As High Dynamic Range (HDR) and Wide Color Gamut (i.e.BT.2020) are increasingly mandated by major industry players like Netflix and Amazon, DOPs in the broadcast realm are under intense pressure to get it right during original image capture. We all know (or learned the hard way) that the amount of detail required to produce an optimal HDR master cannot be recreated or effectively added downstream.
Fundamentally, HDR is about managing a much wider range of light levels and color. Consumers may marvel at the blinking ‘4096I got 4K!’ message when firing up their Big Screens, but what they really see and appreciate is HDR’s more compelling pictures. Increased resolution plays a role in this of course, helping them to see the enhanced detail in the highlights and shadows, but resolution is not the whole picture, or even most of the picture, when it comes to producing the optimal viewer experience.
Currently, most DOPs and DITs utilize a three-screen strategy for capturing HDR. The typical workflow comprises a waveform, vector display, and dual HDR-SDR monitors, like the Sony BVM HX310 and Sony BVM E171, that can be toggled for side by side comparison on set. The approach, while workable, can be awkward and inconvenient for DOPs in actual practice.
The toggling of monitors is inefficient and not particularly informative, especially when viewed from afar across a busy set. Needless to say, DOPs have a lot to think about in the normal course of things, from indecisive directors to narcissistic actors, so the constant toggling of monitors comparing HD and SDR signals should not be, cannot be, a top priority.
The dual-monitor approach can also be highly subjective, as it is (as in all things when it comes to HDR viewing) dependent on the ambient viewing conditions. Stray light from a house ceiling fixture or an LED work light in video village can strike the face of the monitor, and easily skew a DOP’s assessment.
Converting to NITs
Ultimately, for capturing proper HDR, a single luminance-based reference monitor is preferable. The system, championed most notably by Tektronix, initially applies an HDR conversion LUT, converting light level voltages to nits so a DOP can more easily relate to what is displayed in the waveform. Depending on the graticule selected, DOPs can readily spot areas of potential clipping, and assess how elevated the highlights in a scene actually are: 1000 nits? 4000 nits?
Today, it is critical to know the highlight values in an expanded HDR range. DOPs can work from a waveform displaying in either nits or stops. Working with the log output from cameras like the ARRI Alexa or Panasonic VariCam, the luminance in stops may be easily adjusted, simply and predictably, up or down, via the lens iris. However, if the inputted signal is Hybrid Log Gamma (HLG) or Perceptual Quantizer (PQ), the display mode of the waveform should be set to nits.
A working HDR configuration like the Tektronix PRISM features a dual-display with the main waveform and two smaller HDR-SDR insert waveforms placed on the left. The configuration has the advantage of displaying both the HDR and SDR waveforms simultaneously, to confirm they are roughly the same shape and their respective highlights fall approximately in the same area. The DIT may choose to display a CIE chart in the left side configuration as well to ensure there are no out of gamut color areas.
The key, however, to this approach, and the most compelling part, is the single monitor displaying false color luminance values. If employing an integrated LED display like in the PRISM system, the monitor is typically placed on the right side of the dual setup. However, any suitable large-screen TV may also be used, even a consumer one, with a suitable HDMI input.
While the luminance-based system supports false color assignments for up to ten regions, most DOPs, for reasons of simplicity and easier visibility, opt to display fewer colors. Using the ARRI luminance values and colors as a reference, we typically assign blue or cyan to 18% grey (17-26 nits in the HDR-HLG environment). We target reference white at 180 nits, then assign yellow and orange bands to represent values from 180 to 203 nits, and red to scene areas above 203 nits to indicate values outside the SDR range. Once light levels and a camera stop are set, the harried DOP should only need to reference the false-color display to verify all is well – with no toggling or unnecessary intrigue!
