Do we have more pixels than we ever needed?

[ErikKaffehr]

Hi,

I love pixels, and want to have enough of those to outresolve my lens: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/78-aliasing-and-supersampling-why-small-pixels-are-good

But, Zacuto has tested several single chip video cameras, including Arri Alexa, RED and an iPhone 4S and they found the iPhone S was actually good enough to be shown at a movie theatre. No question, the real cameras run circles around it, but with good lighting it was almost good enough.

Now, what my article shows is that small pixels are needed for good rendition. I shoot 24 MP on DSLR and 39 MP on MFD. Both are on the short side of what I would call proper rendition. But, I seldom print to make those images justice. One reasons is that those images are not good enough. Both 24MP and 39MP look great at A2 size, which is what I normally print. I would like to print larger, but wall space is limited. Very few images get printed larger than A2, and those images are often stitched panoramas.

A nice way to present images would be a 4K screen on the wall. Images can be switched, so I can show many images on limited wall space, nice! But 4k is just 8MP, OK 9.8 MP if we take form factors into account. So around 10 MP is what we actually can show, except in large prints intended to viewed close.

To make justice to modern day cameras it is more like 8K needed. Now, 8K may be around the corner but it may take long time to arrive.

My next investment may not be a new camera, but a new projector at 4K. It will not make my images justice, but it will be far better than the 2K projection I have right now.

For the cost of a 4K projector I would be able to make 80-120 70x100 cm prints. But wall price is even higher than 4K projection costs.

Best regards
Erik

[Gulag]

You ain't seen nothin yet. NASA has been using Giga-Pixel cameras for a little while by now and they want to go Tera-pixel in a few years.

[Petrus]

But, Zacuto has tested several single chip video cameras, including Arri Alexa, RED and an iPhone 4S and they found the iPhone S was actually good enough to be shown at a movie theatre.

We have been watching 600 to 800 lines of vertical resolution for half a century without complaints. That is the real resolution of 35mm movie projection in a good theatre.

http://www.motionfx.gr/files/35mm_resolution_english.pdf

[AlterEgo]

To make justice to modern day cameras it is more like 8K needed.

8k = ~  7680 × 4320 = 33mp x 3 RGB light emitting elements, so w/o interpolation you might want a ~100mp (actually 120mp, 2 "greens" combined together in RG1BG2 bayer) sensel bayer camera... w/o interpolation... to do a real justice, no ?

[Telecaster]

Quoting self from a few months ago:

Just personally: I no longer care much about printing. I still do it occasionally but it's not the focus of what I do. I am interested in making photos with 4K display in mind, though, and particularly in taking control of the R, G & B sub-pixels of each full-color display element. A nice thing about this interest is that it imposes some hard limits. A 4096x2160 image, if you provide non-interpolated RGB data for each display pixel, requires a Bayer sensor with 8192x4320 photosites (an R, G & B for each display pixel, with the two Bayer Gs averaged in some manner). That's 35.4mp.

I personally doubt that at a ~50–60" display size, viewed at typical TV-watching distances, going from 4 to 8K will make enough of a visible difference to be worth the bother. We may get 8K at some point anyway, of course, but let's see how 4K does first…

Anyway there's no harm in oversampling at the camera end, so bring on the mps. 

-Dave-

[ErikKaffehr]

Hi,

No reason to watch either 4K or 8K at typical TV-watching distances. The company I am working with now is testing the use 4K as a common computer display. Before that we typically used 4x24" screens stacked. Optimal viewing distance for computer screens is probably around 80 cm.

What I have on mind is replacing say A2 size prints with high resolution screen. A quite often used criterium is that 180 PPI is needed for a really good print. Would you have say a 20x30" screen, the needed resolution would be 3600x5400 pixels which is 29 MP.

The reason I mention 8K is that it is in the pipeline. NHK is planning to broadcast the Tokyo Olympics 2020 in 8K.

On the other hand, my primary interest may be projection, where I find today's HD lacking, and 4K projector prices are a bit above my acceptance level.

A final point is that we are of course going to show todays images on tomorrows media.

Best regards
Erik

Quoting self from a few months ago:

I personally doubt that at a ~50–60" display size, viewed at typical TV-watching distances, going from 4 to 8K will make enough of a visible difference to be worth the bother. We may get 8K at some point anyway, of course, but let's see how 4K does first…

Anyway there's no harm in oversampling at the camera end, so bring on the mps. 

-Dave-

[ErikKaffehr]

Hi,

Screen resolution tend to increase with a factor 2. We had HD, next is 4K and NHK will send the Tokyo Olympics in 8K.

Getting back to your reasoning, I don't agree. Clearly there are cases where a subject is single colour, like a red flower and such a colour will be resolved at 25% of the nominal resolution. The lack of colour resolution can create artefacts with high resolving lenses. Good reasons to go to small pixels.

On the other hand most colours are not pure, OLP filters help eliminate colour artefacts and demosaic algorithms are good at interpolating luminance.

Now, human vision is not very good at discriminate colour but is very sensitive to luminance. Try yourself. Convert an image to LAB, duplicate the layer, select "a and b" channels and apply say z a median filter of size of 3. Try to flip that layer on and off, are you able to a difference? Now do the same experiment on the L channel. Great loss of sharpness will result.

Best regards
Erik

8k = ~  7680 × 4320 = 33mp x 3 RGB light emitting elements, so w/o interpolation you might want a ~100mp (actually 120mp, 2 "greens" combined together in RG1BG2 bayer) sensel bayer camera... w/o interpolation... to do a real justice, no ?

[BernardLanguillier]

Yep, we have known for years that the days of paper are counted.

