Why 4K?

[Martin Stevens]

David... nothing delivered at 1080P measures full 1080P resolution either. So the argument needs to be "what does the industry consider enough resolution packaged in a 1080P (or 4K) wrapper that "is good enough" in terms of percentage of output resolution.

My opinion, not that it is worth anything, is that roughly 80% of packaged resolution is probably good enough. That means for 4K and that means for 1080P... as these specs have already been adopted for 1080P.

So what qualifies at these specs for 4K? Film (4K scan) and RED (so far).

Jim

While I agree with the the principle idea that more resolution is always preferable....as in 3.2K Red One resolution looks fantastic when projected 4K, there will always be the rare exception, such as "The Blair Witch Project" type of low resolution movies.

One of the real beauties to me of high resolution acquisition, is that if you ever wanted to lower the resolution in post to say simulate old 8mm movie film you could.

 

[Julian Gerretsen]

This is primarily a photographic article but worth a read:

http://www.kenrockwell.com/tech/mpmyth.htm

My personal opinion (for what it's worth), is that 4K is, generally, probably going to be around the maximum general delivery format that we are ever going to need (in general*)…. (generally speaking, that is).

4K is around 9.4MP (assuming a 16:9 frame): anyone with even a 6MP quality digital SLR and good lenses will know that you can get stunning images. I regularly blow my 6MP Nikon Raw's up to 30X20 inches with incredible results and no visible pixelation, even under close scrutiny.

There is a limit to what our human eyes and camera lenses can resolve. Honestly, I have really had to look hard to see any individual pixels in a 2K projection and generally only off CGI type animations that are very very sharp. For the most part, where stuff has had to pass through a lens first before arriving on the sensor, it is very hard to pick up. Maybe it's just my codgerly 38 year old eyes…

Yes if you are sitting 5 feet from even an 8K projection that is 100 feet wide, you are going to see pixels. But to see the whole screen, you are going to have to sit pretty far back, without having your brain fried by all the movement.

I have a cinematography book (Davids' is better) by a British cinematographer (Paul Wheeler) which came out just before digi intermediate's were starting to become the norm. Paul spoke about the final resolution of a release print (from 35mm) being equivalent to about a 1.3K image because of the degradation along the way through inter-negs and lots of optical printing lenses etc. Even though you were starting out with around 4K on your negative, you lost a lot along the way optically. So even a 3K (true) off 4K RED is a vast improvement on traditional film prints.

Acquisition on the other hand is another story.... Ignoring the limitations of the resolving qualities of glass, higher resolutions in acquisition are obviously going to give you more to play with downstream, not just in resolution, but also in re-framing options.

As an aside to that (sorry, this paragraph delves into the ancient history and lore of HDV), I have always been astounded at how much I can reframe HDV that is being re-delivered to SD. For any SD work (less and less nowadays of course) involving slow zooms, I don't even use the zoom on the camera anymore as I can add it in post without losing any quality and keeping all my options open. I've even been able to "cut" (i.e. choose another part of the frame) on exactly the same scene using re-framing without any jump-cut problems... that's how much you can move things around.

If you are delivering 2K off a 4k (3? true) Red file, you also have many reframing options. Higher acquisition resolution would of course open those options up massively. Nice for those dramatic slow zooms too.

Imagine then the time when RED gives us 8K or even higher acquisition resolutions (and the lenses to take advantage of it)… and we put those out at 4K delivery. Imagine the possibilities of reframing, stabilising, cropping etc. that that would open up?

* Of course, there are always very highly specialist areas like Imax that will require more resolution, but they will always be highly specialist areas requiring highly specialist equipment. I think.

 

[Rick Burnett]

To make an analogy with audio recording, I do everything in 24-bit regardless if I know the final destination is going to be 16-bit. I just like the extra dynamic range to work with.

But, I do not record higher than 48kHz. My equipment will do 192kHz and I have done plenty of tests on my high end monitors and don't feel the difference in quality (which I cannot perceive any personally) is worth the absolute hit on hardware performance / disk space. 6 streams of 192kHz mono audio for 5.1 is quite a bit compared to 6 at 48kHz.

