5 ms·

I recall in the 90s when they said a 640x480 image was the equivalent of a 3x5 photo print. I think it's naive to assume our vision can be diluted down to a fi

I recall in the 90s when they said a 640x480 image was the equivalent of a 3x5 photo print. I think it's naive to assume our vision can be diluted down to a fixed resolution at a given distance. No, I cannot see the pixel grid of my 55in 4k OLED TV at a viewing distance of more than about a foot. However I can see a single pixel illuminated at about 25ft. So to say there is not benefit to 8k is like saying we'll never need more than 640k of RAM.

ajross 8y ago
> I think it's naive to assume our vision can be diluted down to a fixed resolution at a given distance. You have an alternative theory of vision where resolution is not limited by photoreceptor density? Look, there are only so many cells on your retina. > However I can see a single pixel illuminated at about 25ft. And you can see a star (which subtends a MUCH smaller angular size than that mere pixel) at hundreds of light years. Yet it takes a skyscraper-sized telescope and mindbendingly crazy interferometric and occlusion techniques to detect even the planets in that system, much less the disk of the star itself. No, this is wrong. Your eyes can't detect the difference between 4k and 8k at routine (i.e. 90 degree width or so) distances. They just can't.
- devwastaken 8y ago
  >No, this is wrong. Your eyes can't detect the difference between 4k and 8k at routine (i.e. 90 degree width or so) distances. They just can't. proof please.
- Retric 8y ago
  People can tell the difference between a 4k and an 8k TV. 4K is still only 2,160 vertical pixels. That’s normally plenty but some images have really obvious issues at that resolution. EX: https://en.m.wikipedia.org/wiki/Moiré_pattern https://en.m.wikipedia.org/wiki/Moiré_pattern PS: Eyes have 120 million rod cells, and 6 million cone cells. Further, those cells are not in a grid pattern. A 4K TV is approximately 8.3 million pixels so it’s approaching the right order of magnitude, but still has some room before you can individually activate each cell.
  - davrosthedalek 8y ago
    Isn't there also automatic eye movement? So in principle there is a super-resolution effect which pushes it even higher.
  - aidenn0 8y ago
    Yes bad mastering is an issue; a properly anti-aliased 4k master would not have any moire effects. By all means shoot in 8k or 70mm film and then master it to 4k, but 8k broadcasts are silly.
- KaoruAoiShiho 8y ago
  But is 8K the final frontier though? It's finally "retina" for TVs at regular distances?
  - reaperducer 8y ago
    I won't be satisfied until my TV goes to 11K.
    - alasano 8y ago
      Wait no longer - https://www.fanaticalfuturist.com/2017/09/samsung-to-unveil-new-11k-tvs-with-natural-3d-effect-in-2018/ https://www.fanaticalfuturist.com/2017/09/samsung-to-unveil-...
- FPGAhacker 8y ago
  We can argue about physical limitations of reality, but all we ever really understand is a model of reality. Sometimes it seems to match predictions well. But it’s always the model we understand. Given how complex reality appears to be and how comparatively simple our models appear to be, it seems likely that the models are incomplete, and probably always will be incomplete. As such, perhaps it would be better to say that given your understanding of the physiology of the optics of the eye, and the receptors and nerves of the eye, that you believe strongly that it’s not possible to detect the difference between 4k and 8k at some common viewing distance. Or perhaps more simply “most” or “nearly all” people can not detect the difference. As with unusual occurrences like tetrachromats, it could be that there are some people with higher resolution perception such that they could see the difference. I don’t have a reference now, but I recall from a Feynman writing that there have been experiments to see how few photons a person could perceive. As I understand it, nobody was able