Yes, if you leave them on. Every overlay in this panel — Show Diagnostics, Show Scopes, and every module's Show ... checkbox — is drawn into the frame by the render, not painted over your viewer. There is deliberately no preview-only mode, because a readout that looked different depending on your viewer quality would be a readout you could not compare against itself. The amber DIAG badge at top center exists to tell you one is live. Check for it before you export.
Because they are measuring different things, on purpose. The Decoloris scopes read the tone-and-color chain at the tap you choose, so grain, dirt, halation, diffusion, MTF, adjacency, dye cloud, local contrast, denoise, vignette, gate weave and flicker are not in the reading — that is what "(no grain)" on every Scope entry means. Your host's scope sits at the end of everything and measures the rendered picture. Both numbers are right about their own question.
By decision, and the FAQ you are reading is the answer. Adobe hosts never showed per-control hover help. DaVinci Resolve used to, and as of v2.158.0 it no longer does: the tooltips popped up over the work and distracted more than they taught. The full guide lives here instead — one place, searchable, every control, no popups over your footage.
Five modules — Grain, Halation, MTF, Adjacency and Dye Cloud — each have their own stock popup, and each defaults to Match Stock, which derives the setting from your main Stock (Negative) selection. That is why a follower popup can be sitting on a value you did not choose. Pick an entry explicitly to override it. It also has a consequence worth knowing: many of the sliders under those popups say "(Manual)" in their labels, and Match Stock is an entry, so those sliders are not reaching the picture until you set that module's popup to Manual.
Six stages, in this order, and the panel is laid out in that order too.
1. Camera decode. Decoloris first has to know what your footage is. You tell it with Camera Input — your camera's log format — and it decodes that into the Working Space, the log space the whole chain runs in. Get this wrong and everything downstream is being asked to grade the wrong picture.
2. Balance. Before any film is involved, you set the scene straight: exposure in stops, and white balance by declaring the light the scene was shot under. This is the scene-referred correction. It is where neutrals become neutral.
3. The negative. The balanced scene is exposed onto a Stock (Negative) — the film stock whose color character the image takes on, plus its contrast, its toe and its shoulder.
4. The print. The negative is printed onto release stock, which is where most of film's contrast and color signature actually comes from. This is also where the lab-timing controls live: printer lights, negative exposure, print timing.
5. Texture and damage. Everything a real strip of film picks up on its way to a screen: grain, halation, emulsion sharpness, adjacency, dye clouds, dirt, gate weave, flicker, archival fade, plus the optical stages that happen at the lens — vignette and diffusion.
6. Output. The finished picture is delivered into whatever standard you are working in — Rec.709, P3, HDR, sRGB, or left in log for you to grade downstream. An imported LUT of your own, if you use one, runs last of all.
The panel reads in that order. The controls you touch most sit ungrouped at the very top; the deep controls sit in twirled-down groups underneath, in chain order.
And one habit ties the whole panel together: diagnose the stage, correct at the stage. The taps and readouts exist to tell you which stage owns a problem — and every control family has its owning stage. Balance owns the input, the stock and print own the look, the texture stages own the surface. Find where it broke; fix it there.
Make neutrals neutral, put skin at the right brightness, then let the skin line arbitrate. In practice:
Step 1 — neutrals. Set Camera Input to your camera's format. Then declare the light: Source (As Shot) Kelvin and Source (As Shot) Tint are what the scene was actually lit at, not what you want it to look like. If you don't know, park on a representative frame and fire Auto Balance (estimate) once — it writes an estimate into both dials and zeroes the Shift dials so you still have a clean zero to trim from. Check it; it is a starting point, not an answer.
Step 2 — exposure. Set Show Diagnostics to Skin Analysis. The bottom band of that overlay prints where the skin in the frame landed, in stops relative to 18 percent gray — the same stops Exposure, Exposure Trim and Negative Exposure are dialed in. Move Exposure until skin sits where you want it. The classic exposure zone is roughly a stop above mid-gray; the heat map's ladder draws that band for you. The overlay measures, it never dials anything for you.
Step 3 — read what the print did to skin, then place it deliberately. Now switch Show Diagnostics to Output Skin Analysis. This one lets the whole chain run — stock, print and all — and measures what came out the other end: a printer-point solve, moving red and blue while holding green, exactly how a lab timer quotes one — and with your Sample Point parked on the face, solved from that face, not the frame's average. Read it as a gauge before you touch anything: the solve aims at neutral through your look, so on a balanced input the numbers are the look's own signature — how your stock and print render skin. That makes them best at matching shots (the same subject should read similar numbers across a scene; trim the drift) and comparing stocks. When you do want to move the print — a taste trim, or holding skin steady after a stock change — dial toward the printed values deliberately; they are resulting settings, not numbers to add, and a deliberate trim is usually one to three points. If skin simply looks wrong, the fix is back at step 1: the print's look is dialed to receive an honestly balanced scene, and the lights can't rescue the input. And remember what success looks like on this screen: stable non-zero numbers. Zero belongs at step 1; character belongs to the print; these numbers are how you keep that character consistent from shot to shot — chasing them to zero just removes the look you chose from skin. Once one shot sings, matching the rest of the scene to it is mechanical — see How do I match skin from shot to shot?
Two honest limits on step 3, and they matter. The correction is solved through the color stages only — grain, halation, diffusion and the other texture stages sit outside the calculation — so treat the number as a measured starting point and then judge the picture. And when the solve won't stand behind its answer, the readout says so in a word rather than showing you a wrong number: NO SKIN, PRINT OFF, NO FIT, or NOT UNIFORM.
Try this: balance with Film Stock and Print switched off and Scope Tap set to Output — that turns the built-in scope into a plain balance scope for your source. Then switch the film back on and go to step 3.
These sit above every group, visible the moment you apply the effect.
It fades the entire Decoloris look in and out as one control. At 100 you get the full look. At 0 you get your original image with only the camera decode and the output transform applied — your color management stays intact, the look simply isn't there. It sits on top of every module's own enable and strength rather than replacing them, so nothing you dialed is lost while you fade. It is not the same control as Stock Strength, which fades the Film Stock module alone. At 0 the picture matches the shipped Neutral (Verified) preset — decode in, output out, no look — so that preset and this fader at zero are two routes to the same reference frame.
Try this: park Master Strength on a keyboard-reachable value and flick 100 → 0 → 100 as your A/B. It is the only true whole-look bypass in the panel.
This tells Decoloris what your footage already is, so it can decode it into the working space. Set it to your camera's log format. If your clip is already in the working space chosen in the Working Space popup directly below, leave it on None. This is the single largest lever on the first frame: a mis-set decode hands every stage below it the wrong picture, and no amount of grading downstream recovers that cleanly.
Try this: if a fresh apply looks wrong in a way you can't place — muddy top end, no highlights — check this popup before you touch anything else.
The shipped Camera Input menu, exactly as the panel lists it — alphabetical by manufacturer, None first:
None (no camera decode) · ACEScct / AP1 · Apple Log (iPhone) · Apple Log 2 (iPhone, AWG) · Arri Log C3 / AWG3 · Arri Log C4 / AWG4 · Blackmagic Gen 5 / BMD WG · Canon Log 2 / Cinema Gamut · Canon Log 3 / Cinema Gamut · DaVinci Wide Gamut · DJI D-Log (Mavic/Air/P4) · DJI D-Log-M (Pkt 3/Mini) · Fujifilm F-Log2 / F-Gamut · Insta360 I-Log · Nikon N-Log / N-Gamut · Panasonic V-Log / V-Gamut · Rec.709 normal (default) · Rec.2100 HLG (BT.2020) · Rec.2100 PQ (BT.2020) · RED Log3G10 / RWG (Nikon) · Sony S-Log2 / S-Gamut · Sony S-Log3 / SG3.Cine
Each entry decodes both the transfer function and its gamut. Camera and log-format names identify compatibility only — see the trademark note at the end of this document.
It is the log space the whole chain runs in. If Camera Input is set, your footage is decoded into this space first; if Camera Input is None, this is what your footage is assumed to already be. It decides what the chain receives and emits — not what the film does: the film stock's character, the Color grade, the Balance tints and the grain zones are all evaluated in one internal reference space, so a stock looks the same whichever working space you pick.
Output is the transform applied on the way out. Log (working space) leaves the image in the working space for your own downstream grade. Everything else delivers a finished, display-referred picture. The two Rec.709 entries are reference Rec.709 for a gamma 2.4 display, which is what a calibrated broadcast or grading monitor expects. In Premiere, turn on Display Color Management or the picture reads lifted on a desktop monitor — that is a host setting, not a plugin fault. Rec.709 (Gamma 2.4, Clip) applies no tone mapping at all and clips above legal white; for log camera sources use Rec.709 Tone Mapped (2.4) instead. Rec.2020 and P3-D65 are tone mapped the same way with their own primaries; Rec.2100 HLG and PQ are HDR and carry their own standard curves; sRGB (Web) is the clipping Rec.709 picture with the sRGB curve instead.
Try this: at a declared source range of 100 the two Rec.709 entries are identical; above 100 they diverge immediately. That is the cheapest way to see what tone mapping is doing.
ACEScct (AP1) is different in kind from the rest: it hands off scene-referred log for another application to grade and finish. It applies no tone mapping and no display curve, Tone Map Max In does nothing on it, and LUT export is not available from it. DCI-P3 Theatrical (2.6) is the theatrical twin of P3-D65 — same primaries, same tone mapping, encoded for a projector calibrated to gamma 2.6. On an ordinary desktop monitor it will look dark and contrasty, and that is correct rather than a fault: judge it on a P3 projector or a display set to 2.6. LUT export is available from it, which is what a show LUT for a DI theater is.
