Resolution Test: 4K (3840x2160) vs. 2K (1920x1080)

Participation in this is voluntary. If you do not like polls and/or do not want to participate in this one, ignore it.

If you do want to participate ...

• You'll need to have a 4K display device
• You'll need to know its diagonal dimension
• You'll need to display a 4K JPEG on it
• You'll need to look at the JPEG from different measured viewing distances and evaluate what you can see

If you don't like the design of the test or the poll, it's too late to change anything. Design one that you think is better.

This thread is not a platform for argument. If you post something to try turn it into an argument, I'll report your post for removal.

So ...

I've seen plenty of threads where people talk about how well they can see the visible resolution difference between 4K and 2K. Talk is cheap. It would be nice to have a tangible reference and factual reporting on this for a change.

I built a test image, available below. This test is strictly about resolution difference and viewing distance for still images. It's not about color gamut or frame rates or refresh rates or compression or anything else.

The upper left quadrant is a straight 1920x1080 crop from a scene I photographed over the summer. The other three quadrants are that same crop, but downsampled in Photoshop to 960x540, then upsampled back to 1920x1080. I used three different algorithms for the resampling: Bicubic (best for smooth gradients), Bicubic Sharper/Smoother (best for reduction/enlargement), and Nearest Neighbor (preserve hard edges).

Testing different resampling algorithms is important because many media devices do their own interpolation while others might not. For now, we can confine the test to comparing the best and worst quadrants - Nearest Neighbor being the worst by far.

When the combined image is displayed on a 4K monitor or TV, only the upper left quadrant is getting the benefit of 4K. The resolution of the other quadrants has been reduced to the equivalent of HD (2K) content.

When I look at this on my own 40" 4K TV, I have to be within just a few feet of it in order to see any difference between the good upper left quadrant and the horrible lower right quadrant. The other two quadrants are in between, and I would have to be even closer to see their differences.

I have average vision (with glasses), but my daughter has better than 20/15 vision. I asked her to do the same evaluation of this image (she is a Photoshop-savvy photographer and technology worker, so she knows how to look at displayed images critically). Her observation was the same as mine. She has to be within a few feet of the screen to see any difference. In our normal viewing positions of 10 feet from the screen, it's absolutely impossible for her to see a difference.

In our cases, our viewing distance must be approximately 1.5x the diagonal measurement (40" x 1.5 = 60" = 5ft) or of the display or closer in order to see the difference between the best and worst quadrants. At 2X times the diagonal measurement (80" or 6.7 feet) or farther, we can no longer see the difference.

Now the poll question: At what distances can other readers here see the resolution difference in this test image with their 4K displays of various sizes?

(I marked my answer #4: 1.5x the diagonal measurement or closer.)

View attachment 2139815
4K Test

I did not participate in your poll because I do not have a 4K monitor. However, I can tell you what I see on my Acer Aspire One notebook screen from a distance of about two feet.

It is possible to see some resolution loss in the downsampled images, and the bottom quadrant image is clearly the worst. The loss of resolution in the bottom quadrant image is quite obvious at a distance of three feet, but becomes undetactable at larger distances. Now, I am viewing your test image at 100%, and scrolling around, but I have a good recollection of the level of detail in the high resolution image, when viewing the others, so don't discount people like me when presenting a test to be evaluated by we great unwashed souls that don't have a 4K monitor handy.

On another note: Box Bicubic would hardly be my first choice for resampling an image. My choices would be, in order of preference, Lanczos and Mitchell.

24IS said:

don't discount people like me when presenting a test to be evaluated by we great unwashed souls that don't have a 4K monitor handy.

Click to expand...

This poll is geared toward 4K display owners because a certain percentage of them have made pretty grandiose claims about what they are able to see 'compared' to 2K displays, and from what distances. Strangely, though, they never really explain how they're doing the 'comparison' or suggest how the 'comparison' can be duplicated. I want to see if some myths will be busted here.

Charles McRobert said:

Box Bicubic would hardly be my first choice for resampling an image. My choices would be, in order of preference, Lanczos and Mitchell.

