REDs new "Juggernaut 32k" brain?

[Will Keir]

http://www.ign.com/videos/2012/01/11/ces-85-inch-prototype-hdtv-has-16-times-more-pixels-than-1080p

Check this out guys, this editor at IGN.com (a game review site) had some odd facts about a new 8k LCD TV Sharp built at CES this week. I started thinking about what he said and it just didn't sit right. So check me on this please:


1) "8k is 8 times the resolution of 1080p."
Odd, I would calculate that 8k gives you more than 16x the screen space as 1080p?

2) "8k is double 4k."
Wouldn't 8k be quadruple 4k?

Now, that's my instinct and experience talking, I could be wrong,. If I am correct, then we need to give this editor a few e-mails before the masses think our 4k images are only double 1080p resolution.

Are there any cameras in the world shooting 8k?

Then they are already talking about 32k!!! Where is the limit? At a certain point, we will see no lines and on most screens 10' and less. Can u see lines on a 4k images projected 20 feet?

What places does 32k have in our future as film makers? Am I going to be upgrading my Epic-X brain next year with REDs new "Juggernaut 32k" brain? At some point you gotta stop waiting for the next tech wave and starting shooting with what you got but with advancing technology where is the sweet spot, I thought it was R1, now it's Epic. Hopefully the camera's get more powerful and smaller, where is the resolution ceiling?

What do you guys think? Where will it end? Do we hold off on 4-5k for the moment to upgrade to 10k a year from now? I've been surprised that there are even 4k theaters that the general public can buy tickets for so soon. EX: "Girl with the Dragon Tattoo" in Thousand Oaks, CA.

What a ride huh?

 

[Laco Gaal]

"Are there any cameras in the world shooting 8k?"
you can scan 8K from bigger format films ( 65mm or 70mm ), just like they did on batman: http://www.studiodaily.com/filmandvideo/currentissue/9703.html

And maybe they will enable 8K recording on the Sony F65, since it already has a 8K sensor.

 

[Solomon Nero]

I read somewhere that the Olympics this year will be shot on propriatary 8k cameras with 8k screens feeds live at the event. I cant find the link anywhere though!

 

[David Collard]

Dalsa had a great sensor before Arri bought them out, but the workflow sucked.
I think you need to focus on the film-like qualities that Epic brings instead of increased discussion about sensor size.
Most of us have more than we'll need for a while. Also, sensor size and pushing pixels around in a state-of-the-art
workflow is one of the brilliant aspects of Red.

 

[Jeff Kilgroe]

http://www.ign.com/videos/2012/01/11/ces-85-inch-prototype-hdtv-has-16-times-more-pixels-than-1080p

Check this out guys, this editor at IGN.com (a game review site) had some odd facts about a new 8k LCD TV Sharp built at CES this week. I started thinking about what he said and it just didn't sit right. So check me on this please:

Consider the source...

1) "8k is 8 times the resolution of 1080p."
Odd, I would calculate that 8k gives you more than 16x the screen space as 1080p?
2) "8k is double 4k."
Wouldn't 8k be quadruple 4k?

We see this quite a bit in a lot of "technical" publication rags like Engadget (I like to pick on them)... To get the 7680 pixels wide that many manufacturers are blasphemously referring to as "8K", you multiply 1920 x 4. To keep the same aspect ratio of 16:9, you also multiply your 1080 vertical number by 4 as well. So you're right x4 horizontally * x4 vertically = 4x4, so "8K" is 16x the information presented with 1080p -- in terms of areal resolution. In terms of linear scale it's x4 on each axis, which is what many people try to reference by, but that's flawed. I have no idea where he comes up with "8 times 1080p". This just reinforces the fact that most "editors" are absolutely terrible at math.

Now, that's my instinct and experience talking, I could be wrong,. If I am correct, then we need to give this editor a few e-mails before the masses think our 4k images are only double 1080p resolution.

Too late. Most people think that anyway, thanks to all the others out there who have the same math skills. And it doesn't help when the marketing departments at companies like Apple and Samsung promote their high-DPI or "retina" displays as having "twice as many pixels" or "double the resolution". Uh, no, dumbasses it's double the number of pixels in each direction, meaning quadruple the resolution.

