Relationship between sRGB, BT.1886 and Rec.709

All,

Because it came up at the last Output Transforms working group meeting, I’ve created the attached diagram to illustrate the relationship between sRGB, Rec.709 and BT.1886.

If a computer monitor were calibrated with an EOTF representing a pure power law function (e.g. gamma 2.2), and the sRGB encoded images (sR’G’B’ code values) were displayed on it, the measured colorimetry from that display would not match either of the displays shown in this diagram.

Hi Alex,

This ignores viewing conditions.
sRGB was created to give the same appearance as Rec.709 for same code values, but in a different viewing condition (lit office instead of darker living room). Charles Poynton can give you the full story AFAIK.
That is, a Rec.709 video played on an sRGB display (unconverted) in the office “will appear similar” to same video played in the living room on a BT.1886 display.
This means that there shall be no value conversion between Rec.709 values and sRGB values.
Also, as the viewing conditions differ a direct comparison of XYZ is not meaningful. I think you have to convert the measured XYZ values to CIECAM for comparison. CIECAM takes viewing conditions into account.
And similarly, converting between sRGB and BT.1886 would need to include a CAM model.

Tricky!

2 Likes

@LarsB I ignored surround and appearance intentionally, but I figured someone would call me out on that.

My intention was to show how one could re-encode colorimetry produced by a set of Rec.709 code values displayed on a BT.1886 display to sRGB code values that when displayed on a computer monitor calibrated with the sRGB EOTF will produce the same colorimetry. In the situation that both of these displays are in the same environment (e.g. the same room) and calibrated to the same peak luminance, color appearance should match. I fully recognize that’s not consistent with the viewing environment specifications for both encodings.

I was expecting the more controversial topic to the use of the inverse sRGB encoding function as an EOTF on the computer monitor. Some feel very strongly the sRGB EOTF should be 2.2 power function.

Is this not what is written in the specification?



1 Like