Resolution – The Truths and the Myths about PPI, DPI & Resampling

Introduction

The terms dpi” and “ppi” are often used (and misused) across the photography, design and printing industries. But what do these terms actually mean?

Here I’ll discuss the terms dpi and ppi (they are not actually the same thing) and will discuss resolution in general terms.  I will try to make sense of what they are, what they affect and how they relate to digital images for the web (screen) and how both dpi and ppi affect your final prints.  It’s amazing how these simple terms are misunderstood and confuse people when in reality it is really quite simple.  First, empty your mind of what you think it’s about. Seriously forget what you think you know and go with me and I will try to make sense of both dpi and ppi. So let’s start from scratch.

Dots & DPI
As you can guess, dpi means dots per inch.  Despite what some may say though, this actually has nothing to do with “digital” images at all.  Your images are not made up of dots, they are made of pixels.

For what we are discussing here, dpi refers in fact to how many dots of ink a printer will place on a page.  It’s a term of output resolution that relates not to the digital image itself but to the physical printed image or document that you would hold in your hand or hang on your wall.  DPI can also refer to how many dots (or samples) per inch in a scanned image/document.  I prefer personally to say they are samples rather than dots but for now we are only interested in the digital image and the print (scanning does follow these same rules though).

So what is the difference between a dot and a pixel?  The main difference is a dot physically exists usually on some sort of media like a photograph or page of paper (in the main) – a pixel however does not physically exist (see more below). The dot is the smallest point that a printer can make when printing.  That’s it. A dot is exactly that – a dot on a piece of paper – Nothing more.

Dots per inch therefore is the number of dots of ink, dye or toner (per inch) that a printer machine places on the paper to make the images – That has no relation to the number of pixels in the digital image though. The dpi is controlled by the printer driver itself.  You (the photographer cannot control the dpi you are printing (unless you are the one with the printer).  The printer just lays down the dots on the page and ideally for photography you want a fairly high dpi – Most lab printers will be printing ay 300dpi or 600dpi.

If you print at home, you may see some exaggerated claims from printer manufacturers that their printers can print at hugely high dpi values.  Epson claims my R3000 printer can print at 5760 x 1440dpi.   Whilst the printer can lay down that number of droplets, it actually needs to place potentially 8 droplets of ink in the same place to create a single coloured dot.  So really those resolution claims can now be divided by the number of inks the printer uses.  So to work out the effective resolution of my R3000, I need to divide the resolution by 8.  Therefore my printer has an effective maximum resolution of 720x180dpi at the top resolution.  Often 2880x1440dpi is more than enough for a top quality print and this makes my effective resolution 360x180dpi.

Maximum X Axis resolution Maximum Y Axis resolution Number of inks  Effective Resolution
5760 1440 8 720x180dpi

(5760/8 = 720) and (1440/8 = 180) = 720x180dpi – still a very high resolution for a printer.

Pixels & Resolution
Now to the main meat as it were.  Digital images are made up of one thing only – pixels. A pixel is the smallest single element of a digital image. The ONLY thing that matters in your digital images is the number and quality of those pixels. They are very precious.

Everything you view on your monitor is made of pixels (except the post it notes if you’re anything like me).  Pixels are the building blocks for everything that you see in digital form.  Be it the text you are reading, the games you play or, as we are discussing here, the digital images you are viewing.  The pixels themselves though actually don’t physically exist in the real world.  They exist purely as digital data located on your hard drive (or other media like memory cards, flash drive, pen drive, CD, DVD etc.). The computer hardware and software that we use on our computers/devices converts the digital data into the pixels that make up the images that you see on your screen.

Pixels are normally rendered as a grid of squares and each of these pixels is given a digital value which represents a single colour – either red, green or blue in the RGB colour space.

When you place these pixels together in hundreds, thousands and millions they make up a variety of sizes of digital images and the colours of these pixels merge together to create the images we see.  I won’t discuss the quality of pixels here as that is dependent on your camera/scanner (input device).

Before I go on, let’s discuss how you see your images on your screen.
Your digital monitor also has a grid of elements that displays image pixels (the number of these elements is determined by the maximum resolution of the screen (measured in pixels)).  The pixels of your digital image map directly to the pixel elements on your screen.  So if you have a 900 x 600 pixel image it will be displayed using 900 x 600 pixels on your monitor – how large it appears on your monitor will be determined by the resolution your screen runs at.  A higher resolution monitor will make the image seem smaller – in reality though it’s still the same size.

