The Best Thing About 4K & UHD Isn't More Resolution: Sayonara Interlacing, Hello Wider Color Gamut

itu r rec 709 2020 hdtv 4k uhd ultra high definition 2As some have speculated, the recent push for 4K/UHD may have as much to do with hype as it does with quality. And, as has been stated time and time again: you may not get a huge benefit from 4K in your home, depending on viewing distance and screen size. There are some other factors, however, that make 'Ultra HD' technology desirable, regardless of clarity so crisp you can't even tell how crisp it really is. These factors are the other important goals defined in the ITU-R's (aptly dubbed) Rec. 2020 spec for 4K/UHD. Namely, they are (larger) color space and (progressive-only) frame rate.

The official spec is properly called ITU-R Recommendation BT.2020, and it suggests a few much-needed updates to the seemingly still-pretty-damn-new but actually-quite-aged Rec. 709 (first approved in the early 90's) for HDTV. It's worth pointing out that Geoffrey Morrison highlighted these benefits previously at CNET -- it's also worth highlighting them specifically for filmmakers.

No More Interlacing -- Progressive-Only Frame Rates

I don't know about you guys, but I really hate interlacing. I'm admittedly kind of 'interlacist.' To me, interlacing represents one of the originally quite ingenious, now entirely vestigial hold-overs from analog broadcast. Along with non-integer frame rates, interlacing fills a need for video that we actually don't even have anymore. After all, you're not reading this, or more importantly watching movies, on a tube. Probably. (Anyone continuing to use CRTs industrially will always be able to PSF-ify and/or cross-convert to their needs). This quote is from a 2004 article by Nick Radlo:

Yves Faroudja founded Faroudja Laboratories, and has spent years inventing ways to improve picture quality in broadcast TV, winning a Lifetime Achievement Emmy in 1998 for his work. "Why are we still talking about using interlace? This is something I've been dealing with for decades and I can't believe we're still considering it in the 21st century," he said. "I'm a complete enemy of interlace. There are some things in an interlaced picture that you just can't de-interlace. When TV began it was a very simple way of reducing bandwidth, but now it is not needed. I'm very depressed when every ten years I see people about to make the same mistakes."

Let the depression be lifted (again, hopefully). So says Rec. 2020, in a bold, progressive (whoa-hoa!) move towards the future: no more fields. 4K/UHD should be progressive-scan only. Quoth the spec:

Picture temporal characteristics: Frame frequency (Hz) 120, 60, 60/1.001, 50, 30, 30/1.001, 25, 24, 24/1.001. Scan mode: Progressive.

Note the specific division by 1.001 in 60, 30, and 24 -- this denotes support for the far more common non-integer 59.94, 29.97, and 23.976 rates we're accustomed to. Also note support for 120 Hz imaging... in a sense, Doug Trumbull foresaw the benefits of a 120-based future with his digital Showscan technology. The good thing about 120 is that pretty much everything divides into that rate evenly -- 30, 60, and yes, 24. This is partially already in play with 120 Hz sets, but media actually delivered in 120 (for whatever reason) means the sets wouldn't have to perform that great 'motion smoothing' trick they do.

The problem with phasing out legacy technology (i.e., shooting true 24 over 23.976 fps) is always backwards compatibility -- which creates a bit of a conundrum, because you want to shoot something that, ideally, anybody can eventually watch, even on their aging home system. Conforming true 24 to 23.976 and vice-versa are solutions, but not ideal. So when are digital cinematographers to be finally rid of these redundant fractional frame rates as well? It's a bit of a vicious cycle, in which cameras and display technology may continue to be stuck for some time. Rec. 2020 is, at least, suggesting a big step in the right direction. Speaking of which...

Larger Color Space -- A Way, Way Bigger Triangle

Rec. 2020 identifies D65 as the white point of its color space, located at [0.3127, 0.3290], the same as in Rec. 709. The difference is the available gamut -- and the fact that Rec. 709 is limited to 8-bit depth, whereas Rec. 2020 supports 10- and 12-bit depth. Rec. 709 defines its outer color limits at R[0.64, 0.33], G[0.30, 0.60], and B[0.15, 0.06], keeping in mind that "Picture information can be linearly indicated by the tristimulus values of RGB in the range of 0-1." Rec. 2020, on the other hand, sets its boundaries much, much farther apart, defining R[0.708, 0.292], G[0.170, 0.797], B[0.131, 0.046]. This apparently equates to a 75.8% coverage of the CIE 1931 color space compared to Rec. 709's 35.9% coverage of it. What all this looks like, when graphed, is this:

Images by Sakurambo & GrandDrake. If you want the wavelengths in nanometers, check out the or Wikipedia pages on Rec. 2020.

One of our commenters pointed something out, in Joe's recent post on ALEXA's somewhat-begrudged evolution to 4K: that this --  an expanded gamut -- is what ARRI should really be concerned with, or even excited about. Geoffrey Morrison generally agrees, though he expressed concern that this would be the most difficult part of Rec. 2020 to achieve across the board, citing potential difficulties (or expenses) in developing display technology. And, of course, there's the ubiquitous question of backwards-compatibility. Might Rec. 709-shot material display all out-of-whack on a Rec. 2020-compliant UHDTV? Will the propositions of Rec. 2020 be universally embraced, with display manufacturers forming a 'unified front' for such improvements? If so, how long will it take for the technology to catch up -- while still being affordable?