They tell the tone map how much highlight your format actually carries. Auto Range from Camera ships on and sets the number for you from the camera curve you picked, using the ceiling the camera's maker publishes where one exists. Four entries have no maker figure, and it is worth knowing which: DJI D-Log-M is derived from DJI's own conversion LUT, and Insta360 I-Log and Nikon N-Log are read off the curve itself, because no published ceiling exists for them; two more involve a standards judgment — HLG is anchored to its BT.2408 diffuse-white scene scale and PQ at 203 nits, matching the HDR output side in both cases. While Auto Range is on, Tone Map Max In (Auto) is set for you and your changes there are ignored — the "(Auto)" in the label is the honesty, not a mode name. Turn Auto Range off to set the number yourself; it needs Camera Input on a real curve anyway, so on None the slider keeps control regardless. Tone Map Max In does nothing at all when Output is Log, ACEScct or Rec.709 (Gamma 2.4, Clip).
Try this: set Tone Map Max In to your camera's ceiling for clip-free highlights — Sony S-Log3 is about 3900, and most log formats land between 2000 and 5000. The shipped 2000 is the highest setting that still leaves middle gray exactly where the film chain put it: raising it holds more highlight range, lowering it gives a punchier picture.
It is the film gauge being emulated. Smaller gauges magnify the negative more, so grain, halation, dye structure and dirt all get coarser relative to the frame as you move from 35mm to Super 16 to Super 8. Note the first entry is 35mm (Super 35) specifically — the aperture the whole tree scales by is Super 35, which is 13 percent wider than an Academy gate, so if you are matching Academy-gate material, know that going in.
It is the film stock whose color character the image takes on. The menu is a list of camera negatives, and the label says so because it is not a list of finished looks — the print stage below is what develops them. Two entries are marked because they are not camera negatives: Teatro 2383 (print) is a print emulsion and Vivezza 50 (reversal) is a transparency. Noir 5222 and Noir 5231 are black and white silver stocks: they convert the image to monochrome inside the stock character, so grain, halation and print all then run on a mono image, and with the follower stocks left on Match Stock their halation goes neutral too — a silver emulsion has no dye layers to color it. Nocturne 200T is a finer-grained, sharper 3200K stock than Vesper 500T.
It blends the Film Stock module against your untouched image — its color character and its curve trims, and nothing else. It does not touch Balance, Print, Grain or any other module. On Noir 5222 or Noir 5231 it cannot bring the color back, because those stocks' monochrome conversion is part of the stock itself, not part of what this fader blends.
It blends the printed result against the negative. Lower it for a flatter, more negative-like image; 1 is a fully developed print. At 0 there is no print at all, which also sends the negative's grain and dirt back to their pre-curve positions. The two faders sit next to each other and do different jobs: the one above dials the film stock's character, this one dials the print.
Overall exposure in stops, applied before everything else. This is the scene-referred exposure correction — the equivalent of having stopped up or down on the day — not a display-side brightness trim. Everything downstream, including where the film stock's toe and shoulder land, follows from it.
They are a declaration, not a taste control. You are telling Decoloris the color temperature and the green-magenta balance the scene was actually lit at — as shot, not as you want it. Decoloris adapts from there to its 6500 K reference, so leaving these at the true source is what neutral looks like. Kelvin runs from 1667 to 20000. On Tint, positive goes magenta and negative goes green.
Try this: to warm or cool the picture, leave these two alone and use Temperature Shift and Tint Shift in the Exposure & Balance group. That way you always have a real zero to come back to.
Show Diagnostics replaces the picture with a measurement overlay. Exposure Heatmap in Stops, Exposure Curve and Skin Analysis read the plate at Balance, before the film look runs, and stop the chain there. The Exposure Curve draws on a ruled scale with mid-gray marked, so its reading sits on numbers rather than guesswork. The two Output entries let the whole chain run and measure what leaves it — Output Exposure in Stops for exposure, Output Skin Analysis for skin. The two readings are allowed to differ, and the difference is exactly what the Output modes are for. Both skin overlays print, in their bottom band, where the skin landed on the exposure scale in stops relative to 18 percent gray — the same stops Exposure, Exposure Trim and Negative Exposure are dialed in. It is a measurement of where skin sits, not an instruction. Nothing is ever written for you.
It is a lab-timer's answer to "what did the stock and the print do to this face?" It is solved on this frame's own skin against the reference chip beside it, and it moves red and blue while holding green, which is how a timer quotes one. It prints the resulting printer-light values — set the Printer Lights in the Print group to what it shows, rather than adding an increment to what you have. Two limits, stated plainly: it is solved through the color stages only, so grain, halation, diffusion and the other texture stages are outside the calculation — treat it as a measured starting point and then judge the picture. And at the Output tap it is the only correctly aimed instrument for a face: Temperature Shift and Tint Shift act before the film stock, so using them to fix what the print did re-balances the whole scene to correct one face.
And keep the three stages straight, because they are the doctrine the whole workflow rests on. One: Kelvin and Tint make the input honest. The print's look is pre-dialed to work best when balanced skin arrives at it, so an unbalanced input rescued downstream fights the very look you chose. Two: verify with the Sample Point — park it on the face and the readings describe that skin, not the frame's average. Three: the printer lights place your skin on the print. With Print enabled and the point placed, the quoted lights are solved from the face you pointed at — which is exactly the photochemical timing workflow: the timer receives an honestly exposed negative and places its skin on the print in points. Order matters: lights before an honest input give you numbers that chase a moving target, and on a frame holding more than one population that chase ends at the NOT UNIFORM refusal.
One more thing about stage three, and it changes how you read the numbers. The solve aims at the neutral chip through your look — so with a film look engaged and the input honestly balanced, the quoted lights describe the look itself: how your chosen stock and print render skin. Read them as a gauge, not a to-do list. Their real uses: matching shots — the same subject should read similar numbers across a scene, and drift between shots is real and worth trimming out — and comparing stocks, where a swap shows you, in points, how differently each one renders skin. Dial values in only to move the print deliberately: a small taste trim, or restoring skin's placement after a stock change. Typing the full quoted correction under a look strips that look's warmth out of skin — occasionally exactly what you want, usually not — and a large quote on a balanced input is the look's signature, not a defect.
So here is the rule to keep, because this screen is misread in exactly one way: on the Output Skin screen, stable non-zero numbers are success, not error. The print is shifting skin toward the stock and print's look — that is what you chose it for — so making adjustments after the fact, chasing zeros for skin, will just remove the stock and print look. Zero belongs upstream, on the balance side. Character belongs to the print. And the output numbers are how you keep that character consistent from shot to shot.
Fix Kelvin + Tint first. Adjust printer lights to taste. Then match with other shots.
1. Balance. Kelvin and Tint on every shot, until the Skin Analysis solve reads zero.
2. Grade the hero. Printer lights to taste on your best shot. Its Output Skin numbers — the R/B points and the stops line — are now your match target.
3. Match. On every other shot of that subject: Sample Point on the same cheek, drive the readout to the hero's numbers. Kelvin and Tint for the bulk, printer trims for the last fraction.
Don't chase zeros. The stable non-zero numbers ARE the target — zeroing them strips your look off skin.
The fine print, for when you have more than ten seconds:
Because a wrong number is worse than no number. NO SKIN means the frame carries too little plausible skin to average. PRINT OFF means the Print stage is down, so there is no print for the lights to time. NO FIT means skin was found but the printer lights cannot move it — the correction does not exist on this frame, and nothing about your lighting is wrong. NOT UNIFORM means skin was found, the correction was applied, and the sample still would not come together: what you sampled is not one population — sand and a face, or two people under different lamps. That last one is the actionable refusal: reframe, or get the face larger in frame. The exposure reading in the same overlay keeps working even while the printer-point band above it is refusing, because a mean has no way to fail that a solve does. The cast reading is a white-patch estimate — it assumes the brightest things in frame are neutral, so large bright saturated areas and blown speculars bias it, and it is not smoothed frame to frame. It is a viewing aid, and it is never saved into a preset.
Your host's scopes sit at the end of the graph and can only ever measure the finished image. This one reads the film chain at a point you choose, so you can look at what the camera decode alone did before Balance touched it. On the vectorscope, the six squares are the 75 percent color targets drawn in their own colors, and the letters beside them light up only when the picture actually carries that hue. Two hexagons, and they answer different questions: the dotted one joins the targets at that same 75 percent level — a saturation reference, not a limit, so chroma outside it is only more saturated than color bars — and the solid one outside it is the RGB gamut edge: anything landing beyond that is color no combination of the three primaries can make at legal levels. A third caption line, GAMUT, prints the share of the frame's samples outside legal Rec.709 at the tap you are reading — at Output, how much of what you are about to deliver is out of gamut; at Camera, how much of your source already was before Decoloris touched it. A bounded output reads 0.0 percent, and that is the answer, not a failure to measure.
Yes, if you leave it on. The overlay is drawn into the frame by the render, and there is deliberately no preview-only mode: a readout that looked different depending on your viewer quality would be a readout you could not compare against itself. The amber DIAG badge at top center is there to tell you a diagnostic is live. The box sits bottom-right at 36 percent of picture width and does not draw at all below about 245 pixels of delivered width, so a quarter-resolution preview shows no box rather than an illegible one. It also stands down entirely while any other diagnostic is reading the frame — Show Diagnostics in any mode, any Show checkbox, the working-space warning — because those answer narrower questions and win. When that happens the frame says DIAG BUSY rather than showing you nothing, so a scope that does not appear is never a mystery; switch the other readout off and the scope comes back.