Of course, given the maximum brightness of today’s displays at 1000 nits or more, there is no way, nor is it advisable, to drive an entire screen at such an elevated level. Indeed, the Automatic Brightness Limiter (ABL) in OLED TVs is intended to reduce the threat of burn-in and damage to the organic dyes from high-brightness images, by dimming the overall picture substantially. The ‘feature’, if that’s what it is, cannot be disabled.
Thus, the issue for DOPs is not just the range of highlight values but also what percentage of the screen is impacted by the elevated values. For this reason and others, it is imperative for DOPs to understand and utilize the luminance readout function in their system, indicating the percentage of the screen affected at any given brightness level.
Needless to say, most DOPs today are already comfortable working with false color as we’ve used the strategy for years to find and maintain critical focus. Using the same approach for capturing and managing HDR, a DOP now only really needs to know what is coming out of the camera - log format (logC, S-Log, etc.), which color space, and the HDR flavor targeted. The false-color display handles the rest, simply and elegantly, providing the necessary feedback we need easily and conveniently, without the hassle of subjectively comparing dual monitors and the associated guesswork that goes with it.
Increasingly, major players like Netflix and Amazon are demanding high-quality HDR masters, and DOPs must respond accordingly. With the advent of HLG, PQ, and Wide-Color-Gamut, professional shooters are obligated to accommodate much brighter consumer displays while still supporting traditional 8-bit SDR output. Whether viewers are watching our work in 4K HDR or HD HDR, the picture improvement afforded by High Dynamic Range is obvious and compelling.
HDR and brighter displays mean viewers can see greater detail in our pictures – which most of us would agree is a good thing. Unfortunately, it also means viewers can see more lens defects, especially chromatic aberrations. More than ever, DOPs shooting HDR must consider the quality of our optics!
With the advent of HDR, DOPs must think in terms of nits - or stops, when referring to camera log output. Traditional analysis tools and scopes that were developed decades ago express scene brightness in millivolts and percentages - hardly relatable to most DOPs today. Adapting a 709 workflow for HDR production, an appropriate conversion LUT must be applied in order to satisfactorily display the expanded HDR signal. Viewing the HDR output from a camera without applying the required conversion LUT produces a useless, tiny and crushed vector diagram. A vector diagram with the appropriate HDR to Rec.709 conversion LUT is shown here.
Figure 4. a) Tektronix PRISM; b) A practical HDR waveform displays in nits or stops depending on the incoming signal and graticule selected. c) A typical configuration places the waveform and CIE chart on the left, and the false-color luminance monitor on the right. Many DOPs will also want to display the HDR and SDR waveforms in reduced size to help ensure a reasonably comparable output across both platforms. On many sets, a larger external false color monitor may also be desirable.
The working DOP should only have to refer to the false-color monitor (a) to verify optimal HDR-SDR settings. Assigning false colors to the LogC output from an ARRI Alexa (b) may produce the following luminance values, which can be placed on a log scale in stops.
|Magenta:||0.0000 - 0.1002|
|Blue:||0.1002 - 0.1223|
|Green:||0.3806 - 0.4014|
|Pink:||0.4523 - 0.4731|
|Yellow:||0.9041 - 0.9290|
|Red:||0.9290 - 1.0000|
Some cameras, like the Canon C700 (a) and Panasonic VariCam LT (b), can already encode to and output HDR-HLG for live broadcast.
You might also like...
In a multi-disciplinary subject such as color space, it is hard to know where to start. John Watkinson argues that the starting point is less important than the destination.
Most people are aware that any color can be mixed from red, green and blue light, and we make color pictures out of red, green and blue images. The relationship between modern color imaging and the human visual system was…
In this thought-provoking missive, Gary Olson delivers his predictions and insights for IBC 2019.
A long chain of events is needed to see a color picture on a TV set. Only by considering every link in the chain can we strengthen any weak links.
Almost since photography has existed, people have pursued ways of modifying the picture after it’s been shot. The “dodge” and “burn” tools in Photoshop are widely understood as ways to make things brighter or darker, but it’s probably less widely…