I know I’ve already got a “Myths” page, but this is the biggest myth of all, so it gets its own article!
Very often, in discussions about printing, I read the advice: “Don’t resample your image. Just resize it to the correct dimensions, and let the resolution fall where it may. Resampling reduces quality”.
This is misleading, and potentially dangerous. If someone gives you this “advice”, smile and thank them politely, because they mean well, but disregard it.
Resampling is, in fact, vitally necessary for good quality prints, for one important reason – sharpening.
The Mechanics of Sharpening
You’re probably familiar with Unsharp Mask. (If not, please read my explanation here, then come back to this article.) If so, you’ll be aware of the critical role that the sharpening Radius plays … too wide, and you get visible sharpening haloes (yuck!); too thin, and your sharpening isn’t effective at all (d’oh!).
Radius is measured in pixels. Therefore, the width of the Radius is dependant on the size of the pixels in the image. In a 300ppi file, a 0.5px Radius is 1/600th of an inch wide (which might not seem much, but is actually pretty good); but in a 500ppi file, that 0.5px Radius is now only 1/1000th of an inch wide (which is close to invisible).
So, you’re beginning to see the problem. Let’s say you shot a portrait session with your ten megapixel camera. When you prepare the proofs for your customer, you crop some of the images a little bit, others a lot, and some not at all.
By the time you finish, you end up with images ranging from ten megapixels down to less than five megapixels.
Then, your client places their order. “I’ll have all those as 6x4s, those as 5x7s, that one at 8×12, and I want that one at 12×18 for my wall”.
So what do you get if you resize all of these images for print, and “let the resolution fall where it may”? A dog’s breakfast, that’s what. You’ll end up with resolutions ranging from less than 200ppi to over 600ppi.
How on earth can you sharpen consistently and effectively in those conditions? Your sharpening settings might look good on one print, but completely invisible on another, and much too strong and ugly on another!
No, my friends, if sharpness is important to you, you MUST resize your print files to a consistent resolution. It’s the only way to get the results you desire.
What Resolution? 300?
300ppi is the most commonly quoted resolution for printing. Personally, I use 250, others use anywhere from 200 to 360. Ask your lab what they recommend.
The point is, it doesn’t really matter what resolution you use. Just pick one and stick to it. All that matters is consistency – the same res every time.
What Sharpening Settings To Use?
My print sharpening class will answer that question for you. Otherwise, do your own tests with various settings, and find out what works best for you.
You see, there are a few unique factors that control the sharpening that you do. First and foremost is your lab – some labs sharpen, others don’t. Results will vary from lab to lab. Pick one that you like and stick to it. (If you change labs, you’ll need to do the sharpness tests again, and change your habits if necessary.) Your screen is also a factor – some screens are sharper than others. Your camera is a factor, particularly regarding its noise. And ultimately, your taste is a factor … you may like your images more or less sharp than the next photographer.
Different Resampling Methods
If you are a real nerd, you should also run test prints of Photoshop’s three main Interpolation Methods – Bicubic, Bicubic Sharper and Bicubic Smoother. “Bicubic” is the original; “Sharper” is purported to be good for downsampling; and “Smoother” is intended for upsampling.
Personally, I stick with plain old Bicubic. In all the tests I’ve done, I’ve never seen enough difference to make me give a damn, and frankly, it’s one less thing to remember. Given that I do nearly all my resizing with the Crop Tool (which requires a visit to Photoshop’s Preferences every time you want to change the Interpolation Method) it’s not worth my time, or my sanity, to bother changing my Method.
But don’t take my word for it. Try it yourself and make the right call for your own workflow.
A Note About Cropping
Just a reminder that this whole discussion about resampling is only relevant for output – ie printing. If you crop or resize during your workflow as part of your post-processing (a very bad idea, by the way), you definitely should not resample. Every pixel of your master file should be preserved.
Another Note About Web Sharpening
So far I’ve only mentioned sharpening for print, but exactly the same principles apply for web images. Make sure you resize to the exact required pixel dimensions before sharpening (remember, PPI is irrelevant for web). If you sharpen before resizing, or at the wrong size, the effect will be lost or distorted.
The great thing about web sharpening, of course, is that no test prints are required. What you see is what you get, so if it looks good, do it!
Yet Another Note About Raw Sharpening
It’s common to apply some sharpening to Raw files. This early-workflow sharpening is called “Capture Sharpening”.
Raw sharpening is a very good idea, but don’t go over the top – a mild amount will do. Again, the important principle is consistency.
Let’s Bust Some Myths!
The “Don’t resample” brigade have authentic-sounding arguments in their case for letting the resolution fall where it may. Let’s discuss a few, and inject some reality …
Argument 1: 300ppi is an out-of-date standard
That’s fine. If 300ppi offends you, use something else. The key to great sharpening is not what resolution you use, but how consistently you use it.
Argument 2: The lab’s software can resample better than you
That might be true, but will your lab’s software sharpen satisfactorily for you afterwards? Maybe, maybe not. Do your testing and find out.
Argument 3: Downsampling removes pixels
This statement always makes it sound as if every second pixel is simply removed, and the remaining ones moved closer together. This is not the case, of course.
All of the pixels are mathematically averaged and merged (using the Interpolation Method) to create a smaller representation of the images in fewer pixels.
Argument 4: Downsampling removes detail
Yes, eventually it does, but again, this statement is misleading. What downsampling really does is shrinks detail until it eventually becomes so small that it is lost.
But the actual loss of detail occurs after it becomes invisible to the human eye anyway, so what does it matter? Have you ever stopped to think how tiny a 300ppi pixel really is? It’s invisible. Heck, we don’t even notice our 100ppi screen pixels most of the time.