HPA Tech Retreat 2011 Day 2
Mayhem, confusion, and chaos continue!
By Adam Wilt | February 17, 2011
Day 2 of the Tech Retreat covered the year in review, CES, cloud storage, broadcasting, pool feed audio, content protection, transcoding, stereo subtitles, and more...
These Tech Retreat posts are barely-edited stream-of-consciousness note-taking; there's no other way to grab all this info in a timely manner, get it published, and still get enough sleep for the next day's sessions!
Introduction & Technology Year in Review - Mark Schubin
This is the 25th annual, or 76th annual, "this is the year of HDTV." Nielsen says 56% households have HDTVs, but only a fraction of viewing is HDTV, mostly sports. AP: "The Pope is going HD". Still no HD press bridges / pool feeds. Super Bowl forst show not shoot-and-protect for 4x3; was 16x9 HD all the way.
2nd or 83rd annual "year of 3D". Quantel: 83 3D channels worldwide. CBS: "No way... unless someone pays for it." Costs 6x as much for a 3D baseball broadcast. CEA's Shapiro: "3D is overhyped", and when the CEA says it, it must be bad. Nielsen survey of world buyers regarding 3D: 28% own 3D or will, but 52% are negative. (USA: 8% positive, 78% negative). Everyone wants 3D sports, but sports is the hardest thing to do in 3D. More people want 3D TVs if they haven't already seen it (!).
NAB T-shirt: "Broadcasting - It is what it is." Super Bowl 2011 had 111 million live viewers. Nielsen: 35.6 hours/week TV viewing per person, "very close to a job", time-shifting increasing, but not affecting brand purchases (so folks not skipping commercials? [Or do the commercials simply not matter? -AJW]). Forrester: online 17 hrs/week, TV 14 hrs/week (but Forrester was a self-selected voluntary survey, online!). 33% stream video.
OTA broadcast ad revenues peaked 2007, cable in 2008. Cable TV: basic subs dropping, digital up, data & phone over cable up; getting to the point where cable revenues from phone/data will exceed those from TV. Cable homes passes 47.1% (down from last year). STBs: cable 42%, satellite 31%.
Tech trends: large-format cameras (DSLRs, AF100, F3); LED lighting; HDR lens (Live Technologies LiveLens); packaged 3D (lens adapters like Zepar, V3i moving-sensor; single-lens ISee3D); virtual cameras (Hego OB1).
Storage: small (Ki Pro Mini, Atomos Ninja); 3D on tape, disk, and SSDs; record directly to LTO-5 (1Beyond); on-site data management (Marvin system); more/faster/cheaper.
Processing: 2D-to-3D; 3D correction; searching; tracking; synching; cloud editing; stitching; acoustic summing. Distro: tiny web encoders; USB 3; low-latency codecs (Fraunhofer: 1 macroblock line latency); 60 GHz; 4G mobile; JVCs ASI-out camera. Presentation: OLED; curved monitors; ultra-bright; glasses-free 3D; multitouch wall screens; ETRI Real-Sense 4D (scents, heat, fans, etc.).
CES Review - Peter Putman, ROAM Consulting
DVD players sell for $20: DVD is probably on its way out. 32" TV $300, name brand 32" TV $350, 42" 720p plasma $449, LG 42" LCD $575, Toshiba 55" LCD $987; 3D 46" LCD with 4 glasses and Blu-ray player $1700.
Ripped from the headlines: TV business "in the toilet", prices dropping 20-30%/year. Net decrease in pay TV customers, biggest quarterly drop ever; cord-cutters switching to satellite, internet. TV shipments are slowing down, though LCD and plasma are up. Best Buy suffering, as neither 3D TVs nor internet TVs are catching on (just smartphones and tablets). 76% probably or definitely won't buy 3D TV. Netflix has more than 20 million subscribers, revenue up, may pass Comcast this year.
CES attendance: over 100K. Lines for the cab lines! Cab fares approaching extortion with $3 credit card surcharge. Yet, no "wow factor" at the show. Attendees largely optimistic. Reduced 3D emphasis, though passive and autostereo displays shown. Over 80 different tablets at the show.
Highlights: Active 3D glasses, passive, autostereo, Internet TVs, Oakley & Polaroid 3D glasses. Multimedia over Cat5/Cat6 cabling, over wireless, over AC power. Power over multimedia wiring: HDBaseT, DiiVA. Wireless HDMI, DisplayPort over structured wiring.
