A light-led skin engine, real print-film tone, and color that stops short of the talent on its own. Built over a decade of commercial work, for the node tree you already use.
They are built to sit in the same node tree without any of them undoing another. Tap one to see everything it does.
Shapes the light on a face before it touches hue: fill the shadow side, put form back, place the glow, roll off the top. Then evens the color underneath. Never blurs, so texture cannot be lost.
Real film stocks only, no invented presets padding the list. Stock Strength and Look Mix are separate, so you can take the tone response without the full look.
Every global saturation or hue move normally lands on the face too, so you build the correction twice. Here the skin protection is carried inside each preset.
The whole suite, the install guide and node order, and free updates on everything. $78 less than buying them one at a time.
Every tool here was compiled into a C test harness built from its own source before it shipped. That harness confirmed each one is a true null at defaults, that no combination of sliders produced a NaN across thousands of sampled states, and that the tone response stays monotonic on every preset. Most tools you can buy have never been through that, and it is the reason these ones do not surprise you at 2am on a delivery.
Every other skin tool starts by painting hue. This one starts where a gaffer starts: fill the shadow side, put form back in the face, place the glow, roll off the top. Then it evens the color underneath. It never blurs, so texture cannot be lost.
Instant download · Mac & Windows · Metal and CUDA · lifetime updates
One node. Magic Mask on the face, Skin.dctl inside it. No plate cleanup, no tracking a window frame by frame.
Chase the hue and you flatten the face. Fix the light and the hue mostly sorts itself out.
A cheek that reads red and a jaw that reads olive are usually not two different skin colors. They are one skin color at two different exposures, sitting on different parts of the tone curve. Even the exposure across the face and most of that color split disappears on its own. That is why the light controls sit at the top of this tool and the color controls sit underneath as support. Work in that order and you need far less of the second group than you think.
Radiance lifts a bell curve through the midtones. Position moves the centre of that curve up or down the range, Width sets how much of the tone scale it touches. So the glow lands on the lit side of the face in front of you, not wherever a fixed preset assumed skin would be. Highlight Soften is mapped to Radiance Position, which means the rolloff follows the glow instead of always sitting at the top of the scale.
Freckles, pores and lash detail come through the move untouched. Nothing in this tool reads a neighbouring pixel, so there is no mechanism by which it could soften them.
Nothing here is a black box and nothing is held back for the sales page. This is the whole panel exactly as it appears in Resolve, then every control explained underneath in the same order.
Evenness, Tone Consistency and Redness all rotate hue. Rather than applying three rotations in sequence, the tool sums their pulls, clamps the total to one, and applies a single rotation. That is what makes stacking them fold-safe: no combination of the three can push skin through a hue it never had.
These three sit at the top of the panel because you turn them on, look, turn them off, and correct. They are diagnostic views, not part of the grade, and they have no effect on your render.
These are the kind of thing usually sold as a separate scope utility. Here they are running on the same frame as the hero shot above.
False color laid over the face that tells you at a glance whether you are on the skin tone line, and which way you are off it. It reads the graded result, so it updates live as you correct, and it only draws inside your mask. Everything outside the selection is left exactly as it is.
The face above is close to solid yellow, which is what a corrected face looks like. The magenta lands on the lips and the eye corners, the green speckle sits in the shadow side of the jaw. Those are the two places to go back and work.
Clean 10 IRE bands laid over the face so you can read the actual level instead of guessing at it. Skin-only mode keeps the bands on the talent and leaves the rest of the frame alone, which is how you catch a forehead running hot against a jaw sitting two bands lower before the client does.
Read the face above from the top down: the forehead and cheekbones are the brightest bands, the jaw and neck drop several steps, and the shadow side falls to the darkest. That spread is what Fill and Dimension are for.
Earlier versions had a built-in hue qualifier with a range and a mask preview. It came out. Resolve's Magic Mask already tracks a face better than any hue window can, so Skin.dctl reads whatever selection you feed it and spends all its maths on the grade instead of the selection. Fewer controls, better result, and no double-keying.
Before release, the tool was compiled into a C test harness built from its own source. That harness confirmed it is a true null at default values, that no combination of sliders produced a NaN across thousands of sampled states, and that the tone response stays monotonic. Most tools you can buy have never been through that.
