This webinar presented how UHDTV will be much more than simply four times more pixels than HDTV—in a couple of years.
This month's webinar from SMPTE (Society of Motion Picture and Television Engineers) was entitled "Quad HD to UHDTV: Making the Difference." The presenters were Hans Hoffmann, Senior Manager and Head of Media Fundaments and Production Unit at the EBU (European Broadcasting Union), and Howard Lukk, a movie director and technologist who worked on Toy Story 3 and co-created the ASC-DCI StEM (American Society of Cinematographers-Digital Cinema Initiative Standard Evaluation Material), a sequence of scenes that is used as an image-quality reference for digital cinema.
Hans Hoffmann focused primarily on the work being done in Europe on UHDTV, much of which wasn't news to me or any AVS members who have been following the development of the next TV standard. In particular, he stressed that "4K" is a marketing term that really refers only to pixel resolution—and confusingly at that—while "UHDTV" encompasses resolution as well as frame rate, dynamic range/bit depth, colorimetry, and immersive audio, most of which is outlined in the ITU-R (International Telecommunications Union-Radiocommunication) recommendation called ITU-R BT.2020. (This spec is sometimes called Rec.2020 because it's an ITU official recommendation; "BT" stands for Broadcast Television.)
When talking about the current crop of 4K TVs, Hans called them "Quad HD," since they have exactly four times the number of pixels as so-called Full HD but no other enhancements beyond HDTV. This is also referred to by the EBU as "UHD Phase 1" or "UHD-1," which retains all other aspects of current HDTV, such as BT.709 (often called Rec.709) colorimetry, dynamic range, and frame rates.
The next stop on the UHD roadmap is UHD-2, which is expected to be implemented in 2016 to 2018 and should include BT.2020 colorimetry, frame rates up to 120 fps, bit depth up to 12 bits for high dynamic range, and immersive audio of some sort. Finally, UHD-3 retains the characteristics of UHD-2 but with a pixel resolution of 7680x4320 (so-called "8K"); this is being promoted mostly by Japan and is expected to emerge sometime between 2020 and 2025.
As pixel resolution increases, so does the recommended viewing angle. Not indicated in this diagram is the 60° viewing angle for 4K. These angles correspond to seating distances of 3H (three times the screen height) for HD, 1.5H for 4K, and 0.75H for 8K. BTW, this graphic is an example of the confusion surrounding the various phases of UHD; what is labeled UHD-2 here is actually UHD-3 in the EBU roadmap. However, SMPTE also uses UHD-2 for 7680x4320.
Of course, increasing the pixel resolution by a factor of four quadruples the bandwidth required to transmit the signal at a given level of compression, and adding HFR (high frame rate) and HDR (high dynamic range, i.e., greater bit depth) increases the required bandwidth even more. Hans cited a rate of 6-14 Mbps for current HDTV broadcasts using H.264 compression and 25-35 Mbps for test broadcasts of 2160/50p (remember, Europe uses a frame rate of 50 rather than 60 Hz) with current implementations of H.265, though he expects that bit rate to drop as H.265 is improved. And don't forget that current HDTV broadcasts are 720p or 1080i, and 2160p test broadcasts do not yet include high frame rate or increased bit depth.
One interesting tidbit from the presentation was the notion that content with HFR and HDR might be downscaled to 1080p to reduce bandwidth requirements, since upscalers are so good these days. I'm not sure I agree with this idea, but something's gotta give, at least until H.265 or some other high-efficiency codec can reduce the required bandwidth to a reasonable level.
This slide reveals the results of a perceived-quality study conducted by the EBU with many viewers; the display was a 56" 4K monitor, and people viewed six scenes at a distance of 1.5H (41") and 2.7m (8.9 feet), which was deemed the most common seating distance in the average home. Interestingly, the viewers slightly preferred the greater seating distance when viewing HD, possibly because of upscaling artifacts, but when shown native UHD content, there was no preference between the two seating distances.
Another perceived-quality study by the EBU focused on different frame rates and shutter settings. Clearly, viewers preferred higher frame rates and a 50% (180°) shutter more than a 100% (360°) shutter. The "Low Anchor" point is 30 Hz; I don't know what the shutter setting was.
According to Hans, the dynamic range of human vision encompasses 12 orders of magnitude ("decades") or 40 f-stops. The current TV system is limited to two orders of magnitude or 6.6 f-stops, while an HDR display would encompass five orders of magnitude or 16.7 f-stops, which corresponds to the capture capability of modern digital-video cameras.
Part of the BT.2020 recommendation is a definition of colorimetry, which includes more colors than BT.709 (the current standard), DCI P3 (the standard for digital cinema), or the gamut of real-life reflected-light surface colors. However, there are no displays capable of reproducing the BT.2020 color gamut, in which the colors of red, green, and blue are monochromatic. This might be possible using lasers or quantum dots.
Howard Lukk's presentation focused mainly on digital cinema and workflows during the production and post-production processes. He did show a few slides that were more germane to the consumer experience of UHD.
In this slide, Howard illustrated how a narrow dynamic range forces you to expose for and display different brightness levels without capturing the entire dynamic range in the scene. Seeing the detail in one brightness range means that darker ranges are underexposed and brighter ranges are overexposed.
In this Photoshopped composite image, the effect of a wide dynamic range is simulated; detail in the bright and dark parts of the image are visible.
As I said earlier, there wasn't much about UHDTV in this webinar that those who follow the subject from a consumer perspective don't already know. As I've been saying for some time, the so-called 4K TVs being sold today do not embody any of the enhancements promised by UHDTV except for increased pixel count, and the content-creation community is not yet geared up to deliver content with these enhancements. We should see these two paths converge in the next couple of years, at which point UHDTV will blow your socks off.
Thanks to SMPTE for hosting the webinar and to Hans Hoffmann and Howard Lukk for their presentations.
Like AVS Forum on Facebook
Follow AVS Forum on Twitter
+1 AVS Forum on Google+