Click to expand...

I chose to present the more pedestrian resampling algorithms rather than the best. That ensures significant differences so we can see what the 'best case' viewing distances are when distinguishing good from bad.

Good idea! I will do the poll later and try to be unbiased, however there are a few major issues. First, the poll is not blinded, so I’ll naturally be inclined to think I see differences when actually I don’t. I’ll try to be honest! The other major issue is that depending on the test image it may or may not be difficult to see differences. So the usefulness of the poll will be somewhat limited on that basis. All the same, again, great idea!

I downloaded the photo, plugged the stick into a blu ray DVD player (I think it will show HD only) I could just make up some difference from in front of the screen (a couple of feet).

Then I plugged the stick directly into the TV (49" 4k panel) and I was able to just see some extra detail in the signs and one particular woman at about 10 feet away but it was marginal.

Even using the zoom up to 8x there was not all that much difference to see.

BTW, I read which was which afterwards.

Filling up 40 degrees of the field of vision, as recommended by THX, requires that one sit within 4 feet of a 40" TV--well within the 1.5X diagonal at which you claim to be able to appreciate differences in sharpness.

Radek Matyas said:

Filling up 40 degrees of the field of vision, as recommended by THX, requires that one sit within 4 feet of a 40" TV--well within the 1.5X diagonal at which you claim to be able to appreciate differences in sharpness.

TV Size To Distance Calculator (And The Science Behind It)

Choosing a new TV for your room can be a daunting challenge. The market has never been more complicated, with dozens of new models released each year and a mountain of marketing jargon to work through.

www.rtings.com

Click to expand...

What's the definition of Mixed Usage vs. Cinema? The Mixed Usage distance of 5.5 feet is right in line with my observations of this still image resolution test.

sybersitizen said:

Radek Matyas said:

Filling up 40 degrees of the field of vision, as recommended by THX, requires that one sit within 4 feet of a 40" TV--well within the 1.5X diagonal at which you claim to be able to appreciate differences in sharpness.

TV Size To Distance Calculator (And The Science Behind It)

Choosing a new TV for your room can be a daunting challenge. The market has never been more complicated, with dozens of new models released each year and a mountain of marketing jargon to work through.

www.rtings.com

Click to expand...

What's the definition of Mixed Usage vs. Cinema? The Mixed Usage distance of 5.5 feet is right in line with my observations of this still image resolution test.

Click to expand...

The recommendations by SMPTE (30 degree FOV) and THX (40 degree FOV) are not predicated on the appreciation of sharpness, but rather on immersion, from what I understand.

However, the link I supplied further notes that "Studies show that someone with 20/20 vision can distinguish something 1/60 of a degree apart. This means 60 pixels per degree or 32 degrees for a 1080p television. 4k UHD TVs double that to 64 degrees."

27" 2560*1440 Monitor @ 500mm distance, Full Screen.
Differences discernable, fence mesh and jaggies on steps.

60" 1920*1080 Plasma TV @ 3m distance, Full Screen
Very hard to pick, fence mesh a tiny bit, but only because I knew where to look.

Good test, thanks for doing this before I bought one !

--
Ron.
Volunteer, what could possibly go wrong ?

sybersitizen said:

This poll is geared toward 4K display owners because a certain percentage of them have made pretty grandiose claims about what they are able to see 'compared' to 2K displays, and from what distances. Strangely, though, they never really explain how they're doing the 'comparison' or suggest how the 'comparison' can be duplicated. I want to see if some myths will be busted here.

Click to expand...

Your test is flawed. A 2mp image upsampled to a 8mp monitor looks better than the same image on a 2mp monitor. How do I know that? I had and still have 2mp monitors and now I have a few 5mp to 8mp ones. The difference is real. At work, I use a 3mp one and after have used my 8mp ones, 3mp looks grainy. 2mp (HD) looks no less than stone age, this is my second monitor at work.

BTW, I have HD and 4k TVs as well and yes, 4k looks much better.

J A C S said:

Your test is flawed.

Click to expand...

