Forum Discussion
VR Pre-Rendered Non-Real-Time
Hi,
I'm thinking about a method to pre-render VR view-points as each view-point being a stereoscopic movie stream. So to speak there is a lot of movie streams, and the VR-Goggles just play back the corresponding view-point-stream in time, depending on the tracking.
That would possibly unburden VR from real-time rendering and could transfer the VR experience from the reduced or no global illumination as we see it so far, to VR experience at the quality of Hollywood movie productions like Transformers or high end photorealistic architectural stuff ... and so on.
To evaluate the necessary resources for such an approach, i was wondering if there is an info available about the 3D resolution of the tracking of the Oculus Rift DK2 ? I'd like to calculate possible viewing-positions form a certain range of mobility. Basically along a path, with a certain freedom of movement. Knowing the amount, i can determine the render resources for pre-rendered scenery.
Can possibly the necessary view-points in space for a smooth and convincing VR experience be lower, than the number of addressable tracking-points in space ? If so, the comfortable density of viewing-points is of interest for me, so to skip unnecessary tracking-points in space.
Is there info about this available yet ? Or is there a viable way to find out with tests ?
Is there a coder out there who'd be interested to setup a multi-stream player with me ... i'm not a coder at all, though interested in this concept a lot.
Regards,
Frank
I'm thinking about a method to pre-render VR view-points as each view-point being a stereoscopic movie stream. So to speak there is a lot of movie streams, and the VR-Goggles just play back the corresponding view-point-stream in time, depending on the tracking.
That would possibly unburden VR from real-time rendering and could transfer the VR experience from the reduced or no global illumination as we see it so far, to VR experience at the quality of Hollywood movie productions like Transformers or high end photorealistic architectural stuff ... and so on.
To evaluate the necessary resources for such an approach, i was wondering if there is an info available about the 3D resolution of the tracking of the Oculus Rift DK2 ? I'd like to calculate possible viewing-positions form a certain range of mobility. Basically along a path, with a certain freedom of movement. Knowing the amount, i can determine the render resources for pre-rendered scenery.
Can possibly the necessary view-points in space for a smooth and convincing VR experience be lower, than the number of addressable tracking-points in space ? If so, the comfortable density of viewing-points is of interest for me, so to skip unnecessary tracking-points in space.
Is there info about this available yet ? Or is there a viable way to find out with tests ?
Is there a coder out there who'd be interested to setup a multi-stream player with me ... i'm not a coder at all, though interested in this concept a lot.
Regards,
Frank
3 Replies
- Correct me if i'm wrong or misunderstand but woulden't this require thousands of possible streams because the tracker resolution is very high?
- This sounds a lot like what otoy is already developing, http://render.otoy.com/newsblog/?p=547
I'm hoping some other companies that are focused on rendering will release similar methods to 'bake' viewpoints/scenes in a sort of point cloud, hologram-type format. - Hi, yes it'd be all possible view points.
Thx for the link to OTOY, the Holo Video is very interesting. I'm still checking into it, didn't find a demo download for the DK2 there yet but asked otoy for info.
As it talks about real video and 3D animation, i'm quite curious how the system actually works and in which boundaries its to be produced with.
Holo Video so far sounds like the right format for spectacular VR experience.
