4K Is Not 4X?

It is often known that sooner or later anybody will need to get professional help. Now I am that anybody.

My deal is that that I spoke way too soon about 4k monitors which i shall be learning a lot about now. And then a customer's son told me that 4k is just double resolution, not four whole times? When i pushed his son back more than once, he then got onto my nerve, and claimed that the heavily-weighted and round-tabled guy said this The marketers are claiming 4K video gives four times the resolution of 1080 HD. It doesn’t. It only doubles. Feeling the burn I almost blurted back with J.H. weight, but I tried to keep his cool, till I ask for what to say?

He, Simon Lambert, (3 roundtable heads down) is the man that came out with thist! Is it the truth be told?


4K is four times the resolution of 1080p. It's double the resolution both horizontally and vertically, and 2x2=4.

See the image below. You can fit four 1080p green rectangles in the red 4K rectangle. Argument over.

Thanks to you, ethan ace! I was never the argumenter anyway, but now i can use your boxes as I please to show the truth, to the troubleing son of a customer.

Lookit this though, his tweeting is still here, but now mocking on phisycs with his 4k double errorr.

I know sometimes ipvideo can tweet blast at someone to learn what the deal is, can you use it to send a question on his maths on the 4k, because he says he is respected and most looked at in 2012 linkendin? He, Simon, could answer because he is not shy i think!

Hold on a minute. Just to play devil's advocate - 4K is double the number of pixels both horizontally and vertically, which yields 4X the number of pixels, but does that equate to 4X the resolution?

Since resolution is not the number of pixels, but the ability to resolve details, (as John beat into our heads during the IPVMU classes), I can see a valid argument that doubling the number of pixels both horizontally and vertically would yield a 2X increase in resolution (assuming all other factors are equal).

I'd agree with you on that point, but I really don't think Mr. Lambert means resolution in anything other than the typical pixel count way. Even if he did, he cannot say in hard and fast, no uncertain terms that it would be double. That's as precise as saying 4x.

I would disagree on what Mr. Lambert means. From the roundtable link, he states,

"The marketers are claiming 4K video gives four times the resolution of 1080 HD. It doesn’t. It only doubles. Is that worth four times the demand on your storage and bandwidth?"

The fact that he says it's 4X the demand on storage and bandwidth indicates that he knows it's 4X the total number of pixels. [Edit: as John states, he's wrong that 4X the pixels results in 4X the bandwidth / storage, but I take it as evidence of where his line of thinking is going.] However, he says that only gets you double the resolution. From the context, I think he is talking resolution as in the ability to resolve detail. He doesn't state that outright, as he should if that's what he means. Consequently, he only confuses the issue rather than clarifying it.

Also, from the context of Mr. Lambert's statement, I think he is talking specifically about 4K cameras, i.e. a 4K camera is not 4X better than a 1080P camera. However, Tedor stated that he is talking to his customer about 4K monitors. When you start talking about 4K monitors, you're talking about different issues so Mr. Lambert's statement does not apply. I think someone is justified in saying that a 4K monitor is 4X the resolution of a 1080P monitor. As you stated, you can put four 1080P images on a 4K monitor and not lose a single pixel.

I absolutely agree though that "Troubling Son of a Customer" is probably misunderstanding and thinks he means it's only double the number of pixels. If so, he needs to be taught basic math.

Tedor, interesting discussion! Richard, good feedback!

To answer this question, we need to first agree on what we mean about 'resolution'. As you all have alluded to above, there are 2 common meanings:

  • Resolving power, i.e., what can be seen
  • Pixel count, the number of parts on a sensor

In this industry, almost everyone uses resolution to mean pixel count. And on that count, 4K is certainly 4x the 'resolution' as 1080p. It's a simple output of counting physical parts on a sensor.

What makes this more complicated is that the traditional meaning of resolution is the ability to resolve details, like looking at an eye chart. It does not matter how big your eyes are or how many rods or cones they have. What matters is if you can see smaller lines on the chart.

On that aspect, the resolving gains of 4K range from negative to maybe double. It will likely be negative in low light, as 4K cameras will generally be worse than their 1080p counterparts (both smaller pixels and less sophisticated image processing). It may also be worse in WDR scenes as first generation 4K cameras will not have true multi-exposure WDR. However, in certain wide scenes with even lighting, you might be able to see more details at farther distances, whether that's 1.5x, 2x, etc., is a matter of debate.

There's only one claim which is absolutely false here; That is Simon Lambert's conclusion that 4K will have "four times the demand on your storage and bandwidth" than 1080p. Bandwidth does not increase linearly with resolution. For example, just like 1.3MP was 4x the pixels of VGA, it was not 4x the bandwidth / storage. This is because one can more efficiently compress more pixels in the same space/scene without sacificing quality.

Bandwidth does not increase linearly with resolution. For example, just like 1.3MP was 4x the pixels of VGA, it was not 4x the bandwidth / storage. This is because one can more efficiently compress more pixels in the same space/scene without sacificing quality.

I agree with this, but I am wondering if I do for the same reasons. Without going into bits of entropy in the scene etc, would you say a fair natural language approximation of this concept would be:

In a typical scene, for any given any pixel, the value (hue/intensity) of any of the surrounding 8 pixels is statistically much more likely to be identical to the center pixel's value than any other single value. This allows these groups of like values to be efficiently encoded as hue/intensity and number of repeating occurences, instead of listing each one seperately.

When you go from 1080 to 4K you can look at it as dividing each pixel into four. And due to the above statistics, where you had a single certain valued red pixel in 1080, now you have four that average out to that same value, but may in fact be distinct. Though a lot of times they will be the same or similar values, leading to encoding efficiencies.

Agree?

*This also would explain the reason that compression efficiency can take a nose dive in low-light, when too much gain amplifies the 'dark noise' to the point that random pixels register as points of high intensity. It is these 'spoiler' values that shorten the contiguous runs of adjacent like valued pixels.

Hi all

Very interesting discussion.

When I'm testing the difference between 2 displays (1080P and 4K) the main factor is the amount of pixels.

A 4K monitor (most of them) have 4 times pixels then 1080P monitor.

For ex' – 1080P from certain company have 1920 by 1080 (2,073,600 pixels), while a 4K have 3840 x 2160 (8,294,400 pixels)

So isn't it a simple mathematics?

BR

I understand what you are saying, but on the other hand if one takes 4X to mean that you would be able to discern four times as many horizontal lines or vertical lines on this TVL chart, you might find that you can only isolate half that many. And TVL is an accepted measure of resolution.

We should not assume that because there is four times as much to see that we see four times as much...

And regardless of whether it ends up being 2X, 3.4X or 6X, in the strict sense resolving power (resolution) is measured by resolving.

(The conflation is probably Microsoft's fault for using the term screen resolution in Windows Control Panel:Display, when really referring to pixel dimensions. It's interesting how, counter-intuitively, the preview screen gets smaller the higher the resolution.)

"You might find that you can only isolate half that many. And TVL is an accepted measure of resolution."

That's presuming you cover the same area with both cameras (1080p and 4K). What the pixel proponents argue is that you take the 4K camera, cover 4x the area and get the same TVL. Voila, 4x the resolution...

Of course, as I've mentioned above, the 'true' answer for resolution will depend significantly on the scenes being monitored as 4K brings with it (today in 2014), some significant limitations in low light and WDR performance.

In the comments of that article, Lambert shows that he is simply basing resolution purely as pixels, not about actual image performance.

Excerpts from the comments:

  • First, he says, "Obviously comparing 12.7px/cm with 6.35px/cm we can see that this 4K image gives twice as much detail in each centimetre of the person's height than does the full HD image. So, this is twice the resolution, not four times."
  • Then when questioned he agrees that, "The resolution of the FHD camera (in px/cm or px/ft, etc) could be extended by a 4K camera to cover an area twice as high and twice as wide so this would mean four times the area is covered."

Unfortunately, he totally misses the other elements that go into producing an image quality beyond mere pixels (like ability to handle contrast, low light, etc.). In this case, it is a huge deal, as 4K cameras will be much worse in these areas (at least for the first generation or two of 4K).

IMHO, the glaring flaw in Simon's argument is that given a new "resolution" with a pixel count of 1920x2160 (doubling the y pixels only), and using his same logic, he would conclude rightly that

Head to foot is now covered by 2160 pixels (this being 1080 x 2). So now each centimetre is shown using 12.7 rows of pixels (because 12.7 = 2160/170). Obviously comparing 12.7px/cm with 6.35px/cm we can see that this image gives twice as much detail in each centimeter of the person's height than does the full HD image. So, this is twice the resolution, not four times.

But even Simon would have to admit that if 1920x2160 is twice the resolution, a 3840x2160 would have to have a least a little more resolution, no? And what would he say about a 3840x1080 pixel count? Simon Says: Same?

Maybe to appease Simon, the industry could change from saying that the resolution of 4K is "four times" to just saying "twice doubled."

*I would be interested on getting Simon's take on Samstretch's 1280H format... NOT!

So under Lambert's logic, a 1920 x 2160 (4MP) camera has the same resolution as a 3840 x 2160 (8MP) camera...