YouTube Is Adding VR Video Support To Streaming Videos

An anonymous reader writes: While YouTube's streaming platform currently supports 3D videos OR 360 degree videos, the combination of the two is essential for properly immersive virtual reality video. Fortunately, the company has announced that they'll soon enable support for 3D + 360 degree videos, bringing more immersive VR video capability to the platform. Currently, 360 degree YouTube videos can be viewed through desktop web browsers and on the YouTube Android and iOS apps, with the Android app being the only one of the bunch currently providing a side-by-side view for VR viewers like Google's Cardboard.

Comments

Re:3D stereoscopy + 360 != VR

[MobileTatsu-NJG]

You do get head tracking with it.

Yay, But...

[Greyfox]

I'm super-psyched about all of this, but I suspect the camera I'm going to have to bolt onto my helmet to film 3D 360 degree VR videos while flying the wingsuit is going to be nasty.

Re:3D stereoscopy + 360 != VR

[MobileTatsu-NJG]

If you're saying you cannot do spherical video in stereo... you're wrong.

Re: 3D stereoscopy + 360 != VR

[Karljohan]

Actually, there are many ways to do 360 video in stereo, however the eye separation (direction/slant) is always determined when the video is generated. This means that when you move/tilt your head, the eye separation will be wrong.

Dome production solve this in different ways but it's never real VR. It's typically based on the principle that the dome audience look mostly forwards.

Re:3D stereoscopy + 360 != VR

[MobileTatsu-NJG]

Turn your head to the right 90 degrees. You now have 2 videos feeds, one from where your left ear is, and one from where your right ear is.

Again, this is untrue, and I don't know where you're getting this information from. You can generate left-eye/right-eye spherical views that are stereo in *any* direction. This has been done both in CG and with live-video, and does not require interpolation... just more cameras. This is not interpolation, either.

There are two limitations. The first is that between the two cameras there are areas that one camera can see and the other cannot, this creates a region of occlusion. This creates a bubble around the camera, virtual or real, that won't be viewed properly. The minimum size of that bubble is determined by the separation of the cameras, anything passing into it will get cut off or distorted.

The second is that the stereo effect doesn't work past the poles because now you're looking at video intended for viewing from the opposite orientation. There is no known solution for this.

This is not theory, these sorts of videos are already out in the wild.

Re:3D stereoscopy + 360 != VR

[MobileTatsu-NJG]

Click that link, it specifically mentions stereo spherical video.

Re: 3D stereoscopy + 360 != VR

[MobileTatsu-NJG]

No, it doesn't mean that. Eye separation does not change when you look around in a spherical video. I don't know where this myth that stereo spherical video is only stereo looking straight ahead, but it's not even close to true.

Re:3D stereoscopy + 360 != VR

[kesuki]

back in the day didn't they just use fisheye lenses to get the desired effect? http://realvision.ae/blog/2014/08/part-2-graduated-stereo-falloff-in-360-the-language-of-visual-storytelling-in-vr/

Re:3D stereoscopy + 360 != VR

[MobileTatsu-NJG]

Honestly I'm unaware of stereo panoramic shots done back in the day... which I'm assuming is sometime in the 90's when products for this purpose started appearing on the market. I didn't see an actual stereo viewer until this year. (Although the place I worked at did experimental stuff.)

I know that video shot these days would involve multiple pairs of cameras, stitched together with stereo in mind. The app that does the stitching, however, needs to be aware that the footage is stereo, otherwise it'll try to 'correct' the seams, not realizing that sometimes it shouldn't.