Is streaming VR content over your browser feasible in the foreseeable future?
Jamieson Brettle of Google thinks so.
At his talk at the VR World Congress 2017 in Bristol, Brettle spoke about the opportunities and challenges of streaming VR content over the web. He also spoke about some of the specifc projects that are trying to bring VR to your web browser.
Web VR makes a lot of sense when it comes to consumers. The barriers to entry for consumers are much lower, apps are easily downloadable, and it is easy to update content to ensure people are always up-to-date.
For providers, however, there are some serious challenges.
The need for compression
The main problem for web VR is the sheer amount of data needed to get VR content moving across the web. This makes compression essential.
Google have taken steps to get make their video compression better, mainly moving to VP9. This has led to consumers getting content faster with less buffering time.
Being able to effectively stream VR over the web, however, will require more compression then this.
This is clear when you begin to some of the data requirements involved.
An average piece of content may consist of 360 stereoscopic video, fully immersive spatial audio where the content is rendered binaurally, and virtual environment rendering and indoor mapping. The latter in particular involves huge amounts of data if point clouds are used.
The challenge of how to get all this data over the web to a browser is being attacked by a number of different Google-backed projects.
The first is OPUS, a compression technology for spatial audio. OPUS utilises a process called ambisonic compression, whereby sound is projected on a 360 sphere around the user. Through a complex process called spherical harmonics, the more spheres you add the better your sound but the higher your bits.
The current level of compression achieved by OPUS is 192kbps for 4 channels and 384kbps for 16 channels.
The second project is Draco, an open-source compression library for meshes and point-clouds. Draco currently boasts a decode speed of 35ms, or essentially real-time.
So, the web is a viable VR platform, but it will require a lot of data. A lot of data will require a lot of compression. Whoever solves the compression problem first is likely to open up a huge new marketplace for the VR industry.