Do you grade one extra version for web/computer displays?

[Joel Arvidsson]

I wonder how common it is to do a seperate grade for the web/computer displays. If you do I bet you have more or less made a preset.

 

[Gunleik Groven]

Now... I know I will get a bit of flack if I said
THAT is exactly why it would have been so nice with ACES.

Even though youtube is probably not highest on their list...

If you grade for wide gamut, you need to do some transforms, or accept blown highlights and crushed blacks.

Should not be too hard to make a LUT for it, really...

Cheers.

G

 

[Nick Shaw]

Baselight 4.4 has what FilmLight call their "generalised colour space" system. This is kind of their more flexible version of ACES. You can use it to implement ACES, but you can also just use it to transform between known colour spaces. You choose what colour space each source clip is, what colour space you are working in, and then specify different colour spaces (or display transforms) for different deliverables.

I need to do more testing with it, but from what I have seem so far, it is very powerful and flexible.

 

[Nick Shaw]

Should not be too hard to make a LUT for it, really...

I have actually done something similar for a client.

But the real issue is that even if you build a LUT to emulate in grading what image data looks like on a given display, the way video is displayed on the web, or in any computer media player is rarely that straightforward. The player does not simply extract RGB values of pixels from the file and put those same values into the frame buffer of the video card. The player pretty much always makes some assumptions (sometimes based on meta-data an the file, sometimes just on guesswork) about the the intent of the image data in the file and also about the colorimetry of the user's display (again sometimes based on the colour profile of the display, and sometimes based on assumed defaults). The player then makes compensations for these factors. The problem is that not all players make the same assumptions, or even the same compensations for the same assumptions.

So there is no one LUT which will emulate "the web"!

 

[Christoffer Glans]

I'm beginning to think about this too. The thing is that for computer displays it needs to be RGB, other then that I see over and over that computer displays have a much colder white balance.
The D65 isn't working if you do things for the web or computer displays, skintones that you graded warm will be neutral or worse, magenta-isch.

I have had so many clients telling me that they aren't happy with the end result, after thinking, "oh, once again, clients who doesn't understand a calibrated display", they keep coming back with: "it looks wrong on every computer they watched it on". So, professionally, we can stick the D65 up our... because out in the real world it looks wrong.

So I'm going to do this, I'm going to do a master grade, made for theatre and such, and then a webgrade made for promotional material or otherwise material only used for the web.
Only thing I will change is the white balance of the image from D65 to something like D75.

So there is no one LUT which will emulate "the web"!

Except a change in white balance, it's the single most destructive thing about displays other then calibrated.

 

[Marc Wielage]

So I'm going to do this, I'm going to do a master grade, made for theatre and such, and then a webgrade made for promotional material or otherwise material only used for the web. Only thing I will change is the white balance of the image from D65 to something like D75. Except a change in white balance, it's the single most destructive thing about displays other then calibrated.

Actually, what I think is the biggest difference with Rec709 colorspace and computer monitor color space is the gamma setting plus 8-bit color. You'll find you're going to chase your tail with trying to come up with a look that works on all computer monitors, because computer monitors in general don't adhere to any one standard, especially when you spread it across several brands and several different vintages of monitors.

I suspect you'd be a lot better off adjusting your client's computer displays to something that looks a little closer to Rec709 than inventing a new standard out of thin air.

You might also reflect that every single major Hollywood studio uploads dozens, if not hundreds, of videos every single day, promoting their new TV shows and film releases. I don't know of too many that spend much time worrying about anything beyond image size, compression artifacts, audio channels, frame rates, and bandwidth.

 

[Christoffer Glans]

You'll find you're going to chase your tail with trying to come up with a look that works on all computer monitors, because computer monitors in general don't adhere to any one standard, especially when you spread it across several brands and several different vintages of monitors.

The most common thing I encounter is a white balance difference, almost all displays, including TV's come out of the factory with the 7200 setting, computer screens as well. This means that if I grade something at D65 and make it neutral it will look cold on those displays. If more people watching television have "colder" screens, then why do we grade to a standard that isn't met by the people making the TV-screens? My grade is always wrong even though my calibration is right.

I suspect you'd be a lot better off adjusting your client's computer displays to something that looks a little closer to Rec709 than inventing a new standard out of thin air.

...and all the thousands and millions of people watching the stuff later as well?

 

[Gunleik Groven]

@ Christoffer

Could be the reasoning behind Vimeos Green Tint, discussed somewhere else here...

 

[Mark Britton]

One of the reasons to grade to a fixed color temperature and try to ignore what the TVs are being set to is to avoid having manufacturers/big box stores from setting the color temperature even "bluer". They think that their displays look better to a consumer if they blue them up a bit. If you reset your color temperature to match what they are currently doing, then grade from there...dollars to donuts, they're going to just go to 8000K or higher.

 

[Björn Benckert]

You should not generalize as "computer screen" is not a standard. Also different players such as Quicktime and VLC and similar are trying to compensate for different screen profiles so make sure that you know what you are looking at before you start to compensate for things that most likely is only on one screen or two but definitely not on all. In a perfect world your colors on your calibrated monitor should just land perfectly on a "computer screen" but so far the world is not perfect.

Quicktime complicate things. With their try to correct for different profiles. So when you view, render to and / or from quicktime files there is big chance that you clip and / or alter your color information. It's all depending how the quicktime is done, whats installed on the viewing machine, what color profile is used and also of course what monitor is used. But I would say that nowadays it's far more likely that quicktime is the cause of color problems for films than the screens or their settings. Basically if you are viewing tif files or jpg files in picture viewer or photo shop or any such programs things will most likely be quite all right... if you start to look at quicktimes, then not so much.

So my advice is to send a "known frame" trough your quicklime pipeline then it should come out exactly the same in the other end and also on that other computer.... you will see that usually it does not translate too good, things will start to alter, even though the known file as tif or jpg will show quite well on all screens.

A good test for such pipeline test is the attached frame. It's no rocket science. basically you should be able to see all 3 dark rings and all 3 almost white rings. And if you screen dump and color pick on your quicktime file holding this frame then the back color should be perfectly 000 and 255. If you got that then your flow is transparant on the computers and screens involved if not then you need to change things. In such situations staying away from quicktime as far and long as possible usually helps. Or you can learn about correction luts and for eachto and from quickitme step in your pipe line make sure that your colors stays the same.

 

[Marc Wielage]

The most common thing I encounter is a white balance difference, almost all displays, including TV's come out of the factory with the 7200 setting, computer screens as well. This means that if I grade something at D65 and make it neutral it will look cold on those displays. If more people watching television have "colder" screens, then why do we grade to a standard that isn't met by the people making the TV-screens? My grade is always wrong even though my calibration is right.

If the calibration is right, then they'd be at D65. Have you read the SMPTE RP-166 spec?

One hopes that if the original color-correction monitor is calibrated to known standards, that when you look at it on a display that's "off," the overall color will all be in the same direction. The reality is that cheap computer displays are notoriously non-linear, especially when viewed off-angle, so it's kind of a crap shoot. I'm not sure it's worth worrying about, because no matter what you do, if you pull out a dozen iPads and tablet PCs, and a dozen laptop computers, every one of them is going to look different and none of them are going to match each other. If you can calibrate them somewhat, they'll at least be vaguely in the ballpark.

 

[Christoffer Glans]

If the calibration is right, then they'd be at D65. Have you read the SMPTE RP-166 spec?.

They are never right, just visit a friend, or your family or go into a store or watch through someones window (not recommending that) and you will see that at least 80% of the television sets out there aren't even close to the D65 standard in terms of white balance. And almost all of them are in a colder region, I have never seen something warmer then D65.

 

[Marc Wielage]

They are never right, just visit a friend, or your family or go into a store or watch through someones window (not recommending that) and you will see that at least 80% of the television sets out there aren't even close to the D65 standard in terms of white balance.

Did you read the part of my message that started with IF THE CALIBRATION IS RIGHT?

Trying to judge color on an uncalibrated monitor is absolutely suicidal, in my opinion.

 

[Christoffer Glans]

Did you read the part of my message that started with IF THE CALIBRATION IS RIGHT?

Trying to judge color on an uncalibrated monitor is absolutely suicidal, in my opinion.

I'm not speaking of judging the color or work with the grade, I'm talking about when it's delivered, correctly graded based on D65 standards. When it is broadcasted to televisions in people's homes, or watched streamed online etc. it will never look right because the television sets and computers out there are never calibrated to D65. They are sold in stores with a much higher white balance "because it looks more pleasent" in stores. I've never encountered an uncalibrated TV that is even close to D65 and most people if not ALL people don't calibrate their sets.
Even if they choose a warmer setting on their TV it's almost always too warm, far lower then D65.

So what I mean is that if you grade for theatre release then absolutely you have to grade according to the standards, but for television and computers/web I have come to realize that it needs to set to a much warmer look to be "correct". Having a node at the end of my grade that warms everything up might do the trick, but calibrating the white balance to a colder state on the screen and regrade gives the best image.

The truth is that there are no calibrated screens out there for regular people and those regular people is the ones who really watch our stuff, not the minority of enthusiasts that really calibrate their home cinema.

 

[Jaakko Rinne]

When going from REC.709 to web I just make sure black level is what it should be and check the uploaded version with a display that's calibrated to sRGB. I don't think you can (or should) do much more.

