We’ve stumbled upon a truly ironic problem for the highly anticipated HDR (high dynamic range) format after watching a few Ultra HD Blu-ray movies on several 2016 4K HDR TVs we’ve reviewed recently. We were experimenting with introducing ambient lighting to see if we could better mask the backlight inconsistencies and local dimming issues in HDR mode on the LED LCD televisions we were testing, but invariably found ourselves asking, “Why does the HDR picture look so unimpressive? We can’t make out any dark detail!”
And then it hit us.
Contrary to popular belief, the purpose of HDR (high dynamic range) mastering is to expand the available luminance range rather than elevate the overall brightness of HDR videos. High-end display calibration software maker Light Illusion has published this exact quote from SMPTE’s ST.2084:2014 standard “High Dynamic Range Electro-Optical Transfer Function of Mastering Reference Displays” on its website:
This EOTF (ST2084) is intended to enable the creation of video images with an increased luminance range; not for creation of video images with overall higher luminance levels. For consistency of presentation across devices with different output brightness, average picture levels in content would likely remain similar to current luminance levels; i.e. mid-range scene exposures would produce currently expected luminance levels appropriate to video or cinema.
What this means is that for most scenes, 4K Blu-ray’s Average Picture Level (APL) in HDR should not deviate drastically from that of a regular 1080p Bluray in SDR (standard dynamic range). Indeed, that’s what we found in our own 4K Ultra HD Blu-ray vs Blu-ray comparisons.
Because we have the luxury of multiple displays (for review) available in our test room, we set up a Samsung KS9000 (hooked up to an OPPO BDP-103 BD deck) beside a Sony XD93 (fed from a Samsung UBD-K8500 Ultra HD Blu-ray player). Both TVs were calibrated for SDR and HDR (D65 white point and their respective EOTF), and periodically we would swap sources to make sure the difference we’re seeing was not caused by the displays (it wasn’t).
We put the UHD and normal Blu-ray discs of the same movie into the Samsung K8500 and the OPPO 103 respectively, and paused at the same frame on both players. Leaving the backlight and contrast untouched (at their maximum values) on the television showing 4K HDR, we then adjusted the backlight and contrast settings on the TV showing 1080p SDR to match the average brightness delivered by both displays for that particular frame. Finally, on the TV displaying SDR, we switched to another HDMI input connected to a signal generator to measure the peak white luminance produced by the adjusted backlight and contrast settings. Since we made sure the settings were shared across inputs, we could get a good idea of the average brightness of that specific frame in the Ultra HD Blu-ray version.
Let’s start with the results for Mad Max: Fury Road. Photos were taken with a locked-off camera, and you can click on each photo to see a higher-res version. The TV on the left is the Samsung UE55KS9000 showing the 1080p Blu-ray from the OPPO BDP-103 in SDR; whereas the one on the right is the Sony KD-65XD9305 displaying the Ultra HD Blu-ray from the Samsung BDP-K8500 in HDR. For now, please ignore any significant discrepancy in colours (which is mainly caused by different colour space remapping), and focus on the average brightness on each screen.
The following scene is from the beginning of Chapter 4. The Samsung 55KS9000 panel on the left was calibrated to 120 cd/m2 peak white with a [Backlight] setting of “4” and [Contrast] of “97“, and you can see that the Sony 65XD9305 on the right delivered a similar level of overall brightness despite having its backlight and contrast maxed in [HDR video] mode:
 |
| Mad Max: Fury Road – timecode 00:30:19 (Left: SDR Blu-ray; Right: HDR 4K Blu-ray) |
The next shot is from the start of Chapter 9. Again, you can see that the APLs from the Blu-ray and 4K Blu-ray discs were very similar – we didn’t have to change the 120-nit [Backlight] value on the Samsung KS9000 at all:
 |
| Mad Max: Fury Road – timecode 01:22:12 |
Moving on to The Martian, from this scene in Chapter 2, we can see that we’d need to up the light output on the left TV showing regular Blu-ray in order to match the average brightness:
 |
| The Martian – timecode 00:03:24 (Left: 1080p SDR Blu-ray; Right: 4K HDR Blu-ray) |
Increasing the Blu-ray-fed UE55KS9000’s backlight adjustment slider one by one, we eventually settled on a [Backlight] setting of “8” whose light output (peak white measurement 220 cd/m2) came closest to that on the Sony KD65XD9305 in HDR mode:
 |
| The Martian – timecode 00:03:24 |
 |
| The Martian – timecode 00:29:15 |
We’ve analysed three other 4K BD films we have at our disposal (The Lego Movie, Kingsman: The Secret Service and Sicario), and obtained a similar result – the APL of a HDR UHD Blu-ray corresponds to the APL of an SDR Blu-ray presented on a display calibrated to 120-220 cd/m2 peak white.