Four instruments, in colorist order — RGB Parade, Waveform, Vectorscope, Histogram — and the menu opens on the Vectorscope, where this plugin's own instruments live. (The transfer-curve plot that used to be the first entry is now its own checkbox, Show Curves, in the Utility group: it reads no pixels, so it did not belong in a list of instruments that read the frame, and it can be up beside any of these.) The Waveform draws all three records on one plot, blending to white where they agree — a neutral image reads as one white cloud, and any color you can see is the imbalance, at exactly the level where it lives. RGB Parade is the three records side by side, densest regions ramping to white-hot cores. Histogram stacks red, green and blue as three filled panes on one shared height scale, on a logarithmic count axis so the toe and the shoulder stay visible against mid-tones a thousand times taller; its height is measured against the tallest code in the frame, so it is a relative reading. Vectorscope plots chroma with the six 75 percent targets and a marker on this plugin's own skin reference, measured at 127.7 degrees rather than snapped to the convention every other vendor draws. When the frame carries enough skin, a solid X marks where that skin actually sits, and the signed number at the line's outer end is its hue offset from the reference in whole degrees — printed alongside the correction in the units of the control you would actually move, so the marker tells you what to do, not just what is. No skin, no X: an empty reading is drawn as nothing rather than as zero. And every tonal instrument now wears a real ruler: numbers at 0, 25, 50, 75 and 100 in a reserved margin the data can never touch, ticks between them, faint gridlines at every tenth. Place skin at 65, hold a shadow at 5 — see it, not guess it.
That the scope reads the tone-and-color chain, not the rendered picture. Grain, Dirt, Halation, Diffusion, MTF, Adjacency, Dye Cloud, Local Contrast, Denoise, Vignette, Gate Weave and Flicker are all absent from every reading. What you are looking at is the transfer at that tap. Out-of-gamut color is plotted outside the solid hexagon rather than clipped, because that is the most useful thing an in-chain scope can tell you.
They point the plugin's entire skin instrument family at your subject. With the point placed, every skin reading solves from the 9-by-9 block of pixels under the crosshair instead of averaging the whole frame: the Skin Analysis screen's Kelvin and tint, the Output Skin Analysis screen's printer lights and stops, the reference patch beside them, and the vectorscope's skin marker — all of it describes the face you pointed at, and the crosshair is drawn on the skin screens themselves so you can see where the numbers come from. That matters because frame-wide averaging is the right answer for a portrait and the wrong one for a face on a beach, where warm sand is accepted too and the average lands between the two. Point at the cheek, and the sand is left out of the arithmetic rather than out-voted. The reading is the median of the block, so a stray highlight or an eyelash cannot move it — more than half the block has to change before the number does.
Honesty notes, because they are the whole point. On the skin screens, a point parked on something that is not skin says NO SKIN for the printer-light solve — while the Kelvin and tint line keeps reading, which is deliberate: parking the point on a gray card is an intended use. If the point cannot be read at all, the analysis falls back to the frame-wide reading rather than failing. The false-color skin paint always stays frame-wide — the point directs the numbers, not the map. On the vectorscope, the directed X remains your assertion: a point dropped on sand reads sand, because you are telling it where the skin is and it believes you. And the point aims the instruments and nothing else — nothing about your render moves when you drag it.
It is a parked-frame tool: balance once per shot, on a good frame. The values you dial persist even if the subject later moves off the point, and the point itself is never saved into a preset. For documentary work, one careful balance travels across a location's clips by copying the effect between them.
One host difference worth knowing. In Premiere, the drag crosshair arms only after you click "Skin Sample Point" in the Effect Controls panel — that is Premiere's own idiom for point parameters, the same click that gives an effect its transform handles in the Program Monitor. In Resolve there is no activation step: drag the amber-ringed handle right in the viewer, and one drag is one undo step.
Diagnose the stage, correct at the stage — the Scope Tap tells you which stage owns it.
1. Degrees first. A warm cast inflates chroma. Fix Kelvin and Tint, and the percentage often falls on its own.
2. Stops next. Darker-exposed skin reads relatively more chromatic — check the exposure band before blaming saturation.
3. Flip the tap.
100 percent is a reference, not a ceiling. The violation numbers live on the caption band — rich-stock skin above 100 can simply be the look.
Where in the chain the scope reads. Output is the delivered image in whatever Output is set to — the everyday reading, live after Balance. Camera is the working-space image the moment the decode ends, before Balance: the only place to see what the decode alone did, and where clipping, headroom and stops read in log. This is the control the whole feature exists for; no host scope can show you any tap but the last one. Flipping the tap is how you route a problem to its owning stage: hot at Camera means it arrived in the source, clean at Camera but hot at Output means a film stage is doing it.
Try this: switch Film Stock and Print off and set the tap to Output — that is a clean balance scope for your source.
Every pixel the output limiter actually moved, drawn as a diagonal white-and-black hatch with half of each marked run left as the picture, so you can still see the color underneath. Your Output setting is held inside the range its own encoding is defined on, all the way to the last pixel handed to the host; the vectorscope's GAMUT line tells you how much of a frame sat outside that range, and this tells you where. What a mark means: that pixel's color was pulled toward its own brightness until every channel fitted — hue and brightness did not move, saturation did. How to read it: a fringe on a siren, an LED or a specular is the limiter doing its job, and you can leave it; marks covering a face, a sky or a whole costume mean the picture is arriving too saturated for the space you are delivering to, and the fix is upstream — pull Print Exposure or the saturation controls, or deliver to a wider Output. On the two unbounded outputs, Log and ACEScct, the limiter is deliberately not applied, so the box says MODULE OFF on the picture under a DIAG LIMIT badge rather than leaving you to wonder. Like every overlay, it will be in your export if you leave it on.
It marks, on the picture, every pixel that decodes to more light than your working space can express — the one-glance answer to "is this clip really the log format I told it?" The rule is exact and has no threshold: after Camera Input decodes your footage, any pixel with a channel outside the working space's 0-to-1 log range is marked, in a hatch running the opposite way to Show Limited Pixels so the two can be up together. Three things cause marks. A Camera Input that over-decodes — the wrong-wrapper case: Rec.709 material read as Arri Log C4 starts marking, and the same footage read as Rec.709 marks nothing. A working space with less headroom than the camera — correctly wrapped Log C4 crosses DaVinci Wide Gamut's ceiling at nine stops over gray while staying inside its own space, so if a bright clip lights up, try the working space that matches the camera. And a genuinely extreme highlight — a filament, a specular, the sun — where the mark is simply right. Every space has a published ceiling worth knowing in stops over mid-gray: DaVinci Wide Gamut reaches 9.1, Arri Log C3 8.3, Arri Log C4 11.4, ACEScct 10.3, RED Log3G10 10.0, Blackmagic Film Gen 5 10.3. Two limits, so you can read silence correctly. It is a wrong-curve detector, not a wrong-gamut detector: it cannot see a primaries mistake in the narrowing direction — Rec.2020 footage read as Rec.709 comes back desaturated but perfectly legal, and no rule can tell that apart from a scene that was simply less saturated. And it needs Camera Input on a real curve: on None nothing is decoded and it says MODULE OFF under a DIAG CAM RANGE badge. It will be in your export if you leave it on.
When you want a fast, checkable starting point on an unfamiliar clip. It estimates what the scene was lit at from the frame you are parked on, writes it into Source (As Shot) Kelvin and Tint, and returns both Shift dials to zero so you still have a clean zero to trim from. It runs once, on that frame — nothing tracks the picture over time, so a cut or a lighting change does not move it until you fire it again. It reads the color of the edges in the frame and assumes they average out neutral. That is a better assumption than assuming the brightest things are neutral, but it is still an assumption: heavy foliage, a single-gel interior, a red wall against a red sky will pull it. A frame with almost no detail — fog, a blank wall, heavy defocus — gives it nothing, and it will decline and leave your dials alone. The label says "(estimate)" for a reason.
The switch turns film grain on. The grain is a physical emulsion model, so its size and structure follow the Film Gauge and the Grain Stock rather than being a fixed noise overlay. Grain Amount is a scaler, not an absolute: 100 percent is whatever your chosen Grain Stock calls for, so a coarse stock stays coarser than a fine one at the same setting. It is live in every mode, including Match Stock. At 0 the grain is off without touching the switch.
The switch turns halation on: on color negative, the red-biased glow real film gets when bright light passes through the emulsion, reflects off the back of the base and re-exposes it from behind (on the two Noir stocks the halo is neutral, and under Print Halation Physical on 2383 it is blue-violet — the print's layers run the other way). Halation Amount scales against your Halation Stock's calibrated glow, where 100 percent is that stock's amount. The response is steep rather than proportional — 200 percent is much more than twice, and 50 percent much less than half. At 0 the main glow is gone, but Halation Aura and Halation Veil are separate lobes and keep whatever you dialed into them.
It turns on the emulsion's sharpness model: the slight softening a real film layer applies, different for each color record. It is off by default and it is the master switch for everything in the MTF group below, including the print's own optics.
