There is a lot of talk about the use of high dynamic range for video. Currently most high-definition video output can only reproduce around 220 different levels of brightness. This is often confused with the range of illumination that a screen can reproduce and the need for brighter displays. The real challenge is to compress the vast range of brightness of a real scene to the limited range of luminance of most displays.
High-definition television inherited a range of limitations. For one thing it is generally limited to 8 bits to represent luminance or brightness, giving only 256 possible values. Of these, the first 15 and the last 21 values are reserved for historical reasons of backward compatibility. So the range from black to white typically runs from 16 to 235, allowing 220 discrete levels. This can result in quantisation or contouring effects, which may be exaggerated by digital compression and the characteristics of displays.
Using more bits allows greater precision, which can result in better image fidelity. Studio systems typically use 10 bits, providing 1024 discrete values, but these are typically lost when the output is distributed.
The real problem is that there is often a large difference in brightness between the darkest and the lightest areas of a scene. If this is represented using only 220 levels from black to white there are only relatively few levels to represent shades of grey in the shadows and relatively few to represent the whites in the highlights.
However, the human visual system does not respond linearly to luminance. It generally has to deal with many orders of magnitude in the difference in brightness between, say, a dark room and a sunny day.
There is a similar problem in digital photography, where there may be a large difference in brightness between areas of shadow and the lightest parts of a scene.
A digital still camera sensor may record 14-bit images in its native format, which can represent over 16,000 different luminance levels. Various techniques can be used to process these as 16-bit images and map the tones to 256 output levels as an 8-bit representation. This can provide enhanced reproduction of the range of tones in the image, allowing us to see more detail in the shadows and highlights.
A similar approach can be employed with video and is generally applied during post production. However, it is not generally applied to live television.
There could be some benefit in transmitting the 10-bit image that is often maintained during production. However, this does not in itself provide higher dynamic range. That requires mapping the values in a way that can better represent both the shadow and highlight details. The challenge is to do this in a way that can maintain compatibility with the existing receivers.
Online services have the theoretical advantage of being able to identify and define the capabilities and characteristics of the display to deliver different encodings where they can be supported. Netflix and Amazon are already embracing HDR distribution.
There is currently some confusion between the ability to represent a wider range of luminance on an existing display and the ability of a display to emit a wider range of luminance levels.
Many conventional flat screens are not very good at reproducing deep blacks and are limited in the level of luminance they can achieve for the brightest whites.
Luminance is often measured in Nits, or candela per square metre. A typical living room screen may deliver a few hundred nits. Some displays are capable of a few thousand nits, but they generally consume a lot of power.
However, the need for displays that can produce ever-brighter whites may be misleading. No display can actually represent the brightness of the sun. So what we are seeing is simply a greater contrast in brightness between the dark and light areas of an image. It may look impressive but it actually be uncomfortable to view, which rather defeats the purpose.
Screen images do not necessarily need to represent the brightness of the source scene. It is not ultimately possible.
Consider a photographic print that has been processed to show detail in both the shadows and the highlights. It is evident that the blacks are limited by the ability of pigment on the paper to absorb light and that whites can appear no brighter than the reflectance of the paper as it is illuminated.
Yet we can perceive a photographic image as a realistic representation of a scene, provided the apparent detail in the shadows and highlights is consistent with our expectations of reality.
Most of the images that we encounter in print and on screens have been output in 8-bit format. In many cases they originated as images with more bits per channel.
So high dynamic range images are not necessarily all about delivering more bits or having brighter displays. But television production workflows may need to change significantly if we want to see a dramatic improvement in the apparent realism of pictures on our screens.