NHK's 8K Chip developed - 120fps

I see this stated pretty frequently, and I'm curious - where is the data? What "most people", what 4K footage, and what "4K cinema screens"?

Anecdotally, cinema professionals (DPs, directors, cameramen, and crew) and relatively laypeople have reported very favorable experiences with private 4K screenings. These are often tied in as promotion of products or services of some kind, so they may be exaggerated somewhat.

What I haven't heard of is a good A/B or butterfly test of some kind between 2D 2K and 2D 4K in theaters where a large cross section of the movie-going audience simply said "meh".

I'd also be interested to know how many of the year's films were shot with a camera that could produce an image that resolves 4K, edited in 4K, and actually sent out to theaters in a 4K DCP.

The 'data' comes from pretty much everyone I have questioned at a couple of short film screenings that have projected films in both 1080p and 4K. People were able to discern the more attractively lit and shot films from the others, but no one noticed any discernible difference in detail, sharpness or resolution between the 4K films and the 1080p ones.

I'm almost certain you'll find that the favourable 'anecdotal evidence' coming from private 4K screenings is down to precisely that - the fact that they were private screenings, and therefore almost certainly held in smaller screening rooms where everyone is seated closer to the screen. Because it's only when people are up-close to the screen that they're going to discern much difference between the two resolutions. In a regular-sized cinema, people's eyes are simply too far from the screen to perceive much of the additional resolution (that's circle of confusion in action).

People can (and will) bicker and argue about this non-stop I'm sure. Personally, I think it's great news - it means that our current HD standard offers no appreciable limitations in terms of resolution output for any medium that we currently distribute moving pictures too. You can throw your HD productions on anything from handheld digital devices to full-size cinema screens and only the quality of your lighting, set-direction, performances, sound and editing will be able to be picked apart - personally I think that's a good thing.
 
The "data" is simply math.
The resolution performance of the human eye is well known and beyond any doubt - whereby most people don't have a 100% performance.
So you can do your math between the human eye, screen size and viewing distance - than you subtract the resolution loss of lens, debayering, recording compression, delivery compression (JPEG2000 in this case), projection lens and if you are anal the screen itself.

So if you sit in a 8k theater and you think you can see things that are way beyond the physical capabilities of your eyes, it´s just your brain making it up.
If they would show you a 2k film, but telling you its 8k, you would probably see the same things.

This has noting to do with technology - I love progress as any guy out there, but it has to make sense.
To employ an other car analogy - why is nobody (but salt lake racers maybe) building a car that goes 1000MPH?
We got the tech to do it. So why can I not buy it?
Right - doesn´t make any sense - where can you go 1000 MPH (and no, salt lakes don´t count).

Personally I think we are hitting a ceiling here in the same way, that most cars are not going to be faster than 130 mph in the future - even if we could make em faster.
It´s not getting much better than an Alexa (maybe an Alexa with 5k - but that´s it)

So what´s the future?
Higher frame rates for slomo - sure. Better codecs - maybe RAW.
Even more sensitive and cleaner sensors, higher color depth.
Maybe 3-chip 35mm cameras without debayering or something like Faveon chips.
Physical shutter or an electronic equivalent (like that filter that mimics a real shutter).
Seamless electronic NDs
I also can see 48 fps coming as a new standard, but not HDR - 14 stops are enough.


my ct2

Frank
 
Last edited:
So what´s the future?
Higher frame rates for slomo - sure. Better codecs - maybe RAW.
Even more sensitive and cleaner sensors, higher color depth.
Maybe 3-chip 35mm cameras without debayering or something like Faveon chips.
Physical shutter or an electronic equivalent (like that filter that mimics a real shutter).
Seamless electronic NDs
I also can see 48 fps coming as a new standard, but not HDR - 14 stops are enough.


my ct2

Frank

You forgot: cost. The future will bring lower cost for all that tech. Today's $5K digital cameras outperform $50K digital cameras of the past. The ultimate limit is the human eye/perception capability. Very soon, cameras are going to exceed those human limits, and at that point it makes no sense to push them further - if you are hitting 100% of the best human vision, what's the point of going on? So at that point the progress will move along two axis: cost and ergonomics. Ergonomics will mean different form factors for different needs, and of course, trying to simplify and make the operation of the camera more intuitive, flexible and customizable. And cost, well, as always, the cheaper the better, lowest cost for the best quality available, always a balance. I suppose once you have a HEP (Human Eye Perception) camera available for under $1K, we can all stop arguing over whatever new tech is available... well, no, there will always be issues of taste, and arguments over ergonomics, but hopefully an end to pixel-peeping :)
 
