I’m always reading tutorials—every day I take in at least one (and often more)—so I can see how other people are approaching things, and how they teach it, there’s always something to be learned. I also like to see if I can “improve” the steps taken to reproduce the effect, and try to make it more “efficient”(at least, for me anyway).
There are dozens of techniques to produce a pencil sketch look from a photo; my personal favourite up until a few years ago (ok, probably a dozen) was the duplicated layer, inverted, blurred and blended using color dodge—you can find that one all over the place on the web. Let’s take a quick look at that:
The Color Dodge Mode Pencil Sketch Look
Take an image—here I’m using “Beautiful Woman with Healthy Long Hair” by koji6aca on Adobe Stock; I’m not licensing the image at this point so I’m going to work with the watermarked FPO at the moment. There is usually a step mentioned here that makes the image desaturated to start with—we’ll stick to colour for now as we can always add that step later. Here are the steps in this method:
- Duplicate the layer (⌘-J)
- Invert the duplicate (⌘-I)
- Set the blend mode of the duplicate layer to Color Dodge (SHIFT-ALT-D with any non-painting tool selected)
- Apply the Gaussian Blur filter to the duplicate; tune as desired
- Optionally, add a Levels adjustment to make it more “contrasty” (and if you do want the monochrome look, also add a Hue/Saturation adjustment layer, with the Saturation value at zero)
The Divide Mode Version (Layered)
More recently, specifically since the advent of the Divide blend mode, I’ve favoured this one:
- Duplicate the Layer; there’s no need to invert the copy, so one step reduced already
- Set the blend mode to Divide—sadly no accelerator key combo for this one AFAIK, such a shame it isn’t SHIFT-ALT-/ as that would really make sense, don’t you think?
- Use the Gaussian Blur filter, as in the previous method
- Optionally use the Levels adjustment layer
The Divide Mode Version, Working Smarter
So what’s wrong with those methods? Well, technically nothing—they both work perfectly well—but they are both destructive. I’ve seen a tutorial today (using the Color Dodge method) while I was fact-checking that did use a smart object on the duplicated layer so that the Gaussian Blur filter could be tuned, but it could be better. In fact, if this were a an actual brief, then it could be a whole lot better.
Let’s pretend, for a moment, that we need to use this technique on a real job, and we need to use a stock image. The previous two methods both then start to crumble as steps 2 onward need to be repeated once you have the final asset in place, so they would have to be reworked. This method holds good for this entire process:
- Convert your existing image to a smart object
- Apply the Gaussian Blur filter (you don’t need to worry too much about the settings, around 3-4 should do for now
- In the Layers Panel, you’ll see that Gaussian Blur appears as a smart filter (you can double-click on those words in the panel at any time to revisit the filter); on the right of the filter is the filter blending icon—double-click that and change the blending mode to Divide
- Just as before, optionally add adjustment layers as desired. There’s no difference in the way it’s applied, so if you’re comparing versions here then I’ll save you the trouble as I used different settings each time.
- Sticking with our “real world scenario” we now put this on to Creative Cloud, and invite our client to comment on the work, and approve the image we have selected. The client approves and all we need to do is license the image, which we can do from the library panel. If this doesn’t update automatically, simply open the smart object and replace the contents with the licensed version—the workflow is best when you place a linked copy to start off with, but I was trying to keep a “level field” for each effect.
- You’re now free to tune as desired, non-destructively