Where I am going with this is that I look at what I consider the 99% mark in terms of decisions of where I push the technology and where the benefit comes from in the audience. I can hear the difference between 24-bit and 16-bit on extremely dynamic music, like finger picked guitar with strumming and such (not so much on dance music).

With video, I absolutely see the resolution being more analogous to audio bit depth in terms of information density. Video has multiple bit depth dynamic ranges and one rate range (meaning, audio rate ie 192kHz is analogous to video frame rate). ie each pixel has a dynamic range of RGB, and then the sensor size has a dynamic range in terms of X & Y (or pixels per mm across the range of the sensor).

For this reason, I absolutely see where Red is going with shooting higher and higher resolution per the same field of view if you will because in the end, it allows you to not only move down to what is available today for 99% of the public, but means in the future when standards are closer to the native resolution, or we've moved passed it, the footage will stay relevant longer.

But it's even more than that. For the same reason that 32-bit or 64-bit audio has shown that the quality is indistinguishable for the average person compared to 24-bit, we've not approached this I feel in video. I've looked at projections at installed sitting positions with 720p and 1080p and I could see a difference in quality. I feel though when we do get to 4k, a lot of what is after that for the home is going to be indistinguishable from anything higher resolution from nominal sitting positions. But for public locations, it's still going to be noticible especially when you are closer to the screen. When I get close to a film projection, I see blur. When I get close to a digital projections, I see large blocky pixels. In this case, I'd prefer the blur because the pixels add attention to themselves (DLP is bad for this).

So essentially, what we are doing is really helping the people who are stuck in the front row because they got to the theatre late!

 

[Graeme Nattress]

Another factor to think of is the immersive nature of a screen that fills your total field of view, and that means you do need more resolution than you'd for a normal cinema screen and viewing distance.

Graeme

 

[Rick Burnett]

This is primarily a photographic article but worth a read:
I've even been able to "cut" (i.e. choose another part of the frame) on exactly the same scene using re-framing without any jump-cut problems... that's how much you can move things around.

And ABSOLUTELY this is one of the BEST things about shooting higher resolution than what you need to use. Good point!

 

[Jeff Kilgroe]

This is primarily a photographic article but worth a read:

http://www.kenrockwell.com/tech/mpmyth.htm

My personal opinion (for what it's worth), is that 4K is, generally, probably going to be around the maximum general delivery format that we are ever going to need (in general*)…. (generally speaking, that is).

And that's what people kept saying about 1080p... And many still do.

Let's put it this way, until the technology progresses to where I can look at a TV and not be able to tell that I'm not looking out a window, it isn't good enough. We can speculate and speak academically about resolutions and how much the human eye can resolve all we want, but I have yet to see anyone put any real numbers on it that actually make sense. Our eyes, and the image processing "algorithms" in the human brain, are incredibly advanced and very capable of determining fine details, even detail that is finer than the highest resolution our eyes should technically be able to resolve. A highly simplistic example is being able to see a white speck from a distance on a dark-colored wall, even though that speck may be smaller than a single pixel on an LCD display that is supposed to be unnoticeable from a distance half as far away as that wall. Or being able to see individual strands of hair at a distance, etc.. Much the same way many people can see the pitch or gap between pixels on an HDTV, even though these pitch lines are smaller than the pixels.

That being said, resolution is only one facet of many to make an image. Resolution isn't everything. I would much rather watch a movie with a very high quality transfer to DVD than a crappy transfer to Blu-Ray. And crappy Blu-Ray transfers are abundant right now.

There are emerging technologies out there right now that show promise of recording photon structures, not just energy strikes into a receptor well (photosite pixel) as we have now. It's too early on, but leads to theoretical sensor designs that may reproduce images without being bound to pixels or specific resolutions at all.

 

[Martin Stevens]

But, I do not record higher than 48kHz. My equipment will do 192kHz and I have done plenty of tests on my high end monitors and don't feel the difference in quality (which I cannot perceive any personally) is worth the absolute hit on hardware performance / disk space.

