Debunking the Frequency Separation Myth

Frequency Separation aka "FS" is an overly abused technique among photographers and retouchers alike. It is an easy path to excessively retouched images because it gives us a false sense of safety.

What is Frequency Separation?

Frequency Separation (FS) is a Photoshop technique that splits an image into two layers: a high frequency and low frequency layer. The high layer is supposed to contain just the texture of the image, whereas the low layer contains the colors and tones of the image.

FS allows you to even out the tones and correct colors while preserving the texture of an image. It allows you to remove blemishes without affecting the tones and colors as well.

FS in action

Debunking the Assumptions

FS is overly abused. Simpler but standard retouching techniques like healing on a blank layer and dodge & burn are neglected as a result.

Look at Figure 1 below. Do you see any major difference between the top and bottom images? I don't. In both cases the texture is preserved. So why are we abusing FS when cleaning up blemishes?

Steps to replicate:

• Heal via FS: I ran FS and then created a blank layer on top of the high frequency layer. On the blank layer, I used a healing brush to remove the blemishes
• Heal only: I created a blank layer on top of the original image. Then I used a healing brush to clean up the blemishes.

Figure 1: Heal comparison with FS

Assumption #1

The low frequency does not contain any texture.

This isn't true because it varies per image. What is texture by the way? Texture in FS is often the fine, grayish lines you see in the high layer. If you do another FS using the low layer as your base image, you'd find it still contains texture. See Figure 2 below. There is texture in the low layer.

Steps to replicate:

• FS round 1: I set my Gaussian Blur at 1.8 pixels. I added a curve so we can see better the texture in the high layer (otherwise all you see are gray images)
• FS round 2: I deleted the background and high layers from the previous round. I used the remaining low layer as my foundation for running another FS. Again I set my Gaussian Blur at 1.8. Notice there is still texture in the low layer.
• FS at blur 16px: This one I used the original background as the foundation for FS and set my blur settings at 16px.

Figure 2: FS comparison

Assumption #2

The texture of an image is uniform.

The texture varies within an image. Sharper parts of the image have more visible texture. Less sharper parts will contain less texture, but it doesn't mean there is no texture. See Figure 3 below.

Figure 3. Exaggerated high frequency layer to show details

Assumption #3

You only need one round of FS.

Again if we go back to assumption #1 and #2, different parts of the image have varying levels of texture. So if you really want to maximize your FS experience, you'd need to set different blur values per part of the image you're targeting of. That means running FS multiple times.

Assumption #4

Dodge and burn in the low layer produces better results than doing dodge and burn separately.

See Figure 4 below. Do you see any difference? Again I don't see any difference. In both cases I dodged the shadow near the nose and the contour near the cheek for the sake of demonstration

Steps to reproduce:

• d/b method 1: I applied FS on my image. On the low layer, I created a dodge and burn curves. 
• d/b method 2: I created a separate dodge and burn curves

Figure 4. Dodge and burn comparison

Conclusion

Should we use Frequency Separation? Yes, but don't abuse it. Mastering the simple heal brush and the dodge and burn techniques will offer better mileage in the long run than FS alone. So when is FS best utilized? I'll leave that question open to my readers. Please comment below to answer.

Credits

Photography by Mark Serrano
Makeup by Vanessa Serrano
Model: Nikki Sterlinski

Check my portfolio to see more of my work at www.markserranophotography.com