Nokia, or at least the Alcatel part of Nokia, seems to be hell bent on following the path towards a virtual reality future, but some research at Bell Labs, shown at a research briefing at Nokia’s Cambridge Labs this week, gives some idea of the price an operator may have to pay to deliver a VR vision.
Bo Olofsson, head of multi-sensory communication research at Nokia Bell Labs, brightened up an afternoon of product demonstrations and technical briefings which mostly related to products which are with us today, to talk a little about the future – which Bell Labs sees as being somewhat further away than most, but most decidedly it has a VR flavor.
The distance away real VR is, comes down to the fact that the world’s networks cannot hope to deliver the massive bandwidth that VR can eat. Olofsson talked about the world of SD video being a 1 Mbps data rate, the shift to HD taking broadband networks to 10 Mbps, the twin worlds of 4K and what Nokia calls 360 video due to its investment in the Ozo camera and its tools, being a 100 Mbps world.
Olofsson then talked future, whether this was 5 degrees of freedom which true depth of field VR, from one fixed position can offer, or the 6 Degrees of Freedom that is the ultimate VR prize of fully free, eye tracked moving VR, which takes us up and beyond the 1 Gbps to a 10 Gbps world.
But he is definitively NOT saying that this is what your access broadband line has to carry, and made it clear where Nokia is investing its VR dollars. In an earlier demonstration, he showed that by only sending key parts of a screen which were being looked at directly in full definition, and the rest in a lower resolution, he firmly believed that these numbers can be cut drmatically. He talked about Captured Data Rate at the camera, being the issue, not the amount of data that is sent.
“A human can only look at one view at a time, and all we have to send at any point in time is the central tile of what an eye is viewing in high definition and the surrounding tiles which are viewed by peripheral vision, in lower definition. But we have to capture the entire experience and hold it in the cloud in case that viewer turns his head and needs to take in a different view,” said Olofsson.
There was a lot we agreed with, in that he felt that the target timeframe was around 2025 no sooner for both realistic VR and 5G, and also agreed that the smartphone would not be the platform of choice for viewing VR – it simply does not have a good enough screen. Also, he did not call it VR, as he felt, like Faultline Online Reporter feels, that mixed reality, where both what is real and what is virtual are viewed equally, is the necessary outcome and that the device for this has not yet been invented.
What will it look like? And who might delivery it? He shrugged and confirmed that Nokia has made a commitment to go into every part of the VR chain, capturing it, editing it with full workflow tools like Ozo Live, and delivering it to the device, but stopping there and leaving the VR device of the future for someone else to invent.
He talked about Magic Leap, and the name is well judged in that its mixed reality viewing platform so far merely demoed and hinted at, is still two years away, and if it works and the $4 billion plus valued start up does not run out of money first, the device would indeed be a magic leap.
Olofsson just knows that we are scratching the surface of VR now and that Nokia plans to be there, potentially dominant, in professional camera experience capturing systems like the 360 Ozo, but also devices which capture sounds, movement (Haptics) and even smell to be delivered across a network to whatever this device becomes.
In a way, this is Nokia taking a magic leap into the market that Sony has occupied for so long, that of the professional and prosumer camera supplier. Nokia sees no future in the consumer end, and the recent financials of players like GoPro testify to this.
But going a step further Olofsson showed the key graph for us, precisely what he anticipates the loading would be on broadband access networks, by calculating what amount of data per month would need to be delivered to each consumer.
He began saying that “I don’t think that longform VR video will be dominant in the next five years. But by 2025 if just 10% of video content was delivered as full VR, that might require 3,360 GB of data per user per month. If it was 20% that would rise to 5,093 GB of data per user per month.”
Of course, this is not just the strain on any single given access network, but the entire network backbone would break under this stress, as networks are today.
By comparison he has already assumed that video will be 80% UHD by that time, which would imply 887 GB per user per month just to start with, a tiny fraction of what it would take to bring VR to every device, and here’s the reason?
“The human eye can distinguish one 60th of a degree so you need at least one pixel for each one 60th of a degree. That’s 60 pixels times 360 degrees both around and over each viewer. We believe that for this you would need 25K video not just 4K, to make VR look realistic, with the density of high quality video throughout.”
As he said, all of this dictates what has to be stored, and the quality it has to be stored in, and just one viewpoint needs to be delivered at a time over the network, which is what drives the calculations on data delivered per month. Storage and processing in the cloud is going to be up by several orders of magnitude as well.
“We are using a 6K camera and taking multiple shots, and then stitching them together in the network. There are no image sensors with any higher resolution than that right now.” For his demos this is delivered from a camera at 1 Gbps to Nokia’s Mobile Edge Computing cloud – effectively its Velocix CDN hardware nodes, where it is processed and then sent at 7 Mbps over LTE or WiFi to the demonstration. And to be honest it is pixelated as hell and made our heads hurt, and a very poor video experience. So, we think his sums on what it will take for this to be a TV like experience are bang on.
Two things worry us. First that Nokia paints this vision of how tough it is going to be, and invites operators to work with it on their own networks visions. Because that’s what Alcatel did around IPTV in 2000 and many operators bemoaned the fact that they have overprovisioned dramatically to get their and lost their trust in Alcatel. Secondly this entire VR vision only comes to fruition if someone invents an idea device for mixed reality viewing. And last time around VR just never happened at all, and it may not this time either.