(Page 1 of 2 pages for this article  1 2 >)

Filed under: Business •Cameras •compression •Distribution •Hardware •Post Production •

HPA Tech Retreat 2010 - Day 3

The Tech Retreat’s third day covered regulatory issues, HDR imaging, using a plasma for reference monitoring, SOA, networking, file-based workflows… and Mo Henry.

(As usual, these are barely-edited notes, scribble-typed as people spoke. Screenshots are copyrighted by their creators; poor reproductions thereof are my own fault.)

Jim Burger, Dow Lohnes, gave the Washington (legal) update. Due to snow, Washington has been closed, so nothing bad has been happening (grin). Elsewhere (internationally), three-strikes laws for Internet piracy and ISP filtering are going strong.

Cartoon Network vs. Cablevision: case involving “remote DVR” functionality. Claims: content was stored in buffers (thus, copied); charge of “public performance.” Second Circuit Court decisions: buffers don’t equal copies (too transitory); playback copies are allowed; private playback is not the same as a public performance.

ACTA: Anti-Counterfeiting Trade Agreement, secret international agreement (!) which includes clauses on ISP liability for third-party conduct, secondary liability. Causing consternation!

Net Neutrality; Comcast vs. FCC (the “BitTorrent case”). FCC says Comcast violated principles by messing with P2P packets; Comcast says FCC can’t enforce “principles”, FCC has no jurisdiction over Internet. FCC rulemaking; old principles: customers can access any lawful content, connect legal devices, use apps/services, have ISP competition. New: ISPs can’t discriminate, and must disclose network management. Rulemaking still in play, rules may come out after 5 March, or may wait for court ruling in Comcast-BitTorrent.

Comcast-NBCU merger: $30 billion deal, Comcast majority partner, GE minor. Comcast promises: increased kids, local, and VOD content, etc. Subject to FCC, DOJ appeal. FCC jurisdiction based on OTA NBC TV station transfer (not cable, where FCC has no power).

Spectrum war: “looming spectrum crisis”, broadband wants 800 MHz. Where from? Broadcast? Satellite? Claim that we need mobile broadband. Stimulus bill requires FCC “National Broadband Plan”, due to Congress in March. Battle cries: CTIA, CEA, T-Mobile, AT&T, Verizon, etc.: we need more spectrum, OTA (over-the-air TV) isn’t watched. NAB & MSTV: we just transitioned to digital, don’t take it away; more folks are watching OTA; wireless isn’t using all the spectrum they already have! CTIA/CEA: move OTA to low-power distributed broadcasting; reduces interference and need for guard bands / white space; opens room for wireless broadband. Broadcasters: still interference with 8-VSB, tower siting issues, hugely expensive. So what happens? Mandatory clearing of broadcast spectrum isn’t likely.

CALM Act: directs FCC to regulate commercial loudness. In 1983 FCC had been getting complaints “for at least 30 years” but declined to regulate. ATSC adopted loudness guidelines in 2009, but House passed CALM on a voice vote, referred to Senate.

Aside from that, “I have never seen so much gridlock in Washington.”

High-Dynamic-Range (HDR) Imaging Panel, Moderated by Charles Poynton, with

Stephan Ukas-Bradley, ARRI - High-Dynamic-Range Image Sensors,

Yuri Neyman, ASC, Gamma&Density - Interpretation of Dynamic-Range Capabilities of Digital Cameras,

Dave Schnuelle, Dolby - The Tyranny of Dark Rooms: Why We Need to Replace the CRT Display.

Poynton: in imaging, DR has a specific meaning. There’s a hard clip in highlights, and noise in the shadows. The standard deviation of the noise defines the size of the “noise-valued increment” and the number of those between shadows and clipping is the DR. RED is ~66dB, 2500:1, or 11 bits, so the 12th bit is always noise (assuming full exposure, of course). In processing, a DR number is uncertain at best. In display, contrast ratio is the preferred metric. Tone mapping used to map legacy low-dynamic-range (LDR) material to new HDR displays, also used to map HRD to LDR displays. In 1990, the iTU defined BT.709; “we screwed up” by not specifying the display gamma (“electro-optical conversion function”).