Mitsubishi 92" rear-projection DLP HDTV: $5000, active shuttered 3D. Samsung transparent OLED 19 (on a shadow-box diorama: why???), 30% transparent 960x540. WHDMI wireless HDMI 5-6 GHz band, using proprietary coding. Kenmore smart appliances: washers, dryers, refrigerators, all controllable from smartphones. JVC 65" passive 3D TV. Toshiba autostereo demo with 9 possible views that require you to move to see the 3D effect (?). Sony 24.5" autostereo 3D OLED. IDT Frame Rate Conversion, converts anything to 60Hz with deinterlacing, motion interpolation, dejuddering; makes even security cameras look good. Polaroid anaglyph (blue/yellow) 3D demo with 32" LCD. Pico-projectors: 286K sold in 2010. Sony, Nikon has 'em in cameras. Has anyone ever seen one in actual use??? Sony internet TV:
Summit multichannel wireless audio demo,uUp to 10.1 channels, uses ultrasonic doppler to see where you are, tune the sweet spot to match. Gesture recognition, mostly from Chinese TV vendors. Tablets: everyone had tablets (BTW, 16% of Galaxy tablets are being returned to the store). Flexible OLEDS! AV over AC wiring: Sigma Designs HomePlug (Note: electrically noisiest AC adapters are those for iPod and iPad). Big screens: Sharp 70", Samsung 75" (both 3D LED TVs) Projectors are losing out to large panels. Mitsubishi 155" tiled OLED display, Samsung 38 foot wide tiled wall, Sony 92 foot wide tiled wall (both 3D of course). Silicon Image ViaPort daisy-chain cabling, up to 8 full HD displays, uses the TV as the hub to connect all AV devices. CES humor: "HDMI was chosen because it's the only 4-letter word left." Internet TV add-ons. 3D MDTV, 800x480 broadcast on side channel. HDBaseT (US) vs DiiVA (China) connectivity formats. Toshiba autostereo with tracking camera to move overlay screen to optimize the visual sweet spot. Weirdest product: iPhone bottle/can opener case.
Global Content Repositories for Distributed Workflows - Ingo Fuchs, NetApp
Benefits: lowered CapEx and OpEx, flexibility, de-emphasize IT infrastructure in favor of focus on core business. Issues: hype, security, standards. Cloud storage: is it a means, or a service? Don't lose sight of your goal: what is the desired result? Hint: it's not "having cloud storage."
Goal: more efficient content / post business. Enable global, distributed workflows. What to expect: efficiency and flexibility; automation (of replication and distribution); metadata (drives the automation); interoperability and consolidation. A global content repository, properly implemented, automatically handles these needs.
Different content type need different service levels. Single unified architecture, with caching, SSDs, disks, tapes behind it.
Make storage part of your IT design process; consider virtualization and leverage cloud-based storage models.; consider bottom line instead of emotional decisions. Actively engage with standards efforts, e.g., SNIA CSI, CDMI TWG, http://www.snia.org/cloud
Note: clouds needn't be externally supplied; they can be developed in-house, too.
New Developments in Content Protection and the Battle Against Piracy Panel
- Moderator: Brad Hunt, Digital Media Directions
- Dean Angelico, Verance
- Wendy Aylsworth, Warner Bros. Technical Operations
- Matthew Gillins (sic), Fortium
Piracy still a problem, including ripping of DVDs / Blu-rays (both at-home rent-rip-return and of Academy screeners) and file-based piracy.
Fortium: advances in security threats whether from DVD / Blu-ray rippers, digital delivery compromises, or screen grabbers for iTunes screeners. 1/3 of piracy is ripping of rented disks, 1/3 online bootlegs, 1/3 street-corner bootlegs. 32% US and 36% UK consumers admit to copying disks in past 6 months; figures indicate these cost 8% of sales. Fortium code works on confusing the ripper navigation, spoofing total disk size, requiring excessive navigation that normal players don't traverse, etc. Also watermarking. Academy screeners ripped this year were analog dubs, 4 weeks after distribution. Option for DVD PIN play; requires PIN to play disk, built into non-bypassable menu. For Blu-ray, extension of Patronus anti-rip tech to make ripped disks un-authenticatable.
Verance's Cinavia is a content protection watermarking system in the audio track. Camcorded videos preserve the watermark; Cinavia-enabled Blu-ray players (shipped in the past year) detect the watermark and halt playback with an enforcement message displayed instead. A Cinavia "trusted source" watermark is similar, but mutes audio instead of halting playback. These watermarks are added in final production after editing/mixing, with plugins for Windows and Mac, e.g., for ProTools.
Forensic watermarking is used for tracking piracy; each disk has its own tracked watermark, so that dubs can be traced back to the source. Forensic watermarking typically done at the projector.
Cinavia Content Protection watermarking is direct piracy intervention, a last line of defense, and may convert some nonpaying viewers to paying customers.