The switch turns on gate weave: the small frame-to-frame wander of film through a camera or projector gate. Weave Amount Pct is how far the frame wanders, as a percentage of frame width — and the numbers are honest about how small real weave is. The slider tops out at 0.5, half a percent of frame width, because that is already a badly worn projector; the shipped 0.075 is under a tenth of a percent, which is the scale at which film feels alive without you spotting why. Small numbers read as life, large ones read as a worn projector.
Save Preset writes the current grade to a .decolorispreset file — set it to Save and a file dialog opens; the control returns to Off by itself. The dialog opens in your Documents\Decoloris Presets folder — the same Documents folder Windows and the installer use, so it is correct on a machine whose Documents lives in OneDrive. The two factory presets, Neutral (Verified) (every module off — the verified pass-through starting point) and Grain and Halation Only, travel inside the plug-in and are written into that folder the first time you open the dialog if they are not already there; a preset you have edited is never overwritten. Load Preset loads one over the current grade. Only the parameters stored in the file are touched, so a partial preset leaves the rest of your grade alone. Presets travel between hosts.
It writes the frame you are looking at to a still image file, exactly as it renders here including the Output transform. Set it to Export and a file dialog opens; the control returns to Off by itself and the file appears once the frame next redraws. Export Format picks the file: PNG 16-bit keeps the full depth of the render and is the one to send to a colorist or a print, PNG 8-bit is for the web, TIFF 16-bit is uncompressed and opens anywhere. None of these settings changes your picture.
It applies a .cube LUT of your own on top of the finished picture. It runs after everything else, including the Output transform — which is what the "(Out sp.)" in the label is warning you about: the LUT must be one made for the space Output is set to. A Rec.709 LUT over a Log output will look wrong, and the plugin cannot tell. With no file loaded it does nothing at all.
Load LUT... opens a file dialog. Once a LUT is loaded the control reads Active; with none loaded it reads Off. Choose Clear loaded LUT to remove it from this clip. Selecting Active by hand does nothing — it is where the control rests, not a command. A copy of the file is kept in your Documents under Decoloris LUTs / Imported, so the LUT comes back when you reopen the project and travels with a saved preset. 3D cubes only, tetrahedral interpolation; 1D .cube files are declined.
How much of the loaded LUT is applied, from none at 0 to the full LUT at 100. It blends against the picture as it looked before the LUT, so 50 is exactly half way.
Export LUT writes the color half of your grade to a .cube file in Documents, at 33 or 65 cube. A LUT is a per-pixel color mapping, so it carries the color pipeline exactly — and it cannot carry grain, halation, diffusion, dirt or any other spatial stage. The popup entries say "no grain" for that reason. Export is not available when Output is ACEScct, which is scene-referred; switch to a display output such as one of the Rec.709 entries to export.
Try this: turn on Show Curves in the Utility group. It evaluates the same chain the exporter bakes, so what you see there is what the LUT will carry.
It turns the exposure and white-balance stage on. This is the scene-referred correction that happens before any film emulation runs. It ships on, because a scene that has not been set straight is not a scene the film stage can do anything sensible with.
It is a fine exposure trim in stops on top of the main Exposure — the same kind of move, with a smaller range for landing it. One stop here is exactly one stop, in every working space and at every brightness. Use Exposure to get close and this to finish.
Contrast works about the working space's mid-gray: below 1 flattens, above 1 steepens, and mid-gray itself does not move. Contrast Pivot (Con not 1) moves the point the contrast turns about, measured in log code — across its travel that is roughly three and a half to four and a half stops depending on working space. Negative pivots low, so contrast opens up the shadows; positive pivots high and holds the highlights. As the label says, with Contrast at 1 the pivot does nothing at all.
These are the trim half of the two-dial white balance: they warm, cool or shift green-magenta without touching what you declared the scene was lit at. Temperature Shift works in mired, the unit lighting gels are sold in, so one step is the same size of correction at every source temperature — about +167 is a full CTO, about −131 a full CTB, and 0 is exactly the white you declared. Positive warms. Tint Shift trims green-magenta on top of the as-shot Tint; positive goes magenta, negative green, and 0 leaves the declaration untouched.
Try this: declare with the Source dials, taste with the Shift dials. Auto Balance zeroes the Shift dials on purpose, so you always have a clean starting point.
Saturation before any film emulation, measured about neutral log luma. 1 leaves the image alone; 0 makes it monochrome. The name says its job: this is the source's saturation — the correction tier, run before the film stock has done anything. Shape the look's saturation in the Color group instead; trim the stock's own in Film Stock.
They are a warm-cool split tone applied before the film stock. Shadow Warmth warms or cools the shadows only, fading out as the image gets brighter; Highlight Warmth is its counterpart at the top end. Positive goes amber, negative goes blue on both. These are a warm-cool axis, not green-magenta — for green-magenta use Tint Shift. Both are measured in the chain's reference log, so the same setting is the same shift in every working space.
When a source needs a correction the temperature model does not reach — a heavy or technical cast the Kelvin and Tint axes simply do not describe. They are direct channel gains in stops on top of the Kelvin and Tint solution, and they still run before the film stock. Three balance-shaped instruments exist in this plugin and reaching for the wrong one is the commonest way to get lost, so: set the scene neutral with the Source and Shift dials; use these Gains for a big cast those axes leave behind; and if the problem only appeared after the stock and print ran, correct it with the Printer Lights in the Print group instead.
Balance Strength blends the whole balance result against the untouched input — lower it when the correction is right in direction but too strong. Disable Exposure and Disable Balance are expert A/B switches: each bypasses one half of the Balance stage while leaving the other running, so you can see what the exposure move alone or the white-balance move alone is contributing. They are for comparison and are never stored in a preset.
How wide a band of hues the skin diagnostic paints as skin. It sets the false color only — the width of the hue window around the reference chip that the Skin Analysis and Output Skin Analysis overlays tint. It does not affect the numbers those overlays print, and the paint always stays frame-wide even when a Sample Point is directing the numbers at one face. Nothing here touches the rendered image.
It is the veil inside the taking lens: a flat lift added in linear light as a percentage of mid-gray, applied before the film sees the scene. It lifts the blacks and leaves the highlights almost untouched, which is what scattered light inside a lens barrel does. It is measured against mid-gray as this instance has exposed it, so moving Exposure or Exposure Trim no longer changes how much veil you get. It does not follow the brightness of the scene itself — that would need a per-frame measurement, and this control is a constant. For scale: veiling glare measured across a whole lens runs roughly 1 to 10 percent on ordinary glass; measured off a chart the same lenses read a tenth of that, nearer 0.1 to 0.7 percent for a modern coated prime and 1 to 3 percent for older uncoated glass. Above a few percent you are past the physics and into taste, which is a perfectly legitimate place to be. Lens Veil (In Camera) is the stage's own on/off — leave it on and use the amount; switch it off to A/B without losing your setting.
Light spatial cleanup of the digital plate, before the film chain adds its own texture. It leads with color noise, which is the half that can be smoothed safely; luma is left alone unless you ask for it. Degrain-before-regrain is standard compositing practice and this is that step, scoped small. It is not a replacement for a dedicated noise reducer on badly degraded footage — do not buy it as a rescue tool.
Auto measures the noise level from your footage itself, every frame, and usually gets closer than a preset ever could because it is looking at your camera rather than someone else's. It errs slightly low on purpose: under-estimating leaves a little noise, over-estimating smears movement. Switch to Manual when you already know the number, or when a shot is unusual enough to argue with — that hands control to the Noise Level slider below.
The expected noise level at mid-gray, as a percentage of mid-gray. Set it to match your source: too low leaves noise, too high smears detail. It does double duty — it is also the yardstick the temporal filter measures neighboring frames against, so setting it too high makes moving objects trail as well as smearing detail. Manual profile only; while Noise Profile is on Auto this slider is not reaching the picture.
Shadow Bias shifts the noise model toward read noise, which is constant, and away from shot noise, which follows exposure — raise it when the shadows are the noisy part. Gain Comp. sets how much of the noise is treated as a fixed gain term rather than as exposure-dependent. Both are Manual profile only.
Chroma Strength is the main control. Color noise is the ugly half and the safe half to remove hard, so it sits at full strength by default; lower it if saturated fine detail starts to bleed. Luma Smoothing smooths the luminance channel as well, and it is off by default, deliberately: pushed luma smoothing is what makes footage look waxy, skin first. Raise it a little only if gray noise still reads after Chroma Strength has done its work.
It compares each frame with the ones either side, so still parts of the shot clean up without losing detail; moving parts are left alone automatically, so it does not smear. It is on by default. It needs full resolution, which is what the label says: below full playback or render resolution, Premiere and After Effects do not hand Decoloris the neighboring frames at all, and draft and thumbnail renders skip them too — so a reduced-resolution preview can differ from the export you approved it by. In Premiere the neighbors arrive on the GPU-accelerated render path; if you switch the Mercury Playback Engine to Software Only, check a frame before relying on it. It also cannot see an edit: on an adjustment layer or a timeline-level node the frames either side of a cut belong to a different shot. Put Decoloris on the clip if that matters. And it costs export time, because the neighboring frames have to be built too.
Temporal Frames is the window size centerd on the current frame: 3 compares with one frame either side, 11 with five. More frames means less noise on still parts, improving roughly with the square root of the count — so 11 is about twice as effective as 3, not five times. The cost is linear, though: every extra frame is built through whatever sits above Decoloris, so 11 makes an export several times slower. It also costs GPU memory — about 4.8 GB at 8K on the widest setting — and if the card cannot hold that, Decoloris quietly renders a narrower window rather than failing, which can look like the control doing nothing. Start at 3.
Temporal Strength is how much of the temporal cleanup to apply. Lowering it does not make the render faster — the neighboring frames are still built and still fetched, only the amount blended in changes. What it is genuinely for is the residual the gate reads as noise: a single-frame brightness or color event small enough to look like grain — a distant spark, a monitor refresh, a practical beating against the shutter. That erasure is exactly proportional, so half strength erases half as much.
Try this: if that flicker is the shot, pull Temporal Strength down rather than narrowing the window.
Both are readouts drawn over the picture, and neither changes the render underneath. Show Noise Fit prints what Auto measured on this frame — the noise level as a percentage of mid-gray (the same number the Noise Level slider takes), the gain component, and how many exposure bins voted. Two bins on a flat sky and twelve on a full ladder are not the same estimate, so the count is worth as much as the value; if it reads MANUAL / NO FIT, the estimator declined and the three Manual sliders are what is actually running. Show Temporal prints what the temporal pass did: the window, how many neighbors survived out of how many were asked for, and the average effective sample count, where 1.0 means every motion gate closed and the result is the same as Temporal being off. If the pass did not run it names which of ten reasons in one line — including BELOW FULL RES, DRAFT RENDER, SOFTWARE PATH and GPU MEMORY. Fewer taps than the window asks for is not a fault; at a cut or a clip boundary, a narrower window is the correct render.
The fall-off in brightness toward the corners that every real lens has. Because it dims the light before the film records it, the corners change in contrast as well as brightness — and by an amount that depends on which film stock you have chosen. That is the difference between this and a corner-darkening mask drawn on top.
No, and the copy should never imply otherwise. The entries — from a gentle modern fall-off through vintage glass to anamorphic — are named for the look they give rather than measured from real lenses, so trust your eye over the label. Manual leaves the controls below to you; every other entry sets Strength, Outer Radius, Inner Falloff and Anamorphic for you.
Strength is how dark the corners get: 0 is no vignette, 1 is the full fall-off of the selected profile. Profile is the shape of the fall-off — Optical cos4 is the steepest and most lens-like of the three, taken from the shape of the cos4 law rather than computed from a real focal length, so read it as a character rather than as a measurement; Gaussian and Smoothstep are softer, more stylized curves. Any preset other than Manual sets Strength for you.
Outer Radius is how far out the vignette reaches — smaller values bring the darkening in toward the center. Worth knowing: on a 16:9 or wider frame most presets already reach past the corner, so raising this changes the falloff across the frame while leaving the corner itself at full strength. Inner Falloff is how quickly the vignette starts once it begins: low values give a long gentle gradient, high values a tighter, more visible ring. Anamorphic stretches the vignette horizontally or vertically, for a lens whose vignette is not circular. All three are set for you by any preset other than Manual.
Centre X and Centre Y move the center of the vignette for a subject that is not in the middle of frame. Vignette Tint tints the darkened corners rather than only dimming them — positive goes warm, negative cool.
A lens effect, and the distinction is the whole point of the module. This is the halo and lifted contrast of a net or a diffusion filter in front of the lens — what you would have got by screwing a filter on on the day. Nothing in the negative itself does this. The film's own light scatter is far smaller than a pixel and lives in the MTF group; lens flare is a different thing again.
The entries are shaped after the classic filter sets: the Black PM grades, White PM, HBM, Black Sat, Golden Glow, Warm PM and Glimmer, plus Custom Manual, which leaves the controls below to you. Out of the box, with the module switched on, the popup rests on Black PM 1/8 — the lightest measured black grade, a whisper rather than a look. The Black PM, Warm PM and Glimmer ladders are measured — their glow, their color and their shadow lift were fitted to footage of the real filters, grade by grade — so those grade numbers mean what they do on set. The other families were matched by eye, and we say so. Warm PM is the warm-tinted member of the PM group: the same scatter as White PM with a fixed warm cast that does not change as you climb the grades. Glimmer is the lightest touch in the list — a faint, near-neutral glow around the brightest sources and almost no shadow lift, even at grade 4.
Try this: grade numbers are per family. Black PM 1/2 and Warm PM 1/2 are not the same amount of scatter, and the two measured ladders cross. Judge a grade inside its family, never across the list.
Because it only reaches the picture on Custom Manual. It scales the whole filter — halo and veil together — but the Diffusion Filter popup replaces it outright on every other entry, and the shipped default is Black PM 1/8, not Custom Manual. Select Custom Manual and the slider comes alive.
How much of the flat milky veil comes with the halo. Lower it when you want the bloom around highlights but not the lifted blacks. The veil is the honest half of the filter: it spreads light rather than adding it, taking out of the sharp image exactly what it lays back down — so raising it costs contrast instead of exposure.
A diagnostic overlay of which parts of the frame are throwing the glow. With Glow Floor at 0 every pixel scatters in proportion to its brightness, so the bands read as contribution relative to a mid-gray pixel; raise the Glow Floor and they read as stops above it instead — and the key strip prints those levels as exact stops, numbers rather than words. A viewing aid only — and, like every overlay in the panel, it renders into the frame, so switch it off before you export.
Scattered Fraction is what share of the light the filter scatters — 0.25 means a quarter of each pixel's light is bent out of its straight path and spread into the glow. The light is moved, not added, so overall exposure does not change; only where the light sits does. For scale, the Black PM rows run about 0.01 at 1/8 up to 0.82 at 2. Glow Floor sets a brightness below which light does not scatter at all — zero, the default, is the physical case where all light scatters in proportion to brightness; raise it to confine the glow to highlights, which is an artistic choice rather than what the glass does. Bloom Softness is how far the bloom spreads.
Glow Warmth is how much the glow's color changes from its center to its edge. Measured against real filter footage, a diffusion glow is not one color: its core is close to neutral and carries most of the light, while its outer edge is markedly warmer. 1 gives the measured amount, which is what every preset uses; toward 0 the edge cools back to neutral while the center keeps its color. Glow Tint Amount is how colored the glow is overall — 0 gives a neutral white glow, 1 the full measured color — and it scales center and edge together. Both cost no brightness at any setting: the glow delivers the same light and only moves color around.
Shadow Lift lifts the shadows independently of the bloom, for the fogged look of a lighter filter grade — it is a straight lift rather than scattered light, which is why the black grades ship with none of it and the white ones with some. Lift Threshold is how far up the tone scale that lift reaches before it fades out. Contrast Reduction flattens overall contrast the way a heavy net does; it is a tone adjustment above mid gray rather than more scattering, so reach for it only after the veil has not given you enough. Texture Soften softens fine texture without blooming it — the skin-smoothing half of a diffusion filter.
It reshapes the halo's fall-off, and it affects the halo only, not the veil. Despite the name it does not extend the glow's reach: measured against real filter footage it slightly tightens the halo, moving weight inward. Leave it at 0 unless you want that. If you want a genuinely wider reach, this is not the control that can give it to you — use Bloom Softness.
The on/off switch and the amount for this module live at the top of the panel; these are its character controls.
How much of the weave is vertical relative to horizontal. At 1, where it ships, the two axes carry the same amplitude — which is what a camera maker's single steadiness figure implies, and what the projection standard's own small-screen row tabulates. Lower it for a gate that wanders further side to side than up and down; well above 1 the motion starts to read as a mechanical fault rather than as life.
The projector wanders further and faster than the camera and adds relatively more up-and-down motion, because it re-seats every frame where a camera holds it on pins. There is also a routing difference that matters: in Projector Gate the film-borne stages — grain, dirt, halation, diffusion, MTF and adjacency — ride the weave within the frame, the way they do on real projected film, rather than staying welded to the screen. Camera Gate renders exactly as it always did. Note that from the shipped vertical ratio of 1, the projector multiplier puts up-and-down motion above side-to-side, which is further than the projection standard's own tolerance table goes — lower Weave Vertical Ratio if you want it to sit inside that table. It is also what an audience actually saw.
Try this: the projector routing is invisible while the grain redraws every frame. Turn Animate Grain off, or set Coarse Frame Hold above 1, and you can watch the grain travel with the weaving frame.
Weave Period Frames is how many frames one side-to-side cycle takes: shorter is jittery, longer is a slow drift. Side-to-side motion is held to a floor of four frames however low you set it, because a real gate cannot weave that fast. Weave Jitter is how irregular the weave is: at 0 it is a clean repeating motion, and at 1, where it ships, the up-and-down component is drawn afresh every frame — which is how a real gate re-seats film — while the side-to-side component stays mostly a slow drift at any setting. That split is why raising it reads as gate weave rather than as noise. Anamorphic Weave scales the horizontal and vertical motion differently, to match an anamorphic capture.
A colorist's grade rendered through the film — the creative pass, not a film process. It sits above Film Stock in the panel because that is where it acts: everything you dial here is then rendered through the emulsion and the print, so the panel now reads the way the chain runs — Color, then Film Stock, then Print. The preset popup picks the look: Faded Matte, Warm Print, Cool Bleach, Teal Orange and B and W Tone each set a whole split-tone and saturation grade; Manual applies none of it and leaves the offset and saturation controls below to you. Color Strength is how far the preset goes; it starts at half strength, because a preset at 1.00 is the full graphic version of that look and most work sits below it.
They offset a channel across the whole image, the way a color timer dials a printer light rather than a lift or a gain — but applied before the film stock, so it is an analogy and not a lab unit (the panel's order now shows you exactly that). They are measured in the chain's reference log, so the same setting is the same offset in every working space. If you want real printer points, use the Printer Lights in the Print group. Any preset other than Manual sets these for you.
