Feel free to discuss this post in the comments!
This past Thursday, SMPTE-NY had a joint meeting with the IEEE and SID-MAC. The topic “Evaluation Grade Monitoring for Cinema and Television”.
In truth, most of the discussion was on Television monitoring. In a separate room was the Dolby PRM-4200 Reference Monitor, a 20-series Panasonic plasma, and (3) Sony displays sitting side-by-side: BVM 24″ flat tube CRT, the new 25″ OLED in the middle, and their 24″ BVM LCD.
Before I get to my notes, some quick impressions on the displays I saw:
- Dolby PRM-4200 LCD: This $50,000US LCD is a monster. When I first walked into the room I though someone had rolled in one of the old flat screen 40″ CRTs or a rear-projection set. That said, it’s a great looking LCD. And listening to the Dolby presentation I got a sense as to why it’s so expensive. For one, it natively displays the full DCI P3 gamut. Dolby goes through aÂ meticulousÂ process to hand-select panels that can meet this spec. Second, the blacks are very very impressive. Third – the darn thing can output (and, IIRC, maintain grayscale linearity) up to 600 Nits to emulate a typical home display (in comparison, most reference monitors are calibrated between 100 – 200 Nits). Â But again – if you need to take this thing with you on-set, it’ll take three men to move it.
- Sony OLED: This is an impressive monitor. One piece of footage they were running was a long 3 minute wide shot of cliffs over water as the sun sets. The deep shadows in the OLED held detail long after the LCD was overwhelmed by the (dim) ambient room light. And in the Sony presentation it was clear, though not explicitly stated, Sony considers this their Grade 1 reference monitor – knocking their LCDs off thatÂ pedestal.
- Panasonic Plasma: The pro-sumer 20-series they displayed was calibrated to Rec. 709 gamut. Like the OLED, it’s just fun to look at. As I’ve written in my notes, they don’t consider the LCDs to be Grade 1 reference monitors for color critical decision making. They also don’t consider their Plasmas to be in that class either – though they think they make fine client monitors since they can be calibrated very closely to Rec. 709. The big issue is how plasmas need to cut the power in extremely bright scenes (think: standing in front of a white cyc). A reference monitor needs to maintain brightness throughout the grayscale.
Some other general observations:
- This was a very technical crowd of around 60 -100 people. One speaker asked colorists to raise their hands, and I think maybe 3 hands went up. Lots of engineering types in the room.
- Green phosphors – in general, displays have trouble getting saturated greens to meet the wider gamuts. This seems to plague all the display technologies.
- There was lots of talk about modern displays not just meeting the specs of CRTs (which had lots of problems) – but going beyond them.
Notes of importance:
- It turns out that CRTs never had an explicit standards for image display (peak white level, gamma response, etc). TheÂ phosphorsÂ were standardized, but nothing else since CRT technology had inherent characteristics that were difficult to change. If you were building a CRT and used SMPTE phosphors there wasn’t much else you could do to the image. Today, that’s all changed and nothing matches CRT, or each other. Standards and Recommended Practices are currently being developed.
- There’s controversy over what should be a “Specified Standard” and what should be a “Recommended Practice”. I think my notes cover it pretty well.
Read On for my notes from this meeting… and feel free to comment at the end of this post.
â€œEvaluation-Grade Monitoring for Cinema and Televisionâ€
Thursday, February 24, 2011
6:30 PM Program Begins
Produced by Mark Forman, Mark Forman Productions, Corp.
No more CRTs! What to use for a reference display? Join us for a technical discussion on the future of evaluation-grade monitoring in the post-production environment for digital cinema and television.
David Bancroft: Bancroft Technical Consulting, IEEE BTS Distinguished Lecturer and Chairman of SMPTEâ€™s Fixed-Pixel-Matrix Display Working Group
- Interchange & interoperability : consistent results, across a range of suppliers. Suppliers have a common spec to target. Consumers get the possibility to get a consistent viewing experience.
Monitor vs signal standard:
- recorded signals have specified standards, monitors don’t. CRT didnt need specs due to the technology – operated in a narrow range.Â Â NOW we need standards since current displays don’t natively match CRT – they used to be deficient to CRT, now they are excessive.
