The 6500 Kelvin standard?

[Christoffer Glans]

So I've been thinking about this supposed standard of 6500 kelvin. As I understand it was THE standard when CRT's where around, everything shown on television looked correct if it was outputted at 6500K. But now a days almost all the plasmas and LCD's around are sold with a much higher kelvin, around 7200-7500K at it's normal setting. So everyone at home view content on TV's and monitors set to that temperature, not 6500K.

What is really the standard? If content should be outputted to a calibration of 7500K then most calibration standards are wrong, certainly the D65 709 standard.

Thoughts? Or am I wrong?

 

[Mark Toia]

The best way to color correct screens / monitors is to use a iPAD 3 retina with a picture you know is perfect.. or a colour card.
Match your monitors to that... They are incredibly color perfect.

I'm finding that 5800k works for me a sa base..

A plome gun is a good way to measure. but seriously.. try the ipad. Its like way cool for calibration.

 

[Christoffer Glans]

Isn't 5800 way too warm for calibration? I mean, 6500K is even colder then 5800 and it's too warm for screens. If the Ipad 3 is close to the Iphone 5 it has around 7000 kelvin (when I do a reading of that) which is more close to the "standard" I'm talking about.
Most TV's are set to 7500, my own is around 7700.

My point is that if you do professional work for television, commercials or other broadcast or even for stores with LCD/Plasmas, it's wrong to follow the 6500K standard, which is outdated (because it's made for CRT:s).

 

[Terry VerHaar]

The best way to color correct screens / monitors is to use a iPAD 3 retina with a picture you know is perfect.. or a colour card.
Match your monitors to that... They are incredibly color perfect.

I'm finding that 5800k works for me a sa base..

A plome gun is a good way to measure. but seriously.. try the ipad. Its like way cool for calibration.

This is a good thing to know. Thanks, Mark. I just happen to own the 3rd iteration of iPad so I am going to check it out!!

 

[Stacey Spears]

D65 is the white point for SD, HD, and UltraHD. Even when CRTs ruled the planet, most where >10,000 degrees kelvin. It makes the image look brighter on the show room floor. ISF and THX calibrations main focus is to adjust the grayscale to come as close as possible to D65. Note that D65 is a specific point on the black body curve.

SD displays had lots of issues. When HD came along, we had more resolution, but the same display issues. Now we have UltraHD with even more resolution, but still they can't get the basics right. D65, display gamma of 2.4, specific x, y coordinates for Red, Green, and Blue. Most of the time they are not accurate on purpose. The sad truth is every display should produce the same image, but they don't.

 

[Thommes Ulfeng]

The best way to color correct screens / monitors is to use a iPAD 3 retina with a picture you know is perfect.. or a colour card.
Match your monitors to that... They are incredibly color perfect.

I'm finding that 5800k works for me a sa base..

A plome gun is a good way to measure. but seriously.. try the ipad. Its like way cool for calibration.

Could you break it down for us Mark? A step by step. THX

 

[Christoffer Glans]

D65 is the white point for SD, HD, and UltraHD. Even when CRTs ruled the planet, most where >10,000 degrees kelvin. It makes the image look brighter on the show room floor. ISF and THX calibrations main focus is to adjust the grayscale to come as close as possible to D65. Note that D65 is a specific point on the black body curve.

SD displays had lots of issues. When HD came along, we had more resolution, but the same display issues. Now we have UltraHD with even more resolution, but still they can't get the basics right. D65, display gamma of 2.4, specific x, y coordinates for Red, Green, and Blue. Most of the time they are not accurate on purpose. The sad truth is every display should produce the same image, but they don't.

The thing is that I find almost all displays I come across, a lot higher then 6500K, I understand the basics of D65 on the black body curve, but when calibrating, it makes the whole image "warm". What I mean is that if I have this point as the base and do a color grade that is supposed to be super neutral, skin tones are warm, but neautral and the background slightly cooler, but still not too much, it's going to get blue and the skin tones are going to get too much magenta in them when viewing on what everyone else is viewing my content on.

As Mark said, calibrating with an Ipad 3 might be a good choice. I actually took a reading with my calibration meter on my iphone 5 and it ended on 7000-7100 kelvin, so I did a calibration of my screen to that. I think it nailed the best mid-way for how an image should look. It will of course always look different on different displays, but now my grade will be much closer to it's original on most screens, on calibrated screens it will be slightly warmer and on non-calibrated consumer screens it's slightly cooler, while on Ipad's and Iphones it's just the same.

If this is a problem and has always been a problem, that the white point and kelvin is off on all TV's out there, then why bother with the "standard" if it's only valid for work with cinema and high end stuff that's going to be seen on calibrated screens? If you do commercials and TV-shows, it's never going to be right when broadcasted if the standard is so off from what people have in their homes.

