Intelligent Sharpen 2.0

I’ve been promising lots of people to explain how I increase image sharpness for Web images. I’ve talked about sharpening before several times, but new techniques come with experience. The following is by no means my own discovery, but I hope it will be useful to some.

I tried to make this article as practicable as I could, so we’ll leave out theoretical considerations and general tips.

About a year ago I published a Photoshop action called Intelligent Sharpen, developed from the materials I’d studied. Today’s piece is another take on the same ideas, so instead of drawing up a new action, I decided to upgrade the old one to an Intelligent Sharpen 2.0.

Download intelligent_sharpen_2.0.zip

About a year ago I published a Photoshop action called Intelligent Sharpen, developed from the materials I’d studied. Today’s piece is another take on the same ideas, so instead of drawing up a new action, I decided to upgrade the old one to an Intelligent Sharpen 2.0.

The new action has two parts:

L Contour Light/Dark USM

Separate light/dark contour sharpening in the brightness channel. This was the original Intelligent method, described in the first version (in Russian) of this piece. Comes with instructions for using the action in Adobe Photoshop.

Light/Dark SS

Separate light/dark contour sharpening using Smart Sharpen. This is the new way described below, and the one I usually use nowadays.

Here we use the Smart Sharpen tool, which can overcome various kinds of blur and within limits can manage sharpness in bright and dark areas of the image. Alik has worked out and suggested Sharpen settings to use in small sizes (~900×600). I’ve done my own research and I believe they’re a good baseline, work well for most images, with tweaks added case by case.

Applying Smart Sharpen at Alik’s values directly sharpens the image considerably — a little too much, in fact. For a more gentle effect it would make sense to blend a “sharp” layer and the original given some transparency. If we use not one but two Sharpen-adjusted layers, in Lighten and Darken modes, with different opacity levels, we can fine-tune sharpness separately for shadows and highlights. Adding masks, if necessary, gives even more options.

The action at default settings gives the following results.

1. The original is resized to 900×600 pixels.

2. Intelligent Sharpen action (L Contour Light/Dark USM) is applied.

3. Intelligent Sharpen action (Light/Dark SS) is applied.

I chose an example image with both strong contours (branches) and textures (water).

The 900×600 image size was taken as an average size, very common on the Web for 3:2 camera. I personally apply the action to images from 300х200 to 1200×800 or so. It works, only in small sizes the excessive sharpening is more obvious. So the smaller the image, the less transparent I make my Darken and Lighten layers. A 300х200 picture gets about 40% and 20% transparency respectively. There are other ways to manage sharpening based on image size, too.

It’s no secret that the resizing method also bears on the final result. I use a simple Image — Resize with a Bicubic resampling method. As far as I can see, different methods, including downsizing the image repeatedly, look pretty much the same.

There are no hard and fast rules for every case, and this article only describes one possible approach. Take the values and settings as a baseline and work from there. I appreciate any feedback and shared experience. Those who want to experiment with my example image can get their hands on the original, processed but not yet resized, here (TIFF, 33 Mb).

P.S. My thanks go to Alik Vaitsiakhovich’s for his timely advice and ideas he published in his article. I drew on it for the Intelligent Sharpen(ing) 2.0 action.

“LIFELIKE: A Book on Color in Digital Photography”

Advertisement