Saturation in one tonal band each. Pulling shadow saturation down is what gives a faded print its gray blacks. Midtone is where skin lives. Pulling highlight saturation down keeps bright colors from going electric. All three are set for you by any preset other than Manual.
No, and this is worth being precise about: it darkens colors in proportion to how saturated they already are — a chroma-keyed neutral density, which is what "subtractive" means in grading practice. Neutrals are untouched, and the colors keep their hue and their relative strength: nothing is added, and nothing shifts. It is a density behavior, not a simulation of individual dye layers — we would rather name the category honestly than overclaim the physics.
Per-color subtractive control: each dial IS the amount of subtractive saturation for its own color region — colors darken as they enrich, and neutrals never move. Six dials — Red, Yellow, Green, Cyan, Blue and Magenta Density — beside Subtractive Saturation, each running 0 to 1 from a resting point of 0. Raise Red Density and red density applies, immediately and on its own; every dial position does something, in its entire range, from a fresh instance.
How the two relate: the master is the everything-at-once amount, and dial and master simply add — each color's amount is the master plus its own dial. Three properties worth knowing:
The highlight shapers below still govern how color behaves at the top of the scale.
Diagnose the stage, correct at the stage:
The scope taps route you: hot at Camera is a source problem; clean at Camera but hot at Output is the film or the grade.
They govern how color behaves at the top of the scale, because real dye layers desaturate as they approach maximum density. Luma Sat Highlight is how much saturation is taken out of the brightest highlights — raise it and highlight color thins toward the top of the scale; at 0 highlight color is left alone. HL Onset Trim moves the point where that desaturation begins — negative starts it earlier (and so widens the region it acts on), positive holds full color further up. Sat Rolloff Str is the strength of the highlight rolloff — how much saturation it removes across that region; 0 switches it off entirely, higher removes more. It does not change how gradual the transition is; the onset does that.
How much color the Color group drains out of the shadows. This has always happened whenever the group was switched on; until it was exposed as a control it was fixed and hidden. The default is exactly the old fixed amount, so leaving it alone changes nothing. Lower it toward 0 to keep shadow color, raise it for a more washed, printed-down look.
Neither pole is "correct" — the choice is taste plus interoperability, and we will not pretend otherwise. ASC CDL Compatible is the default and matches the operator the ACES and CDL specifications define, so a grade you build here reads the same way in other tools that follow them; it works on log code, and because of that a saturation boost also brightens strongly colored areas. Luminance-Held instead holds each pixel's actual brightness while the color moves, so reds and blues stop gaining exposure as you saturate them. Neutral grays are untouched either way. This affects the Color group and its saturation controls only — Balance and Film Stock always use the first.
The switch turns the film stock's color character on. With it off, the Stock selection above still names a stock but none of its character is applied. The other direction is worth knowing too, and it explains a common surprise: a stock's tonal character — its contrast, its toe and its shoulder — is carried by the negative, and only reaches the picture once the Print stage is on. With Print off you get the stock's color and none of its curve, so two stocks that differ only in tone will look identical.
They trim the stock's own curve without leaving the stock. Contrast Trim negative flattens toward a softer negative, positive steepens toward a punchier one. Shadow Density Trim sets how dense the stock's shadows sit — positive deepens the blacks the way an underexposed negative does, negative opens them up; it is a shift of the negative's speed point in stops, the same number of stops in every working space. Highlight Rolloff Trim positive rolls the highlights off earlier and more gently, negative holds them straighter for longer.
Trims the stock's saturation. It has no effect on Noir 5222, which is a black and white stock and is converted to monochrome inside its own character — there is no color left for the trim to act on.
Each trims one layer's toe — the point where shadow detail starts to disappear. Positive lifts that color out of the blacks, negative buries it. They are not interchangeable: green carries most of the luminance, so the green toe moves shadow detail as well as shadow color, and blue is the noisiest record on most stocks, which makes the blue toe the shadow-color control with the most visible reach.
Each trims one layer's shoulder, where highlights compress. Positive rolls that layer off earlier. Because red and blue sit opposite each other, the practical effect is a color move at the top end: rolling red off earlier cools bright areas, rolling blue off earlier warms them. Green again dominates luminance, so its shoulder changes how the whole highlight rolls, not only its color.
They add a split tone on top of the stock's own — warm on positive, cool on negative, in the shadows and the highlights respectively. Like the stock's own split tone they are measured in the chain's reference log, so the same setting is the same color shift in every working space.
It draws a readout over the picture of the stock you have selected — named in the strip, so you always know which emulsion the curves belong to: its characteristic curves for red, green and blue, and the contrast, shoulder and toe values actually being used — your trims included, not the untouched table values. Every printed number carries a one-letter provenance tag: C where it was fitted to a published chart, V where it was set by eye, S where it is invented. The year appears only where we hold a dated source for it and reads YEAR NO SRC otherwise, because a year attached to a coined name is a claim about which real emulsion it is. And because the curves are the negative's, and the negative is only developed inside the Print stage, the panel says PRINT OFF when Print is down — the curve you are looking at is not reaching the picture. Diagnostic only; it changes nothing about the render underneath it.
They simulate developing the negative longer or shorter than normal, which changes contrast and shadow density together. Push Pull Stops is how many stops: positive pushes, raising contrast and coarsening grain; negative pulls, flattening the image. Push Pull Strength blends the pushed or pulled result against the normally developed negative.
It leaves the silver in the print instead of bleaching it out. Contrast climbs, blacks go dense and color drains, in the way the real process does. 0 is a normal print.
The Grain switch and Grain Amount live at the top of the panel; these are its character controls.
Which grain character is used. Each entry carries its own measured amount, size and tonal weighting, taken from that film's granularity data — so the differences between entries are the differences between the emulsions, not one fineness dial with different labels on it. Match Stock derives the choice from your main Stock selection, so the grain belongs to the film you picked; choose an entry explicitly to override that.
Because Grain Stock sets both on every entry except Manual — and Match Stock is an entry, so on the shipped default these sliders are not reaching the picture. Set Grain Stock to Manual and they come alive. Grain Size is how large the individual grains are: larger reads as a smaller gauge or a faster stock. Grain Softness softens the grain the way a lens or a scan would; at 0 it is as crisp as the model makes it.
Because of delivery. Neutral grain survives compressed delivery far better: colored grain puts fine detail into the chroma channels, which streaming codecs subsample and quantize hardest, so it can break up into a moire-like pattern on web platforms. Turn it off for per-record color grain on a master you are not compressing. Under Match Stock on Noir 5222 or Noir 5231 it is forced on, because a silver stock has one record and cannot produce chromatic grain.
Animate Grain decides whether the pattern changes frame to frame — turn it off to freeze a still, or to grade a frozen frame without the grain crawling. Grain Seed Offset shifts the random seed; change it when two instances of the effect are producing visibly identical grain. Coarse Frame Hold holds the coarse layer for a number of frames before it changes, which slows the largest structure down and is closer to how projected grain reads.
Where the grain sits in the chain. Printed (post-print) puts it after the print stage, where a release print's grain sits. Negative (pre-print) puts it before, so the print stage develops the grain along with the image. Honest caveat: the negative position only images the grain through the print when Print MTF is on and the delivery is wide enough — 3840 and up at 35mm, 1920 and up at Super 16, always at Super 8. Below those widths, or with Print MTF off, the two positions differ in tone rather than in imaging.
It was removed in v2.157.0, and here is the honest reason: the overlay described a grain-amplitude model the shipped grain no longer used — its code never read the grain that actually renders — so it was drawing a picture of nothing. We would rather remove an instrument than let it claim a measurement it does not make. If you had it ticked in an old project, nothing about your rendered grain changes; only the overlay is gone.
Aspect Ratio stretches the grain horizontally or vertically; set it away from 1 to match an anamorphic capture, whose grain is squeezed with the image. Grain Streak pulls the grain into directional streaks rather than round clumps; 0 is the isotropic default. Grain Color is how much the grain differs between the color records — negative pulls it toward neutral, positive lets each layer's grain read as its own color. Grain Color is Manual-only, like the other manual sliders in this group.
Midtone Grain is how much grain lands in the midtones relative to shadows and highlights: raising it fills the mid-scale in, so it changes where the grain sits on the tone scale rather than how much there is overall (the fader at the top of the panel is the control for overall amount). It also sets the level at the very bottom of the scale. Fine Grain is the amplitude of the fine layer, which carries the high-frequency sparkle; Coarse Grain is the slower, clumpier structure that reads at a distance. All three are set for you on every Grain Stock entry except Manual — and Match Stock is an entry.
They set the amplitude of each dye record's grain: the cyan record lands on red, magenta on green, yellow on blue. The shipped defaults are the fitted per-layer ratios, and green is the reference layer the other two are normalized against. Yellow is the loudest layer on most stocks, which is why blue grain reads first. Changing one alone shifts the grain's color.
Nothing, and the label says so. It fed a density-based grain amplitude model that Decoloris used through version 2.87; the grain tone response is now weighted by shadow, midtone and highlight zones instead, and nothing reads this value. It is left in place so existing projects and presets keep loading unchanged. We would rather tell you it is dead than let you chase it.
The Halation switch and Halation Amount live at the top of the panel; these are its character controls.
Which stock's halation behavior is used. Match Stock derives it from your main Stock selection; pick an entry explicitly to override that.
