A/V Veteran

On September 19, 2015, SMPTE (Society of Motion Picture and Television Engineers) published a Study Group Report titled “High Dynamic Range (HDR) Imaging Ecosystem.” If you’re in the mood for some light reading (insert your own appropriate emoticon here!) you can download it at smpte.org, though it took me several tries before my MacBook Pro could get through without being blocked.

The emphasis is on the production side, as you might expect from an organization with its emphasis there, but display capabilities do turn up at key points. The text is heavy with industry technobabble, not all of which I pretend to follow (not being involved in the content production side of the fence), but the bottom line appears to be this: The industry is still engaged in a protracted study of HDR, with no current standards for the production and display of content in that format.

It’s an extremely complex process involving a number of factors, including...

• Content origination (which might well compromise the creative process—more on that later.

• HDR mastering.

• Possible metadata to make HDR-mastered content compatible with SDR (standard dynamic range) displays and with displays using a different HDR format (again, likely involving compromises, if possible at all).

• Lighting conditions in the room where the mastering-to-video process takes place.

• What HDR level is used in the “grading” process during video production and how this is affected by the playback conditions, the lighting conditions in the consumer’s room, and aspects of human visual perception.

Got all that? We’ll also have to account for how HDR interacts with color; a new term called “color volume” turns up here and there in the report; think of this as the color gamut in three dimensions (the third being luminance). All of this is enough to make an HDR developer consider transferring into a simpler field, like computer modeling for climate change.

Let me repeat the date of that report: September 19, 2015. As of that date (essentially yesterday in a standard-setting timeline), several HDR displays were either already on the market or in production awaiting their launch. But there are several formats vying to dominate the HDR Ecosystem (SMPTE seems to like the term Ecosystem—it’s so…eco). The most widely discussed are DolbyVision and HDR10. None of the available displays, to our knowledge, is compatible with more than one HDR format.

Questions abound.

What will happen, for example, if you feed a source mastered in DolbyVision into set designed to display HDR10? Will the set display it properly? (Unlikely.) Will it instead simply display the incompatible source in SDR? Or will it revert to a blank screen? Until we have sufficient HDR program material to check for this, we simply don’t know.

Mulling Metadata
The holy grail here, of course, is coding HDR program material with appropriate metadata. The source would handshake with the set to determine its HDR capabilities, then use the metadata to convert the material to an HDR format compatible with the set. But here are two comments on such metadata, quoted directly from that SMPTE report:

Throughout this report various references are made to metadata…. Content dependent metadata is intended to solve a problem of interoperation of HDR signals with SDR output displays or with HDR displays having less peak white capability than the source [and, we might add, with less color capability than the source calls for—or for displays compatible only with a different HDR format. —TJN]

As noted in many sections of this report, it’s uncertain whether [for] future HDR content delivery/transmission systems, consumer display interfaces and displays will be capable of supporting dynamic, content-dependent metadata.

One of the issues with most HDR formats is that they cannot be performed on-the-fly—an issue for live broadcasting. According to video expert Joe Kane, however, an HDR proposal made, not surprisingly, by two broadcast services (the BBC in the U.K.) and NHK in Japan), might offer a usable alternative. Called HLG (for Hybrid Log Gamma), it provides a single signal that works on both SDR and HDR displays. It requires no metadata and may already be compatible with some existing sets. But reportedly its HDR capabilities are not as dramatic as from other HDR formats—a downside for set manufacturers looking for something so irresistible that it will have their sets blowing out the doors of your local big box video store.

There's also a concern that the use of such metadata (if even possible), might involve compromises that will render a result different from that intended by the original content provider. Whether these compromises will be subtle or significant remains to be seen.

Also ruminate on the following composite quote from the paper, signaling how HDR, while offering new options to filmmakers, might also put serious restrictions on their creativity:

Edit pace may need to be different in HDR versions than in SDR versions, since fast cuts with high brightness elements may cause viewer eye strain…Limits in the speed of [the] human visual system’s adaptation to overall brightness changes may require producers to either avoid, or perhaps prepare the viewer for abrupt changes in average picture level or the presence of multiple HDR highlights elements. In other words, viewers’ eyes might need to be adjusted slowly, in advance, via a progression of scene brightness, for a known, large upcoming change in picture or highlight brightness.

Back on the issue of compatibility of different HDR formats on a single Ultra HD set, I’d love for some resourceful manufacturer at a trade show (CES, perhaps!) to set up two identical sets side-by-side displaying the same material, one created in DolbyVision, the other in HDR10—using both HDMI and wireless streaming. While there are other potential competing formats, I suspect that one of those two (or both) will dominate the market.

Buyer Beware
In the meantime, check our reviews for discussions of compatibility, though as noted earlier these will necessarily be severely limited until we see more HDR content. If you’re in the market for a new Ultra HD set with HDR (and remember that most current UHD sets will not do HDR—it’s not a mandatory UHD feature), be prepared to ask about such compatibility at the point of sale. But don’t be surprised if you get a blank stare. Manufacturers rarely point out what their set’s won’t do. I wouldn’t discourage anyone from looking seriously at a good UHD display, but go into it with your eyes open. HDR is a genuine advance, but still in its Wild West days. Again, from that SMPTE report:

For purposes of this study report, exact targets of peak luminance and achievable black level for consumer HDR displays have not been agreed upon and are largely considered an optimization issue for display makers who must evaluate issues such as panel capability, image quality, color gamut, cost, power consumption, manufacturing, etc. Further discussion and evaluation will be needed with other industry groups as well. It is expected that deployment of HDR will occur through gradual improvement and a migration over time to enhanced consumer equipment.

Wild West days, indeed.