Ukas-Bradley: ARRI’s new Alexa cameras use 3.5K Bayer-mask CMOS with 10% overscan (for viewfinder look-around area). 35mm sensor size, compressed onboard recording.  EVF models (Alexa EV, ALEV) use 16x9 EVFs, optical VF (OV) uses 4x3 finder for use with anamorphics. Target output formats are HD and 2K. Larger pixels have higher sensitivity and lower noise; 8.25 micron pixels with 800 ISO, 70% fill factor and 80% quantum efficiency. Dual-gain architecture (DGA), highlights constructed from low-gain path, shadows from high-gain path, Output is 10-bit HD or ARRIRAW. ALEV III Show Reel, looked pretty impressive. ARRI DRTC (dynamic range test chart, demoed at HPA last year) can show more than 15 stops; ALEV reads 11.6 stops on this chart.

Neyman (DP on “Liquid Sky”): published camera latitude numbers don’t match cinematographer practice. Best possible dynamic range in video is 6.3 stops. Hypergamma 3 (Sony F35?) shows 10.3 stops. Using ARRI DRTC charts, 5.7-5.8 stops measured. 5217 negative film: -3, +4 stops. Printed, get 6.5 stops total (practical range for negative is 7.5 stops, print 5.5 stops). Human eye can only handle 5 stops when resolving details / optimal visual contrast. Steps are not stops and this may be where the confusion sets in. RED ONE 4.7 stops. ODR (observable DR) vs PDR (potentially recoverable DR). Testing film cameras and digital cameras for speed: F35 is ISO 2000, Viper is 170, RedSpace is 35, RED raw is 78. A plea for objective measurement. (Very fast-moving—and more than a little confusing—presentation; need to research this to see why Mr. Neyman’s assertions differ so markedly with everyone else’s numbers!).

Schnuelle: replacing the CRT. Color CRTs with us a long time, but they’re a poor match for current monitoring needs. Several Bluy-rays had to be be recalled because of too much noise, excessive sharpness seen on consumer LCDs & plasmas, stuff not seen on mastering CRTs.

Fixed-pixel displays lets things pop out that CRTs hide with their gaussian MTFs. Gamuts are all over the map in consumer displays (not wide-gamut, but wild-gamut, according to Poynton), they’re not the SMPTE C or EBU phosphors on mastering CRTs. Local contrast is much better on consumer displays; no flare. (However there’s a trend in some consumer displays to highly reflective panels, so environmentally reflections can be problematic; Poynton notes that these specular reflections aren’t the same as diffuse reflections, which degrade black levels). “A real reference display has no knobs, but this assumes a matching reference viewing environment.” Important to know the creative intent, and we have a standardized display environment, but that’s a CRT in a darkened room—we’re still standardized on the limitations of early CRTs.

Discussion: immediate questioning of Neyman’s figures; Morton’s figures (discussed by Neyman) included specular highlights. Poynton: latitude in old days was margin of error for exposure setting, now we say total stops top to bottom. Question: how do we move HDR images around and convert to/from 10 bits? Peter Symes, quoting Dave Bancroft: what we do with new displays should not cause us to look at existing programs and say “ah, we need to change that.” Poynton: yes, but not to the point of limiting things. Schnuelle: fine, but no one looks at old content on old-style displays, it simply doesn’t happen. Director should be able to see something representative of the displays people are going to see the content on. Q: is the Alexa’s sensor cooled? A: yes, D21 cooled to 32 degrees Celsius, Alexa heated or cooled to 32 degrees. Putman: measured consumer sets at 450 nits or less, 200 nits in movie mode, so mastering at 350-400 nits is probably OK. “Tanning lamp mode” or “torch mode” is 450 nits, not as high as published figures of 500-1000 nits.

Pete Putman, Science Experiment: Low-Cost Plasma as a Reference Monitor. Yes, it’s time to toss the CRT.

A pro reference monitor must be able to track consistent color, neutral gray, wide dynamic range without clipping or crushing, consistent viewing over wide angles, precisely calibrate-able. Grayscale is the hardest thing, especially in digital (Pulse-width modulation (PWM) as used on plasma & DLP is the hardest). Shadows are always difficult; grayscale may wander around the ideal. Consumer sets often white-clipped, S-curves, tend to be too hot / too bright / high midtones, inconsistent black levels. Rolling back to 100 nits helps a lot. CCFL backlights bad for color, LEDs much better. Plasma pretty good, too. Plasmas use PWM for tonal control, >600Hz.

