Ask the internet how to retouch skin properly and someone will say frequency separation, usually with a little bit of attitude. It is the technique retouchers reach for, it genuinely does protect texture, and it has a reputation for being the "real" way to do it. Then you try to point it at a 40 minute talking-head recording and things get awkward fast. Here is what frequency separation actually is, what happens when you move it from a still to footage, and where a masked smoothing pass makes more sense.
A real VanityFilter result at close-up range, not a cross-product test. Drag the handle and look at the pores across the cheek and the fine hairs at the temple. That surviving detail is the whole argument frequency separation fans are making, and it is the bar any automatic pass has to clear.
Short version: frequency separation is a still-image technique. The working video version needs a freeze frame, hand painting, and a planar track, which makes it a surgical repair tool for one shot rather than a beauty pass for a whole episode. For footage, a per-person skin mask plus texture-aware smoothing gets you the same "keep the pores" outcome without doing it a thousand times.
What frequency separation actually does
The idea is simple and kind of elegant. Any image can be split into two stacked layers. The low frequency layer holds broad information: color, tone, shadow, the blotchy red patch on a cheek, the shape of the light. The high frequency layer holds fine detail: pores, stubble, fine lines, individual hairs, the grain of the sensor.
Retouchers build it by blurring a duplicate of the image to make the low frequency layer, then mathematically subtracting that blur from the original to isolate what is left, which is the texture. Once they are separated, you can paint out a blotchy patch of color on the low layer and the texture on top stays exactly where it was. That is why the result does not go plastic: you never blurred the detail, you only fixed the color underneath it.
This is also why it has a cult following. It is honest about the fact that skin problems are usually two different problems (uneven color and unwanted texture) that deserve two different fixes. Our guide to smoothing skin without looking fake makes the same argument from the other direction.
Why it does not port cleanly to video
Frequency separation has no concept of time. It operates on one image. There is no frame two.
You can absolutely run the math on footage, and effects that do a high-pass split per frame exist. The problem is what you do next. The value of frequency separation comes from a human painting on the low frequency layer with a brush, sampling nearby clean skin. That is a hand-made decision about one specific patch of one specific image. Do it per frame and you are rotoscoping. Do it once and hold it and it slides off the face the moment the subject moves.
Nobody working in video does it that way. One published Fusion workflow inside DaVinci Resolve is instructive: you freeze a reference frame, separate that frozen frame into high and low layers, paint the color fix and clone the texture by hand, mask the repair, then run a planar tracker so the cleaned still gets match-moved back over the moving shot. You can see that specific node chain in this Fusion frequency separation walkthrough. This is an editorial tutorial, not a native one-click Resolve feature. Blackmagic’s current Fusion documentation separately confirms the node-based retouching workflow and built-in planar tracking.
But notice what it is not. It is not something you run on 40 minutes of two-camera footage. The tracker works from a single reference frame, so heavy head turns, hair crossing the face, and changing light all fight the clean plate. It is a VFX repair, priced in minutes per shot, and it does one blemish at a time.
The tell
If a tutorial promises frequency separation on video and never mentions a freeze frame, a clean plate, or a tracker, it is either doing a plain high-pass sharpen and calling it frequency separation, or it is a photo tutorial with "video" in the title. Both are common.
What a masked smoothing pass does instead
The modern video approach attacks the same problem from the mask side rather than the frequency side. Detect the faces, segment the actual skin per person, then run a smoothing operation that is built to keep high frequency detail while flattening the low frequency unevenness underneath it. The mask moves with the subject on every frame because it is recomputed on every frame, so there is no clean plate to slide off.
The important part is what stays outside the mask. Eyes, lips, teeth, brows, hair, glasses, clothing, and the background never get touched, which is what stops the frame from turning into soup. In VanityFilter that runs per person, so a two-hander does not get one shared setting smeared across both faces, and you can dial one guest heavier than the other.
