Frequency separation explained
If you've spent any time in portrait retouching, you already know that heavy-handed skin work is the fastest way to make a subject look like a rendered game character from 2009. Clumsy skin retouching is the tell. The fix isn't a lighter touch. It's a smarter one. Frequency separation lets you work on texture and tone independently, which means you can smooth out blotchy, uneven skin tone without accidentally ironing away every pore and freckle that makes the portrait feel real.
It's one of those techniques that sounds fancier than it is. Once you've used it a handful of times, you'll wonder how you ever retouched without it.
What frequency separation actually does
Every photograph contains two types of visual information layered on top of each other: texture (fine detail, pores, hair, fabric grain) and tone/color (skin blotches, shadows, uneven pigmentation). Normally these live on the same pixel, which means any correction you make to one inevitably messes with the other.
Frequency separation splits these two types of information onto separate layers. The high frequency layer carries the fine detail. The low frequency layer carries the tone and color. Once separated, you can attack each independently without the usual collateral damage.
In Affinity, the split happens automatically via Filters > Frequency Separation in the Pixel studio. Under the hood, a blur filter is applied to create the low frequency layer, and a High Pass filter set to Linear Light blend mode handles the high frequency layer. You don't need to set any of that up manually.
It's also available as a tool in the Compositing Studio, though that studio is hidden by default. To access it, open the Studio Manager and enable it manually.
A basic frequency separation workflow in Affinity looks like this:
- Go to Filters > Frequency Separation.
- Choose a radius that separates skin tone from skin texture without flattening facial structure.
- Pick a blur method: Gaussian for most portraits, Median or Bilateral for edge-sensitive work.
- Retouch tone and color on the Low Frequency layer.
- Retouch pores, hairs, spots, and texture issues on the High Frequency layer.
- Keep the frequency layers grouped so the edit stays easy to review, mask, or remove.
The radius slider: where most people go wrong
When the Frequency Separation dialog opens, the first thing you'll deal with is the Radius slider. This controls how much blur gets applied to the low frequency layer, which determines the balance between what lands in texture and what lands in tone.
The instinct is to crank it up. Don't. A radius that's too high shifts too much detail into the low frequency layer. You end up blurring structural features like the nose bridge or eye socket, and subsequent retouching starts to look sculpted and weird.
Set the radius so the Low Frequency preview shows blended, smooth color and tone, but still retains the major features and contours of the face. You should see something that looks like a slightly dreamy, out-of-focus version of the image. If it looks like a watercolor painting, back it off.
Choosing your blur method
Affinity gives you three blur methods for the low frequency layer. Most retouchers default to Gaussian and never look back, but the other two are genuinely useful in specific situations.
Gaussian (default): Smooth, weighted blur. Works well on most portraits. A safe, reliable starting point.
Median: Broadens color regions and retains edges better than Gaussian. Useful when your subject has clearly defined features you don't want to bleed into adjacent tones: strong jaw lines, graphic eyebrows, that sort of thing.
Bilateral: Preserves high-contrast edges while blurring within them. When you select Bilateral, the Tolerance slider becomes available. Use this to control how aggressively major features are preserved when you brush over them. It's slower to render but worth it on complex, detailed portraits.
There's no universally correct method. Try Gaussian first, switch to Bilateral if you find yourself losing structural definition in problem areas.
Working the low frequency layer
Once the separation is applied, the low frequency layer is your color and tone workspace. This is where you even out skin tone, reduce redness, and knock down shadows without touching the skin's surface texture.
The Dodge and Blur Brush tools are your workhorses here. For broader skin retouching, the Healing Brush also performs well for patchy, uneven pigmentation. It samples surrounding tone and blends it in seamlessly. Because you're working on a layer that contains no fine detail, you can afford to be relatively aggressive with your brush size and softness. Just keep your opacity moderate; stacking corrections gradually is always cleaner than one heavy pass.