Although 8K would be useful for reframing, VFX, and future proofing, I don't see its general adoption even by 2020. A lot of people don't even have HD sets yet, and for many of those who do, a significant percentage of the content they watch is still SD. And is it me, or are retailers pushing more 720p sets than they did a couple of years ago? I don't mean 32-inch sets, for which the value of 1080p is debatable, but 42- and 50-inch plasmas. These are lower-cost ($400-$600) models, and for a large percentage of the population, price apparently trumps resolution, especially in a down economy. The technical hurdles to 8K will be overcome in the lab much sooner than they will in the consumer market. Standards, like HDMI, have to in place to allow it, and they seem to move at a snail's pace.
 
The "data" is simply math.

Yes, the human eye resolves about 1/60 of a degree. A comfortable viewing distance is about 35 degrees. 35 x 60 = 2,100.

Engineers are free to explore the limits of technology and build 8K chips, just as they are free to send people to the moon, just as you are free to climb Mount Everest, even though the practical benefits are few.

On the other hand, if we are talking about making 8K, or even 4K, the Standard, that's when I get upset. There is intermittent heckling in this forum whenever we talk about what we want in a camera, that we will never pick up a camera, always want more, never have enough. For some, that is true, but for most I think it will at least taper off after a certain point. For me, it is:

HD
10-bit
4:2:2
120 FPS
ISO 1600
12 F-stops
global shutter
$4,000 camera
$10/hour media costs

In other words, once a camera has all these things, I will be much less likely to ask for anything else. If had 10-bit, I wouldn't go along with those asking for 12. If I had 2K, I wouldn't ask for 4K (which bears out right now, if you want proof). If I had 12 F-stops, well, I might gladly take up to 16 or so, but after that the returns are strongly diminishing. In another post about this same topic, I said it in terms of (1) it works, (2) I'm satisfied, or (3) icing on the cake.
 
I guess I can see other applications for the 8K sensor. Other than re-framing, possibly better color and noise levels when resized. Maybe inventing better 3D lenses so make use of the 8K sensor in side by side 3D (hence turning the sensor into a 4K) or something like that.
 
Camera technology has to trick the eye, not match the eyes performance. So, by that definition, the camera must out perform the eye.

Dislpays are what needs to be improved on. Whether it is the increase viewing comfort or to increase suspension of belief in the image.

(I am sorry to all the people that have bad eyes, but 1080p is annoyingly pixelated in theaters, at home, and on my laptop. HD will be around, so not complaining, but it was implemented because of infrastructure reasons, not preference. )
 
Perception of any image is both empirical and essoteric, conscious and subconscious. When we view images from any camera the "Look" is every bit as powerful as the resolution. And before anyone gets confused or angry, I'm not talking about the content being produced just the ghost in the machine leaving it's imprint on the image. Recently I've been shooting on a lot of mixed formats(MiniDV, 16mm reversal film, betacam SP, etc...) and I found myself saying over and over things like: Man, I don't care how many hours I spent in the grade, I'd never get that look. 8k is just another tool to create images, people will either take to it or they won't. I'm interested as an image maker to see what the technology brings to the process and what doors it will open up, not angered by the persuit of it. People will always demand change, arguments for new and better technology are made as soon as new and better technology is released. The technology that people complain about today are things people dreamed about being able to have not very long ago... Just my 2 cents.
 
I guess I can see other applications for the 8K sensor. Other than re-framing, possibly better color and noise levels when resized. Maybe inventing better 3D lenses so make use of the 8K sensor in side by side 3D (hence turning the sensor into a 4K) or something like that.

This is my thinking. Not saying necessarily 8k but, the higher your resolution, the less aliasing you are going to have when you bring the image down to a lower resolution target. This is what I love about the C300 in concept. I really am fine with 1080p footage, but, I want that to be THE BEST 1080p footage we can make. I really would rather the camera take all that data and reduce it before committing it. Having higher resolution does allow for more advanced noise algorithms as well. Although I question if it reduced the ability to store more dynamic range because at some point, the area on the sensor used to separate each pixel becomes more significant compared to active area. It's finding that target resolution that provides the best compromise of all the sensor parameters.

