Why blur color layer when doing frequency seperation

[photomike]

This is a question out of curiosity more than anything else. I am somewhat new to using Photoshop and using frequency separation for retouching images. Every tutorial that I have seen on it says to blur the color layer? What is the purpose for doing this?

 

[thebestcpu]

Hi @photomike
More forum members may jump in with their thoughts (as I am not the strongest expert on frequency separation).
The background as I understand is that the human eye sees luminosity changes with the highest resolution and much less resolution for color.
So when trying to extract frequency components of the image (in particular the higher frequency) the focus is on the luminosity component of the image.
Blurring the color Layer is just to help with focusing on the luminosity.

Yet you have made me curious. Could you share a link to the frequency separation technique you are using. I have seen some variations and would like to investigate this more for my own learning as well.
Thanks
John Wheeler

 

[Rich54]

The technical science behind Photoshop is not my strength, but here's my layman's understanding of how frequency separation (FS) works:

• In general, the purpose of FS is to extract and separate the texture from the color. The classic use of FS is for skin retouching, where you want to smooth-out blotchy or damaged skin while retaining the intricate pattern of pores, wrinkles and the non-damaged areas of skin texture.
• To achieve this, FS employs the Subtract blend mode (as a setting found within the command Image>Apply Image). The Subtract blend mode does a mathematical calculation—subtraction—of the brightness levels between the two layers that are being "applied" to each other. If you have two identical layers and subtract one from the other, the result will be pure black because you are subtracting a given level of brightness from itself, leaving "nothing" behind... zero brightness.
• In order for the Subtract blend mode to do anything useful, the two layers you are subtracting must have something different about them—different levels of brightness—so that there is something left behind that is non-black. This is where the blurring step in frequency separation becomes essential.
• Let's say we have this image below and we want to extract the sandy texture from the green background by using FS. We first create two identical layers. Then, in order for the Subtract blend mode to "identify" the sandy texture, we need to blur the bottom layer to make that texture different enough from the top layer so that it is distinguishable. If we now apply the Subtract blend mode (within the Apply Image command), the green areas subtract from themselves and leave nothing behind (i.e., black), but the grainy parts are now different enough—from one layer to the other—that the grainy pattern remains behind.
• In the Apply Image command, we must specifically set the "offset" box to the number 128. A brightness level of 128 is exactly 50% Gray. So now, instead of the Subtract blend mode resulting in black whenever the two layers are identical, it results in 50% Gray instead. We can then make 50% Gray disappear in the top "texture" layer by applying a blend mode of Linear Light. Other blend modes (Overlay, Hard Light, etc.) will also make 50% Gray invisible, but Linear Light is the recommended one. Once you've made all the 50% Gray disappear, you're left with just the pure texture pattern in your Linear Light top layer.
• Most FS tutorials I've seen don't tell you exactly how much to blur the bottom layer. What I've learned through trial & error is that you first identify the textures you want to retain via FS. Then, on the bottom layer, you blur just enough until that texture disappears or is smoothed away, but not so much blur that you introduce too much brightness differences in the "color" portions of the image that you're not interested in.

Hope this makes sense and is not to wordy.

 

[Rich54]

Should have said... "hope this is not too wordy".

 

[polarwoc]

There is an Optic Illusion...

...taken from this web page that deals with Frequency Separation. This image is obtained by having Albert Einstein in the High Frequency layer which is sharp while there is an image of Marilyn Monroe in the Low Frequency layer which is blurred. You would be able to see Marilyn Monroe when you move away from the screen and see the image from a distance or just see the image with focus being off the screen while squinting.

The blurring is done in Low Frequency layer to retain the general colour in patches while forgetting the finer details. It is the responsibility of the High Frequency layer to remember the finer details. Together, they complete the image but this separation gives you the option to deal with colour issues without affecting finer details in one layer while dealing with finer details in another layer without affecting colour.