HDR in Pixotope

Introduction

High-dynamic-range video (HDR video) describes video having a dynamic range and gamut greater than that of standard-dynamic-range video (SDR video).

The process of HDR video involves capture, production, content/encoding and display. HDR capture and displays are capable of much brighter whites and an overall much higher contrast ratio. To accommodate this, HDR encoding standards allow for a higher maximum luminance and use at least a 10-bit dynamic range in order to maintain precision across this extended range.

While technically "HDR" refers strictly to the ratio between the maximum and minimum luminance, the term "HDR video" is commonly understood to imply wide color gamut (WCG) as well.

The goal

The goal of HDR (high-dynamic-range) and WCG (wide color gamut) is to:

• increase image quality

• achieve more natural-looking images

• approach towards photorealism

Additionally, it allows for:

• linear compositing, which greatly enhances realism when combining graphics and video

What HDR requires

• An HDR-capable camera

  • to capture many more f-stops in a single exposure than a traditional video camera (greater dynamic range)

• An HDR-ready transportation medium

  • to ensure optimal bandwidth and compatibility without losing extra information captured by HDR cameras

  • to encode the video into a format that can transport and store the HDR values, such as various LOG formats (sLog, clog, vlog), HLG, and PQ

• An HDR-ready color management system

  • to ensure that the captured light information is interpreted, processed, and encoded in a predictable and correct way

• An HDR display

  • to show brighter highlights without compromising shadows and details

  • to accurately judge what you see

Reference white and brightness levels

When working with HDR it is important to understand that "white" is not the brightest value in the signal. The reference levels to remember are:

• SDR reference white: 100 nits

• HLG reference white (also called diffuse white, graphics white or "paper white"): 203 nits, which corresponds to a 75% signal level on a waveform monitor

Values above reference white are "super bright" values, reserved for highlights, speculars and light sources - things that in the real world are brighter than a white piece of paper. The thing you want to look like white (a logo, a graphic background) should sit at reference white, not at the maximum signal value. A graphic created for SDR with large areas at 100% white will appear uncomfortably bright when reused in an HDR output, because its white now lands in the super bright range.

PQ vs HLG

For broadcast, two HDR transfer functions are relevant: PQ (Perceptual Quantizer, SMPTE ST 2084) and HLG (Hybrid Log-Gamma). They are similar in many ways - both use the Rec.2020 gamut and comparable curves for compressing a wide brightness range into a 0–1 encoding - but they differ in one fundamental aspect:

• PQ is an absolute mapping. A given code value always corresponds to a specific light output, regardless of display: a value of 1.0 always means 10,000 nits, 0.5 always means ~100 nits. This makes PQ ideal when you want to describe exactly how much light a display should emit - for example on an LED wall

• HLG is a relative mapping. The signal does not know the display's maximum brightness; the maximum code value maps to whatever peak the display (or LED processor) is configured for. HLG is designed and balanced around a nominal 1000 nits display, but adapts to the capability of the screen

• HLG is backwards compatible with SDR. The lower half of the HLG curve closely resembles a Rec.709/BT.1886 gamma curve, with only the highlights compressed into the top - so an HLG signal shown directly on an SDR monitor looks a little flat, but entirely watchable. This is one of the reasons HLG is preferred for live broadcast

HDR in Pixotope

For most broadcast productions, 2 flavors of HDR are most common:

1. Camera native LOG formats that capture HDR dynamic range and wide gamut.

   1. Example formats would be:

      • For Sony: S-Log3 S-Gamut3.Cine

      • For Arri: LogC3

      • For Canon: CanonLog3 CinemaGamut D55

      • For Panasonic: V-log - V-Gamut

      • For RED: Log3G10 REDWideGamutRGB

   2. Hybrid Log Gamma (HLG)

      • HLG exists in several different flavors, especially on the camera side. Optimally, you would have a colorspace conversion specifically for how the footage was encoded, but in practice, this is not readily available, so using a single standard HLG conversion is sufficient for now

As the workflows for LOG formats are better understood and standardized than the various HLG formats, we highly recommend shooting and outputting the graphics in LOG format, with conversion to HLG/PQ downstream if possible.

Regardless, the workflow in Pixotope is the same:

1. Use the same color conversion on inputs and outputs, unless you specifically need to create an SDR output for preview.

   • Ie. Your camera feed is encoded in sLOG, so your main output should also be configured as sLOG to ensure that all colors in the video plate are unchanged, and that the graphics are tonemapped in the same manner. However, the sLog will look very flat on a Rec. 709 monitor, so if you can not convert it in the monitor or using an external conversion device, Pixotope can convert the signal to Rec709 by choosing that as your output space

   • Similarly, on your PC monitor, the image will look “wrong” (typically very low contrast) if you do not use a preview display conversion in the Editor. If you have an HDR monitor, choose the appropriate HDR lookup (ie, PQ/HLG), or if the monitor does not support HDR, choose an sRGB preview, which will convert the HDR signal to SDR for preview

   • For HLG input/outputs, use the “Rec.2100-HLG ”

2. When outputting HDR on SDI, make sure the SDI signal range is set appropriately (see section on SDI signal range).

3. Turn off “Film” tonemapper

   • The LOG formats will all heavily tonemap the content

   • HLG has a less aggressive tonemapping, but it is highly recommended that you do not use the “Film” tonemapper, and rather rely on the built-in curve and light accordingly

4. Preferably, work with a scope to ensure that the white levels of your graphics are set appropriately.

   • This is particularly important when working with motion graphics elements, such as Logos, where specific colors and values are often desirable

   • As an example, in HLG, a 75% marker on a waveform monitor references HDR Reference White (203 nits), which is also diffuse white and Graphics White. That means that values above 75% will be super bright values that are brighter than the reference white

   • In general, it is highly recommended that users familiarize themselves with the “Guidance for operational practices in HDR television production” documentation (ITU-R BT.2408-7): https://www.itu.int/pub/R-REP-BT.2408-7-2023

5. When using inverse ODTs for converting input video into linear space, you can linearize an SDR signal and reconstruct plausible HDR-like values from it. It will be a relatively crude approximation, but it can give some of the nice effects of HDR when doing compositing, even for output

   • Be aware that linearizing an SDR signal does not make it HDR. The dynamic range that was never captured cannot be recovered - the conversion only maps the existing SDR values into the wider container. Converting a Rec.709 source to log or HLG does not turn it into an HDR source