Do we need 8K video?

By David Meyer posted 08-06-2018 02:22

  
**UPDATE August 30th** For those who have read this before, more readers were interested in the 8K info below than they were in the HDMI 2.1 CTS announcement, so I simply changed the title. 

On August 1 HDMI announced the release of the HDMI 2.1 compliance test specification (CTS). They did so with zero fanfare — no press release, just a short few lines on hdmi.org. So why does this announcement matter? Well, HDMI devices need to be compliant before they're released to market, and that can't happen if there isn't a compliance spec against which to test them. That's what's out now. It's the gateway to seeing HDMI 2.1 enabled products released.

I tweeted this news on Friday, expressing that I'm excited about it, and got a response saying "especially if you can't wait for 8K displays". The sarcasm wasn't lost on me. But it's not 8K that I'm excited about. Why, you may ask? Well I crunched some numbers about visual acuity and viewing distance, which we'll take a look at before circling back to what it is about HDMI 2.1 that I am excited about.

8K and Visual Acuity

Say we're installing a 65" display. At 4K its pixels equate to just under 68 DPI, or around 135 DPI for 8K. How much of that we can see depends on two factors: how good our vision is, and how far away we're sitting. CEDIA CEB23A recommendations calculate to a viewing distance of 8ft/2.4m for a 65" display. THX standards call for 40° viewing angle, which actually works out a little closer at 6.5ft/2m

Normal human vision is expressed as 20/20. The very best vision is 20/8. Think fighter pilot. At the closest recommended distance (THX), with normal vision we can resolve 44 DPI on the TV, whereas the super human with 20/8 can see up to 110 DPI. Move to the more pragmatic CEDIA distance and 20/20 vision can see 35.8 DPI, or 89.5 with 20/8 vision. So even with 20/20 vision at the VERY close sitting distance recommended by THX, we can only see about half of the DPI required to fully resolve 4K. That is, it's closer to 1080p. In practicality, many living spaces may have a smaller TV (55"?), place a display of that size a little further away than proposed here, and also many people have poorer than 20/20 vision. So in normal home viewing, any benefit purported by 8K really is questionable.  

8K will benefit three main applications;

  1. VERY large screens. ie; video walls,
  2. Large interactive screens. Turn the 65" display mentioned above into a touch-screen and stand within arms reach. Now we're talking!
  3. VR and AR. That is, very small screens with ultra-high pixel density.
No hurry.

HDMI 2.1 Benefits

So back to the HDMI 2.1 CTS, and the two things I am excited about are eARC and a broadening of HDR.

Firstly, eARC — this is a complete re-imagining of audio return channel (ARC). eARC will enable full hi-res immersive audio to travel upstream through HDMI from a display back to the AVR, for example from the Netflix or Prime app in the smart TV. It doesn't use CEC at all either, instead self-discovering and configuring with its own autonomous "heartbeat" that it issues systemwide. Expect this to be one the first available features in HDMI 2.1 products, with Teledyne LeCroy and other test instrument manufacturers already boasting test protocols to aide in the development of eARC enabled products. 

Secondly, HDR. HDMI 2.0 limits HDR to 4K/60 4:2:2, but seeing as there's not really any 4:2:2 content out there, in real terms it's 4K/60 4:2:0. To get 4K/60 HDR with 4:4:4 (eg; gaming) HDMI 2.1 is a must as the data rate reaches 24Gbps. This will comprise 4 channels at 6Gbps each, whereas existing 18Gbps HDMI cables are 3x 6Gbps. Any that are constructed with 4 identical lanes should, in theory, be able to support the 24Gbps level no problem. This simply means having the clock channel built the same as the main AV data channels. Many decent cables have that already.