Threshold in Stops (Manual) is how bright a highlight has to be before it starts to halate, set in stops over mid-gray — the same scale the Show Halation Indicator's key strip prints. Lower it and mid-tones begin to glow, raise it and only the brightest speculars do; the range runs −3 to +4 stops and the default (+1.85) is the same brightness the old scale called 0.65. Note the "(Manual)" in the label: Halation Stock sets it on every entry except Manual, and Match Stock is an entry, so on the shipped default it is not reaching the picture. Real film has no threshold at all — this is a look control, and the honest default is the stock's. Halation Softness is how far the halation spreads — larger values give a broader, softer red bloom around the same highlights — and it is live regardless.
Chromatic Dispersion is a look control: it lets the halo's color separate toward its edge by giving the green and blue parts of the halo different reach. It is not a model of how the records scatter in the base — real halation's red comes from the layer order, not from different scatter distances — so treat it as styling, not physics. Hue Offset rotates the halo's hue away from red — real halation is red because the red record sits deepest, so move this only deliberately. Halation Sat (Manual) runs the other way from what its name suggests: 0 is the reddest halo the engine can make, and raising it moves the halo toward neutral — a fully neutral halo is not reachable from this control. The shipped 0.8 is the model's own calibrated color set, so leaving it alone is the physical choice; move it only to stylize. All three do nothing at all under Match Stock on Noir 5222 or Noir 5231: a silver emulsion has no dye layers for them to steer.
Halation Aura is a second, wider glow around the main halo: the look of a stock with its anti-halation backing removed, where light returns from much further away. 0 is off. Aura Size (needs Aura) widens it and only does anything when Aura is above 0. Aura Green is how much green that wide aura carries — real no-backing halos are not purely red, and this is that residue. Halation Veil is a soft red lift that follows the light: it averages the brightness over the halo's own radius around each pixel and adds a little red-weighted fog there, so it lifts the blacks near bright areas rather than glowing around highlights — and it is local, not a uniform wash across the whole frame. Aura Green and Halation Veil both go silent under Match Stock on the two Noir stocks.
A diagnostic overlay of which highlights are driving halation, its key strip printed in exact stops rather than words. A viewing aid only, and it renders into the frame — switch it off before you export. Judge halation at 100 percent view: a fit-to-window preview downsamples the frame first, which averages away the near-clipped pixels the effect keys on, so the halo you see small is not the halo you get delivered.
Falloff Skirt controls how much faint glow survives far from a highlight. Brightness Spread widens the halo of brighter sources: the hotter a highlight is above the threshold, the further its glow reaches. It does not gate dimmer sources out — anything above the threshold still halates at its normal size. Skirt Shape switches the skirt from an exponential fall-off to a power law, which keeps a faint reach much further out; the power-law tail is our own fitted shape for the far reach, not a measurement of film scatter. Halo Model (in the group, beneath Halation Stock) is the one-click version of the physics, and Ring is the default: the physically grounded halo — light returning through the film base cannot come back closer than twice the base thickness, so a real halo is a ring with a dim center, sized from the cited base geometry. Bloom is the alternative: the soft near-field glow most tools ship. For the ring plus a little core softening, pick Bloom and raise Halation Annulus (Bloom) toward 1 — that slider is the expert blend between the two shapes and only applies under the Bloom model; Ring sets it fully to the ring. Under Print Halation Physical on 2383, the ring carries the print's blue-violet ordering too. Red Record Source decides what the halo is generated from: 0 uses overall luminance, 1 uses the red record, which on color negative is physically what re-exposes the emulsion (under Print Halation Physical on 2383 the deepest record is blue, and this control still reads red — a camera-negative control, left as is).
It sets which way the halo is colored on print stock, and it only does something when Film Stock is Teatro 2383 and Halation Stock is Match — every other combination ignores it. Print film is coated in the opposite order to camera film: the manufacturer's own layer diagram puts the blue-sensitive layer nearest the base, then red, then green, so light bouncing off the base hits blue first and a print halo is blue-violet, not red. Physical follows that layer order. Stylized Red keeps the red halo the plugin shipped before, which is the look people know from camera negative — a style choice, not a second opinion about the physics.
It selects which stock's measured sharpness curves are used; Match Stock derives it from your main Stock selection. The label carries the honesty: the emulsion blur radii are whole pixels, so below 4K most of these entries deliver the same picture as each other and some deliver nothing at all. Judge a choice here at delivery resolution, not on an HD preview. Noir 5222 selects a channel-uniform row, because a silver image has one record and not three.
Emulsion Softness is how soft the emulsion is overall — larger values scatter light further inside the layer, which is what a faster stock does. Acutance is the adjacency-style edge enhancement development produces; it puts back some of the apparent sharpness the scattering takes away. Acutance Radius is how wide that edge sits: small radii sharpen fine texture, larger ones put a visible halo on edges.
Interlayer Scatter is how much light scatters sideways between the color layers; it softens color more than luminance, which is what stacked emulsions do. Emulsion Skirt (tail) blends in a heavy-tailed point spread instead of a plain gaussian, keeping a faint softening far from an edge as measured film point spreads do — and it does not change overall softness: the kernel is re-solved so the delivered blur width stays the same at every setting, and only the far tail moves. If you want more blur, use Emulsion Softness. Highlight Scatter makes the softening depend on exposure so highlights scatter more than shadows; a stylistic control, not a measured one. 0 is off on both.
It adds the printer's own optics to the chain, applied on the negative's density as a contact print. Contact print is selected out of the box, but the MTF switch at the top of the panel decides whether any of it runs: with MTF off it does nothing, and ticking MTF gives you both halves of the film's softening — the negative's and the print's — in one move. It also needs the Print stage to be developing something. The gauge decides whether it is audible, and that is physics rather than a setting: the frame has to be wide enough for the blur to reach a whole pixel. At 35mm that is 4K and up, so HD at 35mm is silent; Super 16 needs about half that and Super 8 less again, so the smaller gauges are live where 35mm is not. The print's blue record is about twice as soft as its green, which is what a release print does, and all three records switch on together, never one at a time. High Quality does not reach this stage: it always runs at the standard tier.
The local contrast real development produces at edges, where exhausted developer next to a dense area leaves a visible fringe. Adjacency Stock picks which stock's behavior is used; Match Stock derives it from your main Stock selection.
Adjacency Amount (Manual) is how strong the edge effect is — higher values give film's characteristic bite at high-contrast boundaries — and, as the label says, Adjacency Stock sets it on every entry except Manual, with Match Stock being an entry. Adjacency Radius is how far the effect reaches from a boundary: small radii read as sharpening, larger ones as a visible fringe. Edge Gate (expert) is how much of an edge is needed before the effect engages; raising it keeps it off flat areas and grain-level detail. Interimage (expert) is where one color layer's development suppresses its neighbors — what gives color negative its cross-channel edge color. 0 is off.
A diagnostic overlay of where the adjacency stage is finding edges, with a legend that declares its scale. A viewing aid only, drawn into the frame — switch it off before you export.
The colored blobs a developed dye layer actually forms — which is what color negative grain physically is. Dye Cloud Stock picks which stock's cloud sizes are used. Match Stock derives it from your main Stock selection, and on Noir 5222 it derives off entirely: a silver stock forms no dyes, so there is nothing for the module to model. Choose an entry explicitly if you want the effect on a mono stock anyway.
Dye Strength is how visible the clouds are — it scales their amplitude, not their size. The three Cloud um (Manual) sliders are the cyan, magenta and yellow layers' cloud diameters on the negative, in microns: larger clouds read as coarser color structure in that channel, and the yellow layer sits on top and is usually the coarsest of the three. All three are Manual-only, and Match Stock is an entry, so on the shipped default they are not reaching the picture. Chroma Independence (expert) is how independently the three layers clump: at 0 they move together and the structure reads as neutral; toward 1 each layer clumps on its own and the noise takes on color.
A diagnostic overlay of the dye cloud field itself, with a legend that declares its scale. A viewing aid only, drawn into the frame.
The dust, hairs and sparkle that collect on a negative and on a print in circulation. Dirt Preset picks how dirty and in what way — the entries run from a clean negative through fresh and worn prints to archive material and small-gauge documentary. Any preset other than Manual sets the density, size, dark/light split, opacity and tint controls for you.
It is the difference between where the dirt was. Bright specks are dirt that was on the negative: it blocks the printing light, and a negative-working print stock leaves that spot clear. Dark specks are dirt and debris on the print itself or in the projector gate. Dirt: Negative (white) is how many particles land on the frame — the single control that separates a clean print from a battered one — and Dirt: Print (black) sets how much of the dirt is dark rather than bright.
It is how large the particles are: small sizes read as dust, larger ones as debris in the gate. And yes — the size is a physical one, so it follows Film Gauge. The same speck is about twice as wide on a Super 16 frame and four times as wide on Super 8 as it is on 35mm, which is four and eighteen times the frame area. Any Dirt Preset other than Manual sets it for you.
Animate Dirt decides whether the dirt changes frame to frame; turn it off to freeze it on a still. Dirt Strength blends the whole dirt layer against the clean image, so you keep the pattern and simply have less of it. Particle Opacity (expert) is how opaque the particles are, bright and dark alike: at 1 a speck blocks all the light it covers, lower thins both populations together, and at 0 the dirt disappears entirely. Dirt Tint (expert) tints the dirt — real emulsion debris is rarely neutral; positive goes warm, negative cool. Opacity and Tint are both set for you by any preset other than Manual.
Dirt Seed Offset offsets the dirt's random seed; change it when two instances of the effect are drawing the same dirt on the same frames. Reseed Dirt does that for you in one click: it sets the seed from this frame's own content, so different clips stop sharing the same specks without you inventing numbers. It is deterministic, not random — the same clip clicked again lands on the same seed — and the value is written into the dial above, where you can see it, undo it and keep it in a preset. It moves only when you click, and if the frame gives no answer the dial is left untouched.
Try this: one click per clip at the top of a reel, and no two shots share dirt for the whole show — through every re-render and every export.
They do nothing, and the labels say so. Both kinds of particle now land at Particle Opacity, on the negative and on the print alike, so there is no blend law left to choose. The entries are kept only so older projects keep their settings.
It turns the print stage on: the negative is printed onto Teatro 2383 release stock, which is where most of film's contrast and color signature actually comes from. The strength of that print is the Print Strength fader at the top of the panel.
How the negative is exposed onto the print, in stops. Positive prints brighter and thinner; negative prints darker and denser. Expert note worth having: +1.0 exposure with 0.20 paper-light trims is the classic highlight-protect print-down.
A printer point is the unit a lab color timer actually works in. One point is 0.025 log exposure at the print — about a twelfth of a stop — so the full ±12-point range here is one stop of printer exposure. That is a stop of the light hitting the print, not a camera stop: other tools quote roughly seven points per camera stop, which is a different quantity and does not convert. Lowering a light puts more of that color into the print, which is the opposite of what it sounds like. Near the print's aim density one point is worth roughly a quarter to a third of a stop in that channel, so a skin correction is usually one to three points, not ten. Green is the channel a timer conventionally holds still and times the other two against — if you are unsure, move red and blue and leave green alone. And one property worth knowing: the lights operate in the film chain's own fixed print domain — independent of the working space and upstream of the output encode — so a printer point means the same thing in every project you will ever open.
Try this: steer with these last. Set white balance at Balance before the stock, fix a large overall cast with the WB Gains, and use the Printer Lights for what the stock and print did to skin — Output Skin Analysis reads that correction out for you in these same points. The order is the doctrine: make the input honest at Kelvin and Tint, verify with the Sample Point, then let these lights place your skin on the print — solved from your point when one is placed. If the correction runs big, the fix is upstream at Kelvin and Tint.
It times the print so mid gray comes out where it went in, the way a lab times every release print to an aim density. On, the print changes contrast and color without also darkening the whole frame, and it holds mid gray at every Print Strength. Off is the untimed print, which sinks the picture by about a sixth of a stop at the default strength. It does not change the print stock itself: deep shadows still fall away, and the slope you measure at a given exposure does shift slightly, because timing a print slides it along a curve. Leave it on unless you are comparing the two.
It takes the print to full strength in the bright highlights only, so you can tame a blown sun or a specular without paying for it in the shadows. Below about three and a half stops over mid gray it does nothing at all, and that is exact rather than approximate. Around 50 removes the hard clip and the green-yellow fringe a clipped highlight develops; 100 overshoots and the highlight goes cool. It has no effect when Print Strength is already at 1, because there is no headroom left for it to take. And, plainly: it cannot bring back detail that was never recorded. A clipped source is a flat plateau; this changes how that plateau and its edge render, not what is inside them.
The dye loss and base yellowing a release print picks up over decades in storage. Fade Stock Era is the print stock era being aged — older chromogenic prints fade faster and pinker, and the Dye Transfer IB entry is effectively fade-free, because imbibition dyes were never chromogenic. Fade Storage is how it was kept: cool dry archival storage barely ages, hot and humid storage accelerates every decay term. Fade Age Years is how many years to apply. The cyan dye goes first, which is why old prints go pink.
Each multiplies one dye's decay rate. Cyan is the least stable dye in a chromogenic print and drives most of the pink shift. Magenta is the most stable of the three, so raising its multiplier is a deliberate departure from real chemistry — worth knowing before you do it. Yellow sits between them. Base Yellowing is the yellow stain the film picks up with age, from leftover color couplers and oxidised developer in the emulsion, so it warms the whole frame including the blacks.
Fade Strength blends the whole aging result against the fresh print — a straight wet-dry mix, so half strength is half the fade on every control in this group. Fade Model picks how the aging is computed. Beer-Lambert is the physical one and the default: it removes a fraction of each dye, so density falls, contrast goes with it, shadows go maroon and saturated blue skies wash out toward white, while anything already at paper white stays exactly where it is. Classic is the look the plugin shipped before, built from a shadow lift and a highlight tint that were tuned by eye rather than derived. It is kept so an existing grade can be reproduced — a style choice, not a second opinion about the chemistry.
The frame-to-frame brightness wander of a projector lamp, a hand-cranked camera or a mains-driven light. Projector Drift and Camera Modern are slow and gentle; Hand-Crank Silent Era is the irregular exposure of a cranked camera; Mains AC Beat is the steady pulse of an AC-driven lamp.
Flicker Amplitude Stops is how far the brightness swings — a tenth of a stop is already visible in a dark scene. Flicker Frequency Hz is how fast it cycles in real time, so it holds its rate whatever the project frame rate is; on Mains AC Beat remember a lamp pulses at twice the mains frequency, so 24 fps on 60 Hz mains divides exactly and a real projector shows no beat at all, while 50 Hz mains at 24 fps gives about 4 Hz. Channel Asymmetry is how differently the three color channels flicker: at 0 it is neutral brightness, and raising it lets the color wander the way a real lamp does — as it dips it cools, with blue falling fastest and red least. Dropout Probability is the chance per frame of a much larger drop, the way a worn projector occasionally stumbles; 0 never drops.
Animate Flicker decides whether the flicker advances with the timeline — turn it off to freeze it on a still. Flicker Seed Offset shifts the seed; change it when two instances of the effect are flickering in step. Flicker Strength blends the flickering result against the steady image.
It is presence added in the mid tones without sharpening edges — and it is explicitly not a film process. Nothing in a real film chain does this; the film's own edge effect is the Adjacency stage, which works at a different scale, with the opposite polarity at an edge. This is a grading control. It is gated by tone: it acts only in the mid range and is fully inert in the deep shadows and the brights. One measured consequence to expect, because it sits downstream of Grain: it amplifies grain as well as detail — about 25 to 31 percent on a flat midtone area such as a sky, a wall or a cheek, and between a third of a percent and 10 percent on detailed footage where the edge rejection dilutes it. That figure does not change with delivery resolution.
Amount is how much local contrast is added in the mid tones; negative values soften instead, which is a gentler route to a diffused look than the Diffusion stage. Radius is the size of the detail worked on, anchored to frame width rather than absolute pixels — so a given setting covers the same fraction of the picture at any resolution. The default of 16 works hardest on detail about 15 pixels across on an HD frame and about 30 across on a 4K one. Small values bring out texture; large values shape broader forms.
Range Sens is how large a tonal difference counts as an edge rather than as texture. It is what keeps the effect off edges and prevents halos, and it is the only thing that does — push it toward its maximum and that protection goes away by design: the filter stops distinguishing edges from texture and starts behaving like a sharpener, halos included. Measured on a step edge, the overshoot goes from nothing at the default of 5 to 13 percent of the edge at 20. Note it works the opposite way round from an unsharp mask's threshold: that protects flat areas, this protects edges. Shadow Gate and Highlight Gate set how far down and up the effect reaches: at their defaults it fades out below the lower mid tones and above the upper mid tones, so noise in the blacks is never amplified and bright detail is never made crunchy.
It is on by default and it should stay on. It sets how finely the textured stages are sampled — halation, diffusion, grain, local contrast and dirt. Turning it off is a faster preview, not a milder look: the halo geometry and the grain amplitude are the same either way, and only the smoothness of the sampling changes. Grain matches the final render at any preview size. Your host may already be lowering this while you scrub; your delivered render always uses the full setting.
It draws a second, separate box plotting the plugin's own transfer curves — what the film stock, the print exposure, the fade and the dye cloud together do to tone at the point Scope Tap is set to. It looks at no pixels at all. Every other reading in the panel is a measurement of your frame; this one is a picture of the controls, evaluated through the same chain the .cube exporter bakes, so what you see here and what an exported LUT does cannot disagree. That also makes it free — sixty-four evaluations of a curve, at any resolution — so leaving it on costs nothing measurable. It draws four strokes — red, green, blue, and the neutral in white on top — so a chain that leaves grays gray reads as one white line, and any color separation you can see between the strokes is the amount the chain is putting there. Use it to understand what a control does to the shape of the tone rather than to judge a frame: pull Print Exposure and watch the whole curve slide; pull Fade and watch the toe lift off the floor. It sits at the opposite corner from the scope, so both boxes can be up at once — and it lives here in Utility because it is a study tool for the deep-color reader, not a grading scope. Like every overlay, it will be in your export if you leave it on.
Internal identifiers, not picture controls — please do not edit them. The plugin writes them when you load a LUT; together they are a fingerprint of the LUT file's contents. If you do change one by hand you will not break anything: a value matching no LUT in your imported folder is simply skipped and the rest of the effect renders normally, and a value that happens to match another LUT you have imported will load that one instead. Use Load LUT... rather than typing here.
Trial watermark, licensing and pricing — deliberately not drafted. No trial, license or watermark control appears in the shipped panel table, so nothing in the walkthrough above needed skipping. If any trial messaging surfaces to the user (a mark on the frame, a dialog, a status string), it needs its own FAQ section written to the same honesty standard, and it is not written here.
Camera and log-format names are trademarks of their respective owners and are used only to identify compatibility. Decoloris LLC is not affiliated with or endorsed by any camera manufacturer.