- Calibrated measurement tool. 1 volt needs to equal 1 volt. Measuring appearance to reveal content we are adjusting, lighting, exposure, color balance, & balance between camera angles, scenes, even if it doesn’t reflect reality. It must “tell the truth, not just pleasing”. Pleasing is for consumer sets.
What we need:
- Same content looks same on every reference monitor and should never looks like it needs to be remastered in another facility.. No serious change in appearance with new tech vs CRT mastering.
What Manufacturers need:
- Ref monitor that should be capable of meeting standard and must have the data to measure.
- May have to set standard to a higher black level that the monitor can attain so operators can actually calibrate. Blacks in OLEDs are literally black – how do you measure pluge?
Standards are codified and should beÂ practical, realistic
Recommended Practice are suggestions
They need to cover things we can control: Inherent properties of the monitor, inherent properties of the viewing room, controllable properties of the monitor, controllable properties of the viewing room.
The following is personal opinion of speaker about how they should be divided:
- Standards are the things the User can’t control in the Monitor and viewing room. These are inherent properties.
- Monitor inherentÂ – gamut limit, intensity range from black to super-white @ color temp, contrast range, pixel counts (spatial response) and 1-to-1 pixel mapping mode (no scaling); temporal response (how quickly can it refresh for acceptable human vision); viewing angle (max allowable variation in luma and color in horizontal and vertical) & surface reflectivity
- Viewing inherent. -Â Â Max low level light ambient light; reflectance of surfaces on walls.
- Things easily controlled by the User
- Monitor setup: engineering setup, operational setup (quick calibrations)
- Viewing: ambient light, monitor surround (10% of peak white?)
Current status of specifications –
- EBU revision – April 2011 tech document 332o. It’s a USER requirement
- ITU has group working on gamma
- SMPTE, 2008 Study group is now Working Group, 2011Q2 / 2011Q3 for monitor inherentÂ standard; 2011Q4 for viewing inherent; then Recommended Practices in 2012
Audience Question: Shouldn’t we go for more colors and wider gamut rather than be restricted by what CRTs could reproduce?
Answer: We need to agree on incompatibility between old and new content. Not a technology problem, requires more bits, maybe noisier pictures.
Call for User Input on SMPTE panel
Pete Putman, ROAM Consulting, Display Consultant, Educator and Columnist
Suitable CRT Evaluation Replacements:
Display wide grayscale at consistent over time
- LCD transmissive, like light shining through window blinds – white crush is problem and is non-linear at very top of response curve
- Plasma, emissive, like staring at sun (my comment, not speaker’s)
- OLED, emissive : low yield, two types SM OLED & P OLED
- FED, killed by lawyers
CRT Replacement must :
- track color consistent
- neutral gray
- wide viewing angles
- calibrate to standards
- wide grayscale without clipping or crushing
Accurate gray scales are the key. Shadow detail is toughest. false contouring
LCD – CCFL are tough to track. Color accuracy is problem (especially greens due to added blue phosphor)
Plasma – False contouring is not a problem
Plasma Vs LCD
- Wider viewing angle – especially up/down axis
- can match 709 and most of DCI
- full white drops with power management dropping light output (from 100 nits down to 65 nits)
- High switching rate (up to 600 Hz)
- Phosphors age (usually in first 200-300 hours).
LCD v Plasma
- Polarizing is a problem and reduces viewing angle
- Led backlights help reduce blacks
- green can be undersaturated – blue red hits 709 and DCI, green 709.
- IPS tend to look better.
- Pro monitors moving to LED backlighting, doesn’t solve viewing angle.
- Motion blur can be problem.
- Put up grayscale and move around to test off axis viewing issues.
Panny TH-42PF20U Plasma
- Covers Rec.709, most of DCI,
- gamma just barely off and fixable (especiallyÂ using something like Davio),
- grayscale track is good as good as many CRTs
- 120-129 nits, 931:1,
- nice deep blacks.
Waiting in the wings – OLED, SM OLED is current favorite. High current Low voltage,
LEM-150, 15″, $6k,
- gamut maps 709 nearly precisely
- not suitable for P3
- nice gamma response
- grayscale track is flat
- error better than many eval. displays.
- no viewing angles issues.