The other aspect of HDR that it VERY important is the introduction of support for dynamic metadata,where HDMI 2.0b was static only. Dynamic allows the scene dynamic range to swing up and down the absolute range to optimize scene-by-scene rendering, rather than fixing it for the full program as static does. Dolby Vision has always been dynamic, but as HDMI didn't support it, Dolby instead embedded the metadata in a proprietary manner. While effective, it caused some interoperability challenges at times, particularly with any intermediary device which processed the video without the ability to handle the embedded magic. This includes AVRs, HDBaseT extenders with compression, scalers, etc. It will be interesting to see of Dolby revert to standardized metadata.

Watch this space.
5 comments
56 views

Permalink

Comments

7 days ago

well now that all the tech talk is out of the way and John certainly has a handle on that i respect  ..   couple things ..  

yes 8K can't be here soon enough so that we can sell ever more larger displays which require us to install vs buying at the double consonant store.

plus we (CEDIA installers) have many clients who just insist on having the latest and greatest and will happily replace all their TVs every couple years.

oh and in nearly all our theaters no matter how large, ie 40' deep,  the homeowner sits in the last row ...  don't ask me why but its a near universal situation .. i am just happy and consider it a victory if i am able to get the last row 3' off the back wall !! .

yes the real world sometimes clashes with the ideal.

25 days ago

I'm with you on that, John! Thanks for your input.

25 days ago

I'm with you on some high pixel density benefits like large walk up touch screens and simulation over large fields of view. 

But when it's cinema content, or an adaptation of it like VR/AR may bring, you're still talking about pixel count per vertical degree. 

VR goggles look like about a 30 or perhaps 40 degree vertical FOV, so 1080 to 2160 would be fine, and we're seeing that on phones today. 
Higher pixel counts beyond 30 or 40 pixels per degree would be no more beneficial than 4K on a big screen from 2 screen heights away. Indistinguishable from 2K, all else being equal. 

The one thing super resolutions bring is the ability to walk up to the screen without a mosaic appearing. That's cool, but still no difference for cinematic viewing distances, 

4K will take us a long way for a long time, and quite frankly if we had 2K projection that did HDR and WCG, the 4K units would get a real run for their money. Joe Kane and I are the only ones even asking for that one I think:-)
Cheers,

26 days ago

Thanks for the extra info and insight, @John Bishop. I think it's fair to say that I'm simply not excited by 8K, at least not for cinema (yet). I'm more keen to see 4K done really, really well, and for the complications with HDR (particularly delivery) to settle into dependability, and dare I say, normalcy. 

What does float my boat about 8K is three applications other than cinema;

  1. For it to be a really compelling stepping stone to 16K in VR/AR, where the "reality" can actually live up to its name. Extreme pixel-dense in tiny screens, 
  2. For use in large interactive/touch screen displays, where the viewing distance is obviously needs to be within arm's reach. I've already seen this demonstrated in industrial applications,
  3. For in-frame panning and zooming within a static frame while retaining UHD, or near-UHD net output. BBC R&D are working on AI tech to follow action, zoom and switch between static, unmanned cameras. Doesn't bode well for producers and camera operators though, whose jobs may be automated in future sports and stage-based events!
As always, I'll continue to watch with immense interest. I love tech and things executed really well, but I am a pragmatist too. 

John, I'd also like to visit your demo rooms once they're complete!​ And yes we're keen to get moving in revising CEB23A.

27 days ago

Great topic and subtext for it.

Regarding the top line question ‘Do we need 8K’ Let me borrow from John Goodman; 8K ‘is as dumb as a box of hammers’. There is no justification and it damages the quality of 4K native content, through up-conversion losses, and worse for HD content. And, of course there is no 8K content. Six years in we have only a few hundred movies at best in 4K, and many don’t have a genuine 4K provenance. Let’s get real here!

More interesting is the argument made for any particular pixel structure, 2K, 4K or 8K based on visual acuity. I consider that a red herring, and the reasons couldn’t be articulated better than in the Christie white paper on cinema projection technology which I distributed to attendees of my class at CEDIA EXPO this year. You can go to the Christie website to download it. My classroom version is full of highlights and notes I made within. There’s a lot of good stuff there on cinema tech in general, and viewing geometry in particular.