And I'd like even more dynamic range for use in the highlights. And, I'd love global shutter. I'd like to be done with rolling shutter altogether myself.
 
the higher your resolution, the less aliasing you are going to have when you bring the image down to a lower resolution . . . . Having higher resolution does allow for more advanced noise algorithms as well

From what I have read, as long as raw res is at least 25% more than target res, then aliasing can be zilch. Any more than that, and you're just throwing away dynamic range and ISO.

And higher res may let you do further noise reduction, but larger pixels simply have less noise to begin with.

One choice that may satisfy a lot of people is a 4K, FF35 sensor. I don't want FF35 any more than 4K. But the crazy people who want 4K are often the same crazy people who want Full-Frame.* Just make sure to give that camera an S35 crop mode. A 4K FF35 sensor cropped to S35 would leave about 2.7 K of resolution, which is plenty for me.

* crazy adj. People that I do not understand, they may not actually be totally crazy :)
 
From what I have read, as long as raw res is at least 25% more than target res, then aliasing can be zilch. Any more than that, and you're just throwing away dynamic range and ISO.

And higher res may let you do further noise reduction, but larger pixels simply have less noise to begin with.

One choice that may satisfy a lot of people is a 4K, FF35 sensor. I don't want FF35 any more than 4K. But the crazy people who want 4K are often the same crazy people who want Full-Frame.* Just make sure to give that camera an S35 crop mode. A 4K FF35 sensor cropped to S35 would leave about 2.7 K of resolution, which is plenty for me.

* crazy adj. People that I do not understand, they may not actually be totally crazy :)

I think it's a bit more complicated than this, I would say it's not that the noise is less, it's that the larger pixels are more sensitive since they are larger and have more surface area to detect photons. Therefore, if you gain up the output of smaller pixels to match the exposure, you gain the noise up. But, the reason you are doing this has nothing to do with the noise, and all to do with surface active area.

The question is, where is that inflection point where having more pixels allows you to detect and average out noise easier? Imagine 8k or 16k sensors where you've got duplicated colors RIGHT next to each other. Then, you measure both, average the result, to try and reduce by taking multiple readings? This is being overly simplistic, but I'm just pointing out there may be advantages we've not really thought about.

Also, given Bayer patterns, the resolution of red/blue is pretty far off. Does that 25% account for that as well?

I agree I think 8k is overkill. I think 4k downsampled in camera at S35 is my dream camera. I'd like it to have Canon highlight handling as well. C300 is getting pretty close to what I'd want. Can't wait till this all trickles down.

I just don't like shooting with FF myself.
 
It's not that big a deal anyway. The Nokia 808 PureView Phone announced today is an 8K sensor, lol.

We'll see how it does during the tests and examinations :)
 
Although 8K would be useful for reframing, VFX, and future proofing, I don't see its general adoption even by 2020. A lot of people don't even have HD sets yet, and for many of those who do, a significant percentage of the content they watch is still SD. And is it me, or are retailers pushing more 720p sets than they did a couple of years ago? I don't mean 32-inch sets, for which the value of 1080p is debatable, but 42- and 50-inch plasmas. These are lower-cost ($400-$600) models, and for a large percentage of the population, price apparently trumps resolution, especially in a down economy. The technical hurdles to 8K will be overcome in the lab much sooner than they will in the consumer market. Standards, like HDMI, have to in place to allow it, and they seem to move at a snail's pace.

I think 8K is aimed at digital cinema more than television..
 
clearly, no one has spoken to american cable companies about this. we haven't got the infrastructure to properly support 1080 bandwidth at the moment, judging by the picture on my television.
 
Given that we've already reached a point of diminishing returns with 1080p for home viewing, it become highly debatable.
Not really disagreeing with you Grug, just adding on top of what you said. Since most of what you are saying pertains to today's standards. And by that measure, I very much agree with you.



8k/4k are useful for bigger screens for home theater enjoyment, and personal computing where we are close to the screen. Shure, 1080 is enough to get things done, by today's standards, but when I look at a leaf, I can never get too close to not see additional detail. So, I think it is very organic to have higher resolution, better dr, and more color depth in image reproduction.

There is more to imaging than viewing a movie from across the room on a 42inch screen. Shall we only have laptop speakers because I can understand what people are saying from them? No, I want the works. Tweeters, mains, and subs. I want as much of the full spectrum and faithful reproduction as possible.

Saying 1080p is reaching the point of diminishing returns is like saying MP3@96k is reaching the point of diminishing returns.

I will always like film for images and vinyl records for music, but digital technology is still in its infancy. I could do with vast improvements in the technical realm.
 
Last edited:
Back
Top