the challenge of processing for orientation – if I’m facing forward and I look up, left and right are one thing. But if I’m facing backwards (from the camera rig’s perspective) and then look up? They’re backwards. And heaven help me if I’m looking to the left, and then look up – the camera’s left and right would be my above and below. That would need to be processed all to heck, to restore the illusion in any orientation.

But hey, we’re humans, we’re smart, we’ll get to the bottom of that eventually.

But so far, I haven’t seen anything like that coming to market.

If people are going to want virtual reality that is even close to virtually real – that is, both all the way around, and with actual depth – someone’s gonna have to cook up a camera like that.

 

Tweet about this on TwitterShare on Google+Share on LinkedInShare on FacebookEmail this to someone

Author: Pete

Editor-in-Chief, Lead Software Developer and Artistic Director @ 3dspace.com

1 thought on “3D Cameras vs. 360-degree Cameras”

Leave a Reply

Your email address will not be published.