I was an original member on the CEB23a committee and frankly, I dropped out after a time partly because of business demands, but also because the document was basing technical details on TV, and not genuine cinema.

 Now it was a 1080p world when it started, but I will say, 4K and 2K are cinematically equivalent when it comes to viewing geometry in conventional cinema, especially when considering 2k DLP, which has the same baseline resolution as 4K LCoS, which is why both are deemed superior to 35mm in the Digital Cinema Initiatives specification.

Here’s what Christie concluded, and what cinema practices today demonstrate. There is no visual acuity limit in either 2K or 4K when viewing cinema content in a cinema environment. One reason is the Snellen chart argument looks at familiar black letters of the alphabet on a white background under 40fL of illumination as if that translates to moving pictures in a cinema environment. Even the white on black rolling text of a film’s credits don’t challenge a viewer’s visual acuity at 1.5H away from the screen in 2K. And remember, 80% of all cinemas today are 2K with seating from 1H to 3H typically.  A cinema engineer was a guest on Scott Wilkinson’s show ‘Home Theater Geeks’ and he told the story of an A/B comparison done at the Chinese Theater where two DLP projectors, 2K and 4K were being compared. He said at one point they yelled back to the projectionist to change back to the 2K projector...and of course the projectionist yelled back to them, ‘you’re looking at 2K!. That was the front row, one screen height from a 90’ wide screen. And these were trained eyes in cinema tech!

The seating in most conventional cinemas, from Cinepolis to the Samuel Goldwyn and Lynn Dunn Academy theaters range from 1H to 3H. Large format theaters have last rows at 1.5H for 1.85 content and about 2H for scope (always wrong for scope to be a smaller field of view than flat…that’s another TV based cinematic failing)

 THX does say 40 degrees on their ‘home theater’ side, but look at their cinema spec. They say 36 degrees minimum…FOR THE LAST ROW! Why would we ever design a genuine home cinema where the view presented was a last row experience?

These are things I touch on in my Architectural Cinema training syllabus, and that was the title of my class at EXPO, (thanks to the attendees, we doubled the CEDIA expected head count😊

I did hear one industry expert on a panel this year proclaim out loud that the ideal viewing geometry for HT was 50 degrees horizontal. First, viewing geometry should be expressed in vertical terms, as they do in cinema. Horizontal is defined by the aspect ratio per the director’s intent. That was never more clearly demonstrated than in 1952 when Lowell Thomas let that curtain just keep on opening as he said, ‘Ladies and Gentlemen, This is CINERAMA’

So what is the ‘ideal’ viewing geometry? There is no such thing, there is a range of views within which most people find their preference, which can vary with content or even mood. Today in the SMPTE world it is generally accepted that the prime seating positions in theaters with a full depth of rows is between 1.5H to 3.0H. That’s roughly 30 to 15 degrees vertical. If the content is flat that’s 57 to 29 degrees Horizontal. If it’s scope the horizontal range is from 72 to 36 degrees; and there it is, the THX spec for the last row.

So when we’re designing home cinemas, I size screens accordingly and when finished it will look like a real ‘Hollywood’ screening room. And indeed, if you get the sound right, sustained SPL and dynamic headroom especially, then you will deliver a genuine cinematic experience. And it won’t be much like HT’s you usually see. CEB23a needs to be redone, and I’ve been asked to help, and will gladly do so. Anything we can do as an organization to elevate our work and differentiate ourselves from the typical Best Buy execution, is good for all of us.

That is why I’ve coined the phrase Architectural Cinema and why I’m building a training center with the James Loudspeaker Company to show several executions of genuine cinemas; a 6,000 cu’ room, a 2500 cu’ room, and a media room. We’re designing all to meet the genuine cinema standards for image and sound, at a fidelity level for both not often seen commercially or in private spaces. I look forward to sharing more as we progress towards completion in early 2019.

Thanks for continuing to post these timely topics.

Cheers,