It's imperfect, but it has at least one virtue: It's a test that people can apply in a uniform way and report on in a uniform way. I've never seen anyone else here even attempt such a thing, although I would welcome it.

As I said in the first place: If you don't like the design of the test or the poll, it's too late to change anything. Design one that you think is better.

sybersitizen said:

J A C S said:

Your test is flawed.

Click to expand...

It's imperfect, but it has at least one virtue: It's a test that people can apply in a uniform way and report on in a uniform way. I've never seen anyone else here even attempt such a thing, although I would welcome it.

As I said in the first place: If you don't like the design of the test or the poll, it's too late to change anything. Design one that you think is better.

Click to expand...

Here we go. I downsized crops to 50%; which is how I view them on my current screen. This is my starting point, not a 100% crop since images are normally displayed at less than 100%.

Then I downsize with bicubic (for reduction) 50% again and upsize with the nearest neighbor to simulate pixels of the size of 2x2 blocks (see also below). I post this together with my starting point. Click for full size and use the arrows of the keyboard to navigate. Hopefully, you will not use scaling in your browser. If you do, save and display them.

BTW, your pixels are not squares, they are more likely colored bars so a real HD image would look worse than my simulated one on a 4k screen and much worse than your simulated one.


starting point


simulated HD


starting point


simulated HD

sybersitizen said:

24IS said:

don't discount people like me when presenting a test to be evaluated by we great unwashed souls that don't have a 4K monitor handy.

Click to expand...

This poll is geared toward 4K display owners because a certain percentage of them have made pretty grandiose claims about what they are able to see 'compared' to 2K displays, and from what distances. Strangely, though, they never really explain how they're doing the 'comparison' or suggest how the 'comparison' can be duplicated. I want to see if some myths will be busted here.

sybersitizen said:

Box Bicubic would hardly be my first choice for resampling an image. My choices would be, in order of preference, Lanczos and Mitchell.

Click to expand...

I chose to present the more pedestrian resampling algorithms rather than the best. That ensures significant differences so we can see what the 'best case' viewing distances are when distinguishing good from bad.

Click to expand...

I know i can easily see a difference from 2x, my screen is 55" and 4k. From about7 or 8 feet i can still see a difference. I can't view farther than that but i suspect i might be able to had i the room.

Either way, im not sure what myths are floating around but i assumed everybody understood the concepts here. Higher rez matters depending on your vision acuity, and the viewing size per TV pixel.

Someday i'd love to get a 70" 4k screen and sit 4' from it. I'd probably benefit from 8k at that distance, but that's another conversation.

J A C S said:

sybersitizen said:

J A C S said:

Your test is flawed.

Click to expand...

It's imperfect, but it has at least one virtue: It's a test that people can apply in a uniform way and report on in a uniform way. I've never seen anyone else here even attempt such a thing, although I would welcome it.

As I said in the first place: If you don't like the design of the test or the poll, it's too late to change anything. Design one that you think is better.

Click to expand...

Here we go ...

Click to expand...

Hoo boy. I was expecting you would start a separate thread/poll with your preferred test. Now we've got two (three?) completely different things going on to confuse people who might want to contribute to this one. <sigh>

It's late here. What I'll say for now is the difference between the two parrot images appears to be on par with the differences visible in my own test; but the difference between the two foliage scenes seems more extreme. Not sure what to make of that; but overall I'm not seeing a huge practical difference between your test and mine. I don't know if others will agree.

Anyway, gotta hit the sack now, but maybe I'll see a new thread tomorrow!

Max Iso said:

I know i can easily see a difference from 2x, my screen is 55" and 4k. From about7 or 8 feet i can still see a difference. I can't view farther than that but i suspect i might be able to had i the room.

Either way, im not sure what myths are floating around but i assumed everybody understood the concepts here.

Click to expand...

That might not be a good assumption. I'll quote from the thread where I first felt a need for verity:

John Sheehy wrote:

"... the visible difference in REAL (not fake, aliased) detail is huge. I can see clear differences in a 4k image and a 2k version of it on my 39" 4k monitor from 18 feet away, and my vision is nothing like it used to be."

