Skip to main content
mptp
Up and Coming
September 30, 2014

3D rendering simplifies 360° video

I feel like the real hard part with creating panoramic, stereoscopic video is that you need to be able to accurately recreate the view for all head orientations. This basically equates to the requirement for many cameras in complex setups to produce acceptable results.

However, in a 3D rendered scene, it's possible to create as many cameras in whichever orientations you like. So wouldn't it be possible to create a beautiful 3D-rendered animated scene, capture it from many different camera orientations in stereo, and present that to the viewer? You could even do separate captures for different head tilts and present the correct one to the viewer based on the information received from the HMD tracker. (this would probably need to be done in software though, since having 180 seperate videos (assuming you have one video per degree) isn't doable.

Just a thought. I'm more a programmer than an animator, otherwise I'd make a proof-of-concept.

    1 reply

    Bccc1
    Participant
    September 30, 2014
    Why would you render per head-tilt? If you want the viewer/camera to rotate, simply render a spherical panorama and load it in a player like liveViewRift or MaxVR.
    If you also want to account the parallax you'd need to render way more videos. Take a 3D Grid of coordinates with 1cm spacing and the size of possible movement within the rifts camera tracker and you have way more videos to render than space on your drive :P
    And this is only using 1cm distance between the rendered videos, so you'd need some kind of live video morph-blending in your application.
    Terms & ConditionsAccessibility statement