Pro Talk: When to Sharpen?

(© Ian Plant) I get a lot of questions from people via email, Facebook, Twitter, Google+, etc. about locations, equipment, processing workflow, and all sorts of other stuff. Unfortunately, I don’t have time to respond to everyone, so a lot of questions go unanswered. That’s why I’ve decided to start a regular feature on my blog called Pro Talk, where I take some time to answer a reader question and share it with everyone—think of it as “Dear Abbey” with more testosterone and less talk about cheating spouses. Unfortunately, my busy travel schedule and lack of office staff means I will still likely be unable to respond to many (if not most) questions which come my way, but whenever I get a question that I think will be useful to my readers, I’ll take some time to put it up here on the blog.

Today, I’d like to share a question from Damien, who asks:

Was reading your tutorial on saving images for the web. As I do HDR images and always sharpen images during the Photoshop process . . . are you suggesting to sharpen them again once converted to a srgb for the web? Just notice in Photoshop they look better then they do on the web. Does this extra sharpening help?

It seems to me that this question raises two issues about sharpening. First, does it ever make sense to sharpen an image during the raw conversion or image editing phase? And second, does it make sense to sharpen an image when posting it to the web? I’ll answer each question in turn.

As for the first question, in my opinion, there is little sense in sharpening an image during the raw conversion or processing phase. Here’s why: your master image file is not likely to be output at its exact size. Rather, you are likely to resize the image when creating a print or posting to the Internet. In the case of making a print, you might need to enlarge the master file. If you have applied sharpening to the master file, then you have likely done so to optimize the look of the file at its native size. If you enlarge that file, however, any sharpening effects will be magnified, in some cases unpleasantly so. What might look good for the file at native size might look terrible when the file is enlarged to make a 30″x40″ wall print. So, as a general rule, I don’t apply much sharpening to my master files—usually, the only sharpening I apply is the default level of sharpening applied to a raw file when I am converting the file using Adobe Camera Raw (which is very mild). That way, I can upsize or downsize my master file without fear of unsightly sharpening artifacts ruining quality—and I can optimize sharpening for my final intended output after resizing.

As for the second question, when posting an image to the Internet, you are typically downsizing the image (often considerably), and a general loss of apparent sharpness can sometimes be the result. I don’t always sharpen an image after I downsize it for the web, and when I do, I prefer to apply sharpening lightly and incrementally in order to avoid over-sharpening parts of the image. Sometimes, I will sharpen the image using layer masking techniques, so I can apply more aggressive sharpening to areas of the image which need it, and apply less aggressive or no sharpening to areas which don’t. Please note that sharpening is to some degree a matter of taste, although an over-sharpened image will have obvious artifacts (especially around high contrast edges), whereas an under-sharpened image might appear visibly blurry.

Here is an example of what sharpening can do for your web images. For the first image, I did not do any sharpening at all, I merely downsized the image and saved as a jpeg for the web (with some mild image compression to keep the file size manageable).

For this second image, I applied sharpening globally to the image. Some parts of the bird look perfectly sharpened, but other parts—especially some of the backlit feathers—look a bit over-sharpened to my eyes.


For the final image, I applied the same global sharpening, but this time I did it on a duplicate layer. I then created a layer mask, and masked out those areas which appeared over-sharpened to me. The end result is somewhere in between the first two versions. The differences are somewhat subtle, but with the final image I managed to take some of the edge off of the over-sharpening applied to the backlit feathers in the second image, and was able to make the eye and beak look cripser than the original. Please note that when posting to the web, you also have to contend with downsizing and compression artifacts, which you have considerably less ability to fix.

Optimal sharpening

I hope this post is helpful! My Creative Digital Processing Videos explain in detail many of the techniques discussed here, so if you are interested in learning more I encourage you to check them out. Thanks!