The title is pure nonsense. I live 20 miles outside of New York. That’s it. The complete description of the distance. Actually, I live 21.19 miles from B&H in Manhattan, which is like magnetic north for many of us photographers. We always feel drawn towards the toy-store. Now that 21.19 miles of course could mean either 30 minutes or 3 hours of driving, depending on traffic through the Lincoln tunnel. But I digress.
I could have described my distance from my house to New York as 20 miles, or disregarding traffic, as 18 minutes at 65mph. But it is nonsensical to describe my distance from New York as 20 miles at 65 mph. The 65mph becomes a superfluous bit of data when stated like that.
So why the strange title for this posting? Because as a digital photographer I see the same kind of nonsense perpetrated on a daily basis when image size is described in terms of absolute pixel dimensions (600 x 400 pixels) and a specific dpi such as 72dpi or 300 dpi also demanded.
It is unnecessary, or worse, confusing.
I am sure everyone has seen image size requested as 600 x 400 pixels at 72 dpi. Or 300 dpi. When an image is described as 600×400 pixels, then the dpi (or ppi) value is of no consequence. It means nothing. It has become the superfluous 65mph description. Since the actual image size is already described in very specific terms, (600 x 400 pixels), there is nothing else to add. That’s the size – the pixel dimension – 600 pixels by 400 pixels. We’re done. No dpi or ppi value needed.
Where the dpi (or ppi) value comes in, is in describing how large the image will be printed. But before we get there … people use dpi (dots per inch) when they mean to say ppi (pixels per inch). Dots per inch (dpi) is mostly a legacy term to describe how printers resolve detail. How many dots per inch it can print. It is largely meaningless to us as digital photographers. ppi however, will describe how large the image will be printed at the current ppi setting. But the terms ppi and dpi are often used to mean the same thing. So it has to be read in context.
There are numerous articles about this on the internet, and a quick search will bring them up. Here’s a start. And another. Even more. There are so many articles on the subject that this posting hardly seems necessary, but it grates me when so many digital photographers clearly don’t understand something as basic as image size and resolution.
Let’s look at an actual image opened up in Photoshop, with the Image >> Size dialog box open:
.
Looking just at the dialog box, the pixel dimension of this particular image is 2592 x 3888 pixels straight from the camera. If we do some quick math, 3888 / 2592 = 1.5 and this means that the aspect ratio of this image is 3:2 which is the usual aspect ratio for 35mm film cameras and most D-SLRs. This ratio of 3:2 also means that the image can be printed as 6×4 without cropping any part of it. You will have to resize it, but not crop it. But 8×10 or 5×7 which are different aspect ratios, will need cropping.

Still looking at the same dialog box as it opens to the default – if I wanted to use 350 pixels per inch, and I have 3888 pixels on the longest side, then simple math with a calculator tells us we can get an image 3888 / 350 = 11.109 inches wide. And there it is .. the dialog box already did the math for us. We have an image that will print as 11 x 7.4 inches at 350 pixels per inch, if we didn’t resize the image from the camera’s resolution.
The camera in this case was the Canon 1D mk3, and the specs say it is a 10 megapixel camera. If we look at the resolution, 2592 by 3888 and pull out the calculator again, we get 2592 x 3888 = 10,077,696 … and that is a touch over 10 megapixels – the resolution of our camera. Without resizing the file in Photoshop, that’s the pixel dimensions that the image will open up with. How large the image will print, as it is, depends on how many pixels per inch we want to use.
To change the size the image will print to, without changing the pixel dimensions, we can uncheck the ‘Resample Image’ button. That keeps the pixel dimension of the image static.

Doing so will allow us to change the size the image will be printed to (measured in inches) without affecting the basic resolution of the image as it came from the camera.
The next two screengrabs are where the ppi value was changed, and the pixel dimensions are then changed accordingly by Photoshop via resampling the image data. ie, the image is made larger or smaller.


The numerical values in this screengrab should be easy to understand. If I want an image that is 20×30 inches, and I want / need the resolution to be 300 ppi, then the basic math tells us that we’d need 20×300 = 6000 pixels along the one side to have an image that size. Basic math and basic geometry with the aspect ratio.
In this instance, the ppi value makes sense … a certain ppi at at required size (in inches).
Describing the print size in this way makes sense. (It is equivalent to traveling for 18 minutes at 65 mph. It is a description that makes sense.)
But when the image is described in pixel dimensions (eg 6000 x 9000 pixels), then the ppi doesn’t matter at all. The image size is described completely by just those two values. (ie, I live 20 miles from New York.)
![]()
If you find these articles interesting and of value, then you can help by using
these affiliate links to order equipment & other goodies. Thank you!
Stay informed of new articles via the monthly newsletter.
Also join us on the Tangents forum for further discussions.
![]()




Amen! One thing I dont understand is the PPI box that Lightroom has when exporting images. What does it do? Is it just attaching a PPI valuie to the Metadata?
Comment by Andrew Kraker — November 19, 2008 @ 12:55 pm
Here is another link describing how dpi (and ppi) does not apply if the image is for screen display only.
http://www.scantips.com/no72dpi.html.
This link is important because people who work with print images get the wrong idea when trying to put those mages on the web. For example, scanning a photograph into the computer and then telling someone to set that image at a certain dpi, which is meaningless for screen display.
Comment by Stephen T. — November 19, 2008 @ 1:58 pm
You are right in almost every respect. Where you are wrong is when it comes to software that uses the embedded pixel resolution to determine how big to display an image. Such software will display an image embedded with 72 dpi very differently than the same image that has an embedded resolution of 300 dpi. Microsoft Office is a perfect example of where this comes to play. Try importing multiple copies of the same image with different embedded resolutions into MS Word or MS Powerpoint and see what happens.
This is indirectly the answer to Andrew Krakker’s question about the Lightroom export dialog. It asks for the resolution because every image has the resolution embedded into the metadata. For most software programs it really doesn’t matter what that resolution is, but for some software it does.
Comment by Sean Phillips — November 19, 2008 @ 5:50 pm
Comment by Neil — November 19, 2008 @ 6:48 pm
Thanks Neil! I often point people to your blog when they ask questions like this on forums and this is just another in a long line of extremely well written, informative pieces from you. Keep up the good work!
Comment by Stefanie — November 19, 2008 @ 7:19 pm
Great post… which reminds me of something I’ve been thinking for the last couple of days…
I’m planning on sending some of my 12MP pictures to an online printing service… they accept anything from 100ppi, being 300ppi their recommended resolution. I’d like to test some 20″x16″ printing which means my original pictures will have to be cropped (from 4288×2848 to 3560×2848) giving me an efective 178ppi resolution on the 20″x16″ paper.
Now, I’m wondering… do you think 178ppi is enough for a casual observer not to notice the lack of resolution?… and also, should I just crop the image and send it with 178ppi or should I crop and resample the image up to 6000×4800 so the resolution becomes 300ppi?… I was just thinking that maybe CS3 resampling is something I could trust (or not?)
Anyway thanks for the reply and keep up the good work!
Alberto
wonder
Comment by Alberto — November 19, 2008 @ 10:36 pm
Wow Neil! It looks like you’ve been reading that “Advanced Calculas for Genuises” book again! Thank you for this post, but it still leaves me with one question. What do you save your images as when you convert from RAW? I use Canons DPP (as do you) and I usually put ppi at 300 and change the size to 6″x9″. Sometimes I have to compromise between resolution and file size. Is this any different than choosing 100 ppi @ 18″x27″? Thanks
Comment by Daniel — November 21, 2008 @ 4:48 pm
Ok, I just realized I spelled “calculus” wrong. Do you have a “genius” book for spelling that I can borrow?
Comment by Daniel — November 21, 2008 @ 4:50 pm
I have always taken my files to a printing service lab and requested a 12×18, which is the size I usually enlarge to.
I shoot with my 4 megapixels D2H but many times, the file comes from the D70S that has 6 megapixels. Both files, when enlarged, have looked great. I leave resampling the image to the service lab.
When “resample image” is not checked the photographer can add the size he wishes to print, like my 18 inches in the widest dimension of the print. Photoshop will add the height and the ppi. The ppi could be much less than the usual 300 and still, if not under 100, the print will be fine simply because, by not checking resampling, the quality of the original has not changed.
If something I have learned from this method (taught by Scott Kelby) is that we do not need high ppi images to have an excellent print.
Comment by William Rodriguez — November 24, 2008 @ 2:26 am
Tam, I don’t really agree with what you want to demonstrate ?
you say : “And seems like everyone is made to believe that its a 27-105mm equivalent” and you say “Its still a 18-70mm (cause it was specifically designed for an APS-C sensoer camera)” and “The multiplication factor comes into play only when you stick a (non-DX) lens of a APS-C sensor camera.”
I totally disagree:
Your 18-70mm IS a 18-70mm because of its optical specifications, even when not mounted on a camera. BUT because of the sensor size of APS-C cameras, your 18-70mm field of view matches the field of view of a 27-105mm zoom with a 24×36 size sensor.
And if it is a DX lens or not doesn’t matter:
a 18-70 DX and a 18-70 non DX zoom will project the same image on an APS-C sensor. The image from the non DX zoom have a wider field of view, but the borders are lost outside of the APS-C sensor. In both cases the captured image matches the field of view of a 27-105 on a 35mm camera.
Hope this help
Comment by kurodo — November 26, 2008 @ 7:40 am
Nice clear explanation Neil – thanks!
Alberto – the boys from Luminous Landscape have a video on printing which discusses exactly your question. They suggest that if your native resolution is between 180 ppi and 480 ppi you should send the native pixels (ie not resampled) to the printer. However in your case for the size you want your ppi drops (slightly) below 180ppi.
Their opinion is that in that case you can easily upsample to 200% with CS3 provided that the image you use is good to begin with (ie tack sharp and well exposed) and properly sharpened.
There is lots of detail in the LL video but Jeff Schewe also wrote an article about this in Digital Photo Pro magazine – The art of up-res. See http://www.digitalphotopro.com/technique/software-technique/the-art-of-the-up-res.html
In your case I’d do a simple experiment. Create the cropped 20″ image with native pixel resolution of 178ppi. Then create another copy where you resample to the required 300ppi. Then after you have upsampled, you need to resize to the desired 20″ (untick resampling).
Finally create one last image with the left or right halves of both test images and send that off to your online printing service. You then have one 20″ test where you can see the effect of both settings side by side.
Heck I may just do this myself on my new Epson printer – sounds like fun for a rainy day! Let us know what worked for you!
Cheers,
Ed
Comment by Ed — November 26, 2008 @ 1:51 pm
Thanks, Kurodo – glad you added your comments.
The DX Dilemma has been a rats nest since Day One. Since someone can put a DX lens on an FX body, it needs to be labeled at its optical spec, even though it isn’t going to present that field of view on its intended (DX) DSLR body. Just confuses a lot of people – always wished they had put both focal length representations on them.
Comment by Brian Patterson — November 27, 2008 @ 3:57 pm
Thanks for your input guys… Ed, I’ll take a look at that LL video you suggested… and I think I’ll try that test image too. I’ll let you know how it goes. Regards, Alberto
Comment by Alberto — November 29, 2008 @ 11:29 am