Oculus VR Develops '5K Super High Immersion Immersive Movie' Using a New Approach

by Nathanael Coyne

Oculus VR , a VR related hardware and software development company, announced that it will experimentally introduce a new "5K immersive movie" in the June 2018 update of the VR headset Oculus Go and Oculus Rift . I explain how Oculus VR's developer blog "5K immersive movie" of the new technology was developed.

Behind the Tech with John Carmack: 5k Immersive Video | Oculus
https://developer.oculus.com/blog/behind-the-tech-with-john-carmack-5k-immersive-video/

June Platform Updates: 5K Video, Embedded Dash Panels, and More | Oculus
https://www.oculus.com/blog/june-platform-updates-5k-video-embedded-dash-panels-and-more/

John Carmack , Oculus VR 's Chief Technology Officer (CTO), is a super programmer who has worked as a lead programmer in the development of games, also known as the origin of the FPS game called DOOM . According to Mr. Carmack, there are various restrictions imposed on the video decoder, such as resolution of the display itself, range of FOV (real field of view), compression. Oculus announces that it re-releases it as "5K immersive movie" which reproduces the short VR animation " Henry " released in 2015 with 5K using the new decoding technology.

In order to project a 5K 60 fps movie (about 5120 pixels) over 360 degrees in the headset VR for 1 second, 5120 × 5120 × 60 actually requires 1.5 billion pixels of information. It is extremely difficult to decode this amount of information and it is possible to reduce the amount of information to a certain extent by sacrificing image quality and frame rate, but Oculus is different from the conventional one in order to realize a 5K movie We examined the approach.

by Ted Eytan

In the VR movie, the screen displayed according to the line of sight of the person wearing the headset is changed. Since the range displayed at one time is only 1/6 of the whole sphere, if you can decode only the part of sight, you can solve the problem of huge amount of information. However, since VCR movies that may change viewpoint at all in order to decode images of a few seconds as a video clump ( GOP ) as a mechanism of video compression, it is not possible to adopt a system that decodes only the visual field did.

In order to change the image to be projected according to the viewpoint, the 360 ​​° image is resampled to 2 2 × 2 K images. The 2K × 2K image is further divided into 10, and 20 images with different focus are produced in a 360 ° movie. By changing the viewpoint of the user, I change the combination of the images to be projected and switch the image that looks to match the movement of the viewpoint.

So Oculus said that instead of decoding all 360 ° movies perfectly with 5K image quality, they adopted the approach of lowering the overall information volume by compromising the image quality directly above and below the sphere. Most users do not see directly above or beneath while watching a VR movie, and in addition to being relatively less important, in addition to being directly under the camera part of the movie, it is often black It may be filled.

Mr. Carmack explained that it was able to distinguish between the part which should be clear and the part which does not need to be necessarily clear and it was able to raise the quality as a whole, "Mr. Carmack succeeded in greatly improving the quality of the 360-degree movie, I am doing it. "