 

[Subhadip Sen]

You raise a good question. The white point for Rec.709, upcoming Rec.2020 and many popular color spaces is D65, or 6503K. But yes, the trend is most consumer displays and increasingly tending towards 7000K. I suppose it looks "more pleasing" and that's what the consumers want. The iPad 3/4 LG.Display panels are consistently in the 6800K-7200K range. Not just that, Nexus 10, iPhone 5 and a majority of consumer LCD TVs, are all tending to ~7000K. Ironically, the original iPad 2 was pretty close to 6500K, though I am not sure about the revised iPad 2,4 selling nowadays. Microsoft's Surface for Windows RT is also calibrated at near 6500K. Somewhat off-topic, but iPad 4's color accuracy is reasonably accurate to sRGB/709 at the high saturation range, but drops off by the white point. No substitute for a spectrophotometer calibration at any rate, though.

The question I assume is, should we start calibrating our displays to 7000K instead of 6500K? For internet / streaming / television oriented content, I think there's an argument to be made for it.

 

[Sam Taylor]

I am really glad that you have raised this subject Christoffer. It is something that I have sometimes thought about, but was not knowledgeable enough to resolve. I look forward to seeing what experienced colorists for tv and mobile devices suggest!

 

[Christoffer Glans]

You raise a good question. The white point for Rec.709, upcoming Rec.2020 and many popular color spaces is D65, or 6503K. But yes, the trend is most consumer displays and increasingly tending towards 7000K. I suppose it looks "more pleasing" and that's what the consumers want. The iPad 3/4 LG.Display panels are consistently in the 6800K-7200K range. Not just that, Nexus 10, iPhone 5 and a majority of consumer LCD TVs, are all tending to ~7000K. Ironically, the original iPad 2 was pretty close to 6500K, though I am not sure about the revised iPad 2,4 selling nowadays. Microsoft's Surface for Windows RT is also calibrated at near 6500K. Somewhat off-topic, but iPad 4's color accuracy is reasonably accurate to sRGB/709 at the high saturation range, but drops off by the white point. No substitute for a spectrophotometer calibration at any rate, though.

The question I assume is, should we start calibrating our displays to 7000K instead of 6500K? For internet / streaming / television oriented content, I think there's an argument to be made for it.

Yes, my point exactly, if all intended displays that will show our content isn't showing 6503K, then why is this the standard today? I expected this to be a thought of controversy of course, because the standard is set to enable all content to be made in the same way. But times change and technology change. If 7000K is more accurate for displays today, then why not change the standard Kelvin to that?

Even my TV at home has three settings, around 5600, 7500 and 8500. There's no 6500K settings at all and 7500 is set as "normal". If content creators use one standard and display manufacturers use something else, there is something wrong. If the higher Kelvin is a business model for display manufacturers to show off more a more pleasing look to sell more displays, then that IS the standard set.

I find it funny that I meet so much criticism for this thought, especially from some who seem to be very keen on getting the right colors. Both colorists and calibrator technicians say to me that I'm wrong, but their calibration and their grades shows a blue tint when people watch it, because the standard is so off from what people have.

Because there is so much difference, there seem to be a better idea to calibrate at 7000K, setting a correct RGB balance, contrast and brightness (blackness) instead. That way, your work will probably be much closer to what you intended the colors to be then with the D65 standard.

Science, standards and technological theory is as everything else, if it isn't working practically it's not correct.

 

[ericyoung]

The trouble is approximately 7000K is NOT a standard (yet), so trying to calibrate and grade to that is a non-starter if consistency is the goal.

And content creators have never had control of what viewing conditions are in the home, or on the road, or what end users do with their display controls. That's the whole point of a "standard". The industry adheres to the standard. What happens outside is usually not controllable. If the industry arbitrarily and individually chases what display manufacturers are doing, then the manufacturers will just move the goalposts again if they deem that "bluey tinted" pictures are desirable.

Of course we can check output on consumer displays at the end too, and tweak as necessary or output an alternative grade specific to the final deliverable, but unless final outputs are ALWAYS to a specific consumer display, it'll be an endless chase in circles.

 

[Marc Wielage]

I actually took a reading with my calibration meter on my iphone 5 and it ended on 7000-7100 kelvin, so I did a calibration of my screen to that.

Every iPhone and iPad I've ever seen is different from the standard and different from each other, plus their linear grayscale values and colorspace values are so unpredictable, nothing you see is real. Buy three iPads and lay them side by side, and you won't see the same pictures. I think iPads are fine for judging focus, performance, and camera framing; very bad for color or exposure.

If this is a problem and has always been a problem, that the white point and kelvin is off on all TV's out there, then why bother with the "standard" if it's only valid for work with cinema and high end stuff that's going to be seen on calibrated screens? If you do commercials and TV-shows, it's never going to be right when broadcasted if the standard is so off from what people have in their homes.