Blown-out highlights are also handled here. Sample a nearby skin color, then paint over the clipped area using the Paint Brush Tool set to Darker Color blend mode with a low Flow setting. It's a simple fix that would be significantly messier to pull off on a merged layer.
A useful keyboard shortcut: press F to toggle between the high and low frequency layers on the fly. Checking your progress in context, without leaving the retouching flow, is one of those small things that makes a session significantly faster.
The goal is not to make skin perfectly even. It is to reduce distractions while keeping the natural transitions that make the face feel real.
Working the high frequency layer
The high frequency layer holds all the fine detail: pores, individual hairs, blemishes, texture irregularities. When you retouch skin texture here, you're working on the surface only, without touching the underlying tone, which is exactly what you want when dealing with spots, stray hairs, or localized texture problems.
The Clone Brush Tool is the go-to here. Sample nearby skin texture with a similar orientation and density, then paint it over the problem area. Because you're cloning pure texture rather than texture-plus-tone, the result integrates almost invisibly. No telltale dark or light halos that betray a patch job.
The Blemish Removal Tool also works well on the high frequency layer for isolated spots and small imperfections. For anything larger or more complex, the Clone Brush gives you more precise control over what you're borrowing and where you're placing it.
Selections and masks work here too
One thing that often gets overlooked: both frequency layers behave like normal layers. You can apply selections, masks, and even live masks to either of them exactly as you would anywhere else in Affinity.
This opens up a lot of targeted retouching options. Create a rough selection around a specific area (the forehead, cheeks, or under-eye zone) and limit your retouching to that region. Use a layer mask to fade an edit in or out around edges. Use the selection tools to isolate a color cast in one section of the face before treating the low frequency layer beneath it.
Treating frequency layers as isolated, sealed-off zones is the wrong mental model. They're regular layers with the same full toolset available to them.
Common frequency separation mistakes to avoid
The mistakes usually come from pushing the technique too far, or working on the wrong layer.
- Setting the radius so high that facial structure starts to blur
- Retouching tone problems on the high frequency layer
- Cloning texture from an area with a different direction or density
- Forgetting to toggle the group on and off to check whether the edit still looks natural
- Using frequency separation to fix problems better solved with lighting, dodge and burn, or color correction
When frequency separation is the wrong tool
Frequency separation is superb for portraits, skin retouching, and any image where surface texture needs to be preserved independently from tone. It's less useful, and potentially counterproductive, in other contexts.
Product photography where even, seamless gradients matter often responds better to conventional dodge and burn on merged layers. Heavily textured surfaces like fabric, stone, or leather can behave unpredictably when the frequency split is applied at the wrong radius. And images with strong noise tend to bake that noise into the high frequency layer, making it harder to work cleanly.
For editorial portraits and beauty work, though, it remains one of the most useful techniques in the retouching toolkit.
Building a non-destructive workflow
The best version of a frequency separation workflow is non-destructive. In Affinity, the original layer stays untouched when you apply the separation. If you need to revisit or roll back any correction, the source is always there.
Using selection tools before painting on either layer means you can also limit the blast radius of any correction. If a patch on the high frequency layer looks wrong, the selection boundary means it hasn't crept into areas you didn't intend. Methodical, but fast once it becomes habit.
The bigger picture
Frequency separation isn't magic. It won't fix a badly lit portrait, and it won't make a muddy image look like it came out of a controlled studio environment. What it does is give you surgical precision over the two fundamental types of visual information in a photograph, so your retouching decisions stay where you put them.
The technique scales from quick commercial corrections to painstaking beauty retouching. Whether you're cleaning up a headshot for a magazine layout or working through a full editorial spread, understanding the mechanics (radius, method, layer behavior) means you're making decisions rather than just pushing pixels and hoping.
Affinity's implementation removes the manual setup that historically made frequency separation feel inaccessible. Set the Radius decision, the blur method choice, and you're in. What you do with the layers from there is entirely up to you.