Aylsworth: MPAA studied piracy in 2004, 2005; major studios lost $6.1 billion to piracy worldwide [note: that's what the MPAA reports; others may dispute the amount of monetary damage. -AJW]; 60% overseas, 20% US. Content Protection won't stop piracy, but will slow 'em down; "I've not been burglarized but I still lock my doors, I lock my car even though a determined thief can still steal it." Here are three ways consumer copying can be allowed.
Digital Copy, launched in 2007, allows free or low-cost copy of a DVD or Blu-ray, additional copies at higher cost. Can be provided as files on disk; transcoded via Internet from DVD content (?). Content provider determines terms, DRM.
Managed Copy with Blu-ray AACS can make additional copy in device-compatible format for whatever device, using approved DRM. Copying machine makes request to authentication server, comes back with available offers. Selected copy is authorized and machine makes copy.
UltraViolet (UV) lets consumer buy UV content online or in a store. UV media authorized by up to 6 registered family members, up to 12 registered devices, up to 3 account-linked streaming devices, 1 unregistered device with per-play authentication. Publisher selects from approved UV devices and DRM methods. Rollouts probably late this year, early next.
Panel discussion: tools are getting easier to use, a couple extra steps in post, but value proposition is high. Working to make sure disk QC works with Cinavia, protection holds off for 20 minutes so that QC can be done without triggering protection (on unfinalized, unencrypted check disks). Is Digital Copy working, and does UV replace all these other copy-allowing mechanisms? Digital Copy uptake a bit slow, but optimistic for the future. Will UV take over? Long range, all these copy-allowance methods may merge. Audience survey: only 5%-10% here at HPA have used some sort of digital copy themselves. What about arms race against the ripping tools? Two teams: pro hacking team and a pro protection team, constantly studying the ripping tools and working on circumvention. Have to balance between playback compatibility and content protection. "One fellow drills three holes in the disk outside the data tracks; it plays, but nothing can copy it!" How inaudible is the Cinavia watermark? Constant tradeoff between high-fidelity and protection robustness. Lots of testing to ensure that the watermark is inaudible, tuning and tweaking the signal to keep it (mostly) undetectable. Constantly updating it. Cyber-piracy growing; what does panel think about dealing with streaming piracy? Working on legislation so that when a pirate is found, can get court order to go through ISP to shut pirate down; international effort. If content protection is moving into the media, will this mean that DHCP can go away? (applause) No, DHCP will remain; it's necessary to protect the uncompressed link.
Subtitling for Stereographic Media - Jonathan Jenkyn, Screen Subtitling (http://www.subtitling.com)
Stereography: both 2D (monocular) depth cues like brightness, contrast, relative size, scale, occlusion, perspective, motion parallax, accommodation; and 3D cues(stereo) parallax and steropsis, convergence (within 10 meters or less). 3D cues fairly weak; if 2D and 3D cues fight, the 2D cues win. Yet 3D cues cause the headaches. Positive parallax: object is in or behind screen; zero parallax at screen plane; negative parallax, object is in front of screen. You need to avoid positive parallax much beyond the interocular distance; actually can get up to 5 degrees of divergence for "super depth" effect (but don't do it for too long!).
Issues: vergence vs. accommodation; blur triggering accommodation changes that fail to resolve (thus, avoid narrow depth of field); a weak depth cue that may lose to 2D cues.
Subtitling: not sexy, "read but not seen". Typically in lower third area. Styles: plain white text, drop shadow, black box, stripe boxing, shadow boxing, colored text, off-center placements. In 3D: where in depth to place it? What happens if screen content comes into negative space beyond ("in front of") the subtitle? Pull subtitle into the most negative space, but now it's floating way out in front; convergence strain, very tiring. Solution: match parallax to the action: same parallax as the focal object, or the foremost object (scene by scene). "Complement the material for the best 3D experience." Requirements: subpixel position in X and Y (audiences can detect a 1/10 pixel move). Breaks existing conventions; may be annoying for 20 seconds, but after 5 minutes it looks OK. Audience learning gradient unknown (for both 3D as a whole and for 3D subtitles).
Display size matters: for a 22" monitor, a 1% disparity is 0.19", just fine; but for an IMAX screen, 1% is 140.62", a definite problem. Material must be mastered for the display size.
3D subtitle prep: acquire 2D subtitle, manually position in 3D space (changing timing to avoid scene changes as required, as well as depth positioning to avoid occlusions or excess disparities). Reposition for each language to make sure things fit. This can be tiresome (4x as long to subtitle 3D as 2D, "and you need a whole pile of ibuprofen handy"); can it be automated? Yes: disparity mapping; computationally intensive, a bit tricky in broadcast where the fine adjustments of the two cameras may not be accurate (vertical and angular displacements, for example).
Screen Subtitling can do burned-in subtitles; some Blu-ray and digital cine projectors can super titles from a data feed, but compatibility is limited so far. Live-event titling is a huge headache.
Future: boxing the title to avoid occlusions; 3D depth in the titling; titles should get smaller as they push into positive parallax; dynamically positioned in broadcast applications; better disparity maps.
Pool Feeds & Other Shared Origination: How Will They Sound? - Ken Hunold, Dolby
Pool feeds used for restricted-access news events, sports, humanitarian events; e.g., "Hope for Haiti Now" concert, "Stand Up for Cancer" telecast, State of the Union address. Multiple feeds, multiple recorded channels. Looked at loudness LKFS of main element; Dialnorm value and actual measured loudness; channel config, e.g., 5.1; dynamic range and LRA (loudness range) measurement (EBU R-128; "'Loudness Range' estimates the distribution of loudness of a programme with statistical tools"), can be used to measure how loudness has changed through the distribution pipeline.
For "Stand Up to Cancer", distribution in both 5.1 and 2-channel audio. Produced in stereo and 5.1. 3 TV channels delivered stereo, 5 as 5.1, 1 upmixed stereo to 5.1, 1 used L/R channels of 5.1 as stereo (center and surround were missing).
State of the Union: pool feed in stereo, but pool feed was just one element the networks mixed in their audio. Each network had a different idea of what they wanted to produce. Same kinds of issues on how the feed got sent to air; some upmixed to 5.1, large variation in loudness range between channels (down to only a 2dB loudness range in one case!), one version had the LFE channel as loud as the main channel, 7 were heavily compressed, one had a completely separate audio source.
Distribution methodology: pool feed to network/channel to home viewer, or pool feed to network/channel to affiliate to home viewer. But some programs also add a satellite bounce or other waypoint.
Digital distro can result in similar sound from channel to channel (but any errors will also be delivered faithfully). Need to ensure consistency, and that program delivery properly assigns channels. Loudness strategies exist and should be followed; feeds can be passed as-is or mixed in to another feed; producers will determine the sound. Don't remaster at every stage; "if it ain't broke (until it is), don't fix it."
Discussion: if you set up the upmixer wrong, you can get a lot of LFE content. The White House pool feed isn't ATSC compliant. Seems like many channels have compressors inline all the time, whether needed or not.
Multi-Language Video Description - Art Allison, NAB
(Disclaimer: the preso doesn't include any recommendations from the NAB!)
Visually-impaired (VI) description is narration of key visual elements inserted into pauses in dialog. Mandate in effect October 2011 for four top networks (ABC, CBS, NBC, Fox) and 5 top non-broadcast networks to provide 50 hours/month. Stations must pass though descriptive audio if they have the technical capability. All DMAs must have this by, maybe, 2037 depending on rulemaking. The law doesn't say what language the description must be in.
There's an HI channel as well (hearing-impaired).
Carry the stereo description in 2 of the 8 channels of Dolby E atop the ATSC 5.1 stream. Two Dolby E streams can hold two 5.1 mixes plus two descriptive channels, e.g., an English and a Spanish feed each with description. No problem through the distribution infrastructure. For transmission, can send multiple audio services for a video channel in an MPEG-2 stream. Cable may need to pick one description for analog NTSC SAP.
The metadata in ATSC (like the PSIP) lets the TV set decode and play various services. CEA-CEB-21, Recommended Practice for Selection and Presentation of DTV Audio, in progress since July 2008 but almost done, will be used to drive set design. Key Issues: user setup and control, explicit language selection, explicit VI and HI selection. Receivers should automatically select the best combination of available audio channels to match user preferences.
Multi-language VI and HI tracks are in our future. Force-fitting into a 2-channel mode is problematic; so plan ahead. Now is the time to think about this stuff.
Discussion: Dolby Digital+ for these services? Gear doesn't exist yet (is this EAC3?). EAC3 part of standard in future, may be a possibility. But still have to deal with legacy equipment. So in very long future, sure, but legacy sets without EAC3 capability need to be accommodated. Or, look forward only, leave legacy sets behind.
Next: Broadcaster Panel, Consumer Delivery Mayhem, Over-The-Top Video, and Loosely-Coupled Transcoding...
Get articles like this in your inbox: Sign Up
freshdv - Fri, May 24 2013 - 2:23 pm
RT @natums: If you are considering going to #CineGear, get your pass today. It’s the final day it’s free! http://t.co/nts0BYEFxw Also, #cinebeer anyone?
freshdv - Fri, May 24 2013 - 2:00 pm
RT @wingrove: Maschwitz law: 'IT’S NO LONGER OK FOR CAMERAS NOT TO GIVE US EVERYTHING THEY’VE GOT' @prolost http://t.co/SlUSFNtaaG