Frequency Separation Rehab

Although there have been countless articles written on the subject of Frequency Separation - including several here on Fstoppers - the current state of retouching has me somewhat concerned about its use. Its widespread adoption, use and overuse has brought us to the point where frequency separation is fast becoming the number one culprit for poorly retouched and cringe worthy work. While I’m not convinced that we need to ban it outright, I do feel that the way we approach it needs to fundamentally change.

There's no doubt that Frequency Separation (FS) is a revolutionary technique that allows us to do lots of things that weren’t previously possible. Sadly, it’s this power that has also allowed people to misuse it by being given a false sense of security. The popular belief is that since the texture and tones are separated, there’s no limit to what can be done on the low frequency layer since texture will always be preserved. While the entire assumption is incorrect given that texture does require mid-frequency information to render correctly, it’s also critical to recognize the importance of the “low” tonal layer in producing a realistic result. If you’re unclear on what I’m talking about in terms of “high” and “low” layers, be sure to check out Julia’s detailed article on the theory behind the technique.

Now let’s be clear, I do believe that FS still has its place in modern retouching and I still use it sparingly in the manner I teach in my photography and retouching course on RGG EDU, but I’ve also reached this point through a series of mistakes on my end. For over two years I relied too heavily on FS and ruined a good number of photos on the process. I’ve since removed nearly all those photos from my portfolio and had to start anew with a different approach. While in some genres of photography the look often produced through frequency separation is acceptable, in fashion photography and most commercial work it’s best to avoid it and retain a certain degree of “rawness” and imperfection. In these genres the first thing we need to accept is that the image doesn’t have to achieve a completely unrealistic standard of beauty and that some “flaws” are OK and even encouraged. If we have to strive for a point of technical perfection to produce a good image, then perhaps the image doesn’t contain enough substance in the first place? So with our mind in the right place, how can we use FS and still leave our images looking great? Here are some dos and don’ts to put you on the right path and hopefully leave you with some food for thought as you approach your next retouch.   

Don’t use it if you can avoid it

My biggest piece of advice is to stop treating FS as an all-around retouching solution and begin viewing it as tool of last resort. Try to go from a state of “OK, now let’s go into Frequency Separation”, to “Is there a problem in this image that I need to use Frequency Separation for?”. This simple question will help you to avoid the trap of working needlessly in it for “just because” reasons. Many of the things that people use FS for can be sorted out with a few adjustment layers, healing or dodging and burning so odds are that you’re using FS for one of the following reasons:

1. I keep hearing about FS and it sounds fancy so I should probably use it too
2. I’m used to using FS so I’ll just keep doing it
3. I don’t understand the other tools available to solve this problem
4. Using any other tool would be far too time-consuming or difficult

Of these reasons, the fourth is the only case where the use of frequency separation is recommended and justifiable, and I’ll touch on some of these scenarios throughout this article. So while my best advice is to reduce your use of FS to a minimum, if you do decide to continue using it in some capacity, try to at least adhere to some of the recommendations below.

Do use it to enhance or fix texture

Working purely on texture based issues outside of Frequency Separation is both challenging and onerous, so this is one case where FS can save a great deal of time and help to produce a better result than through other approaches. In the below image, if you look at the area of the cheek close to the nose, you’ll notice more enhanced texture on the after image through the use of texture grafting under FS. For a demonstration and discussion on texture grafting, check out my article and video on the subject here.

Do use it at the start of your workflow

One of the main reasons that professional retouching houses frown on FS is because it’s destructive and can’t easily be changed or eased back the way adjustment layers can. Because of this, there’s really no ideal place in the workflow to incorporate FS, but the best compromise is to do so towards the start of the retouch prior to any color adjustments, corrections or dodging and burning. Since our goal is to apply it sparingly, the FS’d layer should contain only subtle changes from the layer below it allowing for some degree of opacity adjustment if needed. If you enter FS later in your retouch then you’ll no longer be able to make changes to your dodging and burning or color adjustments since the FS layer is rasterized and will cover those layers. If you're retouching for yourself then you may not find it necessary to maintain a non-destructive workflow but it's still a good principle to adhere to. 