Yes. The Input control covers Rec.709, ARRI LogC3, DaVinci Wide Gamut, RED Wide Gamut, ACEScct, Sony S-Log3 and Panasonic V-Log. Set it to whatever the node is handed and the tool converts in, does the work, and converts back to the same space. Verified: a LogC plate and the same plate pre-converted to Rec.709 grade to within 0.0004 of a code value.
No, and it structurally cannot be. There is no spatial operation anywhere in it. Every control works on the pixel it is handed with no reference to neighbouring pixels, which is the only way a tool can soften texture. Skin gets even while pores stay crisp, even at extreme settings.
Yes, and the light-first design is why. The tool works from a target hue you set and a luma position you choose, not from a fixed assumption about where skin sits. Radiance Position moves the glow up or down the tone scale so it lands where the midtones actually are on the face in front of you.
Skin Tone is a destination and Skin Hue is a push. Skin Tone names an angle and does nothing on its own; Evenness and Tone Consistency then rotate pixels toward it, which collapses variation. Skin Hue offsets every pixel by the same number of degrees, which preserves variation and just relocates it. Converge first, trim second.
The DCTL itself runs in both free Resolve and Studio. Magic Mask, which is the cleanest way to select a face, is Studio only. On free Resolve you do the selection with a power window or a qualifier and the tool works the same inside it.
Inside a node with your face selection, after your primary balance and before your look. Set Input to your timeline space and it converts in and back out on its own, so it works at clip level in log with no Color Space Transform around it. The install guide has the node order laid out, and it sits alongside the other tools in the Suite without any of them undoing each other.
Everything is computed in float and the rolloff curve is monotonic, meaning it never reverses direction and creates a step. Grade in a float timeline and this tool will not introduce banding.
This is the tool I reach for on paid commercial work, in the same state I use it in.
Get Skin — $119Five measured stock responses plus Custom, applied to luminance only. Hue and saturation come out exactly as they went in, so a film curve can never quietly shift your color.
Instant download · Mac & Windows · Metal and CUDA · lifetime updates
One node, whole frame. Contrast up by half, the black point down and the white point up, and every hue in the picture still sitting where it started.
If it changes hue while you are shaping contrast, it is a look, not a tone tool.
Most film emulations bundle a tone response and a color cast into one preset, so the moment a shot needs 2383 contrast but not 2383 warmth you are back to building it by hand. This one does not have that problem, because it has no color behaviour to bundle. The curve is applied to luminance and the pixel is rescaled to the new luma, which holds hue and saturation exactly unchanged. Contrast here, color in Color.dctl, and neither one contaminating the other.
No invented presets padding out the list, and the numbers behind each one come from how the stock actually behaves: print stocks carry high contrast with deep blacks and a soft shoulder, the negative is low contrast, flat and lifted with a long toe.
Nothing here is a black box. Each control does one named thing, and they are grouped in the order you would use them.
The trims sit on top of the preset rather than replacing it. Leave them at default and you get the chosen stock exactly.
The curve is not a lookup table and it is not a polynomial someone hoped would behave. It is built from a slope function that is always positive, which is then integrated to produce the transfer. A curve assembled that way is monotonic by construction, not by testing: it cannot reverse direction at any setting, which is the specific failure that puts a step into a sky or a wall. On top of that it was compiled into a C test harness from its own source and checked for null identity at defaults and NaN-free output across thousands of slider states.
Because the curve is applied to luminance and the pixel is then rescaled to the new luma, hue and saturation are held exactly unchanged. You can push contrast as far as you like and your color will not move with it.
The Contrast control is calibrated rather than estimated. Normalising a curve divides by the integral of its own slope, so asking for a given steepness and getting it are two different things, and most tone tools simply do not get it. Here that relationship turns out to be exactly linear, which makes the inverse closed form: the tool solves for the internal shape that delivers the slope you asked for. Verified across the range, the error is zero to three decimal places.
No, and it used to try. An earlier build converted log to Rec.709 inside the node, which meant it was choosing a tone map on your behalf before you had touched a control. Resolve’s own Color Space Transform does that job properly. This tool works in Rec.709 and sits after yours, which is both simpler and better looking.
Yes. Put your Color Space Transform on an earlier node and this one after it. It wants Rec.709 in and it hands Rec.709 back, so it behaves like any other look node in the chain.
