10

I think I have figured out why the DPI/PPI/LPI/Pixel/resolution (for simplicity lets refer to all of these a DPI in this discussion) discussion always devolves in a very big problem. The problem is 3 fold:

  1. To understand this quickly one needs to be able to have 2 conflicting models of how things work in their head.
  2. It is really easy to fool yourself into seeing something that is not part of the equation. For example one can see filtering and as a result deduce that everything is fine. SO one can make experiments that lead you onto wrong tracks if you are not careful.
  3. There is a marketing push saying higher resolution equals to better quality. But understanding that what applies to devices does not apply to images is a hard thing to swallow.

Point one is especially troublesome because it is really hard to pierce the human mind with conflicting information. Humans in essence defend their world view, this is why we have many religions. Point two makes point one very hard to go trough because its very easy to find conflicting evidence. And point 3 is just making these questions more common.

Ok, so how do we devolve this situation? There are many that say that we can not. But there are a few quite easy remedies to this problem.

  1. The first is to stop people thinking of DPI as a metric for quality

    • DPI is not a metric of quality
  2. The second is to stop people for mixing pixels and DPI like conversion factors. Either think in physical units (pt, mm, inches, cm, furlongs etc) or think in pixels

    • Never bring in DPI to discussions about pixel dimensions
  3. Images are fundamentally different from imaging devices! Changing image DPI value does absolutely nothing to an image. A image still only consists of samples, just because i have attached a processing istruction to it does not change anything.

    • DPI value is meaningless for the content of the image file

Now this should be able to put into a human readable friendly message.

| |
  • 1
    Perhaps i should add this into <rant> and </rant> around this – joojaa Sep 23 '16 at 8:01
  • Yes please. I share your frustration. (it's not going to be an easy thing to do though...) – Cai Sep 23 '16 at 9:58
  • @Cai: check out chat from here for more info – PieBie Sep 23 '16 at 10:00
  • Since when is PPI meaningless for the content of an image file? – Ryan Sep 23 '16 at 12:50
  • 1
    @Ryan since forever. Its not part of the image its just a not that says i wish this file would be of this size. It affets nothing, all it does is sets the default size of the image when placed in a page layout app or sent to printer. But that is not intrinisic data of a image. You coudl you know tell the printer/layout to fit the image or whatever. – joojaa Sep 23 '16 at 13:29
  • PPI has a major effect on one typical usage scenario: Joe Jane User is told to create an image of 9x11 inches. They set off to do this in an application that offers inches in the new image dialog, proceed to design their image, give it to the printer - and get told it is much to small to print, by a factor of about 4. – Michael Schumacher Dec 19 '16 at 13:23
  • @MichaelSchumacher thats manufacturing, it has no meaning for the image itself – joojaa Dec 19 '16 at 13:25
  • So you argue this is something a designer should be ignorant about? – Michael Schumacher Dec 19 '16 at 13:27
  • @MichaelSchumacher No i argue that the confusion is because they mix context. Resolution only affects manufacturing its not a property of an image. Its a property of the imaging device. In practice many designers do not know what DPI is. So clearly not knowing works fine. – joojaa Dec 19 '16 at 13:29
6

OK, one more try, totally different approach. Here we go:

Attempt on answer 0.3b

TL;DR

PPI is only meaningful in the context of Printing or perhaps when buying a display panel. It is there to be helpful to you doing task of printing. If the image is not printed then PPI is not needed only pixels.

Before we begin

Observe that you can not just blindly search for the term PPI or DPI fill it in with a value and be absolutely certain that it always means the exact same thing in every situation. It measures a similar quantity but whether the meaning of these quantities is the same depends on the context in which it is used in. Let us explain this by observing how another, much more common and another less common, measure may behave in different circumstances.

Case 1

A car moving at 80 km/h (80 kph, just expressed in SI- notation), is on a straight road. How much has the car traveled in on hour? Answer: 80 km.

Case 2

A paper machine producing paper at 80 km/h (80 kph, just expressed in SI- notation), is on a straight road. How much has the paper machine traveled in on hour? Answer: 0 km, the paper machine is not moving!

WTF. What does this have to do with PPI/DPI? Well, it shows a central property of measures you can not blindly stare at the quantity and deduce things form it. You have to have a understanding of the system, in this case the technical system around you.

This kind of thinking happens a lot with people transitioning form Photoshop to Illustrator. Illustrator has no PPI or DPI value, simply it is not dealing with discrete elements so it does not need one. However, many people see that there is a option labeled "Raster Effects" in document settings and it has a PPI ending and conclude that its the same as "Resolution" in Photoshop. It is not, if it were it would probably read "Resolution".