Do use it in the safe zones

There are areas on your subject where FS can be used fairly safely and others where danger lurks. If you notice patches with color shifts, FS can speed up the process of eliminating them. While these can also be resolved with healing or adjustment layers, along some areas on your subject there’s often no harm in tackling them with FS if you're in a pinch. The safest areas tend to be relatively flat surfaces without a large number of luminosity shifts across them. These include the forehead, chest, legs and parts of the cheeks. Conversely, areas to avoid are those containing lots of luminance shifts that define the contours of the face. These are the periphery of the mouth, nose, eyes and eyebrows, hands, certain surfaces of the neck and the nasolabial folds. The below image shows a subtle yet important problem often caused by using FS in these “danger zones” whereby necessary contours are inadvertantly flattened.

Do respect the direction of light and shadow

Regardless of whether you’re working in the safe or danger zones, one of the most important things to keep in mind is the direction of light and shadow and the overall contours of the face. Changes to these contours by altering the luminosity levels is one of the greatest factors for a failed FS application. Retaining the changes across these surfaces requires the following:

• Appropriate brush size - small brush for areas with lots of changes, larger brush for flatter surfaces (shown in the image below)
• Low flow/opacity - make changes gradually
• Correct paint direction - always paint parallel to the light change and never perpendicular to it (see image below)
• Constant resampling - paint only small areas with the same color and resample as you move along the surface
• Lots and lots of practice and restraint

Do use blending modes and backup layers

In a perfect world, frequency separation should never change the luminosities of your image and work on color only. Luminosity adjustments are best saved for dodging and burning (D&B) which is a much more controlled and safe environment for making them. To avoid unnecessary changes to luminosities in your low layer, use a color or hue blending mode on your paint layer (see image below) to smooth out color transitions and leave luminance shifts for later adjustment in D&B. It may seem as though you’re not making much progress, but by creating a clean color base you’ll set yourself up for an easy retouch using other non-destructive techniques like D&B. One thing to keep in mind is to sample color from areas of similar luminance to your target area since the saturation levels behave differently in areas of shadow, midtone and highlight. Using a hue (instead of color) blending mode on your paint layer can combat this problem but may require further saturation corrections later on. Using backup layers is simply a pragmatic approach since it will allow you to recover areas that begin to look incorrect as you work through your FS application.

Don’t use it on black and white images

Since we’ve already established that FS shouldn’t ideally affect luminosities, using it on black and white images (which are luminosity only) is generally a no-no. B&W images should be worked on purely with dodging and burning and healing and hence represent a good exercise for practicing and perfecting these tools/techniques. The one exception to this rule is if you're using it to operate purely on texture based issues such as those we described above. 

Don’t use it for large color shifts

Although we’ve talked about how to use the tool safely through hue and color blend modes, this shouldn’t be taken as a free pass for color matching large areas. If for example your subject’s neck and face don’t match color wise, make these corrections with a single adjustment layer and mask in as opposed to painting color over it inside of FS. Remember that whenever possible, we want to make an equal shift in color across all the surfaces (luminosities) of that area as opposed to completely replacing the area with a single color value. This is precisely why using multiple adjustment layers and masks is superior and safer than frequency separation; the former is a shift, the latter is a replacement. Let’s consider the case of a surface like the neck. The neck isn’t flat and has several surfaces across it that pick up light differently. These areas not only reflect different luminosities but also different saturation levels. By altering it with an adjustment layer, we’re retaining those initial saturation and luminosity levels and shifting them in one direction or another. When we begin to paint globally across them with a single color under FS (even in color blend mode), we’re replacing them with a completely different value and discarding (rather than shifting) the original. This is sometimes desirable, but often isn’t and represents one of the most dangerous elements of FS. If this concept seems unclear or confusing to you then I dare say that you probably shouldn’t use FS until you understand it in full. I hope you can now see that although FS seems like a relatively simple concept on the surface, lurking beneath it are many nuances that need to be understood and respected before you dive into it.

Do believe there’s a world outside FS

This for me was one of the most difficult tips to apply as I weaned myself off frequency separation. Frequency separation is a bit like a drug as it represents convenience and power in one simple package. We can do so much with a single brush stroke that adjustment layers begin to feel like a dated and archaic approach. While lots of these other techniques may seem antiquated as they’ve been around for years - some even in the darkroom days - they’ve stood the test of time for a reason and aren’t going anywhere. Start with a tapered approach where you give yourself a time limit on FS application. If you’re currently spending 30 minutes on FS, try to limit yourself to 20, then 10, and so on. Soon you’ll find that a five minute application gives you all you really need to complete a great retouch, your results will improve, and you’ll become a more well rounded retoucher in the process.

You can connect with me via the social media links below