As Morrison states: "Unfortunately, it's hard to say what of the current version of Rec. 2020 will survive to future revisions, and when, if ever, any of these revisions may come to fruition." I certainly hope they that they do. These definitions stand only to improve the palettes of digital filmmakers, or allow them to image in ways they've always wanted to. Despite the concerns, I think there's only one way to upgrade: bite the bullet, accept backwards compatibility risks (where not completely self-defeating), and push onward toward a bigger, brighter, more vividly colored future.

What do you guys think? What negative implications, should these specs take hold, have I missed? What about the benefits?


You Might Also Like

Your Comment


If you have a wider color space, that fills the old color space completely, there will be no problem what so ever displaying old material in it.

Look at what is currently done with DCP's. DCP's are stored in X'Y'Z', which covers ALL colors (mainly for archive reasons), and it's displayed in P3 at theaters. But most films are graded in rec709!

You will only get problems when you try to display an image that has a wider color space than your display. Because than the display has to decide how it will show you the colors, that it can not reproduce.

What I really don't understand is the use of x/1001. Because nearly every (if not every) film you watch at a theaters is presented in 24p, but recorded at 23,976p! This is just stupid as it is the same as interlaced. It is a relict form the past. It was necessary to be backward compatible with B/W TV, but it's not anymore! So why not just drop the drop frame Timecode altogether?

July 18, 2013 at 2:59PM, Edited September 4, 8:21AM


Consider the formats that we will be recording in and then displaying in theaters and on broadcast networks. It's going to be very exciting to see!

July 18, 2013 at 4:17PM, Edited September 4, 8:21AM


Back in the 80's and 90's, the well heeled amongst us would build a home theater using a front projector from the likes of Barco, Vidikron/Runco, etc. Those units, compared to the sub-$10K systems like Mitsubishi had a built-in line doubler, usually from Faroudja. Sony once tried a similar technology with a 27" TV but with a less expensive/powerful chip. In today's terminology, it'd be a 480p. Back then, it was billed as Improved Definition and the 27" Sony ID ran ~ $4K retail (a similar XBR set was ~ $1K). The picture was a lot denser than a regular tube but there were horrible motion artifacts. It may have been tolerable for movies at certain distances but definitely not for sports.

As to Rec.2020, I am pretty sure it'll be a major sales push as well.

July 18, 2013 at 4:25PM, Edited September 4, 8:21AM


Interlacing is one of the worst aspects of digital distribution today. Many of the "films" made in the past 20 years will never look great and some may never even be distributed on some of the most ubiquitous digital platforms due to interlacing issues that are ever prevalent in the DV and similar movies made by amateur (and even some professional) filmmakers worldwide. Always shoot progressive, that is one thing I cannot stress enough to people who ask about equipment and process.

Then you just need to get people on the same page regarding frame rates, etc., but with the advent of higher frame rates and future interoperability, hopefully most devices will play well with all sorts of content.

July 18, 2013 at 6:02PM, Edited September 4, 8:21AM


Most current display panels couldn't even cover 100% sRGB, Rec.2020 colour space is more like a beautiful dream.

July 18, 2013 at 7:49PM, Edited September 4, 8:21AM


I saw Philip Bloom say interlacing is a dirty word in his house.

July 18, 2013 at 8:43PM, Edited September 4, 8:21AM


These aspects are however completely independent from 4K. A new colour space is highly wellcome, but even this has been tried before, with xvYCC and "deep colour". Both could and should have been integrated into the existing Blu-Ray standard years ago.

July 18, 2013 at 11:48PM, Edited September 4, 8:21AM

Thyl Engelhardt

I'd like to see them add 48p to the rest of those frame rates. As Peter Jackson proved, 48fps is good for smoother motion simialr to 60fps but can easily be converted to 24fps as well.

July 19, 2013 at 10:15PM, Edited September 4, 8:21AM


Seems like this article suggests ARRIs move to a 4k sensor expands the colour gamut of their camera from that of REC709 to REC2020. But the camera can already exceed the colour gamut of REC709 so long as you capture in raw or log. REC2020 will allow us to broadcast more of the colours that the Alexa can capture, but the move to 4k really bears no relationship to this new standard.

The other thing to consider here is that whilst the 2020 standard expands the potential colour gamut, it does nothing to ensure that that colour gamut will be reproducable on any 4k display. We are at least going to be waiting for 4k OLED displays before we actually see the whole range. The first panels coming out wont output much more than 709.

July 25, 2013 at 1:43PM, Edited September 4, 8:21AM


There are some practical problems with wide gamuts that crop up, unfortunately.

Number one is that most systems designed to output video simply don't handle it correctly. Neither do real-time 3D rendering systems. Our office has a slew of wide gamut AdobeRGB monitors that are a pain because display calibration is not respected with GPU based rendering or video decode. There ARE sofware workarounds, but they are a hassle and require a fair bit of legwork. The other option is to drop into sRGB emulation modes, which are globally set, inaccurate, and not quick to change when you've got three screens and calibrations for all of them. For this reason, it's almost always a bad idea to run wide gamut screens connected to a PC.

Number two, these wide gamuts stretch 8 bits per channel far too thin. Even in full 24 bpp RGB (no subsampling), the distance between values becomes wide enough to start creating problems in midtones. This is something that requires 10 bit processing by the monitor to correct. (A 10 bit panel is not necessary; adaptive dither to an 8 bit panel works fine.) You also need to have 10 bit output from the device driving the display, which requires particular hardware and software. And - you guessed it - it doesn't work with video or accelerated 3D graphics.

We bought quite a few wide gamut Dell Ultrasharp monitors (27 and 30 inch IPS panels, all 10 bit) for the office, and they've created a lot of problems in dialing in colors properly. I would not do it again. The rest of the technology stack is just not set up for it.

July 21, 2014 at 3:50PM, Edited September 4, 8:56AM