Frequency separation technique in photoshop


Totally unrelated photo

If you are wondering what frequency separation is, it’s basically just separating your colors and your texture. Now why would you do that? Well, it’s especially useful if you are editing a portrait photo and you want to do skin smoothing, removing blemishes etc. With this technique you can do all that without loosing texture in the skin. This technique is used a lot with model photos as it’s really important to keep the photo looking “real” and that is easy when the texture isn’t on the color layer.

So how is it done? Well it’s really easy.

1. Open up your image in Photoshop.
2. Make 2 copies of the background layer Naming the bottom one “Low” and the upper one “High”
3. Select your Low layer and go to filter – blur – Gaussian Blur nd blur this layer at 10px (guideline – You can do more or less if you wish)
4. Select your High layer and go to Image – Apply image and depending on whether you are editing a 8 bit or 16 bit image.

8 bit image:
Layer: low
Blending: Subtract
Scale: 2
Offset: 128

16 bit image:
Layer: low
Blending: Subtract
Scale: 2
Offset: 0

5. Now set the blending mode of your High layer to Linear light.
6. That’s it. Now your high and low layer should give you exactly the same as the background layer. Now you can begin to edit your high and low layer separately.

I’m sorry I didn’t have a photo that fit this technique. I will upload a couple of images when I find a fit 🙂 Please try this out on your next portrait edit. And then after you have done this, you start using your dodge and burn on you low layer to get rid of those blemishes.

That’s how I do Frequency separation in Photoshop. Be sure to tell me how it works out.

11 Responses to “Frequency separation technique in photoshop”

  1. Michael Says:

    Sounds interesting, but it doesn’t work for me in practice. What I end up getting is a black screen using this method. What photoshop are you using? I’m using CS5.5

  2. Thomas Says:

    Hey Michael,
    I’m pretty sure this is still working in photoshop CS5. In step 4 make sure the “layer” in the popup box is set to the layer you called “low”, instead of merged, which might be the default setting on yours. I think that will solve your problem.

    Let me know if it doesn’t help 🙂

  3. borden Says:

    I tried and got the same as Michael – a black screen. I’m positive I had it all set up right.

  4. Joe Gunawan | Says:

    Make sure when you’re on the Apply Image, you check the “Invert” box by the Layers. That should fix it.

  5. Joe Gunawan | Says:

    And for 16-bit image, you need it to be set on “Add”, not “Subtract”

  6. Rohan Says:

    Hi Thomas,

    Wonderful technique, used it and I really like it.
    Now my texture of the skin doesn’t distort while I remove the discoloration from the skin.

    The high and low names confused me so I changed that a bit.
    named it “texture” and “color” respectively.

    Thanks again!

  7. Thomas Says:

    Thanks Rohan,
    Glad that you found the technique useful and a great idea for naming of the layers.

    And thanks for the comments Joe, some great pointers 🙂

  8. Cathleen Says:

    Hi Thomas and everyone else,

    I’ve seen tutorials for this technique on a couple of sites but each tutorial is pretty specific to one piece of software. Do you know if all the features needed for this are available in Photoshop Elements? I know that the Gimp methods I’ve found do not work as well, so I do have to purchase software. I just want to know before I spend $100 if I’m just going to have to get the full software anyway at seven times that.

    Thanks for any answers you can give.

  9. Thomas Says:

    Hi Cathleen,
    I haven’t used photoshop elements for quite some time now. But a search around the web seems to indicate that you can get good results with an adapted technique 🙂 seems to produce some nice looking images.

    I think(might be wrong) that adobe offers an upgrade price for photoshop elements to photoshop. So you might want to look into that too before buying anything.

    Hope this helps a bit.

  10. Alayjah Says:

    […] mind!). I achieved this using the frequency separation method which you can find an explanation of here Attachment 69520 The pose itself is usable as long as it's not so extreme. For example the only […]

  11. Sara Says:

    Hi there,

    Thanks so much for the article but I am SO confused!! How can you tell I’m just starting out right!? The part where I get confused is 8 bit image:
    Layer: low
    Blending: Subtract
    Scale: 2
    Offset: 128

    16 bit image:
    Layer: low
    Blending: Subtract
    Scale: 2
    Offset: 0
    How do I know if I’m using an 8 or 16 bit image?What’s the difference etc. The blending scale etc where does that come in to play? What does that mean? Is it options on the box that opens up? I’ve never seen that any where. What’s the difference etc. Can you explain that part in a little bit more detail!! Sorry, I know that wasn’t the point of the article but I’m thinking I will need to know this in order to use the technique!? Thank you so much!