Cheers,
Bernard

[Telecaster]

No reason to watch either 4K or 8K at typical TV-watching distances. The company I am working with now is testing the use 4K as a common computer display. Before that we typically used 4x24" screens stacked. Optimal viewing distance for computer screens is probably around 80 cm.

Yeah, 80cm is about my viewing distance for my 22" monitors (I use two, typically splitting them between UI and image data). But for a 50–60" screen? A bit too close IMO.    Having seen how good 4K video downsampled to 2K display can look I'm all for 8K capture whether or not we eventually get 8K display.

-Dave-

[RSL]

36 mpx in my D800 can be overkill. If I'm doing the kind of thing I used to do with my old view cameras I love the D800, but if I'm shooting the dress rehearsal for a play, I don't need 36 mpx, and the 12 mpx D3 is my camera of choice. There are specialized situations where the sky's the limit on desired resolution, but not in everyday photography.

[Hans Kruse]

Next step is 8K displays, but not sure when they will be moving to market at a price people will pay.

[BJL]

Next step is 8K displays, but not sure when they will be moving to market at a price people will pay.

I am puzzled by the 8K video development, beyond the "because we can" rationale of some engineers and marketing people.  What, if anything, is the evidence that viewers actually perceive a difference between 4K and 8K?

In the document https://pro.sony.com/bbsccms/static/files/mkt/digitalcinema/Why_4K_WP_Final.pdf Sony is arguing the advantages of higher resolution projectors (4K vs 2K in this case), including looking at the extreme requirement of matching what "better than 20/20" vision can make out when carefully, slowly viewing a high contrast still image (eye test chart), which is a more demanding requirement than with moving images (I know the reply: "some people watch movies of still text and examine it closely"!).  And even from that commercially motivated advocacy of higher res. projectors, and then replacing its "2K vs 4K" by "4K vs 8K" it seems that only sharp-eyed people sitting in the first one or two rows of a cinema, or people siting as close to their big-screen TVs as the _height_ of its screen and less half the screen width would perceive any imperfections in 4K, and even th front rows look to be handled perfectly by 6K at most.

What is more, even if there is an argument that a few people can benefit from "a bit more than 4K", the useful threshold might be at some intermediate level like that 6K. It might be easiest technologically for cinema projection equipment to advance by doubling from 2K to 4K to 8K, but with content only ever needing to be recorded and processed at some intermediate level like 6K, and then up-ressed to match the display hardware.  Note that my example of 6K video could be got as crop from a mere 24MP sensor in 3:2 format.

[davidedric]

I don't think I have 8k eyes 

[Ray]

I have a 65" plasma HDTV which I often use to display my still images down-sampled to HD resolution, about 5.9MB.

As a result of my own testing, I find that in order to see all the detail displayed in these 2k images, I have to sit at a distance from the screen of no greater than  2.5 metres, wearing appropriate spectacles for the distance.

I'm interested in the new 4k UHD displays, because I love detail and resolution, but I can't see the sense in getting a UHD screen so small as 65" diagonal. Surely in order to appreciate the extra resolution that 4k provides, I would have to sit even closer than 2.5 metres from such a screen, say about 1.25 metres.

From a viewing distance of 2.5 metres, I would need a 130" UHD screen to enable the appreciation of all the detail in a 4k still image. If that 130" screen could display 8k, I would be back to a minimum viewing distance of 1.25 metres. Is this not the case?

The Samsung 110" UHDTV seems close enough to the right size for viewing 4k material, but it's a bit expensive at $150,000. 

[Colorado David]

A few years back, a couple produced a movie about a husband and wife that were lost at sea while scuba diving on a dive trip when the dive boat miscounted and left the area, accidently leaving them there to die.  I can't remember the name of the movie.  It was shot on a Sony PD150 Standard Def NTSC DVCam camcorder.  It had a pretty broad theater release and was fairly successful as I recall.  Standard Def DVCam would be 720x480.

[Diego Pigozzo]

...It was shot on a Sony PD150 Standard Def NTSC DVCam camcorder....

For moving images, the all HD/FULLHD/ULTRAHD/GAZZILLIONK fuzz is just that: fuzz.
The human retina has "high resolution" only in the central part of the field of view, so having high resolution on all the movie frame is physiologically useless.

[bjanes]

For moving images, the all HD/FULLHD/ULTRAHD/GAZZILLIONK fuzz is just that: fuzz.
The human retina has "high resolution" only in the central part of the field of view, so having high resolution on all the movie frame is physiologically useless.

That assumes that the viewer fixes his view on only the central portion of the screen. However, what if she directs her gaze at the edge of the screen?

Regards,

Bill

[Diego Pigozzo]

That assumes that the viewer fixes his view on only the central portion of the screen. However, what if she directs her gaze at the edge of the screen?

Regards,

Bill

Still, having high resolution on the full frame would be physiologically useless: the viewer would see high resolution only on the edge.

[Isaac]

Still, having high resolution on the full frame would be physiologically useless: the viewer would see high resolution only on the edge.

Wikipedia: Saccade

[BJL]

Wikipedia: Saccade

Indeed: as I understand it, our fine visual resolution depends in part on using saccade to detect edges, and that ability is at its peak for the detection of high contrast edges, leading to the peak angular resolution as measured with eye charts.

Moving images do not allow this, which is one reason that we have distinctly less resolution of moving images than of stills: note how bad the resolution looks on still frame grabs compared to the moving image they are grabbed from.

That is why I am betting that the difference between 4K and 8K video will only be noticed:
a) When reading the fine print at the tail end of movie credits.
b) In Warholesque movies with long still shots of eye charts and other finely detailed, high contrast subject matter.
c) When used to display still images and then viewing them from a distance of about half or less the width of the screen.