DPI and Output Size – what you need to know

4 02 2012

Ok, so I know that I promised I’d come up with a post about how to pull all the previous discussed techniques together, and trust me, I will.  However a conversation at the office yesterday arose and I thought it’d be a really good thing to post about…output size and how large should we be saving our images.

Keys to remember:

1)   It’s all about pixels!

Remember this, as it’s very important and the common factor between most of what we’ll talk about for the rest of the post.

2)  There’s going to need to be a bit of math, but trust me…it’s not that hard.

Go back to your days of Algebra I and you’ll be fine.  We’ll be using the equation DPI * Inches = Pixels.  From this equation, if we have 2 of the 3 variables, we can always solve for the other.  More on this to come later.

3)  It’s about the distance at which the object is going to be seen.

The human eye can differentiate pixels at about 120-150 dpi when being held and read in one’s hand.  As you step further away, less definition is needed.  Think about those various pixelated images that when you get close you see they are a montage of small pictures, but when you stand back appear to be one image.  This is exactly the same principle.

Image courtesy of Sabri Farouki at http://sabrifarouki.com

Also, as you get older and are in lower light, you will be able to differentiate less dpi.  The other thing to keep in mind is that vector based information done in a raster based program will show effects dependent on the dpi.  The best example of this is text in Photoshop.  While working with a 72 dpi image, the text will look fairly pixelated and jagged.  The reason is because the linework of the font has to fit within the pixel.  For this reason, I’d suggest working with text in either Indesign or Illustrator.  Working in Photoshop will force you to use a higher resolution (200dpi-300 dpi) because of the text issue which will force you to render or use a larger image, negating the speed of using the smaller rendering or image.  See below for an example of the pixelated text and how the text needs to fit within a pixel as opposed to remaining vector based.

Text in Photoshop - click to see full size

4)  What is the output media going to be?  Print? Video? Billboard?

Typical video/computer/projectors use a 72 dpi standard.  HD video uses 1920 px X 1080 px (hence your tv is a 1080).  The computer monitor or projector will have their own resolution.  The laptop I’m writing this on for example is 1400 x  900 px.   As previously stated, printed material for hand outs should be min. 150 dpi and then printed presentation boards could go down to 100 dpi.  If you’re looking to create a billboard or rendering for site signage, you can get down to 50-25 dpi even.  I have even seen freeway billboard images that get down to 5 dpi and look fine from 100′ away.


5)  Think about what you are going to use the image for, both now and in the future.

If you need it at one size now, will you need it at a larger size for print/projection/etc later?  It’s always easier to reduce the file and reduce than it is to enlarge.  The reason for this is a bit technical but essentially, the algorithms that a program uses to “guess” what color a pixel needs to be when it’s enlarged struggle to execute effectively as they are asked to enlarge and compute more and more pixels.  Therefore if you’re asking Photoshop to increase an image 110%, it might do a decent job.  If you ask to increase that image 200% however, you’ll notice the image struggles to create accurate results.  Essentially, the enlargements that you see on NCIS, CSI, etc don’t really exist…sorry to burst your bubble!

Yes, there are ways that you can enlarge an image in Photoshop using “Resample Image” selection but remember, this still is an algorithm calculating it’s best guess as to what the color of the image should be.

The software will guess what color goes in each red square based on the color and surrounding colors from the white squares.

Just remember that these algorithms can cause problems if you reduce and image just to enlarge it again.  See what I mean below as I take an image, reduce it and then enlarge it to it’s original size.  the problem is that when going from small to large, it looks blurry or more pixelated because the computer is guessing what the pixel color should be.

Original Image Size - click to see full size

File reduced to 520 pixels - click for full size

Reduced file enlarged again to 1040 px - click to see full size


Now when you are working about to create a rendering, it starts off by figuring out what size the rendering needs to be.  For example, let’s say I need a rendering for a hand out.  This tells me that I’ll need at minimum 150dpi.  Go back to our pixel equation from above and we can calculate that our output should be 1650 pixels (150 dpi X 11 inches) by 2550 pixels (150 dpi X 17 inches).  What this also tells me is that with the same output or image, I can crop the image to fit my 1400×900 pixel laptop screen.

You don’t want to create an output too much larger than necessary because time is always going to be an issue.  Remember doubling the output isn’t actually doubling it.  A 11×17 image at 300dpi is actually 4 times (twice the pixels in height times twice the pixels in width)  the pixels as an 11×17 image at 150 dpi.  This means that it will take 4 times longer to render the 300 dpi image than the 150 dpi image!

White square = 1 px @ 150 dpi, Red Square = 1 px @ 300dpi

Again, because it’s easier to reduce rather than enlarge, I can reduce that same image to 8.5×11 easier than enlarge it to 18×24.  However, if I needed that image on a board that would be seen from 5-10 feet away, I know I can get away with about 100dpi which would allow me to enlarge the image to 16.5 inches (1650 pixels/100dpi) by 25.5 inches (2550 pixels/100dpi).

You can now understand how with a decent knowledge of this information a lot of time can be saved without reducing the quality of the image.  I know this was a bit off topic, but hope it helped.