Why would you do this? You can't create data that isn't in the original in the first place and in the second place you can use Gaussian Blur on a layer.
I think in principle such an image-analytical operation might be possible (with potentially significant issues regarding edges) but I’m not convinced of the merit.
I usually just use Pattern Overlay set to Linear Light with a few custom Noise Patterns on blurred content in situations like the OP seems to describe.
Yes there is a retouching method that uses "so called" high frequency and low frequency layers. There is a great discussion in the retouching section on a site called Model Mayhem. Some pretty clever retouchers hang out there. You can use this effect to smooth skin for portraits, and you can vary the smooting from just a small amount to some of the images used in makeup advertising where the model has a perfect yet sort of plastic looking skin. I use this technique now and again working on portrait subjects that have less than perfect skin.
Basically you make two copies of your opening layer, name the first layer "Low Frequency" and the second layer "High Frequency".
Select the Low Frequency Layer and apply a small amount of Gaussian Blur around 2.9 is a good starting point
Now go to the High Frequency Layer and go to Image and select apply image. In the drop down select your low frequency layer and check the invert box and set the blending mode to add with a scale of 2 offset 0 and apply then change the Layer Mode to Linear Light. You now have two retouch layers you can work on and you still maintain some texture in the skin.
To finish off, select the Low Frequency Layer and duplicate and name as smooth skin. Apply a blur, surface blur works best with a starting point of Radius = 7 and Threshold = 6 apply and then create a black mask, Now painting with white over selected parts of the image will smooth out the skin.
Experiment with the settings to get the effect you are looking for. Also, when cloning and healing on the high and low frequency layers look to see what is effected which will give you a better idea on the seperation technique.
Here is a quick example of this technique
The Before Image
The After Image
Now go to the High Frequency Layer and go to Image and select apply image.
Quite frankly I have a hard time taking a technique seriously that utilizes Apply Image instead of simply going with Layers.
Have you tried this technique or just making an observation?
There are many talented retouchers who use this technique and there are many articles and demonstrations out there
And what are those retouchers’ reasons for working destructively in this case instead of using Layers (or Smart Objects even)?
Ar least it seems the result of the Blend Mode and settings you described can be achieved with Layers.
Forgive me not answereing the question: I only did a quick check to see if the Apply Image result could be achieved non-destructively.
And I usually use a different approach with skin retouching.
you devided image into two separations, high and low. The original question was seporation image on any amout of frequences. I also wondering if there is a smart method of doing this.
Yahor, to be honest I have never tried nor delved too deeply into this technique. I tried it on some portraits and it worked well for me. There is a massive discussion on this technique here http://www.modelmayhem.com/po.php?thread_id=439098 you might want to check this out and if you are a member you could post your questions here.
Bleh, there's that whole "non-destructive" thing coming up again.
Some very adept Photoshoppers I know quite often adjust pixel values directly. There's nothing special about always having to do things "non-destructively" in layers. Layers are good for some things and pixel editing is good for others.
Try not to be swayed to an extreme by a mantra. What may seem irresponsible to you is expedient for someone else.
Not everyone is in a situation where they are likely to have to tweak things over and over again either because they have persnickety clients or aren't good enough to get things right the first time.
And you're smart enough to know that if all you do is make a new layer where you've baked changes into the pixels then mix that layer creatively with the original you have your cake and eat it too.
Never save over your original file - there, that's as non-destructive as you can get.
Actually I would say the technique I described (lifted from the link I provided) is very much a non-destructive approach. None of the original pixels are touched as the retouching and changes are done on the high and low pass layers - well that is my description of non-destructive editing. But then again who does not work on a copy of a file - that lesson can be discovered the hard way........
I also agree with Noel's comment, what works for one person may not work or be acceptable to others - who cares if the end result is achieved.
But this is what makes the Internet and forums like this an exciting place; there are soo many experts out there with their I am the only one that is right perspective and just provide negative comment rather than constructive feedback.
I’ve voiced my opinion, I do try to stay polite, if my feedback is perceived as negative – so what?
I like to point out what I see as improved options with regard to non-destructive working because I am under the impression that there are a lot of tutorials about that are either older or employ a needlessly destructive approach.
And while having the original layer when the result is close to impossible to recreate exactly qualifies as non-destructive it falls short of what I strive for in image editing.
But like Noel indicated the client-situation (and also the necessity of collaborating with other image editors) influences my attitude.