A 1920×1080 pixel resolution screen has 1920 pixels across by 1080 pixels high.  A 900×600 pixel image will take up around a quarter of the area of your screen.  But if that image is straight out of say a high resolution camera there could be closer to 6000×4000 pixels.  Viewed at 100% therefore you would only see a small portion of that image because your screen can only see 1920×1080 (and some of those are used to display your operating system).

As stated above, digital image sizes vary greatly.  An image from something like a Canon 5D MkIII has 22.1Mp.  That is 22.1 million pixels.  That resolution is easily calculated by multiplying the pixels on each edge together.  My 5D MkIII produces an image that is 5760 pixels across by 3840 down – 5760×3840 = 22,118,400 pixels (so 22.1Mp).

Compare that to a web sized image that is perhaps 900 x 600 pixels in size.  That is just 540,000 pixels. (0.54Mp)

Here’s two images shown at 100% magnification.  You can see the difference.

This image is resampled from 5760 x 3840 pixels down to 900 x 600 pixels. It’s 900×600 pixels in size.

This image is also 900×600 pixels.   It is a 100% crop of a portion of the original 5760 x 3840 pixel image – We are just seeing a small section of the full image

Both images above are the same size (900 x 600 pixels).  The first is a web sized image (reduced number of pixels only) down-sampled from the full resolution image.  The next is a crop of the original image.

If you then try to resample the first web image (by cropping the same area and resampling to 900×600 (adding pixels not there) – you can see the quality is lost because we are croppig from a drastically reduced number of pixels.

 

This is obviously an over-exaggerated example but it shows that pixels are extremely important.  Adding random pixels (resampling) can and does degrade your image although to be fair, on very high res images  you would be pushed to see the difference – but that very fact makes resampling a little bit of a waste of time (and disk space).  If you can’t see the difference why do it at all?  You can’t “magic” extra detail from images by adding random pixels although software like Genuine Fractals can do a very good job – but really for most purposes it’s just not required.  If you can size images properly for print and sharpen correctly you will not need to make your images bigger which effectively slows your whole workflow down adding steps not required and making larger images which is a drain on your computer’s resources.

 

Pixels are extremely important.  Adding random pixels can and does degrade your image although to be fair on very high res images you would be pushed to see any difference – but that very fact makes resampling a little bit of a waste of time (and disk space). It slows down uploads to your labs when you are sending images for print and offers no real benefit.

If you can’t see the difference why do it at all?  You can’t “magic” extra detail from images by adding random pixels although software like Genuine Fractals can do a very good job – but really for most purposes it’s just not required.  If you can size images properly for print and sharpen correctly you will not need to make your images bigger which effectively slows your whole workflow down adding steps not required and making larger images which is a drain on your computer’s resources.

Pixels Per Inch
Now let’s discuss Pixels per inch (ppi).  It is often over complicated and although I’ll go into detail below really all pixels per inch is is the number of pixels of your print in each inch of the print.
It is perhaps because we can’t figure out in our minds what a pixel actually is that this may seem too simple?  Think about the term.  Pixels per inch. It refers only to the number of pixels (horizontally and vertically) that are actually printed in one square inch of an image.   Note that I said “actually printed” because ppi has no relevance to anything that is not actually printed.  PPI controls ONLY print size from any given number of pixels.  It does not control the quality of your digital images and it’s not actually a fixed number (it can be changed in any image at any time to control the print size).  The higher you pack the pixels together the smaller the actual image will be.

A 1 inch square printed at 300ppi would have 300 pixels across and 300 pixels down.    Sounds very simple – and it is.  That is all there is to it.  If you take the same 300 x 300 pixel image and print it at 150ppi, you have effectively spread out those pixels a bit….. and now the image would be printed 2″ x 2″.  You have not changed the file, nor the file size by doing this.

There’s an equation that allows you to find any of the three variables.  The ppi calculator below is very simple and can also help you understand the relationship of the three variables.  You need to do the calculation for each axis of an image (so twice per image).  The equation can be written 2 other ways (depending on the variable you want to find).