This has been debated for decades. Even in the 1990s, after SMPTE came up with the SMPTE RP-166-1995 standard for monitors, not everybody agreed on the D65 standard. In particular, there were debates on whether to use 30fL or 32fL as a brightness setting. The issue is, if you don't have some kind of standard, then all bets are off. You can never predict how something will translate when going to an unknown color space, unknown brightness setting, and an unknown gamma range. I can tell you from years of experience that if you do use a calibrated monitor, you have a better chance of hitting a reasonable middle ground with many kinds of monitors around the world.

I know of a major blockbuster feature where the director was paranoid about how his film was going to look on different consumer monitors. To placate him, the facility went out and bought about two dozen off-shelf consumer monitors and placed them in a room down the hall, which they christened "Best Buy." The staff engineers tweaked these monitors as best they could to sort of get in the ballpark of D65 and the SMPTE standard. Sure enough, the final transfer held up pretty well... but the director still wound up making a few changes based on how these low-cost monitors looked. No one could convince the director that he was chasing his tail in trying to use multiple standards as a reference.

In reality, it's very similar to the problem faced by sound mixers: a mix intended for a large theater will sound different on a TV set, in a car, on a boombox, and on iPod earbuds. And the standards for sound room calibration are actually looser than they are for picture (despite efforts by the AES, SMPTE, THX, and Dolby). Surprisingly, though, I think the best rooms in LA actually sound more similar now than they used to, and everyone has pretty much come to similar conclusions as to which speakers to use and how acoustics work.

The standards for P3, XYZ, and Rec709 colorspace also work pretty well for their intended purpose. But to throw these standards away is pure folly. Try to color-time a movie on a $300 Dell monitor and let me know how it looks in a theater.

 

[George Butterfield]

So, the guys who are getting good results, how are they doing it? Or is it as you say, you can't be good on all monitors.

 

[Christoffer Glans]

I'm actually not speaking about cinematic release because most often then not, feature films are shown on more calibrated places then commercials. For me, I'm doing colorist work mainly for commercials and TV and within that work the colors are almost always different from the D65 standard. The "middle"-ground is what I mean is set too low with D65, the middle ground for Kelvin is most of the times more close to 7000K instead.

If it is such a bad idea to change a standard, then how do we bang the display manufacturers in the head to really release every display with a "industry standard" setting that is the normal state of the screen? So even if people in stores keep jogging up the Kelvin to sell more, the ones sold have the industry standard set out of the box.

Maybe manufacturers could just lock the settings? Making it only avalible to people who knows what they are doing (with calibration)

 

[Dermot Shane]

easy to sort with something that allows custom settings like the Dolby 4200, i grade set to either P3 or 709, then switch to custom white point, plasma emulation, LCD emulatin, on and on

Even the DreamColor has 709 and sRGB pre-sets, i se the sRGB when doing a trim pass for web deliverables.. but i have a 709 grade to start the trim pass from

 

[Christoffer Glans]

I have dont some tests on the projects I worked on. Whenever I have my screen calibrated to rec 709 D65 I always hear from clients that it's too cold, to magenta-isch and so on. I mean, this comes from clients who do not know about stuff like this, their spontenous reaction to it. Whenever I calibrate and then change the white balance to a more K7000 look, they are happy.

So, is it my display that's wrong, even though my meter and Calman 5 is saying otherwise? Is it better to recalibrate my display to my "own" D70 or D75 standard?

 

[Elsie N]

cool!

 

[Angelo Lorenzo]

I think it may be best to take a cue from audio mixing. A record is mixed with flat responding speakers in a room that reduces sound reflections for a clean "dead" sound. Where do you listen to that record as a consumer? In your headphones, on your home stereo, on your iPhone, in your car, etc. A mixing engineer will check their mix for acceptable results on real-world stereos.

While I think we should maintain D65 and our normal Rec 709, P3, etc calibrations, I do think knowing that consumers are currently using warmer displays should at least inform the color pallets and techniques we use within a calibrated space. Does that mean some compromises? Yes, but isn't that the point of having a good color treatment? One that looks moderately good on a number of consumer displays?

 

[Subhadip Sen]

After quite a bit of testing and pondering on this matter, my conclusion is it does not really matter. The gap between 6500K to 7000K is so easily covered by chromatic adaptation in most situations that the difference is only noticeable when viewed side by side. Given the range of ambient lighting we use mobile devices in, I think the 7000K mark is justified. Indeed, the much larger issue is the vast range of ambient lighting people use these displays in. For a well calibrated display, under 3200K tungsten white is perceived far removed from D65. The next step would be to do accurate auto white balance using a more advanced ambient light sensor targeting a perceived 6500K. But till then, 7000K seems like a reasonable compromise.

 

[Christoffer Glans]

The diifference is huge, My graded skin tones goes magenta on colder displays, green and warm colors goes cold and blueish. In the end my grade looses a lot of the look that was intended.