Are there any cameras in the world shooting 8k?

Yes, the NHK prototypes. They're still the size of a small refrigerator and even though are [almost] 8K, the overall image quality has been rather poor. They have been showing them since 2006, I believe. NHK is planning for 8K to be the next broadcast standard, not 4K and they intend their official roll-out to be in 2020, so in another 8 years. Who knows... Saw an 8K demo last year and it was the first time it actually looked good to my eyes. But most people won't be interested until they can make a camera that is as compact as the EPIC or ALEXA-M systems and as easy to operate and post.

Then they are already talking about 32k!!! Where is the limit? At a certain point, we will see no lines and on most screens 10' and less. Can u see lines on a 4k images projected 20 feet?

It will happen. There are also new advancements in photon vector inflection capture or whatever. Essentially a technology that could lead to sensors that don't have pixels or defined resolution. But that's serious future tech at this point... There are also other factors to consider, like sensor size. RED previously announced the EPIC 645 and 617 format cameras. They actually had *bigger* photosites than the current M/MX sesnors, but larger sensors. The 617 matched the form factor of 6x17cm panoramic film and was to have a 28K (250MP) sensor. On a 17cm wide sensor, you gain some interesting optical properties that you can't achieve on 35mm or smaller sensors. Focal lengths that are usually considered to be getting into telephoto range (like 90mm) are actually somewhat wide to work with. Additionally, the large format, huge resolution, etc.. make these an intriguing prospect for VFX work, large presentation video walls or whole-building projections. I can think of lots of things a camera like this could be used for. The smaller 645 was in-line with 6x4.5 medium format cameras and I was excited for this one as well -- coming from shooting mostly medium format photography when given the choice. But with a 9K sensor and the larger sensor area, this camera could be used to shoot a variety of formats ranging from 35mm cinema and FF35 frame size up to 65mm cinema and even larger -- it was only 12% smaller than a 15-perf 70mm IMAX frame. At 9K, presumably resolving somewhere around 7.5K nominal resolution, it would definitely give IMAX a run for its money. Actually it would beat the pants off of IMAX because it would mean no 250lb reels of film to manage and equal (or superior) image quality.

So yeah, there are applications for such beasts of cameras. And if someone makes them, you can bet that there are enough creative people out there to find a way to use them!

And maybe they will enable 8K recording on the Sony F65, since it already has a 8K sensor.

Uh, no... Sony can call it whatever they want, but there is no 8K in there. They also have this funny theory that if you tilt the photosite arrangement by 45 degrees, you magically get a better image. So, by their theory, if I tilt my EPIC 45 degrees and then re-crop to a proper view rectangle, I'll have a superior image.

Anyway, I shouldn't pick on them. Other than RED, they're really the only other 4K game in town with a good product. If I couldn't shoot on EPIC, my next choice would be an F65.


Sony does have 8K in their pipeline and the F65, or that series of camera, is intended to accommodate it through future releases and/or upgrades.

 

[Blair S. Paulsen]

K sera, sera...

K sera, sera...

I believe it was Aaton not Arriflex that scooped up the IP and some of the engineers from Dalsa. IIRC, that sensor program was intended to be the core of the Penelope Delta back.

Would love to know how that's coming along after seeing a prototype at PPW at NAB in 2010. RED takes a lot of flak for being "late" but at least the EPIC was a shipping product in 2011 (yes, lots of bits on backorder, but a shootable rig). The Aaton Penelope is a terrific concept and they had elegant solutions designed into the prototype but it's still in the lab, not on sets.

Jim and the Team set such aggressive goals and are so gung ho that the narrative often fixates on delays and backorders, rather than appreciating the comparative speed of their product development cycle compared to other companies in the space.