Q: is a $2000 industrial plasma good enough for critical monitoring? (LCD problems: costly, high black levels, off-axis color/tone shifts, bad color gamut on CCFL backlights). Took stock Panasonic TH-42PF11UK and tweaked with calibration tools. Got stable gamma with a couple of “speed bumps”, consistent if not perfect RGB tracking, great blacks. Color gamut exceeded 709, even large portion of P3. Green was a bit shifted towards cyan (for brightness). Gamma at 120 nits was 2.5; looked very smooth (movie mode), a bit of a bump in 2.2 gamma. Max gray drift was 145K in the shadows; bit of a blue bump around 70%. Brightness 100-120 nits (29-35 ft-Lamberts), contrast 1189:1 (checkerboard), 11370:1 sequential (gamma 2.2), black level 0.124 nits.

Wanted better; got a Cine-tal Davio using a 3D LUT to correct residual color and gamma errors (as seen in the demo room). After calibration, color accuracy was comparable to reference-grade CRT. Best of all: a very cost-effective solution.

Sara Duran-Singer & Theo Gluck, Disney, Classic Animation Restoration. Disney started in ‘37 on nitrate, transferred to safety film ‘52-‘56. In ‘93, one of the first DIs at Cinesite. Now in multi-year resotarion of all animations, 4k scans and digital cleanup. Fantasia now, the Winnie the Pooh. Six years ago made decision to archive everything filmed out to successive-exposure B&W films. Issues: nitrate deterioraion; how to scan? Steamboat Willie orig neg lost, some 1934 dupe neg also no good.

When Snow White was remastered bad tape splices, lots of perf damage. Thus need to be very aggressive in preserving library. Recall all nitrate, scan, fixup, film out to successive-color B&W film. Technicolor: 125 lbs, 750 lbs blimped. 1932 short “Flowers and Trees”, was 1st Technicolor release. Alignment of 3-strip records problematic.

Successive exposure developed at Disney, red, green, blue frames one after the other.

Sleeping Beauty: Technorama (sideways) 7.62 files of film, new 7.1 mix from original 3-track sound. Horizontal scratches confuse stock anti-scratch software! Original AR was 2.55, new restoration recovered that from the nitrate neg (not seen in previous re-releases, from 1987 film dupes scanned in 1993). Scanning from the successive B&W neg fixed RGB registration errors from previous re-releases.

Screened raw scans / fixups from Bambi, Pinnochio. Freakin’ gorgeous.

Rescanned nitrates sent out to new successive-color B&W films for archiving without fixups; don’t want to bake in today’s technology in case the future has better restoration tools. Fixed-up films saved at 2K to LTO-4 tapes and a new color neg is struck.

Mo Henry speaks - Mo Henry is a negative cutter, appearing in more film credits than just about anyone else ever.

Mark Schubin introduces Mo Henry.

Mo Henry speaks.

She told us a bit about her life. Babysitter for John Wayne’s kids. Tallest kid in 12th grade. Started at 18 in neg cutting, the 3rd generation Henry in neg cutting. First gig: Jaws.

Got bored, left the business, was real estate agent in Beverly Hills in the ‘80s: big hair and a Cadillac convertible. First big client was a Mafia kingpin selling a house: had to go in to his house before a customer and hide all the guns, cash, and drugs, then replace them after the customer left. When kingpin went to buy a house, Henry was hired, and was told that if anything was missing after the move, she would be buried in the back yard! After that, became producer for commercials and rock videos. Worked on “Fruit of the Loom” spot; was attacked by a kid with a loaded gun on location but a cop deflected the kid. Henry, pushed to the ground, thought: What a way to go: lying on Hollywood Ave surrounded by guys dressed like fruit!

Having OCD is very helpful for a neg cutter!

Lived next to Timothy Leary; had a neg cutting business in London until partner had heart attack at dinner and collapsed face-down in the pasta. Left-handed (but has to cut with right hand) and a voracious reader despite being half-blind from neg cutting.

IMDB is incorrect; has only 60% of the films she’s done. Also worked under the alias “Ruby Diamond” when cutting porn: hey, a negative is a negative…

Now working in archiving at Sony and Warner.

Next: SOA, collaborative networking, file-based workflows, sync & timing…

(Page 1 of 2 pages for this article  1 2 >)

Just a note of thanks to Adam for providing this wonderful coverage.  I’m too close to the event to see what’s going on.  I’ll be reading and re-reading this coverage to find out what happened.  It has been a jam-packed four days!  Thanks, again, Adam!

Posted by tvmark  on  02/19  at  12:31 PM

---

Name:

Email:

Location:

URL:

Smileys

Remember my personal information

Notify me of follow-up comments?

Submit the word you see below: