Anyone mastering images for later viewing on any display must use an accurately calibrated grading display. There is no alternative to this, regardless of the expected viewing environment or relative accuracy of the final displays the material will be viewed on!
This includes material expected to be viewed in a theatre; on home TVs; on the web; on mobile phones & tablets - ANYWHERE!
All Image Mastering Must Be Performed on Calibrated Displays
We're not grading for customers with high-end monitors, so we have to adjust the grading for the consumer's TV, to meet them half way.
The above statement, and others very similar, are fairly common regarding calibration of the display when performing grading or mastering with the knowledge that the final viewing will not be on professionally calibrated displays.
The problem is this approach misses the simple fact that even cheap, uncalibrated, displays are manufactured to attempt to adhere to industry standards, even if they cannot actually meet them. They all aim for the same colour target.
If you deliberately skew the grading in any way, to counter for an inaccurately calibrated grading display or to make the images look better on a specific alternate display, the result will never be accurate except on the display you have skewed the grade to accommodate.
The above images are just one example of possible variations with inaccurate displays. It shows colour errors, caused by gamut variations or inaccurate white balance. Gamma and luminance errors are ignored, as are saturation variations, which all add to the possible errors.
If the grade is skewed to accommodate the display with the magenta error, all other displays will show a greener image.
If the grade is skewed to accommodate the display with the blue error, all other displays will show a yellower image.
And so on...
Only by grading on the central, accurately calibrated display, are errors minimised on the surrounding inaccurate displays.
The same is true with inaccurate gamma/luminance.
For example, if your grading display is crushing blacks it will cause you, the colourist, to lift the blacks to overcome the crushing. So when the graded images are viewed on a correctly calibrated display the result will show the inaccurate grading, while things will get much worse if the final viewing display has lifted blacks, as the images will show grey blacks with washed-out detail in the shadows.
The issue with crushed blacks is an inherent problem with (W)OLEDs being used for grading when the black is not set to a sensible level. All (W)OLEDs have potential problems coming out of black, as they have a hysteresis issue that causes black to stick before detail can be seen. To overcome this the black level needs to be lifted.
Additionally, as the proportion of home TVs that are WOLED based is approximately 2% of the market, grading for (W)OLED viewing makes little sense.
The amount of lift required depends on final delivery being SDR or HDR.
For SDR black should be around 0.03 nits.
For HDR the aim is to prevent the black hysteresis sticking issue, nominally 0.005 nits.
Grade on Calibrated Displays
From the above image you can see that by grading on a well calibrated display the worse any final error can ever be is just that of the un-calibrated viewing display.
Equally importantly, the viewer will be comparing your graded images with material they are used to seeing on their specific display, so will be normalising the look of any images, as the comparison will be with different footage also graded on a calibrated mastering display, making the comparison relative to their normal viewing experience.
If you have graded on a poorly calibrated display, or skewed the grad to accommodate a specific alternate display, their relative judgement on their specific inaccurate display will define your images being yet more inferior, as the colour errors will be enhanced.
A relative comparison is just as valid as a definitive comparison, as it will always show the original material to be inferior.
Wide Gamut vs. Standard Gamut (Or why Apple cannot be relied on)
An issue that is relatively new is Wide Gamut displays vs. Standard Gamut, with Wide Gamut being a relatively new concept for home viewing.
Realistically, for TV/Broadcast applications, wide gamut is intrinsically linked to HDR and Rec2020, and so should not exist outside HDR imagery. Unfortunately, display devices that are not home TVs often abuse the use of wide gamut colourimetry, and can incorrectly display standard gamut imagery on a wide gamut display, without correctly restricting the display's gamut.
This gets yet worse as some manufacturers, such as Apple, have bastardised what are defined standards, making it very difficult to know what you are really looking at when not using a well calibrated mastering display. For example, Apple have their own version of P3. P3 is a standardised colour space that is well defined within the film & TV industry, but Apple have what they call displayP3 colour space, which incorrectly combines the P3 gamut with the encoding sRGB compound gamma - not even the correct display defined sRGB gamma, which is a power law 2.2 gamma.
This obviously causes major issues with broadcast material viewed on Apple wide gamut displays... and means they must NEVER be used to skew any grade to accommodate their bastardised colour space.
While Apple's displayP3 colour space is highlighted here, any non-broadcast device - be it an uncalibrated PC display, mobile phone or tablet, or laptop screen - should never be used as a reference for any grading work - NEVER.
Continue Grade on Calibrated Displays
The above is why a colourist must never deliberately alter a grade for viewing on an uncalibrated/inaccurate display, no matter what any unknowledgeable client/producer may ask for. Altering a grade so it becomes inaccurate on a calibrated display will show as being yet more inaccurate on any other uncalibrated/inaccurate display, as the relative comparison to other accurately graded material will make the deliberately altered grade appear even more inaccurate...