Cheers - #19

 

[Blair S. Paulsen]

(snip)
We see this quite a bit in a lot of "technical" publication rags like Engadget (I like to pick on them)... To get the 7680 pixels wide that many manufacturers are blasphemously referring to as "8K", you multiply 1920 x 4. To keep the same aspect ratio of 16:9, you also multiply your 1080 vertical number by 4 as well. So you're right x4 horizontally * x4 vertically = 4x4, so "8K" is 16x the information presented with 1080p -- in terms of areal resolution. In terms of linear scale it's x4 on each axis, which is what many people try to reference by, but that's flawed. I have no idea where he comes up with "8 times 1080p". This just reinforces the fact that most "editors" are absolutely terrible at math.

I have started using the megapixel spec many people have become familiar with in the DSLR world.

1080P is essentially 2MP (MP=megapixel), and 4K is 8MP. I find this shorthand effective in dealing with the mathematically challenged ;-)

BTW, one of my favorite jokes is to refer to the state lottery as an extra tax for people who can't do math. (In most US states that have lotteries, the chance of winning the grand prize is far less than being struck by lightning or killed in an auto accident).

Cheers - #19

 

[J. Eric Camp]

And maybe they will enable 8K recording on the Sony F65, since it already has a 8K sensor.

Whoa, I'm gonna have to stop you right there.

The F65 does NOT have an 8k sensor in the "traditional" sense. Someone decided that they could say it has an 8k sensor. An 8k camera (as convention would dictate) with the seemingly new standard of a 1.9:1 sensor aspect ratio would have 33.6 million photosites. (8,000 x 4210) If you went with double the current 4096 for your 8k horizontal (8192 x 4320) would have 35.3 million photosites. The F65 has 17.7 million, about 16 million short. Totally respectable, considering the Epic is 13.8 million, but not 8k as all previous "K" claims have been calculated. The engineers even knew this when they released the first marketing material for the F65 as it stated the sensor was 6k. Only once the Marketing and Sales juggernaught came in did the claim of an 8k sensor come out. Now I'm not saying the camera wont take pretty pictures. I am just calling the marketing speak out. You have a different and new demosaic (for digital cinema cameras) and you have a very respectable 6k e equivalent sensor. No need to hide that. Speaking of their demosaic, they also like to use funny phrases like 1/4 of the photosite dedicated to this pixel. You can't quarter a photosite. Thats marketing talk for, "This output pixel samples from that photosite to determine what its value should be for said color channel." IE what a bayer does. The only difference is that in the Q67 pattern each output pixel gets a dedicated green photosite. Now that photosite is half the size of it's comparable bayer sensor. Call that good or bad, its up to you.

So again at the end of the day, I am not saying anything bad about the camera, just the plie of trumped up words around it.

 

[Manuel Wenger]

NHK has protoype 8k Cameras and is demoing them since at least 3-4 years.
If seen a demo at IBC 2008 or 2009, not quite sure.
It was a 8K projection on a aprox 10m diagonal screen and a 22.1 surround system.

 

[Robin Balas]

Consider the source...
We see this quite a bit in a lot of "technical" publication rags like Engadget (I like to pick on them)... To get the 7680 pixels wide that many manufacturers are blasphemously referring to as "8K", you multiply 1920 x 4. To keep the same aspect ratio of 16:9, you also multiply your 1080 vertical number by 4 as well. So you're right x4 horizontally * x4 vertically = 4x4, so "8K" is 16x the information presented with 1080p -- in terms of areal resolution. In terms of linear scale it's x4 on each axis, which is what many people try to reference by, but that's flawed. I have no idea where he comes up with "8 times 1080p". This just reinforces the fact that most "editors" are absolutely terrible at math.

Well keep in mind that resolution is a one dimensional figure, pixels or dots per length, i.e. inch or mm - like in dpi. There is no pixels per square inch as a resolution figure.
8K have twice the resolution of 4K given everything else equal, as there are twice as many dots along the length. There is no such thing as an area resolution, that is what the megapixels number is used to illustrate.
Twice the resolution is quadruple the pixels.
The concept of resolution is not limited to the motion film industry so I find it hard to accept this 'area resolution' and 'linear scale' concepts often seen here.
I think in this particular case the editor doesn't grasp the fact that 1080p refers to height or lines and the 4K or 8K refers to width or length.
I think it would only confuse more people if one keeps this 'area resolution' term alive - resolution is resolution, nothing new here.
IMHO

 

[Blair S. Paulsen]

Metrics like line pairs per millimeter, or TV lines per picture height are often cited in spec sheets.

