Irecently had the pleasure of sitting on a panel at Light Reading’s 2018 Big Communications Event in Austin, Texas. My fellow panelists,
Arianne Hinds, Ph.D. (principal architect, video and standards strategy at CableLabs), and
Ozgur Oyman, Ph.D. (from Intel and a board
member of the VR Industry Forum group), provided a very lively discussion on the topic of
VR, MR, and AR. During the course of the talk,
Hinds explained how CableLabs is working on
a container for VR based on existing technology pioneered in Hollywood’s visual effects
First, let me state that I am a big fan of technology. What Hinds and CableLabs are doing is
very innovative and nothing short of remarkable. During her explanation of the container
format, she said it will massively reduce the
amount of bandwidth required to deliver VR
while still maintaining incredible resolution.
And, on top of that, it won’t need a headset to
render. Of course, it’s still very nascent and relies on display technologies that have not yet
been commercialized. But it’s a bold step forward in an industry that is still fraught with
significant challenges to consumer adoption.
Yet, as she talked about this new development, I couldn’t help but cringe. VR, as the next
generation (or even evolution) of video content,
has numerous issues that need to be solved
right now. First, there is production. It requires
significantly more manual processes to create and publish VR video content than it does
2D. For example, in a 360° environment, how
do you hide the camera crew and equipment?
And, much of the production requires specialized stitching software (still in its infancy) to
make the final product. Outside of expensive
volumetric video studios (like Intel’s recently
announced Intel Studio) that can create more
fluid VR content without the need to stitch, creating high-quality VR videos is still hard, complex, and expensive.
Second, there is delivery. If you want to stream
VR video, you really need 8K or more resolution
(to reduce the pixilation), which requires next-
generation, head-mounted displays. And even
with the newer codecs (i.e., HEVC and AV1), we
are still talking about 10GB or more in band-
width. And when that VR video is in a chun-
ked HTTP format, there’s the latency to con-
tend with—which can, unfortunately, result in
Finally, there’s the cost. Yes, there are some
reasonably priced prosumer cameras hitting
the market, but it’s still an investment. But
with VR camera technology changing at such a
rapid pace, gear acquired this year could eas-
ily be obsolete in 12 months’ time. And these
are only some of the technical challenges!
But, now we are talking about developing a
VR container? Is that even a problem to solve
Although, again, I laud CableLabs’ efforts
here, there is a danger in trying to bring standardization too soon to an industry that is still
trying to figure itself out, one in which the technology is evolving at breakneck speed. Perhaps
this container will help galvanize the industry,
but I fear that it will divide people into camps—
those who support the CableLabs container and
those who don’t. Will we end up seeing the kind
of container fragmentation we have in HTTP
chunked streaming? If we do, VR video adoption may be severely hampered.
And, on top of that, there are no content owners involved in the efforts. One of the requirements to be a member of CableLabs is to be
an operator. So, it’s possible that members like
Comcast could bring their NBC folks to the table
but, in speaking to Hinds, that is not happening.
But it’s not the operators who will implement
this container in video production workflows.
Shouldn’t content owners have a say in how the
container is developed, given that they’ll be the
ones who employ it?
I see the merits of the effort—the VR industry
needs technology guidance. I just don’t agree
with the timing.
In VR Video, Is the
Tail Wagging the Dog?
Jason Thibeault is the executive director of the Streaming Video
Alliance. Follow him on Twitter @_jasonthibeault.
Comments? Email us at email@example.com, or check
the masthead for other ways to contact us.