While I agree with Christoffer that home TVs and computer displays tend to be on the blue side by default, I wouldn't go as far as changing my monitoring white balance. I do tend to grade stuff that's just for broadcast a tiny bit warmer than I do for something that's going out to theaters or festivals.

 

[Danai Chutinaton]

I'm not speaking of judging the color or work with the grade, I'm talking about when it's delivered, correctly graded based on D65 standards. When it is broadcasted to televisions in people's homes, or watched streamed online etc. it will never look right because the television sets and computers out there are never calibrated to D65. They are sold in stores with a much higher white balance "because it looks more pleasent" in stores. I've never encountered an uncalibrated TV that is even close to D65 and most people if not ALL people don't calibrate their sets.
Even if they choose a warmer setting on their TV it's almost always too warm, far lower then D65.

So what I mean is that if you grade for theatre release then absolutely you have to grade according to the standards, but for television and computers/web I have come to realize that it needs to set to a much warmer look to be "correct". Having a node at the end of my grade that warms everything up might do the trick, but calibrating the white balance to a colder state on the screen and regrade gives the best image.

The truth is that there are no calibrated screens out there for regular people and those regular people is the ones who really watch our stuff, not the minority of enthusiasts that really calibrate their home cinema.

Christoffer,

I totally understand your point. You'r right that TV's and computer monitors in general are not calibrated and in general are too blue. However at the other end of the spectrum, there are monitors that are also too yellow, especially older ones that have a back light that's very dim. So I would like you to consider this alternative view of the uncalibrated TV/monitor problem. All the material that everyone is watching is graded to the same standard. Which means if you graded to the same standard as everyone else it will look "correct" to the viewer. Or you could also say it looks the "same wrong" as all the other material they are seeing. Our brains are able to compensate for an incorrect white balance of the TV screen. It takes someone with above average color "knowledge" to detect the blue, yellow, red, or green tint in their TV's or computer monitors.

The standard for computer monitors is generally accepted to be D65 with a gamma of 2.2. Web browsers generally expect RGB images to be in the sRGB color space which has the same color gamut as REC709.

Its often said that REC709 is supposed to have a gamma of 2.4. But my take on the 2.2 vs 2.4 gamma debate is to calibrate a monitor to a gamma of 2.4 when viewed in a dark grading suite and 2.2 when viewed in a typical office/home work space with lots of ambient light. This is my long winded way of explaining that REC709 and sRGB are basically the same thing. So I don't really see the need to do anything different from grading for regular REC709 output.

Now the quicktime gamma shift issue is another whole can of worms. Thats a problem you'll have to solve in the encoding/compression step. The way different video players and web browsers display the video is also another huge problem, but this is not something you can control. All you cna really do is test it with different players and recommend people view it in players you find acceptable.

 

[Björn Benckert]

They are never right, just visit a friend, or your family or go into a store or watch through someones window (not recommending that) and you will see that at least 80% of the television sets out there aren't even close to the D65 standard in terms of white balance. And almost all of them are in a colder region, I have never seen something warmer then D65.

Christoffer, Are you using "Adtox" to get your stuff to the stations? If so you need to know that it's not a straight process, again to make sure your pipeline is straight push this jpg trough it...

As you jump between quicktime renders (creating adtoox files is one of these steps) you are likly to push your histogram around, when that happens it very much can look like your white balance is off.

However I do agree that not two screens are the same, not even two freshly calibrated screens of the same professional kind, but to me those small differences are for the gay people, as I do not believe blockbuster hit or miss are made in the nuances of white balance.

 

[Michael Dalton]

does anyone have a good tested pipeline for web delivery? What are software are people using to convert and their settings? Do you start with Proress HQ/DNXHD175 and then convert?

 

[Christoffer Glans]

However at the other end of the spectrum, there are monitors that are also too yellow, especially older ones that have a back light that's very dim. So I would like you to consider this alternative view of the uncalibrated TV/monitor problem.

Yeah, I understand that point, but at least in Sweden, the adaptation to new LCD/Plasmas has been greater then in many other countries. So here there are, at least to my knowledge, more cool/bluish displays around. Even though I know computer displays should be D65 at 2.2, the truth is it isn't that common. It's always more blue in them and my grades almost always gets a very bad looking pink/magenta color in skin tones when shown on those screens. It goes from a nice warm skin to skin full of blemishes and so on.

Christoffer, Are you using "Adtox" to get your stuff to the stations?

Depends on the project, but most of the time we don't go through adtox, we do a direct delivery to the stations broadcast system. Other times I just deliver a RGB master file to the customer. But for all problems you can have with colors shifting among displays, this is hands down the worst problem I have with grading. A standard that works for mostly 10-20% of the displays out there in public.

There have been times I really did create my own standard and to be honest it looked good everywhere.