After your output Color Space Transform, before your creative look. The install guide lays the order out.
No. Both controls can push tones below zero, and rather than clamping them into one flat value the bottom of the curve rolls off asymptotically toward black with the tones still separated. The picture keeps going down all the way to zero instead of hitting a wall.
No. A LUT is a lookup table with fixed resolution and interpolation between the points it stores. This is a DCTL, so the transfer is computed per pixel in float. That is why it holds up when you push it and why it does not band on gradients.
There is no film look to remove. This tool only reshapes luminance, and the pixel is rescaled to the new luma afterwards, so hue and saturation are held exactly unchanged by design. Color work belongs in Color.dctl.
Yes. DCTLs run in both the free version and Studio, on Mac and Windows, across Metal and CUDA.
Take it on its own, or pick up the whole chain and save $78.
A color engine where skin protection is carried inside each preset, so a global move stops short of the talent on its own instead of waiting for you to mask around them.
Instant download · Mac & Windows · Metal and CUDA · lifetime updates
Saturation across the frame goes up 74 per cent. The valley gains 97, the rock 108, the grass 73. His legs and forearm gain 87 and 57, and brightness across the whole picture moves 0.2 per cent, so none of this is a contrast trick.
You are not building one correction. You are building it twice, and the second one has to track.
Every saturation push, every hue rotation, every bit of chroma work you do to a frame lands on the skin as well. So the normal workflow is to make the move, watch the face go with it, then build a second version of the correction masked around the talent and track it for the length of the shot. Do that on forty shots and it is most of your day. The fix is for the color tool to already know where skin is.
A preset here is not just a color move. It also sets how hard the skin key holds the face back out of that move, sized to what the move actually does. The harder the push, the more protection it carries.
Twenty-four parameters. Nothing here is a black box, and the groups run in the order you would actually work through them.
The rotation basis is a cone model rather than a plain hue wheel, built on the approach worked out by Zambrano, Leiho, Yedlin and Stoopman. That is why a rotation here keeps its luminance instead of getting darker as it moves.
Hue rotation uses a cone basis rather than a flat hue wheel, following the approach worked out by Zambrano, Leiho, Yedlin and Stoopman. In practice that means a hue can be moved without dragging its luminance along with it, which is the thing that makes ordinary hue-versus-hue work look muddy when you push it.
The protection is measurable rather than a claim. On the frame above, an 86 per cent saturation push doubles the blue in the sky while the suit comes up 44 and his face 45. The sky moves more than twice as far as the talent, from one preset, with nothing masked and no key here for you to tune.
Skin still moves, because a face that stayed frozen while the world changed around it would look pasted on. It moves roughly a third to a quarter as far. That is the difference between a grade that includes the talent and one that lands on them.
The presets stay live rather than baking down to fixed values, so every control below a preset trims it instead of overwriting it. There is also no internal mix control: use the node blend in Resolve, which is where a mix belongs.
Yes. The Input control covers Rec.709, ARRI LogC3, DaVinci Wide Gamut, RED Wide Gamut, ACEScct, Sony S-Log3 and Panasonic V-Log, so it runs at clip level before your output transform. Verified across all eight presets: a LogC plate and the same plate pre-converted to Rec.709 land within 0.06 of a code value of each other.
The curves are excellent and they have no idea where skin is, so every move you make on them lands on the face too. Here the protection travels with the preset. There is also a cone rotation basis underneath, which keeps luminance steady as a hue moves rather than letting it darken.
No. There is no hue window for you to tune and nothing to re-key when the talent turns their head. The protection is part of the maths of the preset rather than a selection sitting in front of it.
They do different jobs. This holds skin back out of a global color move. Skin.dctl actively corrects the face: shaping its light, evening its tone, fixing redness. Use this to stop a grade landing on the talent, and Skin.dctl to grade the talent.
Additive scales a color away from grey, which is what most saturation controls do and what clips first when you push it. Subtractive removes rather than scales, the way print film does, which stays denser and holds together further. Both are here, plus a six-way subtractive set so you can work one hue family at a time.
Take it on its own, or pick up the whole chain and save $78.
Tone, color, looks and skin, built to work together in one node tree. This is the set I reach for on paid commercial work, in the state I actually use it.