Here is a ppi dpi calculator.

Print Size = Pixels / PPI
So if you have an image that is 600 pixels by 400 pixels and you set it to print at 300ppi then the resulting print would be 2 inches x 1.333 inches (600/300 = 2) and (400/300 = 1.333).  Simples

Pixels = Print Size x PPI
If you have an image you want printed at 10×8 and you want to print it at 200ppi (which will look fantastic) then you need (10×200 =  2000) and (8×200 = 1600)  –  2000 x 1600 pixels.

PPI = Pixels/Print Size
If you have an image that is 4500×3600 pixels and you want a 20 x 16” print what do you need to set the PPI value to get that (without resizing)?

(4500/20 = 225) and 3600/16 = 225). So you set the ppi to 225 and you will have a 20×16 print with no need to increase the resolution to 300ppi!

As should be obvious from this then, PPI ONLY AFFECTS THE ACTUAL PRINT SIZE – NOTHING ELSE.

I hope that makes sense.

Generally the gradation of colours and tones in an image will become smoother (as the number of pixels increases).  This is obviously useful when printing.  However adding random pixels (upsampling) is likely not to always provide the better quality image – random additions can equal degradation of your image and in most cases it is just not required.  You can easily print large images down to 100ppi and get a superb image back printed still at 300dpi on your lab’s machine.

Some software is better at upsampling than others though – but in reality with today’s modern cameras you can print almost any reasonable size with no upsampling required.

Because digital images are only made of pixels we should always be using the term Pixels Per Inch (ppi) when discussing print resolution. Adobe get it right with Photoshop and Lightroom but some software still incorrectly states dpi!  No idea why!  Just remember when talking about digital images (that is any image that appears on any display) it’s ppi that will be used if discussing resolution.  Printer resolution is in dpi.

Here’s a few questions and answers that I’ve provided over the years.

Q1.   Many seem to use the term “dpi” to indicate the quality of a “digital” image. Is that right?
Actually dpi is purely a printer term – It’s how the printer lays down the dots on the photograph being printed.  (I mention only photograph as we are discussing photography).   It therefore only affects how the image is physically printed (not how it appears on the screen).  See the chat on dots above.

Q2.   People say my images have to be printed at 300dpi?
Firstly 300dpi is a printer term (not a digital image resolution). It’s likely people are using the wrong term here and mean ppi.  I appreciate people mix the terms – but even if the term ppi was used, it’s still wrong to advise people to print at 300ppi.

The resolution can actually be any number you choose to get the required print size.  PPI affects only the print size.  If you have a set number of pixels (say 3000 x 2000) the ppi you set will change the print size of the image. 300ppi is generally the optimal resolution for small images where you have the required amount of pixels already (the eye itself cannot resolve any more detail).  But the larger you print, the further away you will stand to view the image (relative viewing distance increases) and so the human eye’s ability to discern the detail decreases.  So you can easily print down to very low resolutions and still get a perfect looking image – the difference even close up between 300ppi and 150ppi is pretty small.  Things like billboards are printed at 10ppi but from the viewing distance they look great.  That is an extreme example though.  At 100ppi and the above 3000×2000 pixel image you can easily print at 30” x 20” and even close up it will look perfectly fine.

Q3.   People say I must save my web images at 72dpi/ppi?
This is totally incorrect.  As stated ppi only affects the print size of your image – it has nothing to do with how an image is displayed on your screen or the web.  So it’s useless information in this case.  The ONLY thing that matters for viewing a web image (or any other image for that matter) is the number of pixels it has. (I’m not talking about any pixels added later).  Web images are usually down sampled to around 800 pixels on the longest edge – At that size they can still be printed though so use of a watermark is advised.

