Video is at the heart of our industry and this three-paper session presents impressive and compelling technical advances spanning encoding, AR/VR/XR and immersivity.

Our first paper uses subjective expert assessments of Just-Noticeable-Differences (JND) in streaming video encoding to train a deep learning model. Results from using that model show impressive bit-rate savings can be achieved.

Our second paper introduces ARF – the Avatar Representation Format, an MPEG activity currently at committee draft. The goal of ARF is to offer an interoperable exchange format for the storage, carriage, and animation of 3D avatars. A technical overview and illustration of ARF is provided along with a discussion of its integration in to a mobile communications system.

Our final paper presents a truly impressive 30K horizontal resolution 360-degree camera system and 15K hemispherical display. The technical design and challenges of the camera, lenses and capture system are shared.