HPA Tech Retreat Day 2

ATSC 3.0, the Olympics in 8K, and more...
Adam Wilt
By Adam Wilt 02.21.13

The Tech Retreat is an annual four-day conference (plus Monday bonus session) for HD / Video / cinema geeks, sponsored by the Hollywood Post Alliance. Day 2 covered design considerations for HDR displays, ATSC 3.0 and the future of TV, broadcasters' concerns, mobile DTV, 8K UHDTV, and more.

Herewith, my transcribed-as-it-happened notes from today's session; please excuse the typos and the abbreviated explanations. (During the rest of the week, you can follow the Tech Retreat on Twitter with hashtag #hpatech13, thanks to various Tweeters in the audience. I will of course post my notes at each day's end.)


Leon Silverman, HPA

This is the 19th Tech Retreat, a gathering originally known as "The Engineer's Picnic".

Introduction & Technology Year in Review

Mark Schubin

The 20th Tech Retreat will be Feb 17-21 2014 in Indian Wells.

27th (or 78th) "the year of HDTV".

3/4 of US households have HDTV sets, but we're still using analog press bridges and are still shooting and protecting 4x3.

4th (or 85th) "year of 3D", but The Verge said 3D is dead (massive applause).

Good year for TV: record election spending, record Super Bowl and Olympics audiences.

Wired: "no one uses Smart TV Internet because it sucks."

Ken Auletta, New Yorker blog yesterday: "Some believe... foremost threat comes from... Dish Network's Hopper... Netflix and Amazon... My candidate is... Aereo.

Radio received by 95% of world; US adults use OTA radio 8x as much as satellite radio, 17x more than streaming/downloading.

32,000 year-old animation (cave paintings; look up "Marc Azema").

Now-ish: phosphor, plasma lighting. 4K@900fps, 8K @ 120fps

Future-ish: ultra-multiview autostereoscopy, metamaterials for computational imaging (metamaterials.duke.edu/node/922), artificial DNA for data storage.

But the 9 foot Lechner distance still allows lean-back and lean-forward; can vary eye-to-screen distance by 3 feet overall.

What Is Actual Customer Behaviour?

Sarah Pearson, 1-3-9 Media Lab

1-3-9 refers to the viewing distances: handheld, computer, TV. 1-3-9 Media Lab studies tech usage patterns, with in-depth micro-behavioral coding and ethnographic surveys (a month at a time), semi-structured interviews. Study running for six years. Detailed analysis, from channel changes, people entering/exiting room, who is in control of the remote control, capturing content of all screens as people text during TV, etc.

We think appetite for simultaneous screens around TV will steady; engagement across platforms is on the rise; behavior change around new tech is socially driven.

In one family, the husband had the remote 75% of the time. Then an XBox and Skype were installed, and the wife wound up using the TV as an entertainment/communications platform much more.

Study with Olympics / Super Bowl: give people mobile apps related to the programs and see what happens: how did peaks of engagement drive use of second screens (anecdotal sample clips).

A day in the life of three young women, recording viewing distances and second screen use throughout the day. Similar study of a young couple with a baby, a TV, and a Windows 8 Surface; another with a family watching the Super Bowl, with iPads in hand (video clips showing screen uses and viewing distances).

[Basically a lot of sample clips of behavioral data gathering; not much in the way of actual results. -AJW]

What Will Win the Future of TV?

Hans Hoffmann, European Broadcasting Union

TV consumption pattern is changing, with increasing second screen usage. Mobile TV is coming back. Context vs. media services:

Market situation for Ultra HD: HD is state of the art and migration is complete. Tech evolution continues, but display market is declining in profitability. New opportunities? Smart/connected vs. immersive. Immersive = perceived better quality.

Why is 3D not working in the home? Eventually all 40"+ displays will be 3D enabled, but consumers don't use the 3D capability. Reasons: uncomfortable glasses, cost, limited content, poor image quality, etc.

Why buy a new UHD-1 (3840x2160) TV? No content, only looks better if you're very close or the screen is very large. Will UHD-1 make it, or be superceded by UHD-2 (8K).

Resolution: 3840x2160 (UHD-1) or 7680x4320 (UHD-2); 10- or 12-bit; 24, 25, 30, 50, 60, 120fps progressive only; color response defined; etc.

HD viewing distance 3x picture height, so UHD-1 is 1.5x and UHD-2 is 0.75x! Can we see a difference between HD and UHD-1 on a UHD screen? 72 participants tested, F65 material cropped to QuadHD @ 50Hz uncompressed, also downconverted to 1080i, 1080p and 720p. Viewed at 1.5x picture height (standard viewing distance for UHD-1), and 2.7 meters (average in UK viewing distance). Viewers rated image quality of clips on a numerical scale.

No statistical confidence that 1080i, 1080p or 720p looked any different, but the UHD material looked better... very slightly.

These tests were with uncompressed material; compression may change things significantly (need to test with HEVC in UHD, h.264 in HD).

Resolution isn't the whole thing; HFR important, too. Higher res seems to call for higher frame rates. We have HFR cameras and HFR displays, the problem is the distribution chain in between. Same thing with HDR.

Even if farther from screen (resolution advantage not seen), blur improvement in HFR, HDR, color improvements would still be seen.

Need production infrastructure (Sony and Panasonic have ideas here), HEVC coding for transmission, new STB. Also need new HDMI standard for single-link UHD at rates above 30p.

EBU UHD-1 test sequences available to public at tech.ebu.ch/testsequences

HEVC test; slight detail loss at 10 Mbps, worse at 6 Mbps, when zoomed in on image.

Relax: HDTV investments are good, Just increasing resolution isn't enough. 3D audio is important, too; we need new standards. 3D and UHD will benefit from each other. Most important: content.

Viewer-Preference Statistics for Shadow, Diffuse, Specular, and Emissive Luminance Limits of High Dynamic Range Displays

Scott Daly (presenter), Timo Kunkel, Xing Sun, Suzanne Farrell, & Poppy Crum, all of Dolby

The usual design for an HDR display is to design what you can build. We want to build based on human visual system (HVS), instead:

Interested in viewer preference limits: important to entertainment viewing.

Diffuse reflectivity vs. highlights: speculars can be a thousand times brighter:

Emissives: light sources 3x to 10,000x brighter. Standard reproduction techniques: Print 1.1x for highlights, video 1.26x, cinema film 2x.

Functional human vision dominated by need to understand reflective objects; not as well adapted for speculars and emissives.

Lots of previous work on HDR displays, and lots of problems and side-effects: contrast changes, boosted black levels, clipping, tone scale changes, local-dimming artifacts, perceptual side-effects.

Most people see the right image as being sharper. It's just contrastier.

Most people see the right image as being sharper. It's just contrastier.

Brighter (contrastier) images also show contouring and compression artifacts more readily.

Goals of experiment: find out what viewers want in an HDR display, for black level, diffuse white max, etc. Results:

The visual system was the limit, not the display.

The visual system was the limit, not the display.

The visual system was the limit, not the display.

The visual system was the limit, not the display.

Highlights: average is 3980 cd/m^2, 13x the diffuse max of 400 cd/m^2.

How to use all this data? 12-bit Perceptual Quantizer, visually smooth, whereas a 12-bit gamma curve shows quantizing at low luminance levels.

How to avoid side effects? Many studies using varying code values vs. varying backlight brightness; variable dynamic range displays. Higher brightness yields higher satisfaction up until higher black levels become objectionable.


Next page: Mobile DTV Deployment at Fox Television Stations, ATSC 3.0, and more...

Prev Next »