What does this mean for a image author

If you are working towards print production this is a convenient way to express how big the image is intended to be when printed. It is also a good enough estimator for whether or not you have enough pixels in normal print publication scales without having to memorize different rules of thumbs for multiple sizes in most of the common range. It is however just a rule of thumb estimate not a fixed truth.

It is also good to understand that anybody down the line, be it a person or a computer may choose to ignore whatever value you have set, for any number of reasons. Usually they do, because this system is much more vertically integrated than many other technical systems, so the designer has more influence.

But if your working for a digital production pipeline, be it for the web, a game, desktop wallpaper, etc. Then this value is of little use. For historical, and practical reasons people do not measure the resolution of a image files meant for screens in physical units. Just number of pixels.

But how then do you know what physical size the image is? Well the simple answer is you do not. The technical system does not give you much guarantees. Some people have bigger screens others have smaller screens. You simply have no control over the screens themselves, they are what they are. More than this, users may have different zoom levels, either by accident or purpose. Not much you can do about this. And in any event forcing a physical size would totally break the user experience of your end user.

Does this mean that screens do not have a PPI value? Well not quite, every individual make and model has a PPI value. In case of modern flat screens a fixed immutable value for that particular model, in case of a projector it depends on many factors. But monitors as a whole do not since you do not usually know what monitor each individual client has. Worse you do not know what kinds of settings their operating system has, this might scale images before sending it to the screen or the screen might scale the image so that the resolution matches the attached projector. Worse the settings may change over time and on whim of user, you simply can not control this. So you do the next best thing you assume the image is relatively sized somehow sanely.

In the end question is of control, in a print setting either somebody else is specifying settings or you are. In which case this is probably known beforehand, or at least could be known by asking. However in the digital realm the situation is much more complex than that, the system in between you and the eyes of your user is much more convoluted and not in your direct control.

In a image format

PPI is not a property of a image, it is a property of a imaging system. It is there to make print processing easier to approach but changing PPI does not fundamentally alter a image at all. In fact a image works fine without PPI information!

enter image description here

Image 1: Picture of what the image file is composed of.

The important bit is that your image has right amount of pixels. Header data is mostly ignored under most circumstances and mostly completely optional (usually only identifier and file size/technical compression info is required). PPI info is thus just a piece of metadata. From this follows that a image with dimensions:

  • 1000 pixels by 1000 pixels at 300 PPI
  • 1000 pixels by 1000 pixels at 72 PPI

    Is the same image! And will display same image in a browser and most digital systems out there.

This leads to a rule of thumb: Only pixels are important. But if you wish to relate them to a physical dimensions then PPI will help you as it converts the values to physical dimensions, but only for print. Never mix physical units and pixel thinking, either think in physical units and no pixels. Also remember since PPI is optional the end user is perfectly within their right to ignore it!

Typical mistakes

  • I need this image for web at 600 px by 200 px at 300 DPI for high quality. Wrong DPI/PPI/Resolution is not a quality metric of the image it is a quality, pretty poor, estimator for print or comparing two monitors. Pixels and resolution in same sentence is usually a symptom of erroneous thinking.

  • ...

Yada yada.

| |
  • 1
    Nice start. I agree with your end comment – Ryan Sep 23 '16 at 16:43
  • I've always thought a great frame of reference for explaining that pixels have no set measurement is showing the humongous pixels used on a large outdoor advertising billboard versus the tiny pixels on common devices. – Dom Oct 1 '16 at 20:56
  • "Pixels and resolution in same sentence is usually a symptom of erroneous thinking." ... Nop, it is not... entirely. This discussion has opened my eyes to something that was there all the time! A cap. I need to work on developing that concept. – Rafael Jan 2 '17 at 15:51
1

A work in progress.

This is are some sketch ideas of a video explanation I am preparing for a video.

It is going to be first on spanish so some lost in translation may occur.

This is going to be a long read, so you might wait for the video relase n_n


Chapter one: Introduction.

In life we deed to measure things, and we use diferent units to measure them. Meters for measuring a house, liters to measure a limonade we should prepare and hours to measure how long women take to get ready.

There are other kind of units that are not simple like the previous ones, but they are a relationship.

Foe example one of the most offen used is speed, veloctiy.

Speed tell us how much distance we travel in some amount of time.

We can have a car that travels 100 km in two hours, but we simplify the result on how much travels in one hour. Then we have that the speed is 50 km each hour. 50 km/h.

It is important to note that the definition is not a multiplication (50km x h) but a division 50km / hr.

Speed then is "The amount of distance traveled in one period of time".

Resolution is the same, it is a relation.

Chapter 2: A definition

My definition is:

"The amount of information units content in each dimensional unit".

We have then, four parts: Amount + Units of information + each + dimensional unit.

This definition adapts itself to diferent situations, for example, we could be used to use the term in images, but the term is also used in sound files for example.

