Srgb Rec709 Viewing Conditions

“Correct” sRGB and Rec.709 Viewing

There is a lot of confusion surrounding the proper way to display Rec.709 and sRGB images and to clarify my own thoughts and hopefully help others I’ve outlined my understanding below. Be warned, I am no expert here and if you see errors please do inform me.

To view images strictly per specification (assuming you’re starting from open domain uniform tristimulus image) the short answer is:

• sRGB¹:
  1. Encode with the two part OETF (Section 5.2 of the specification)
  2. Display on a monitor with a 2.2 power function EOTF in a dim surround (Section 4 of the specification)
• Rec.709²:
  1. Encode with the two part OETF
  2. Display on a monitor with a 2.4 power function (BT.1886 ³) in a dark surround

That is the correct way to view images. However as most images are generally not viewed under ideal circumstances things get a little more complicated

Implicit vs Explicit Management

Most modern imaging software uses what could be called explicit colour management, every step of the workflow is (to quote Merriam-Webster) “fully revealed or expressed without vagueness”. Well in theory at least. Images are generally decoded into some working space and then pushed through various technical and creative transforms onto a display. This is often implemented via OpenColorIO ⁴ for example.

Before this a much simpler system was used, an image would be encoded with a Opto-Electronic Transfer Curve (OETF) such as Rec.709 and then displayed on a monitor with a power law Electro-Optical Transfer Curve (EOTF). This OETF -> EOTF system is not a no-op however and there is a slight mismatch between the functions that has very beneficial side effects. Some people have labelled this set up as implicit appearance management. It is worth noting that with a little work this system can be implemented in a explicit system.

Viewing Conditions

The effects of viewing conditions are many and complex but the essential point is that the lighting conditions of the environment you are seeing an image in have a large effect your perception of said image. It can be quite easily seen for example that the same monitor that appears bright and contrasty in a dim room looks dim and washed out in a bright one. The fact that we have a much brighter reference point (say the sky outside) and that there is a lot more light bouncing of the screen (viewing flare) causes the image to appear to have much lower contrast.

This is where the OETF/EOTF mismatch comes in, the slight difference between the two functions causes a small change in hue, contrast and chroma which neatly adapts the image for the viewing conditions.

The Rec.709 specification only defines an OETF and does not define a corresponding EOTF or reference display. Historically skilled users would adjust the power function of the monitor to make the image appear correct for the given viewing conditions. The sRGB specification however does very clearly define a EOTF and viewing conditions (one could even argue that an image cannot be a true sRGB image until it is displayed on a correctly calibrated monitor in a correctly illuminated room). The specification defines a power law of 2.2 and a “dim” viewing context which approximates the expected environment one would view sRGB content in.

Confusion

The sRGB specification was approved in 1999 and it wasn’t until 2011 that a canonical EOTF was decided on for Rec.709 (BT.1886 with a 2.4 power function). This is potentially where the confusion came from as in that intervening decade vast amounts of of video was encoded in Rec.709 but viewed under a sRGB compliant monitor. The sRGB specification in fact has a whole annex (Annex B) on compatibility with Rec.709 and states that:

“This sRGB standard provides a clear and well-defined reference display for ITU-R BT.709-3 for a dim viewing environment”

It becomes apparent that if one wishes to “correctly” view Rec.709 images one must take into account the current (or expected) viewing conditions and select the appropriate EOTF, as summarised below:

Surround	EOTF	Note
Dark	2.4	Official BT.1886
Dim	2.2	REC.709 pushed straight to an sRGB display. This is what a lot of people expect to see as it’s what a lot of basic video software does
Average	2.0	Rarely used
Bright	1.0	Rarely used

Whilst the only correct way to view sRGB is using a 2.2 EOTF as per the specification it does not seem unreasonable to suggest that if you know the expected viewing conditions of your image it may prudent to use a different power law to compensate.

Implementing An Implicit System Explicitly

As I mentioned earlier it is possible to simulate any of these EOTFs even if you are stuck with, for example, sRGB monitor. Lets say we want to view a Rec.709 signal under a BT.1886 EOTF as we are in a dark surround (say a grading suite) but we are stuck with an sRGB monitor and it’s 2.2 EOTF. We need to figure out what EOTF we need to apply to our Rec.709 image, that combined with our monitors 2.2 EOTF will equal 2.4 We end up with the following equation:

Therefore if we encode out Rec.709 image with a power function of 1.09 and push it to a sRGB monitor, it will roughly colourimetrically match displaying the Rec.709 image on a true BT.1886 monitor. A simple way to remember this is:

1. Whilst the sRGB specification is officially only available for purchase from the IEC online store it can be found elsewhere with some creative searching. The following link however, gives a good overview of the details https://www.color.org/srgb04.xalter . Wikipedia incorrectly (in my view) states that the EOTF should be the inverse of the to part function and references an article on the wonderful Colour Science website and whilst I’m wary to argue with someone far more informed than myself I think in this case they are wrong ↩

2. https://www.itu.int/rec/R-REC-BT.709-6-201506-I/en ↩

3. https://www.itu.int/rec/R-REC-BT.1886-0-201103-I/en ↩

4. https://opencolorio.org/ ↩