to say they saw the photon (or bundle of a few photons), but they were asked to press a button whenever they thought they might have seen a photon. The distribution came back slightly biased towards affirmation that they could see the single or few photons, rather than the distribution mere chance would predict. Nobody could tell you what that photon looked like, and yet somehow they did perceive it x% of the time, where x was something like 10% (or the experiment had a flaw that cued the subject—I don’t know if it’s been repeated). Now, I’m not suggesting this says anything at all about resolution, it’s about intensity, but it is still it is interesting. And the idea of the study is worth considering for learning about resolution perception. Perhaps nobody would pass the Pepsi challenge comparing an 8k Mona Lisa to a 4K Mona Lisa at some distance. But if subjects were shown a series of many images and asked to push a button when they would thought it was 8k, maybe would tell us something interesting. There are lots of things to control for, like a correlation among all the 8k images or 4k images, but I think it could be done. I don’t really know the details of how the hypotheses are constructed for a good study, but if someone has some suggested reading, I’m interested. Perhaps we would say: * All subjects are shown the same N images * The image is randomly (normal? uniform?) selected to be the 8k version or 4k version. * If a version of the image has been shown to the subject, it will not be shown again to that subject in either resolution * Subjects that have seen the images can not discuss with subjects that have not And million other thing like the coloring of the images, subject matter, and so on. Maybe have a version that provides feedback on correctness.
hoppelhase 8y ago
You should watch this video addressing some of your arguments: https://youtu.be/VxNBiAV4UnM https://youtu.be/VxNBiAV4UnM
- ardit33 8y ago
  Read the comments, pretty much most people agree that they can tell the difference between normal HD and 4k displays (and content) at and normal viewing distances.
  - aidenn0 8y ago
    That's inconclusive. For example, most people can tell the difference between the SACD and CD recording of a song. However if you take the SACD and convert it to 16 bits, 44KHz nobody does better than chance at telling the difference; the SACD and CDs were mastered differently, as is UHD and HD content. Of course with HD vs UHD, we are usually talking about sampling rate, not dynamic range (HDR is a thing though). For downsampling, proper anti-aliasing is needed. One good way to check would be to take the UHD content, blur it and then present it on the same 4k display without changing the resolution. Then do the A/B testing and see what the results are. If you just downsize to HD, then you will introduce aliasing artifacts, and then more artifacts will be introduced when the TV upscales to 4k again, and it's pretty well known that people can see aliasing artifacts at HD resolution.
cma 8y ago
> However I can see a single pixel illuminated at about 25ft. Can you distinguish if that pixel is illuminated at column 1279 vs 1278? Or two pixels at (gamma correct) half brightness?
- cptskippy 8y ago
  Can I do that with my 1080p 12inch laptop display sitting at 18-24 inches? Probably not, but I can see the jaggies from the anti-aliasing in the Y-Combinator favicon which is a 16x16px square.
  - cma 8y ago
    At 12inch@18 inches is about 9 times bigger in visual area than your 55" @ 25ft 4K TV example, which only has four times the total pixels.
    - cptskippy 8y ago
      I wasn't citing my actual viewing distance for the 55" TV, just indicating how far back I could stand and still see a single illuminated pixel on it. As for the 12" display, ~18" is the proper viewing distance. The proper viewing distance for a 55" TV is about 7-8'. Even for a monster 85" TV the viewing distance is only around 12'. At a viewing distance of 25' the display should be close to 200".

