How Many Screens?

June 17, 2015, SMPTE Entertainment in the Internet Age, Stanford, CA—A panel looked at the issues associated with multiple screen viewing. Colin Dixon from nScreen Media moderated the panel. [panelists included Mark Adams from Accedo, Chris Cukor from LG Electronics, Will Law from Akamai, and Jim Monroe from Net2TV.

Dixon opened with comments about the changes in consumption patterns, mobile is up 130 percent, and connected TV is up 380 percent. A lot of viewing is on iOS or Android devices. The many devices start with smartphones in the morning, PC during the day, and TV or connected TV in the evening. During prime time the average household has 1.6 screens in use. The multiple screens are causing problems for advertisers, they cannot mount ads across all the screens.

File preparation and adaptive streaming? It seems that many versions are needed for the many screens.
Law stated that the drivers are the formats for delivery, containers, codecs, and digital rights management (DRM). A contraction is coming in DRM with a common encryption which allows multiple DRM from a single file. Open standards like HLS are helping the containers and the codecs have mostly settled with H.264 while some are slowly moving toward H.265. Transcoding is a compute-intensive task, but new packaging is changing the containers. New architectures place the packages in the network at the edge to eliminate a 30 percent performance hit if the calculations are done in the core. Their servers are all white box with no custom hardware for transcoding except for live events. The drivers are standards and commonality.

Creating apps for ore OS?
Monroe noted the number of devices and proprietary operating systems is a challenge, but they try for the greatest level of commonality. Hollywood is now picking which OS to support in their content. However, Web, Roku, LG, Samsung and others have little in common. The best ones are Amazon Fire, WebOS, etc. that are moving towards HTML5, but still have only 50-60 percent commonality.

IP-TV opportunities changing content?
Adams stated that they see the views and try to anticipate future use. The new users are looking for short form and summaries, search capabilities, and curation. There is a difference between a remote and a keyboard in use models and user expectations. By anticipating the control device and audience they can add customization, view the user data and time, and work to extend the viewing duration. Unfortunately, there is still a lot of device and audience fragmentation. Better metadata helps for dynamic programming and personalization. They don’t use pre-rolls, but do use teases. They tell the viewers they will get two units or 60 seconds of ads max. the transition to ads is easy in some formats, and hard in others. There is a 3-4 second tolerance before an ad shows, so they need content integration.

Multiple protocols for files and control for smart TV?
Cukor commented that solving the fragmentation problem requires media packaging, app development and better coordination across devices to work with any viewer setup. The control for the TV is an attempt to create a launch point, even if the content is not on TV. Google is OK and HTML5 is 60 percent than other formats. HD, UHD, and HDR are not easily scaled as other platforms. It is possible to put mobile and Chromecast content into a Web container for TV viewing. Interoperability is also possible through protocol extraction from Apple, Fire, and Roku. A single API call discovers and connects the other devices to the smart TV. The connect SDK finds devices and pairs to them.

WebOS for fas, high quality 4K?
Law declared that 4K is not the same as UHD. The frame rates and HDR require higher bit rates. They think 15Mbps is the threshold for throughput. In the US, one in four houses are 4K capable for real-time streaming. The alternatives are to download, pre-buffer, or have big buffers of a few minutes for non-real-time delivery. In 3-4 years, the industry will be able to stream everything. The UHD files can be cached at local sites and servers to give about 20 ms latencies. Currently, average bandwidth is 3-4 Mbps so 15 Mbps will be hard. This shows up in popular content at prime time, for example Netfix. Caching locally helps a little. Fortunately, there is little native 4K content.

HDR, etc. and other emerging standards?
Cukor acknowledged that the issue is education to show consumers the value of the new products to generate demand. The technologies need to get to a critical mass, like investment for more OLED panel production, but manufacturers are still waiting for the market, content, and infrastructure to exist.

Is UHD more for an upscale market?
Monroe offered the challenges that devices change, the apps change and like the move from SD to HD, there is a need for some transition use.
Adams contended that content offers users new choices and control. it seems that quality is not critical. Viewers are not happy with pre-cache processes so the market issue is that, once again, technology leads content just like HD.

Biggest technology challenge?
Monroe noted closed captioning is hard to get working.
Law opined that the change to 4K will require a full supply chain revamping, from capture to process pipeline.
Adams said that the data causes issues. The device is known, but not the person using it.
Cukor added platform fragmentation and legacy devices make leveraging data harder.

Chromecast on WebOS?
Cukor claimed that you can make a webgen in the OS that results in “Chrome-less” browser without bars. The bonus content and social technologies are at a crawl stage in interactive TV. Some algorithms and fingerprinting generate trigger points to launch HTML5 content that is based on what is showing on the screen.

Increasing engagement. About 5 percent of viewers access the bonus content which spreads through social media to many others. Best use of multi-screen?
Monroe declared that the iPad users are moving to Roku and other alternatives. The ideal would be to follow the viewer from screen to screen. The ad user identification needs more development for analytics and tracking functions.

Lots of information?
Law responded that big data is the driver. They have to handle 40-50 million requests per second and double that for log lines, so the need to evaluate all inputs at about a PB per day at the low end of the daily data volume.

Would a UHD stream provide a constant user experience?
Law suggested adaptive profiles cold provide a minimum 1080P picture. People prefer buffering to downloads if the UHD TV can switch out of HDR mode, etc. If a Blu-Ray bit rate encoding is used, the quality will not be linear, but good enough. Minimum bandwidth is 15 Mbps and higher data rates are not really noticeable. The best compromise would be 1080 P and HDR.
Most will be unicast, since there are no global multicast standards and any format will be router specific. It is possible to do some multicast on UDP in a peer-peer, tunnel, ATSC 3.0, etc. which alow alternative ways to deliver the content.

Link across streams? The business case holy grail would allow cross screen campaigns with every screen in appropriate time and place. Technical issues?
Monroe agreed that a log-in once and then follow across screens would be wonderful.

Fiber to the x versus DSL?
Law commented that it is an unequal world. New providers will come on line to address this situation. Cell phones get over 5 Mbps and the advent of 5g will increase the data rates even more. Satellite gets even more bandwidth, so there are already alternatives in place.

Synchronization?
Cukor noted that ACR keeps the view and engages with the content on the screen. The second screen is better suited for control, then transfer the content to the TV. The current mode of navigation for TV is very outdated, so a way to change the user interface would improve discovery.