May 3, 2018

Light Iron Founder Spells Out Why 8K is Coming and That's OK

Founder of color and post house Light Iron, Michael Cioni, claims that 8K and beyond will be survivable—and even beneficial—formats.

For many, the transition from SD to HD was meant to be the last big switch for a while.  After working with standard definition for decades, and a very slow transition from analog to digital video, the long, expensive, and years-delayed transition from SD to HD felt like a major technical moment after which we could rest. However, in the last decade, we've seen 4K go from a niche product (many didn't believe RED could make the camera work at all when it was announced) to a somewhat standard feature that most of us have on our phones.

Post house Light Iron has been heavily involved in that process, supporting high-end filmmakers (David Fincher is a regular client) and currently owned by Panavision. The company's founder and current Senior VP of Innovation at Panavision and Light Iron, Michael Cioni, spoke recently at the "Faster Together" conference put on by LumaForge as part of NAB about the reasons that this has happened so quickly, and how much further we're going to be going in the near future. Beyond 8K is definitely happening.

The video is worth a look for anyone struggling with the constant rhythm of resolution upgrades and for some insight into the hurdles that have been jumped over to get us there. Particularly interesting is the 4K monitor they were using during the era of early 4K projects, which was actually four 1080 monitors, rigged together with gaf tape used to cover the bezel and create a cohesive image

Credit: Light Iron

Cioni's argument is built around an extension of Moore's Law, while asserting that it's not necessarily a law and it only talks about transistors, but he sees it as a useful guide for computing speed.  Basically, processing 8K footage right now "feels" like processing SD did back in 2002.  SD to HD took about four cycles, and SD to 8K took about eight cycles of Moore's Law, to get to a similar amount of processing power.

Cioni shares several reasons why, even though it seems too soon to think about it, 8K is not only worth considering, but inevitable and necessary, not just as a capture format but also as a release format.  In addition to some self-explanatory reasons (lower noise from less pattern noise interfering with sensor noise, less DoF due to higher resolution and thus smaller circle of confusion), one of the most interesting to us is the possibility of reframing after deliver.  While we tend to think of reframing mostly as a tool in the post suite, when a director doesn't like the way a shot was framed, or wants to get a close-up out of a medium shot, there is the possibility of allowing some reframing for the end user without significantly noticeable degradation. The ability as an audience member to zoom into areas of the frame in a sporting event, or your favorite horror or thriller, and still see 4K resolution at 200% zoom could be fascinating in how it changes the relationship between the viewer and the content.

Credit: Light Iron

It's an especially interesting argument since many of us in the tech industry tend to think the other direction about why this is all happening.  Cioni is pointing out that, in the wider computer industry, technology like affordable SSD, Thunderbolt 3, and faster CPU and GPU processors make 8K doable, so of course we should bump the resolution up in the camera. Most of us working in the field tend to feel it's driven the other way: that the folks on set keep bumping up the camera resolution so we have to buy faster hard drives, computers and cables to keep up. But in some ways Cioni is right; even if cameras stayed comfortable at 4K, computers would keep getting faster (if only to play cooler games), and if we have the power to handle it, in some ways it does seem worth capturing more resolution.

Our personal take is mixed on 8K; currently, the primary innovation that seems to be creating a noticeable increase in quality is larger sensors, with the images from VENICE, RED Monstro, and Alexa LF being truly impressive.  Even the Hasselblad H6C gave beautiful footage at 4K resolution, though it did have rolling shutter issues.  Cioni is arguing, and is almost definitely correct, that 8K is coming for us whether we want it to or not, and the sooner we start thinking of creating ways to use it and embrace it the easier the transition will be.  Conventiently, it's hard to see the difference between HD and 4K on a 24" monitor, so it's really only big monitors where it's going to be worth the trouble.

Give the whole video a watch and let us know what you think down in the comments.     

Your Comment

16 Comments

For IMAX, 8K is kind of the starting point really. I stopped going to IMAX theatres after the projectors went digital, as I saw little benefit. If the resolution and quality started matching IMAX film again I would go back.

May 3, 2018 at 1:34PM

0
Reply
Batutta
378

I hate to say it but many of the examples in the benefits of 8k are just wrong.

1)Lower noise floor - Not necessarily. On the same size sensor the pixels will be smaller and more noise (like dslrs) but the noise will be finer due to more pixels.

2) More DR Mapping - Uh no. There will not be more Dynamic Range

3) Reframing - Yeah that's true

4) Less DOF? - No that's determined purely by sensor size, focal length, focal distance, and aperture

5) New Lenses - That has nothing to do with 8k.

6) Magnification - I think you mean reframing.

So there's one benefit of 8k. Reframing.

May 3, 2018 at 10:01PM

0
Reply
avatar
Zachary Will
Cinematographer
895

Zachary, you forgot benefit number 7. You make more money if you have shitloads of 8K gear to sell.

May 3, 2018 at 10:38PM

0
Reply
avatar
Jonathon Sendall
Stories
1883

Jonathan-
I actually think this point is really excellent and appropriate! Making money is an important part of the equation because it means we can finance R&D so we can try new things and make things better! Plus, on the consumer side, it means we can invest in new technologies and use those to fill gaps in the market created by luddites which allows for entrepreneurial growth! It also means DIY groups can invest in these tools to elevate their quality levels, often resulting in charging more for their services.

May 4, 2018 at 10:55AM

0
Reply
avatar
Michael Cioni
SVP Innovation Panavision & Light Iron
88

Personally, I'd like to see the consumer side catch up. I'd like to see more 4K content, higher bit depth in TV's, tablets and higher bitrates over networks.
I'm glad as an industry we're transitioning from Super 35 to "Large Format" (which is a bad name for it as it makes medium format sound like a smaller format *cough* Arri *cough*.) but from a sharp image standpoint,we're definitely rapidly approaching the law of diminishing returns after 8K.

May 6, 2018 at 6:38AM, Edited May 6, 6:48AM

0
Reply
K W
412

My point Michael was that financial considerations aren't always the best motivators towards better techniques or technology. They can help but they aren't the primary. 3D is the perfect example of this. 8K has it's place of course but it isn't the main reason for better film or tv productions.

May 8, 2018 at 3:12PM, Edited May 8, 3:12PM

1
Reply
avatar
Jonathon Sendall
Stories
1883

The slide was specifically to do with large format 8K so that's where most of those benefits are derived from.

May 3, 2018 at 11:15PM

30
Reply

Hi Zachary-
In the interest of helping explain these points so you can understand them better, let me briefly elaborate:

1) Any sensor (regardless of its sensitivity) will experience a lower noise floor through the process of super sampling. Since today's exhibition formats are 4K and HD, you are forced to scale down 8K material by 1/2 or 1/4 size. This process results in automatically lowering the noise floor due to the law of averaging. Sensors that start at lower resolutions are unable to super sample and thus do not experience the same noise reduction characteristics as higher resolutions like 8K. If you want an absolute measurement, 8K supersampling down to HD is the equivalent of a noise floor reduction of -6db. You can see an example of this here:
https://vimeo.com/260678147

2) The operative word here is "mapping." A sensor can only map dynamic range to what it can see. A greater pixel count means the camera can map more dynamic range in the real world to the captured image. Since a pixel can only have a single value (or represent a single stop) more pixels allows for more values (and more stops) to be mapped to the image. You can learn more about that here @ the 43min mark:
https://vimeo.com/248235757

3) We agree:-)

4) Your ingredients that contribute to DOF are correct, but your assumption that you can achieve the same DOF results in large format is incorrect. Larger sensors will always have less depth of field potential and also allow for lens speeds to increase at speeds 35mm cannot achieve due to its smaller geometry. Because of spacial light mapping, 35mm sensors will always have more DOF than medium or large format (provided you creatively desire less DOF). You can learn more about that here:
https://vimeo.com/253322347

5) The slide you are referencing is 8K Large Format, which is referencing a 46mm diagonal, not S35. By this, S35 lenses do not cover 46mm which means you use new lenses in order to cover the large format image circle. So this has everything to do with 8K in the context of large format. This opens up a world of vintage large format lenses, as well as a slew of new ones such as the CP3, Thalia, S7, Artiste, and Signature lenses. All of these are brand new and driven by large format cameras, mostly thanks to 8K RED since they mass produce 8K large format.

6) No, I mean magnification, which does not refer to reframing. You can learn more about that in the link I provided for #4.

I encourage a good discussion and I think your points are good because honestly, this stuff is new and that means it can be challenging and often open to some interpretation. I hope these answers help clarify these points and moreover, I'm confident if you shoot 8K large format yourself, you'll come to these same conclusions I have in my own work.

May 4, 2018 at 10:52AM

9
Reply
avatar
Michael Cioni
SVP Innovation Panavision & Light Iron
88

I love Panavision lenses and gear, but why the push for the RED platform? Almost every AC and DP I have worked with still enjoy working with the Alexa system more from the Alexa65 to the Mini.

May 4, 2018 at 12:19PM

0
Reply

Great question, Nick. Every camera (Alexa iterations included) utilize parts from companies all around the world. No camera company builds every component within a camera, not unlike how Ferrari integrates parts from other car manufactures when building their cars. They key differentiator is how those parts work together. Ferrari is known for their quality and have a distinct top-end target market, very similar to Arri.
RED cameras are designed for the mass market and serve a wider array of customers. Alexa65 and AlexaLF are only designed for a select group of professionals.
When we examined options for electronic partners to go with our lenses and accessories, we wanted to work with the most advanced sensor and fastest evolving electronics program. That's RED. How we optimize that is like taking the best parts and tuning them to Panavision's target market. This doesn't at all diminish how well the Arri platform is designed, rather we set out to accomplish what Arri and RED (and others) do well and elevate them to a previously unmatched level for top-end professional cinema. RED Monstro, REDCode, and RED's custom ASICs sit as the core of this mission, and once you shoot with it, you'll instantly know the difference.

May 4, 2018 at 12:44PM

0
Reply
avatar
Michael Cioni
SVP Innovation Panavision & Light Iron
88

copy that

May 4, 2018 at 1:22PM

0
Reply

So, can we finally start to consider a 4K worflow, from acquisition to DCP, as the minimum bottom line, at least for features? We have yet to see a Marvel or Disney (except for Tomorrowland and Star Wars) or Pixar movie released in 4K!

May 4, 2018 at 4:38AM

7
Reply

"The ability as an audience member to zoom into areas of the frame" ... WTF. How about writing a good compeling story so the audience will be so immersed into it that won't have time to play with your TV while watching the movie. Enough with this technology BS. The cameras we have now are already good enough. Time to focus on screenwriting (which is at its lowest now).

May 4, 2018 at 5:51AM

0
Reply
Rod P
162

the Ks dont matter. there is no visible difference between reds new cameras. all new films might as well be shot on their 5k dragon. large format? dont matter. you can achieve a blurry look by shooting at 2.8 in daylight with a 50mm.
only thing of interest in new cameras is low light sensitivity and insane dynamic range.

May 4, 2018 at 11:50AM

1
Reply
avatar
Vincent Gortho
none
962

CVS Health Corporation, together with its subsidiaries, provides integrated pharmacy health care services. It operates through Pharmacy Services and Retail/LTC segments.

cvs caremark login |
cvs caremark

May 4, 2018 at 11:57PM

0
Reply

He's right on many things. I have been telling people these things for many years. I suggested Red target 4k in around 2004. But forums have a lot of idiots that just want to argue against real sense. Unfortunately, they are often the neurotic ones out the front, as a legion of naysayers and supporters dominating forums, who will waste 90% of your time trying to argue long after the reality is evident.

But he is extrapolating moores law out when silicon increases are supposed to be coming to an end (again). As I estraporated more than a decade ago, there was enough left in moores law to get low powered 8k consumer video on a phone, which looks like it will come, even though things didn't go so smoothly with performance increase (the run out generation for silicon shrunk a bit). You can do a rather hefty 8k PC systems for processing, maybe 16k is possible very high end. But the shift to gpu processing will probably shift 16k down to sub $8k pricing (also note the nvidia chipset in the Z-Cam E2 camera is already encoding enough native h264 to do pro 8k video. Depending on the chipset, it could be programmed to do that in raw compressed). 16k of course, is basic 4x8k 3D multipoint processing power. Something desirable. But the resolution increases go far past this and silicon is coming to finale generations, So, 16k might be the last thing that is affordable for a bit. The leading edge NVIDIA chipsets will probably reach 16k of processing on affordable systems in future for instance. Now, custom silicon ASIC could extend that a bit (ambarella being a leader on the low end), but don't hold your breath for more than 16k silicon on your smart phone. A simple trick is to look at the claimed framerate at the highest regular video resolution, of the best ambarella chip (they know their stuff), the power consumption, and the maximum data rate, to give you an idea of what is possible. So, when you see 8kp60 come up in future, that is half of pixel rate of 16kp30. You move across to the maxinum datarate looking for between 584mb/s-1200mb/s h264 (half for h265) per 8kp60 worth of pixel rate. Finally look at what the power consumption is to see what size of equipment that could fit in. When they can do descent 8kp60 video you know two things, one that they only have to double performance, likely by doubling the circuitry and power consumption, or enough silicon generations, to get to descent 16kp30 performance. I don't know what silicon generation they have in their latest chip, but if it is the latest they might hit 8kp60 before silicon runs out, and that might be at over a watt of power. But because I'm out of the loop, maybe another doubling of the performance could turn up but at over 2 watts. Anyway, you see, it is a stretch to fit 16kp30 on the phone (though silicon nanowire batteries are expected to 10x the battery life. But one hot phone. So, it probably fits something bigger more). But I'm working on a codec system proposal that can drastically reduce processing power consumption, and datarates to levels much lower than redray.

But, this is only the start of resolution increases. The 720p Sony Holographic filming test bed needed a 16k sensor to get that 720p. That is a ratio of around 12.5x the resolution. So, 4k resolution is over 36k, 8k is over 72k. 360 degree surround is around 432k of resolution. Now, in certain applications you may want to times these resolutions by four. So even a straight 36k to get 4k resolution is a nightmare on camera, or in post. Plus, the holographic calculations are likely going to need a lot. But this is where post silicon comes in. Some of these things are expected in the next 3-5 years. How long before they will be affordably available is another thing.

The 10x expected speed increase of various carbon based processing structures is likely to barely cover it to 36k professionally. Magnetic computing is aiming at 150ghz speeds and up to 1 million times less power, which means you can suddenly pack many times more processing power in parallel. That can cover things better. But we are expecting to deal with things in 3D and with artificial intelligence. But some versions of magnetic computing might go into the terahertz range. That would cover things. You start to reach a level where the information and interactional film making can be like the fictional holodeck with magnetic computing. But there are likely gaps here between silicon peeking out and highest performance alternatives being more affordable outside studios. Maybe only studios might be able to afford anything in the 150ghz+ range for years. So there might be decades of camera development to come.

As far as storage goes. There is a lot more storage to come. We have not reached writing atoms or the end of volume based storage yet, which are outstanding storage densities.

May 5, 2018 at 12:34AM, Edited May 5, 12:47AM

3
Reply
Wayne M
Director of a Life
260