We can record the voice of a person, the vibration in a mic can be recorded a lot of times per second, or less times, recording less information. We can have then a high resolution recording or a low resolution recording.


Chapter 3: A dot.

Imagine we use shugar to draw a cat on a table. We could count how many grains of sugar we have in each linear inch.

But if we use fluor to draw the same cat of the same size, we can imagine we have a lot more of grains than the sugar, on the same linear inch.

On the contrary, if we use rice, the definitio will get worse.

This is a first unit we need to understand. A grain represents one Dot, and this relationship is the first we need to understand: how many dots we have in one inch. Dots per inch or dpi.

But we need to understand the characteristic of this "dot". Either we have one grain of sugar or we don't. The information a grain of shugar is: "I am there or I am not there". There are no "gohstly" grains of shugar, that are "almost" there and almost disapear... unless someone sneezes on our image.

The information a single dot can carry is binary. This is: either you have a dot or you do not have a dot. Of course you can put two or more dots to have more meanings, but i am refering to a single dot.

But using shugar to make drawings is a bit complicated, so normally we use ink.

Chapter 4: A pixel

Chapter 5: A line

(To be continued...)


Some other chapters are like the dot, the pixel, the line, halftone, stochastic (error difussion), distances, electronic mediums, projected images, camera, screen, eye function, CSS pixels, etc, even images viewed form space n_n.

Some to explain the concept and some to explain an aplicated case.


Some misconceptions are due the bad usage of the units. For example a box of cereal should have some amount of cereal measured in weight. And boxes says so, that the content height may vary for the manipulation of the boxes.

If an elephant steps on your box of cereal it still weight the same (after removing the elephant of course) but the level of the box will be now diferent.

(There will be a chapter mentioning the elephant of course)


Edited 1.


What are we talking about?

Units and perception.

There is an old Indian story about 7 blind man that are exploring what is an elephant.

One concludes that an elephant is a big tree, because he hugs a leg, another concludes he is a snake because he touches the elephant's trunk, and so the other 5 men "decides" their own conclution.

Although the moral of the story is a little diferent there (to percive the elephant as a whole) The point it is needed to explain is "What units are we talking about?


And also, I will talk about how information is transfered, or translated, or aplied, or modified across a system, and how this "resolution units" then change and mix.

| |
  • I think there is a problem. A high resolution sound is better quality. For sound its pretty straightforward... But due to historical context there are two different things we call resolution in a image. For monitors we call pure pixel count resolution, for print we call PPI resolution. Now obviously they are not the same. So depending on context only one of these definitions are relevant. Increasing PPI when pixel count is relevant does not make a image better quality and this is the problem. – joojaa Jan 2 '17 at 5:24
  • It is the same, Resolution has a "cap" where it is percived or not, sound, density, distance. But I will take notes on that. – Rafael Jan 2 '17 at 8:59
  • No thats just it they are not. I dont think you get it images you have 2 different things that are resolution! Whet Adobe calls resolution and what your monitor settings for example in windows call resolution. The other is measured in DPI/PPI and the other is measured in pixel dimensions. They are not even remotely the same concept and nearly all problems come form the fact that people think PPI/DPI is meaningfull in their context while thats not true in many cases. Setting higher DPI is only a quality metric if you print. But most stuff is made for screeen only. – joojaa Jan 2 '17 at 9:07
  • T H A T is the point. Resolution is whatever you put in thoose units! Sound samples per second, pixels per screen, pixels per sensor, droplets of ink per inch, dots per centimeter, lines per inch, rice per cm. And yes. Chapters 4, 5, 6 explains that. – Rafael Jan 2 '17 at 15:42
  • "They are not even remotely the same concept." They are exactly the same concept. Ammount of + Units of information + per + Units of dimension. – Rafael Jan 2 '17 at 15:46
  • No you fail to undetstand the problem. 1900x1200 is called resolution. Pixels per inch is called resolution (as are DPI and LPI these are same) The other is a unitless unit and the other is a frequency unit 1/Distance unit which are this by definition of units is different. That is the corner of the problem – joojaa Jan 2 '17 at 16:01
  • Its not a question that people fail to understand what The terms mean or how they work. They fail to understand the relevancy of this info. – joojaa Jan 2 '17 at 16:04
  • PPI is resolition, DPI (a diferent unit) is resolution, total pixels per file is resolution, samples per second is resolution, total pixels per screen is resolution. The point is that they are diferent units (not unitless) that measures diferent things... if they can. In you answer I pointed a concept I must incorporate. CAP. a medium has a cap, human perception has a cap, a system has a cap. And also there is another concept, that i still need to "clarify" information-travel, information-translation, information-projection. (I probably need to incorporate this into the partial answer) – Rafael Jan 2 '17 at 16:12
  • Let us continue this discussion in chat. – joojaa Jan 2 '17 at 16:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .