For a few years frequency separation has been a great way to air brush portraits. I am not a big portrait photographer, I prefer working with landscape and still life more, but as an avid learner of Photoshop I did read up on how this technique works and was happy with its outcome. Recently however I have been using frequency separation for my landscape and architectural work, especially with merged (HDR) images.
Frequency separation allows you to “clean” an image without creating the patchiness or the smudged look you get from using the clone and heal tools. Frequency separation works by separating the low frequency (Colours and light tones) details from High frequency (textures and fine details) details, allowing you to edit them separately. In portraiture this allows you to remove wrinkles without effecting the skin tones and remove blemishes without smudging the skin’s texture.
It is not uncommon in landscape images for there to a be a strange patch of colour, lens flare and with HDR images especially, strange textures appearing in the sky and clouds from too much micro contrast. Recently I was editing a shot that I took in a church in Uniejow that had some issues and I used frequency separation to rescue it.
The picture is a 32bit blended image that was edited in Lightroom. On the front two pews on the right there was a strange chromatic aberration hot spot really killing the image. I tried a few things to remove it and nothing was working. I thought why not whip out the frequency separation workflow and see if it will help. I thought it should work because I would be able to remove the colour cast and then separately remove the edges or the artefact.
I opened the image in Photoshop and duplicated the layer twice. On the second layer (the first duplication layer) I applied a Gaussian Blur of 5 pixels and named the layer blur. This layer is where I will edit the low frequency part of the image.
I clicked on the top layer and named it texture, then went Image-Apply Image; I set the layer to the blur layer, blending to subtract, scale to 2 and offset to 128. Usually when I have finished this step I would change the blending of the layer to Linear light but this was too strong so I switched to Overlay.
This has created frequency separation, if you were to delete the background layer it would make no difference as all the information is now in these two new layers. I added a new layer above blur, with the clone tool I selected a soft brush and set sample to Current and Below. I cloned away the problem areas using the colour at the bottom of the pews.
With the high frequency layer (which I called texture) I duplicated this layer and worked on a copy once again using the clone tool, this time I setting the sample to current layer. The reason I don’t clone on a new layer as I did with the Blur layer, is because it doesn’t work in the same way and just multiplies the intensity of the texture.
As I worked I did switch between them and zoom out to make sure everything looked natural and then zoomed back in.
Click on the before and after images below to see the difference.
Once finished, I sharpened the image as the process had softened it slightly (because I used overlay instead of linear light). I also selectively cooled down the pews so that they would fit with the overall tone of the image. It took about 45mins to clean this area, but if I had used just clone and heal tools or even content aware I feel that the area would look obviously edited whereas here it is hard to notice. I am planning for a future ”before and after forum” post to show this workflow off with portraits as there are some different steps in editing that you can take.
Let me know in the comments what you think of this process, is it something useful for you, is it something you want to give a go?
Remember if you liked this post to; like, share and subscribe.
If you wish to get notifications when I post on my blog, you can follow me on Twitter@apertureF64, on Facebook.com/aperturesixtyfour or alternatively be emailed by subscribing below. All images are the Copyright of Benjamin Rowe , ALL RIGHTS Reserved unless credited to another photographer. For more information please read my Copyright Statement
13 thoughts on “Frequency Separation with HDR”
This was very useful and I’ll definitely try this technique. Very good results judging from the before & after photo.
Thank you for taking the time to comment. It is more useful than it may seem but not something for every image. So far I have had quite good results. Give it a go and us know how it went.
Thanks, will have to try this out.
I would Victor, I have never used it with macro shots but I am sure it would rescue a a few shots.
I may try the photoshop in the near future. Great transformation you made!
Photoshop is worth the subscription esp with the photographers package, you can always use the 30 day trial if you want to test it out.
Interesting technique. I’ll have to experiment with this. Thanks!
For areas with a lot of fine detail and you need to rescue colour it works great. Also you can convert the layers to a smart object and add additional filters.
Great work. Thanks for sharing. I’m using Lightroom and sometimes GIMP. But perhaps PS is worth it!
Hi, If you are using Lightroom then Adobe’s Photographers package is well worth it to have Photoshop as well.
Hello, Ben. I am a little embarrassed. I can’t tell the difference between the before and after photo. Would you please circle the area that is different? Please… 😉
This is very interesting technique, but I am afraid for a beginner like me it’s a bit difficult to follow. I wrote this down in my notebook though. Hopefully I will come back to it when I get to know more about Photoshop. Thanks. Helen
Hi Helen I have updated the before and after screen shots with a circle showing where I worked on the image. Thank you for the feedback. I will have new post in the coming weeks with this technique being used with portraits and I am sure you will see the difference in that one.
Good morning, Ben. How did I miss that? (I am laughing loud..). I looked and looked and looked. Thank you so much! 😉 Helen