HDR: Analyzing PQ and HLG


At TM Broadcast we have already published a number of articles about HDR (High Dynamic Range), although all of them always within the context of UHD (Ultra High Definition). Today we would like to go a bit deeper on HDR and, in particular, with regards to the different gamma curves existing at present: PQ and HLG.

By Yeray Alfageme


As TV has been progressing, large changes in both standards and technologies have dramatically improved image quality. After the momentous introduction of color, a series of developments followed, such as the switch from analogue to digital, from 4:3 to 16:9, and from SD to HD. With each further step, broadcasters have had no alternative than cope with challenges in workflows and compatibility issues with existing equipment. As modern broadcasters have been switching from HD to UHD and from SDR to HDR, this change has become even more significant.

As it happens with any new development, several approaches have been adopted for HDR production. Two main formats exist: PQ (Perceptual Quantization) and HLG (Hybrid Log Gamma). Being very differing concepts, each of them offers unique advantages for different applications.

PQ is capable of representing a higher luminosity range for a given bit depth -10 bits, for instance- by using a non-linear transfer function designed to adapt to the human sight system to perfection. The goal is to reach a system with no harmful visual effects, such as quantization errors. There are several PQ coding variants such as HDR10 –which uses 10 bits and static metadata- and its elder sibling, HDR10+, -including dynamic metadata- which improves representation of hue in each image.

The main manufactures supporting HDR10 are Panasonic, Hisense and Sansumg. On the other side we have Dolby Vision, a proprietary implementation supporting 10-bit and 12-bit representations in addition to static and dynamic metadata. This format is licensed by most manufacturers, both for TV sets and Blu-Ray devices, and many producers have adopted it as a cinema production standard.

On the other hand, HLG is based –at least in part- on the transfer curve of the older CRT monitors. As said curve is not optimized for the human sight system and some coding areas are not considered -especially the darker parts in the image- a lower brightness peak is available. Therefore, HLG cannot achieve the same dynamic range as PQ, although it is easier to combine by the existing TV sets at a standard luminosity level.

HLG also attempts to provide backward compatibility with HD-SDR and 4K-SDR. One of its biggest advantages is that no metadata accompanying images are required for representing the image properly. Carrying metadata through traditional linear broadcasting channels –either terrestrial or satellite- or through OTT platforms is not an easy task, not to mention the issues associated to changing content, as it may be the case when switching to commercials, overlapping graphics and similar items.

Another major difference between PQ and HLG is the fact that PQ is referenced to the screen, while HLG is referenced to the image, but what in the world does this mean?


The difference between a screen-referenced transfer function and an image-referenced transfer function

Under SDR (709) the power signal that used to be carried (or still being carried for the few remaining SD broadcasts) was referenced to the brightness of the image, which results in an image-referenced transfer function. Under PQ, however, the power signal carrying the image’s luminosity information is referenced based on the screen in which the image is to be displayed and this obviously results in a screen-referenced transfer function. This is because the PQ gamma curve is referenced to the luminosity being perceived by viewers and it is optimized for this purpose. The human eye is capable of adapting to the image’s luminosity. We do not see in an identical way in daytime or during the night and the same occurs with the PQ gamma curve transfer function as well.

HLG inherits the image-referenced transfer function as the SDR (709) standard already did, this being one of the reasons for its backward compatibility. But this is not optimal for the human sight system, although it is agnostic in regard to the setup of levels each viewer may have at their own home set, which allows to display a HDR image that is valid in any environment.

Graphic 1 is a representation –in a theoretical fashion- of the concept differences between both standards -PQ and HLG-, being the former screen-referenced and the latter -as it was the case with SDR (709)- image-referenced.



Graphic 1


Some clarifications on graphic 1:
• OOTF (Optical-Optical Transfer Function): transformation that takes place from the camera to the screen.
• OETF (Opto-Electronic Transfer Function: transformation that takes place between the camera and the signal.
• EOTF (Electro-Optical Transfer Function): transformation that takes place between the signal and the screen.


In a PQ environment, if content has been mastered in a 2,000-nit monitor and finally viewed on a 500-nit OLED TV set, said content must be adapted from the source 2,000 nits to the target 500 nits, which is awkward and costly in terms of processing. This is one of the main reasons why HLG has become so popular in live production environments.

The other reason being that under HLG there is no need to generate two simultaneous -one SDR and one HDR- signals regardless of definition, of course, but just one HDR-HLG is also valid as SDR signal. The only thing that needs to be done is ignore luminosity ranges outside the 709 standard and voila, we will have our SDR image.

Format conversions

One of the issues having the biggest impact on the final look of an image is the conversions undergone by the same from production to display. Most current production environments use 8-bit or 10-bit based systems. In a 10-bit environment, coding errors in the image remain hidden behind the electronic noise generated by the cameras, so in practice it is regarded that 10-bit production environments are adequate for HDR images. Theory says that 12 bits are required for generating an HDR image as set forth by the BT.2020 standard, but hardly any production facility or mobile unit are conceived for this.

As it happens between HD and SD -much more evident in this instance- using gamma conversions between PQ and HLG and vice versa is not recommended, as image artifacts could be inadvertently created. Just a couple of conversions are enough to create obvious coding errors, especially in the darker and lighter areas of the image. This causes us to choose our production standard from the outset and keep it throughout our entire production chain in order to avoid problems.

Color implications are now left aside, as an UHD image with WCG (Wide Color Gamut) forces a remapping of colors for conversion to 709, but we will deal with this some other time.



The PQ standard is much better in regard to image quality and luminosity range that can be displayed. However, if the whole broadcast chain, from camera to final display is not properly monitored, maintaining a proper representation of the image becomes complicated and costly. Its main representative -Dolby Vision- is a proprietary standard, which requires payment of royalties, which does not help to promote a widespread use.

In HLG, the gamma curve is based on the older SDR (709) standard, extending the same in order to be able to display broader luminance ranges. This is not optimal for the human sight system but offers backward compatibility and simplicity as far as production is concerned, especially in live events, which has boosted its adoption by broadcasters and producers all around the globe.
To sum up, PQ is ideal for environments such as cinema or fiction production such as series, while HLG features a much simpler model for adoption in live productions and for general broadcasters.

Carlyle acquires Liv
Quicklink Studio all