“This is all very fascinating,” we hear you ask, “but what’s your point?”
Here’s the rub: because [Backlight] and [Contrast] are already maxed out on HDR televisions during the playback of 4K Ultra HD Blu-rays, there’s no straightforward way to increase the display’s light output during the day to avoid the picture being drowned out by competing ambient light – a good analogy is how you need to boost your smartphone’s screen brightness under the sun for the text to remain readable. Even with only a moderate amount of ambient light in our test room, dark HDR scenes (such as the dust storm sequence in The Martian) became extremely difficult to watch – our constricted pupils just weren’t able to discern sufficient shadow detail in the presence of room light.
For regular Blu-rays, this isn’t a problem, because they don’t force displays into HDR mode with maximum [Backlight] and [Contrast] settings to achieve higher peak brightness. You can easily increase the [Backlight] or [OLED Light] on a TV playing Blu-ray (or any other non-HDR material for that matter) to obtain a brighter picture for daytime viewing – there’s ample headroom available since SDR content is typically mastered to a peak white of only 100 to 120 cd/m2.
To compound the issue, the ST.2084 HDR standard specifies an absolute luminance EOTF (electro-optical transfer function), which means every input signal value should be mapped directly to the same output luminance level on all HDR displays. SDR, on the other hand, uses a relative luminance system where each signal value is allowed to produce different luminance levels on different displays with different peak white brightness, making it easier to adapt to brighter viewing conditions.
To be fair, the HDR presentation from UHD BD is not without its advantages. Owing to the extended dynamic range, highlight details appeared brighter, more realistic and better-defined, as can be observed from this shot in Mad Max: Fury Road (timecode 00:59:55):
 |
| Left: 1080p Blu-ray in SDR; Right: 4K Blu-ray in HDR (click to enlarge) |
Even cranking the KS9000’s [Backlight] to its maximum value of “20” couldn’t extract as much brightness and clarity (specifically the sun) from the Blu-ray disc as we could from the 4K HDR Blu-ray:
 |
| Left: Blu-ray; Right: UHD Blu-ray (click to enlarge) |
It’s not only specular highlights that benefit from HDR. At the other end of the contrast ratio spectrum, dark sequences reveal more detail too. For example, there were more stars visible against the night sky in this The Martian scene (timecode 00:28:54) on the 4K Bluray version, even after we bumped [Backlight] to maximum for the 1080p Blu-ray:
 |
| Left: 1080p BD in SDR; Right: 4K BD in HDR (click to enlarge) |
Let’s recap the problem in a nutshell: the HDR presentation in current Ultra HD Blu-ray films is not bright enough for viewing under moderate/ strong ambient lighting, because the HDR metadata forces compatible TVs to be driven at their maximum backlight capacity, leaving no room for higher light output. Of course, there exist other avenues (such as gamma adjustment or dynamic contrast trickery) to brighten the on-screen image for daytime viewing, but these are generally insubstantial compared with the most effective method of raising backlight luminance.
This issue is not confined to 4K Blu-ray – any HDR content that pushes backlight/ contrast to the limit for peak brightness while maintaining an SDR-like APL is going to suffer from the same problem. One potential solution is for manufacturers to provide the option to disable HDR mode either from the display or the Ultra HD BD player (so you’ll get 3840×2160 in SDR instead), but that would mean missing out on the most attractive component in the next wave of UHD (ultra high-definition) development.
Whilst it took the arrival of Ultra HD Blu-ray (so we could compare against standard Blu-ray) for us to spot the problem, it seems the video industry was already aware of this potential banana skin. A white paper titled “HDR/WCG Systems Survey: Emerging UHDTV Systems” published by leading calibration software developer SpectraCal here (email registration required to download) contains this illuminating snippet:
For HDR, the industry is considering 5 nits (cd/m2) to be a desirable surround luminance level.
5 cd/m2 is very dark, roughly the amount of light generated by five lit candles; any switched-on room lamp is likely to exceed this figure. With its usage pretty much restricted to a dark room (not dissimilar to 3D if you think about it), 4K HDR Blu-ray is a niche format that may become even more niche, although in fairness most video enthusiasts would do their critical viewing in a dimly-lit environment anyway. It will be interesting to see if HDR broadcast can succeed in the average living room which is typically not light-controlled…