Max Iso said:

Higher rez matters depending on your vision acuity, and the viewing size per TV pixel.

Click to expand...

Indeed it does. If that poster's visual acuity is nothing like it used to be - what was it originally?

sybersitizen said:

J A C S said:

sybersitizen said:

J A C S said:

Your test is flawed.

Click to expand...

It's imperfect, but it has at least one virtue: It's a test that people can apply in a uniform way and report on in a uniform way. I've never seen anyone else here even attempt such a thing, although I would welcome it.

As I said in the first place: If you don't like the design of the test or the poll, it's too late to change anything. Design one that you think is better.

Click to expand...

Here we go ...

Click to expand...

Hoo boy. I was expecting you would start a separate thread/poll with your preferred test. Now we've got two (three?) completely different things going on to confuse people who might want to contribute to this one. <sigh>

It's late here. What I'll say for now is the difference between the two parrot images appears to be on par with the differences visible in my own test; but the difference between the two foliage scenes seems more extreme. Not sure what to make of that; but overall I'm not seeing a huge practical difference between your test and mine. I don't know if others will agree.

Click to expand...

None of them resembles the real thing but mine is closer.

I actually had a 4K (8mp) and a 3mp (a bit more than 3) screen next to each other for a while. The difference is very real. Ironically, it was more visible with text.

J A C S said:

I actually had a 4K (8mp) and a 3mp (a bit more than 3) screen next to each other for a while. The difference is very real. Ironically, it was more visible with text.

Click to expand...

I actually had the 40" 4K and a 32" 1280 side-by-side for a while, comparing still image content. Maybe text too, can't remember. The smaller screen size made the 1280's pixel pitch equivalent to 1536 for comparison. As I recall, the required viewing distance for me to see the resolution difference was something similar to the present single 4K test image I use. I assume differences in manufacturers, generations (the 1280 had been in service for more than a decade), component designs, software, etc. must also have been factors one way or the other, so I could easily have been comparing irrelevant aspects. Whatever that was worth as an experiment, I won't be doing it again since the TVs are now integrated with other components in separate rooms.

Ron Poelman said:

27" 2560*1440 Monitor @ 500mm distance, Full Screen.
Differences discernable, fence mesh and jaggies on steps.

60" 1920*1080 Plasma TV @ 3m distance, Full Screen
Very hard to pick, fence mesh a tiny bit, but only because I knew where to look.

Good test, thanks for doing this before I bought one !

Click to expand...

Before drawing any conclusions from this test, remember, the original is a 100% crop, which is already upsampled for color and has softening due to noise reduction (and in some cameras, an AA filter), most likely, so the maximum possible difference isn't even present. What would be better would be to take a crop of a 50% resample or bin as the original.

There is more to lose, downsampling that a further 50%.

sybersitizen said:

This poll is geared toward 4K display owners because a certain percentage of them have made pretty grandiose claims about what they are able to see 'compared' to 2K displays, and from what distances.

Click to expand...

You really think they are going to publicly admit being made suckers by the 4K/8K hype?

J A C S said:

Ironically, it was more visible with text.

Click to expand...

Probably because text has lots of important high frequency detail, and natural photographic images usually don't. JPEG compression, as you know, throws out a lot of high frequency detail if it can, which hardly makes a difference for photographs, but which is pretty bad for text and line art.

Aliasing is a big problem with text as well, and so operating system makers made antialiased font renderers to prevent the 'jaggies' and most especially to suppress the color artifacts due to the varying subpixel layout among the different screen manufacturers. Of course, having lots of resolution helps all of this considerably.

From what I remember with computer monitors, getting high quality text was of supreme importance in the old days. The earliest monitors I knew of from the 1970s were text-only and monochrome, while viewing text on the oldest color monitors usually led to a lot of eyestrain due to the large subpixels, and the text size was larger as well, making them less suited for word processing.

Quality text was an important consideration for early pocket computers, like the Palm Pilot, and so the first models of these were also monochrome to maximize resolution; and it was mobile devices which really spearheaded high density displays out of necessity.