ajross 8y ago

> I think it's naive to assume our vision can be diluted down to a fixed resolution at a given distance. You have an alternative theory of vision where resolution is not limited by photoreceptor density? Look, there are only so many cells on your retina. > However I can see a single pixel illuminated at about 25ft. And you can see a star (which subtends a MUCH smaller angular size than that mere pixel) at hundreds of light years. Yet it takes a skyscraper-sized telescope and mindbendingly crazy interferometric and occlusion techniques to detect even the planets in that system, much less the disk of the star itself. No, this is wrong. Your eyes can't detect the difference between 4k and 8k at routine (i.e. 90 degree width or so) distances. They just can't.

devwastaken 8y ago
>No, this is wrong. Your eyes can't detect the difference between 4k and 8k at routine (i.e. 90 degree width or so) distances. They just can't. proof please.
Retric 8y ago
People can tell the difference between a 4k and an 8k TV. 4K is still only 2,160 vertical pixels. That’s normally plenty but some images have really obvious issues at that resolution. EX: https://en.m.wikipedia.org/wiki/Moiré_pattern https://en.m.wikipedia.org/wiki/Moiré_pattern PS: Eyes have 120 million rod cells, and 6 million cone cells. Further, those cells are not in a grid pattern. A 4K TV is approximately 8.3 million pixels so it’s approaching the right order of magnitude, but still has some room before you can individually activate each cell.
- davrosthedalek 8y ago
  Isn't there also automatic eye movement? So in principle there is a super-resolution effect which pushes it even higher.
- aidenn0 8y ago
  Yes bad mastering is an issue; a properly anti-aliased 4k master would not have any moire effects. By all means shoot in 8k or 70mm film and then master it to 4k, but 8k broadcasts are silly.
KaoruAoiShiho 8y ago
But is 8K the final frontier though? It's finally "retina" for TVs at regular distances?
- reaperducer 8y ago
  I won't be satisfied until my TV goes to 11K.
  - alasano 8y ago
    Wait no longer - https://www.fanaticalfuturist.com/2017/09/samsung-to-unveil-new-11k-tvs-with-natural-3d-effect-in-2018/ https://www.fanaticalfuturist.com/2017/09/samsung-to-unveil-...
FPGAhacker 8y ago
We can argue about physical limitations of reality, but all we ever really understand is a model of reality. Sometimes it seems to match predictions well. But it’s always the model we understand. Given how complex reality appears to be and how comparatively simple our models appear to be, it seems likely that the models are incomplete, and probably always will be incomplete. As such, perhaps it would be better to say that given your understanding of the physiology of the optics of the eye, and the receptors and nerves of the eye, that you believe strongly that it’s not possible to detect the difference between 4k and 8k at some common viewing distance. Or perhaps more simply “most” or “nearly all” people can not detect the difference. As with unusual occurrences like tetrachromats, it could be that there are some people with higher resolution perception such that they could see the difference. I don’t have a reference now, but I recall from a Feynman writing that there have been experiments to see how few photons a person could perceive. As I understand it, nobody was able to say they saw the photon (or bundle of a few photons), but they were asked to press a button whenever they thought they might have seen a photon. The distribution came back slightly biased towards affirmation that they could see the single or few photons, rather than the distribution mere chance would predict. Nobody could tell you what that photon looked like, and yet somehow they did perceive it x% of the time, where x was something like 10% (or the experiment had a flaw that cued the subject—I don’t know if it’s been repeated). Now, I’m not suggesting this says anything at all about resolution, it’s about intensity, but it is still it is interesting. And the idea of the study is worth considering for learning about resolution perception. Perhaps nobody would pass the Pepsi challenge comparing an 8k Mona Lisa to a 4K Mona Lisa at some distance. But if subjects were shown a series of many images and asked to push a button when they would thought it was 8k, maybe would tell us something interesting. There are lots of things to control for, like a correlation among all the 8k images or 4k images, but I think it could be done. I don’t really know the details of how the hypotheses are constructed for a good study, but if someone has some suggested reading, I’m interested. Perhaps we would say: * All subjects are shown the same N images * The image is randomly (normal? uniform?) selected to be the 8k version or 4k version. * If a version of the image has been shown to the subject, it will not be shown again to that subject in either resolution * Subjects that have seen the images can not discuss with subjects that have not And million other thing like the coloring of the images, subject matter, and so on. Maybe have a version that provides feedback on correctness.

hoppelhase 8y ago

You should watch this video addressing some of your arguments: https://youtu.be/VxNBiAV4UnM https://youtu.be/VxNBiAV4UnM

ardit33 8y ago
Read the comments, pretty much most people agree that they can tell the difference between normal HD and 4k displays (and content) at and normal viewing distances.
- aidenn0 8y ago
  That's inconclusive. For example, most people can tell the difference between the SACD and CD recording of a song. However if you take the SACD and convert it to 16 bits, 44KHz nobody does better than chance at telling the difference; the SACD and CDs were mastered differently, as is UHD and HD content. Of course with HD vs UHD, we are usually talking about sampling rate, not dynamic range (HDR is a thing though). For downsampling, proper anti-aliasing is needed. One good way to check would be to take the UHD content, blur it and then present it on the same 4k display without changing the resolution. Then do the A/B testing and see what the results are. If you just downsize to HD, then you will introduce aliasing artifacts, and then more artifacts will be introduced when the TV upscales to 4k again, and it's pretty well known that people can see aliasing artifacts at HD resolution.

cma 8y ago

> However I can see a single pixel illuminated at about 25ft. Can you distinguish if that pixel is illuminated at column 1279 vs 1278? Or two pixels at (gamma correct) half brightness?

cptskippy 8y ago
Can I do that with my 1080p 12inch laptop display sitting at 18-24 inches? Probably not, but I can see the jaggies from the anti-aliasing in the Y-Combinator favicon which is a 16x16px square.
- cma 8y ago
  At 12inch@18 inches is about 9 times bigger in visual area than your 55" @ 25ft 4K TV example, which only has four times the total pixels.
  - cptskippy 8y ago
    I wasn't citing my actual viewing distance for the 55" TV, just indicating how far back I could stand and still see a single illuminated pixel on it. As for the 12" display, ~18" is the proper viewing distance. The proper viewing distance for a 55" TV is about 7-8'. Even for a monster 85" TV the viewing distance is only around 12'. At a viewing distance of 25' the display should be close to 200".