Where this genuinely loses to frequency separation: a masked pass makes a global decision about the skin region. It will not selectively rescue one blotchy patch on a jawline while leaving the rest of the cheek untouched. It is a tone-and-texture evener, not a spot-repair brush.
| Frequency separation | Masked AI smoothing | |
|---|---|---|
| Works on | One frame at a time | Every frame automatically |
| Handles motion | Needs a planar track and a clean plate | Mask is recomputed per frame |
| Texture | Untouched by construction | Preserved if the tool is built for it, so verify |
| Best at | Removing one specific blemish from a hero shot | Evening skin consistently across long takes |
| Cost | Minutes of manual work per shot | Set once, then processing time |
| Multiple people | Repeat the whole setup per face | Separate mask and strength per person |
When frequency separation is still the right call
- A single distracting mark on a close-up that viewers will lock onto.
- A near locked-off shot where a tracked clean plate will actually hold.
- Product or beauty work where one hero frame gets pulled as a still anyway.
- A fix that is about color and blotchiness only, with texture that is already fine.
If you are already grading in Resolve, doing it on the Fusion page keeps everything in one project. Our Resolve skin smoothing comparison covers where Resolve's own beauty tools fit next to that.
A practical workflow for long footage
- Correct exposure and white balance first. Frequency separation and smoothing are both worse at fixing skin that is being lit badly, and neither one recovers a highlight already clipped to pure white.
- Run the even-the-skin pass on the source files before you cut, so every clip instance in the edit inherits it.
- Keep the strength low enough that pores survive. Judge it at normal viewing size, not at 300 percent zoom.
- Cut the episode. If one close-up still has a mark that pulls focus, that single shot is where frequency separation earns its keep.
- Export a high-bitrate master. Platform compression softens fine detail on its own, so a pass that looked mild in your editor can read heavier after upload.
Doing the correction before the timeline is the part that saves the most time on long-form content, and it is worth reading up on separately in our pre-edit workflow guide.
How to check you did not flatten the texture
Whatever route you take, the test is the same and it takes 30 seconds. Play the result at speed, at the size people will actually watch it. Then step through a few frames and look at four places: the pores across the cheek, the fine hairs at the temple and jaw, the edge where skin meets hair, and the background just past the face. Texture should be readable, hairlines should be crisp, and nothing outside the skin should have moved.
Then watch for the thing stills never show you: temporal wobble. If the correction breathes or shimmers as the head turns, the mask or the track is unstable, and that reads as cheap far more than a slightly heavy setting ever will.
FAQ
Can you do frequency separation on video?
Yes, but not the way you do it in Photoshop. One published DaVinci Resolve Fusion workflow freezes a reference frame,
separates that frame into high and low frequency layers, paints the fix by hand, then
uses a planar tracker to match-move the cleaned plate back over the moving shot. It is a repair technique for a specific
blemish in a specific shot, not a per-frame beauty pass you run across an episode.
What is the difference between frequency separation and AI skin smoothing?
Frequency separation splits an image into a low frequency layer holding color and tone and a high frequency layer holding
texture such as pores and fine hair, so you can fix blotchy color without touching detail. It is manual and frame-based.
AI skin smoothing detects faces, builds a skin mask, and applies a texture-aware smoothing pass inside that mask
automatically on every frame. Frequency separation gives more control per frame; a masked AI pass gives consistency
across thousands of frames.
Does frequency separation avoid the plastic skin look?
It avoids it by design, because the texture layer is left alone while you work on color and tone. Plastic skin comes from
blurring everything at once. A modern masked smoothing pass can reach the same goal if it keeps high frequency detail and
stays inside the skin, which is why you should judge any tool by whether pores and facial hair survive at normal viewing
size.
What is the fastest way to retouch skin across a whole video?
Correct the source files once, before the edit, rather than applying an effect to every clip on the timeline. A batch pass
over the raw camera files means every cut you make afterward inherits the correction, and you avoid re-tracking or
re-rendering the effect for each clip instance.
See the texture test on real footage →
VanityFilter runs offline on Windows, masks each person separately, and has no subscription. See current pricing.