Q4. What is the maximum size I can print from my camera?
Actually this is hard to answer and to an extent depends on the number of pixels you have.  It may be that you have edited and cropped the image which will reduce the maximum size you can print.  However in reality, even a 6Mp image (3000 x 2000 pixels) can be printed in excess of 30″ x 20″without having to resample (add pixels to the image).   A post that I have directed many to is

Q5. Why does viewing distance make a difference?
Over distance, the ability of the human eye to distinguish fine detail diminishes.  THis is good because it means you do not need to print all images at 300ppi (if you have 300ppi available without resampling I do recommend using that).  For larger images print at whatever ppi value you can (without resamnpling).   Some people resample their images (by adding pixels) so they can print at 300ppi!! Because 300ppi is seen as the best resolution to give the highest quality prints people resample – but what they end up doing is slowing down their whole workflow and have huge files that absolutely not required.  Worse still the assumption that 300ppi is best is very much flawed as they are probably damaging their files by adding pixels that are not there.  You can’t magic extra detail like this and in fact it can and does degrade the quality of your images.  It is better to print the pixels you have at a lower pixel per inch. The human eye would still  see an image printed 30″ x 20″ at 100ppi perfectly well from normal viewing distances.
A website I have recommended people look at to get an idea of image resolution and how a large image you can print with available pixels is here.  I don’t use Smugmug but I do appreciate their information on how many pixels their labs need to provide high quality prints – And it’s VERY surprising to most – This is the same for many professional labs.  A web sized image for many is around 960 x 640 pixels – Would you believe you can print this at around 10″ and get a prefectly decent print!  With just 3300 x 2200 you can get a 60″ x 40″ print!!  that’s 5 feet wide!!
http://help.smugmug.com/customer/portal/articles/93359#minprintres

In the recommended settings of the page they discusses an image that is printed at 80ppi (yes the incorrectly state dpi).  But the point is 80ppi will give a perfectly acceptable print.

Q6.  What about compression?
Lossy compression (used mainly in JPG images) is something that affects the quality of the individual pixels in your image.  Even high-resolution images can be strongly affected by compression.  This can  be a good way to limit how large an image can be printed but care needs to be taken to get the amount correct.  High compression (a lower quality number) means that the images reduces the detail of pixels and surrounding pixels – It does not affect the number of pixels.  For example it will  average the colours of a pixel and surrounding pixels in a bid to reduce the detail and overall size (in KBytes or MBytes) of the image.  For print we want a high quality so low compression is desirable.  In Photoshop that would be quality 12 when saving a jpg file or in Lightroom quality 100%.

A small amount of compression though is unlikely to make any discernible difference to your prints so most will save at quality 10 in Photoshop or 80-90%  in Lightroom.  This will barely affect how the image looks but greatly reduces the file size on your hard drive.

Print Size
Just to show that PPI only affects print size, I’ll look at this again with a specific example from a Canon 60D camera.  (Could be any camera though).

Let’s say you have an image that is straight from the 60D.  It’s an 18Mp (megapixel = million pixels) image.  The dimensions straight from the camera at full resolution are 5184 x 3456 pixels which provides 17,915,904 pixels (so pretty much 18Mp).

If you want to know what size print you can get from a full resolution image all you need to do is adjust the PPI value of your image.  Starting at 300ppi……

So Print Size =  Pixels / PPI

You do this calculation for each edge (axis) of your image

Print Size (PS)

Pixels (Px)

PPI

Print size

?

5184

300ppi

17.28”

?

3456

300ppi

11.52”

Or to use a different ppi value (with the exact same image)

?

5184

200ppi

25.92”

?

3456

200ppi

17.28”

 

So you can see ppi only affects the print size

Digital images are measured only in pixels (not in inches for that very reason)  If we change the PPI value to 100 things change even more.

?

5184

100ppi

51.84”

?

3456

100ppi

34.56”

The Pixels in each of the above are exactly the same. It’s the same image – but the print size changes as the ppi changes.  So you can easily get a nice 50″ print without any resampling.

Cropping into your image removes pixels and reduces the overall size of the image. If I know I want to print an image at 10×8 and I have enough pixels to print at 300ppi, I will use Lightroom or Photoshop to reduce the image size to 3000 x 2400 pixels.  (300 x 10 and 300 x 8 = 3000×2400)

To keep saying, the only thing that matters in your digital images are the pixels.  I see many posts on the web asking “what is the maximum size

I hope that makes some sense and shows why stating a ppi value for web images means nothing and how you can use PPI to suit the image size you want to print.

Please ask questions and I’ll attempt to answer what I can.

Regards
Jim

PS I just came across a great video on Youtube that explains a lot of the above in a 13 minute video.  Very useful as it shows exactly what I discussed above.

no comments
Add a comment...

Your email is never published or shared. Required fields are marked *