I always record at 192kHz when I know I need to slow the sound down later
in post for a special effect. 192kHz is roughly like a 30ips reel-to-reel tape recorder.

 

[Rick Burnett]

I always record at 192kHz when I know I need to slow the sound down later
in post for a special effect. 192kHz is roughly like a 30ips reel-to-reel tape recorder.

I could certainly see that, and would in fact do the same if needed as many time-scaling algorithms produce horrible sounds!

Same thing with video which is why I am excited about the frame-rates of the 2/3 brains!

And pertaining to resolution, absolutely, there are SO many other factors involved. Seeing will be believing. I'm already ready to invest in 4K in my future Red purchases. The question for me isn't 'Why 4k' it's 'When 4k!'

With video there will be a point of diminishing return on investment on all technology, but as Jeff stated, there's a lot of theoretical discussion. I don't think we've reached that point and even with my opinion of 4k being the max of what consumers will want/need at most homes, I could be wrong depending on how not only the images look, but all the other technologies other than resolution that get better at the same time.

It's important to remember that in many cases technologies become adopted because of other technology advances that 'come together' if you will (not that they need to be, but that they end up happening that way).

 

[Jannard]

One last point... IMHO, the easy rule should be "if it doesn't look any better at 4K than it does at 1080P... it shouldn't be released in 4K".

Pretty easy stuff.

Jim

 

[Jeff Coatney]

The market has historically preferred a higher resolution in cinemas than the consumer can get in the home. Using this established metric, 1080p acquisition will not cut it for theatrical release. The success of the IMAX exhibition model, with it's higher per-screen average box office underscores this fact. Consumers want an experience they cannot get in the home environment. Even if 3D and 4K exhibition technologies propagate through to the home, cinemas will push past these technologies in order to offer consumers a premium experience in a communal viewing environment. This debate about resolution has migrated past what the filmmaker has available to them and has now moved into the realm of what audiences will pay a premium to get. It's no longer what is technically best for the existing paradigm, its now come to what is best for the future. With digital projection, the entire exhibition model can now change rapidly to adopt higher resolution formats. In the past, this was not possible because exhibition was tied to the huge infrastructure required to mass produce prints. By taking the analog print out of the distribution equation, cinema owners are free to innovate in order to provide new exhibition models. Once you traverse the digital Rubicon, the word "standards" as we've come to know it, goes out of the picture-- literally. In fact, once tape-based workflows have fully vanished from the market, digital acquisition can then expand its resolution beyond 1080 and take advantage of higher data rates. 1080 will be with us for a very short time-scale, it is a transitional format.

 

[David Mullen ASC]

So if we are saying that 4K scans of 35mm film and 4K RAW Red One footage, even though both measure closer to 3K, are future-proofed for 4K projection, then where is the resolution dividing line? 3K? 1080P is near 2K, but that's not good enough, but 3K is good enough? Is that what we are saying?

I just don't see the future in such black & white resolution terms. All it takes is a 4K RAW feature to shoot with a diffusion filter to suddenly not make this "3.2K" resolution threshold. Do you really think that "The Informant!" shot with a Double-Fog filter would measure out at 3.2K? What about "Che: Part 1" and "The Girlfriend Experience" shot with 2X anamorphic lenses, meaning that only around 2.8K RAW was really used for 2.40 after cropping, meaning probably the horizontal resolution measured would come in closer to 2.2K? So all of these Soderberg movies are not good enough for this 4K revolution??? This is why it gets silly to be absolutist in regards to resolution and not take real-world issues into account.

 

[Jannard]

So if we are saying that 4K scans of 35mm film and 4K RAW Red One footage, even though both measure closer to 3K, are future-proofed for 4K projection, then where is the resolution dividing line? 3K? 1080P is near 2K, but that's not good enough, but 3K is good enough? Is that what we are saying?

I just don't see the future in such black & white resolution terms. All it takes is a 4K RAW feature to shoot with a diffusion filter to suddenly not make this "3.2K" resolution threshold. Do you really think that "The Informant!" shot with a Double-Fog filter would measure out at 3.2K? What about "Che: Part 1" and "The Girlfriend Experience" shot with 2X anamorphic lenses, meaning that only around 2.8K RAW was really used for 2.40 after cropping, meaning probably the horizontal resolution measured would come in closer to 2.2K? So all of these Soderberg movies are not good enough for this 4K revolution???

Good question... my answer is not necessarily what the industry will adopt. But typical measured resolution of 1080P output is between 1.5K and 1.8K. 1.5K is 80% of 1080P. Just using this current "standard" (although no one has actually defined it as such)... it makes sense to me that a similar intent would make sense for 4K. One thing I do know... 1080P at 4K delivery will look no better than with a 1080P delivery.

Jim

 

[christopher witzke]

This is primarily a photographic article but worth a read:

http://www.kenrockwell.com/tech/mpmyth.htm

My personal opinion (for what it's worth), is that 4K is, generally, probably going to be around the maximum general delivery format that we are ever going to need (in general*)…. (generally speaking, that is).

4K is around 9.4MP (assuming a 16:9 frame): anyone with even a 6MP quality digital SLR and good lenses will know that you can get stunning images. I regularly blow my 6MP Nikon Raw's up to 30X20 inches with incredible results and no visible pixelation, even under close scrutiny.

There is a limit to what our human eyes and camera lenses can resolve. Honestly, I have really had to look hard to see any individual pixels in a 2K projection and generally only off CGI type animations that are very very sharp. For the most part, where stuff has had to pass through a lens first before arriving on the sensor, it is very hard to pick up. Maybe it's just my codgerly 38 year old eyes…

Yes if you are sitting 5 feet from even an 8K projection that is 100 feet wide, you are going to see pixels. But to see the whole screen, you are going to have to sit pretty far back, without having your brain fried by all the movement.

I have a cinematography book (Davids' is better) by a British cinematographer (Paul Wheeler) which came out just before digi intermediate's were starting to become the norm. Paul spoke about the final resolution of a release print (from 35mm) being equivalent to about a 1.3K image because of the degradation along the way through inter-negs and lots of optical printing lenses etc. Even though you were starting out with around 4K on your negative, you lost a lot along the way optically. So even a 3K (true) off 4K RED is a vast improvement on traditional film prints.

Acquisition on the other hand is another story.... Ignoring the limitations of the resolving qualities of glass, higher resolutions in acquisition are obviously going to give you more to play with downstream, not just in resolution, but also in re-framing options.

As an aside to that (sorry, this paragraph delves into the ancient history and lore of HDV), I have always been astounded at how much I can reframe HDV that is being re-delivered to SD. For any SD work (less and less nowadays of course) involving slow zooms, I don't even use the zoom on the camera anymore as I can add it in post without losing any quality and keeping all my options open. I've even been able to "cut" (i.e. choose another part of the frame) on exactly the same scene using re-framing without any jump-cut problems... that's how much you can move things around.

If you are delivering 2K off a 4k (3? true) Red file, you also have many reframing options. Higher acquisition resolution would of course open those options up massively. Nice for those dramatic slow zooms too.

Imagine then the time when RED gives us 8K or even higher acquisition resolutions (and the lenses to take advantage of it)… and we put those out at 4K delivery. Imagine the possibilities of reframing, stabilising, cropping etc. that that would open up?

* Of course, there are always very highly specialist areas like Imax that will require more resolution, but they will always be highly specialist areas requiring highly specialist equipment. I think.

Oh please don't reference from Ken Rockwell.... he is the Chance the Gardener of the photographic web community.

More is always going to be more no matter what math tries to prove it wrong.

 

[Jannard]

Also remember that 3K is 400% more resolution than 1.5K. There is quite a bit of elbow room to work with here.

Jim

 

[Jeff Coatney]

I'm saying that there is no longer a "good enough". Resolution will increase continually until it reaches an as yet to be determined plateau.

I have another prediction for you that will blow your mind: movies acquired on film will be remastered to a beyond 16K resolution using largely automated CGI to basically re-shoot the negative. For each shot, the software will rebuild the set or the location, sythesize the actors, and replicate the camera movement in order to re-render it all at higher resolution-- or actually move into the film in order to view it from any angle. Imagine directing Casablanca from the comfort of your sofa. We'll see this begin to emerge within 20 years. It already exists forensically. Soon it will be utilized to re-monetize classics.

 

[Jeff Coatney]

My "soon" is not the same as your "soon".

 

[Julian Gerretsen]

Oh please don't reference from Ken Rockwell.... he is the Chance the Gardener of the photographic web community.

More is always going to be more no matter what math tries to prove it wrong.

Oh right. I wasn't aware of that. The article seemed to have a reasonable enough premise and links to quite a few others and the general consensus in all of them seemed fairly reasonable, that there is an upper limit to what our eyes can perceive in general viewing conditions (i.e. viewing a picture from a normal distance or viewing a movie in a normal theatre - non Imax etc.).

I guess that I didn't explain myself very well in trying to say that we are all trying to find the balancing point between maximum resolution that is actually useful vs cost of acquisition, editing and delivery, and the fact that a lot of that has to take place within the current framework. E.g. What is the maximum that 35mm lenses can resolve usefully without completely changing your system to a larger format of sensor, lenses, body, everything. 35mm film experience would tend to suggest around the 4K mark.

On the delivery side, the cost of implementing even 2K (quality) projection systems for smaller independent theatres is in many cases almost out of reach (in the UK anyway, it might be a better situation in the US). How many will be able to install 4K equipment? Or 8K? Our local multiplex is still only running 2 out of 12 on digital projection (OK, it is North Yorkshire, where sheep outnumber people 50:1 and tractors are classified as SUV's). There are always the exceptions and specialisations like Imax, but for the more "normal" distribution channels, what is realistic in the near future?

 

[M Most]

Good question... my answer is not necessarily what the industry will adopt. But typical measured resolution of 1080P output is between 1.5K and 1.8K. 1.5K is 80% of 1080P. Just using this current "standard" (although no one has actually defined it as such)... it makes sense to me that a similar intent would make sense for 4K. One thing I do know... 1080P at 4K delivery will look no better than with a 1080P delivery.

Jim

I think we're confusing resolution with sharpness. Resolution is a measurable characteristic of a particular optical, imaging, and delivery system. Sharpness is a metric that is not as easily measurable because it primarily relates to human perception. Sharpness can be exaggerated, positively or negatively, by various techniques, such as increasing or decreasing edge contrast and other techniques. It is also highly influenced by optical devices, such as lenses, filtration, and other items in the optical path of both the capture system and the delivery system. One can easily have a "soft" image on a 4K delivery system, and one can just as easily have a very sharp image on an HD delivery system. Which in many ways supports David's view that "resolution", at least as a measurable metric, only relates to devices, not to the material they happen to be capturing or displaying, and is therefore not something you want to so rigidly attach numbers to. Measurable resolution, perhaps. Perceived resolution, no. It is for this reason that studios are looking at archiving using 4K formats - not because it creates a sharper image in and of itself, but because there is less loss in terms of retaining any information on the original, making it a more faithful archive. It has nothing to do with sharp vs. soft, but it has a lot to do with avoiding loss.

 

[Rick Burnett]

One thing I do know... 1080P at 4K delivery will look no better than with a 1080P delivery.

Jim

Will 1080p material shown on a 4K system look worse? (And by this I mean, upscaled by the display system, not upscaling through previous rendering to a 4K format).

On many systems, consumer level, any conversion in the hardware isn't as good. I have no 4K display system experience, but many of the algorithms employed in hardware create some awful results to my eyes.

Just curious.

And the truth is, with 128K projection, there's still gonna be some horrible quality out there

In the audio world I come across people who are absolutely gear whores (and I am guilty of this with synths), but they are always spending $10k+ on boutique amps and such. I remind them, a lot of AMAZING music was made on equipment that cost 1/100 as much. Sometimes the artistic focus gets lost in the technical battle!

I will say this, with Red making 4K more affordable for the independent filmmaker, it's going to make the push to 4K easier if a lot of people already have their content as such!

 

[christopher witzke]

Let's bring back the word " Fidelity " !!