- 100-120 nits
- black level .001 nits. Beyond range of testing device
Gary Mandle, Senior Product Manager, Sony Electronics Inc.
OLED Eval Monitors
Started in 1994, first showing in 2001, 2004 small OLED
$203 million plant just for OLED
First pro OLED was 7″ and now 24″
Simple process, tough execution (process didn’t seem simple to me)
Explained science of OLED
- Low energy
- uses phosphors
- entire panel is 1/16″ thick (some layers are less thick than a water molecule)
To get to P3 gamut they block low energy emissions and enhance high energy emission (esp. in green channel) to get expanded saturation.
Several different OLEDs : Stacked, transparent (HUDs), PLED (flexible), active matrix (that’s the reference OLED)
OLED beats CRT at
- low light
- black performance is whatever you want it to be
- pixel speed from white to black is nearly instantaneous, have to slow it down because the speed produces flickering
- tuned to balance smear vs flicker
OLED gamut accuracy from black to white, LCD loses accuracy near blacks.
Building 17″ 25″ panels. 10 bit, P3, surpassed CRT life 30,000 hours (failure is sudden toward yellow), 12 bit hdmi, display port
Half weight of LCD
2k for cinema, displays interlace (1080i), plugin card for Harris rasterizer
Burn in can be problem, same as CRT but not any worse, auto shutoff after 10 minutes
72 watts, no fans, no heat problems,
Headroom over 200 nits
- It measures color balance and luminance to keep panel stable as phosphors decay, using the headroom to slowly compensate for phosphor decay. When it can no longer compensate it does a sudden failure. You ‘ll know it!
Steve Mahrer, Senior Technologist, Production and Media Services, Panasonic
If we replace CRT, don’t just emulate, improve – CRTs were not quite 709
Ideal CRT replacement
- high res (2k and beyond)
- great Scaler for non native resolution (don’t introduce artifacts)
- 1:1 pixel mode
- user controls for gamut / gamma management
- good off axis viewing
- “native” options
- no 120hz 240hz
- Bigger is better – content errors missed on 19/ 25″ devices are fully visible at 50″
- 3d capable? Probably.
- black not quite OLED
- good gamma especially with LUT management
- green phosphor decay has been fixed for 3d and looks great in 2d – especially in next generation 30-series plasmas
- 80% P3
- better than 709
- LUT presets for 709 601, EBU
- At best are probably Grade 2 devices
- large venue
- high contrast
- great for film outs as cinema display.
25″ vs 50″ size reveals issues
In 2 years all plasma will be 3d for cost of 2d
Panny 42″ 50″ 65″ will all be 3d later this year
At NAB releasing 30series
- better blacks closer to reference
Don’t make smaller than 42″ Plasma since it tough to make pixels small enough
No announcement about OLED.
Floating white points on Plasma at >130 nits can be better controlled, no comment on when it’ll happen.
David Schnuelle, Senior Director, Image Technology, Dolby Labs, Chairman of the SMPTE Digital Cinema Technology Committee
PRM-4200 Refernce Monitor
- better than CRT
- Fixes CRT deficiencies
- Dark blacks
- built in 1d 3d LUTs
- Rec709, P3 (and more)
It’s all about Math – digital processing to model the standardized display
10 bits not enough anymore – Image Interchange Format is 16 bit OpenEXR over 12bit log.
Dithering needs to be removed
Can match consumer brightness. (500 – 600 nits)
How do you do your initial passes without being in the Big Room? Must meet P3 , not a simulation.
10 bit outputs need 12 bit displays
1500 x 3 = 4500 individual RGB LED, 10 bit no dithering,
Balloting finishes on proposal for math to turn electrical impulses into an image on the screen next month
Characterize, calibrate in factory, compensate
Getting accurate color readings at low levels is very difficult and EXPENSIVE
LCD LED life testing. 50,000 hours stay within 1% of their spec.
LCDs are individually qualified because of getting to P3 (mostly in green channel). Primary and white color accuracy better than most CRTs they’ve measured.
- .005 nits for black. They tend to ride it higher – If they lower it expert viewers complain
- 2.4 gamma
- peak deviation are minimal
- color tracking highly precise
$1k / inch @ 50 inches !
Discuss in the comments…