Yes, shifting from a vertical to horizontal axis reference in quoting resolution is misleading to say the least.

I am not sure that I would agree that within the motion media industry a one dimensional figure for resolution is always assumed. Due to differing aspect ratios, various imaging technologies (3 CCD vs 1 CMOS with Bayer mask), scans of analog records vs digital clones, etc I think both dimensions need to be expressed for meaningful comparisons.

In terms of the genuine resolution of the F65 it is my understanding that they reached that 17MP number by having 8,192 photo sites horizontal and 2,160 vertical in their Q67 rotated bayer-ish mask tech. After processing, it creates a 4,096 by 2,160 image that is touted as full 4K. I would expect it to resolve fairly close to a 5K FF 5,120 by 2,700 Epic frame de-mosaiced and down sampled to 4,096 by 2,160 (yes, that would include some cropping due to aspect ratio differential). In terms of zone plate resolution testing the 5K FF Epic frame should exceed 3,800 horizontal even in the most conservative evaluation assuming a high quality electro-optical-processing pathway. Even if the F65 manages to exceed that slightly, I don't expect the pure test chart resolution metric to be much of a differentiator between the two systems. Optical low pass filtering, internal processing of the data set provided by the sensor, shutter artifacts, compression, decoding algorithms, etc are much more likely to delineate the "look" and perceived resolution of the respective systems. I am expecting something like the difference between LCOS and DLP, but until we can directly compare in a butterfly (split screen) on a serious display that's just conjecture.

If you are talking about an 8K system like the NHK UHD implementation, that's a different beast. FWIW, if/when NHK's 8K standard reaches widespread use I would expect 4K digital material and high res scans of modern film stocks (particularly EI 250 or less) to look quite nice, particularly with next generation uprez tech. OTOH, I would expect 1080 material to require extremely effective uprez routines to avoid looking soft. Time will tell...

Cheers - #19

 

[roryhinds]

Does anyone know the physical sizes dimensions of the RedOne MX sensor & the Epic MX sensor?

 

[Blair S. Paulsen]

I am pretty sure it is the same chip but the Epic can use more of the real estate.

A search should lead you to a graphic showing the precise dimensions at various settings.

The 5K FF 5,120 by 2,700 option on the Epic is the biggest bite to date. I seem to remember a 5.6 micrometer pitch being noted but I would not put money on it.

Cheers - #19

 

[Will Keir]

If you are talking about an 8K system like the NHK UHD implementation, that's a different beast. FWIW, if/when NHK's 8K standard reaches widespread use I would expect 4K digital material and high res scans of modern film stocks (particularly EI 250 or less) to look quite nice, particularly with next generation uprez tech. OTOH, I would expect 1080 material to require extremely effective uprez routines to avoid looking soft. Time will tell...

Cheers - #19

So does that mean we can expect old movies shot on film to be available in 4k+ sometime? How far back can we go and "re-post" the film to bring our old favorites into the modern age?

I am surprised going back and watching films I enjoyed 15 years ago with how low their quality looks. The films are still great, timeless but will they get a refresh or the potential to have a refresh? I'm sure the lower quality I see in "Braveheart" is due to compression issues with the DVD. When Braveheart screened in the theaters, what resolution, if any, were we seeing then? I actually never did see it in the theater, too young to take myself. Does film work different in the sense that there are no lines to be seen in the technology?

 

[Blair S. Paulsen]

If the original film elements were well preserved and the stock wasn't high in grain most films over the last 50 years could be re-mastered at 4K with a meaningful improvement over 2k (1080). Something like the 6K Northlight, combined with well designed algorithms and plenty of careful processing should be able to pull enough real detail into the re-mastering process for 4K-sih results. My friend Garret over at Paramount has done lots of remastering from archive elements and could give you a far more detailed account of the process and the challenges therein.

Be aware that there are lots of potential issues that could severely limit the perceived difference between film A re-mastered at 4K and film B. Not least of which is pixel count and bit depth of VFX plates, compositing techniques, ISO (ASA) of the original stock, DI pipeline for the re-mastering, final output format (DPX vs JPEG2000), filtration used in front of the taking lens, the maximum lp/mm of resolution the lens is capable of at the stops used (My guess is that GWDT went Master Primes due to their performance wide open and their desire to DI at 4K), post cropping, etc.

Looking forward to seeing for myself and hopeful that at the very least the 4K DCP spec is used for distribution. IIF/ACES 16bit OpenEXR during DI, etc.

Cheers - #19

 

[Jeff Kilgroe]

Well keep in mind that resolution is a one dimensional figure, pixels or dots per length, i.e. inch or mm - like in dpi.

It depends on semantics. Resolution can be measured in varying dimensions and often is, depending on what is being measured. When speaking of 2D images and imagers, linear representations are the most straight-forward. If anyone wants to know the areal resolution (yes, that is a real term) or areal density, they can derive it. And yes, in terms of imaging systems, that is where megapixel counts come into play as you point out.

Were you tend to see multi-demensional and/or non-linear resolution qualifiers and quantities is when dealing with organic or non-uniform structures where linear representations may not work. None of which really applies here unless you're trying to compare dispersion of wavelength reactants within film negatives or something.

I think in this particular case the editor doesn't grasp the fact that 1080p refers to height or lines and the 4K or 8K refers to width or length.

Yep, that's what it looks like.

I think it would only confuse more people if one keeps this 'area resolution' term alive - resolution is resolution, nothing new here.
IMHO

I think the reason it keeps coming up is because many people don't understand the relationship between the linear representation and the areal quantities. When speaking in common linear terms and we say "4K is twice the resolution of 2K", we are right in terms of that linear scale and also by making the correct deduction that the quantities are uniform (pixels are square and remain constant between formats), because we know this to be so. However, we see far too often a great many people who incorrectly assume that 4K therefore has twice as many pixels as 2K and only requires twice as much bandwidth, storage, etc.. I think that's why the areal representation gets driven home so much.


As for the F65 resolution numbers, it will be interesting to see how the camera actually resolves. Supposedly the first one, serial #001, is being delivered TODAY!!!

 

[geoffboyle]

I believe it was Aaton not Arriflex that scooped up the IP and some of the engineers from Dalsa. IIRC, that sensor program was intended to be the core of the Penelope Delta back.

Would love to know how that's coming along after seeing a prototype at PPW at NAB in 2010. RED takes a lot of flak for being "late" but at least the EPIC was a shipping product in 2011 (yes, lots of bits on backorder, but a shootable rig). The Aaton Penelope is a terrific concept and they had elegant solutions designed into the prototype but it's still in the lab, not on sets.


Cheers - #19

Aaton have always delivered late but they have also always delivered amazing kit.

I've been using their cameras for over 30 years and can't wait for Penelope D

 

[Patrick Grossien]

I believe it was Aaton not Arriflex that scooped up the IP and some of the engineers from Dalsa. ...
Cheers - #19

What about Deanan?

 

[Mark Andersen]

On the 4K vs. 2K things, yes there are four times the pixels and four times the real estate. But it could be argued that it is only twice the resolution or better yet, only twice as "sharp" because when you see the image you see two pixels were you used to see one, or twice and much info in any part of the image, one more line both vertically and horizontally than there used to be. Twice as sharp or clear, but four times the pixels. So in theory 8K would be twice as sharp as 4K. IMHO

 

[Steven Caesare]

I have started using the megapixel spec many people have become familiar with in the DSLR world.

1080P is essentially 2MP (MP=megapixel), and 4K is 8MP. I find this shorthand effective in dealing with the mathematically challenged ;-)

BTW, one of my favorite jokes is to refer to the state lottery as an extra tax for people who can't do math. (In most US states that have lotteries, the chance of winning the grand prize is far less than being struck by lightning or killed in an auto accident).

Cheers - #19

Well, don't shortchange 4K...if you are gonna quote the 16x9 ratio pixel specs for 1080P @ 2MP, then 4K is really more like 9.5MP.

Sounds more impressive too! :wink5:


-sc