Film emulation in color grading: what it copies

Film emulation is the attempt to reproduce, on digital footage, the measurable behaviour of a photochemical film chain: the density curve of a negative, the steeper response of a print stock, the halation that light produces inside the emulsion, and the granular structure of the silver image. It is four separable effects, and only one of them is a color mapping — which is why a single LUT never gets all the way there.

Knowing what each part does is the difference between a look you can steer and a slider marked "film" that you either accept or reject.

What film emulation actually is

Film emulation is a model of how film responds to light, applied to an image that was never on film. The useful version has four components — a tonal response curve, a print rendering, halation, and grain — because those are the four things that can be measured on real stock and published as data.

Manufacturer data sheets are why this is possible at all. Kodak publishes sensitometric curves for each stock, and its own description of them is the whole basis of tonal emulation: they "determine the change in density on the film for a given change in log exposure" (Kodak H-1-5219, read 13 September 2026).

One note on data. Published characteristic curves exist to be read and used; third-party emulation LUTs are separate works with their own licence terms, some copyleft. If you are building a model rather than buying one, derive it from published sensitometry or your own measurements.

The characteristic curve is most of the effect

A characteristic curve — also called an H&D curve, after Hurter and Driffield — plots "film density (log of opacity) versus the log of exposure" (en.wikipedia.org, read 13 September 2026). Its shape is the largest single contributor to what people recognise as a film look, and it has three named regions: a compressed toe at "the low end", a straight middle whose slope is the gamma, and a shoulder where "the curve rounds over".

Those two soft ends are the point. Digital sensors respond close to linearly and then stop; film's density climbs slowly out of the toe and flattens gradually into the shoulder, so highlight detail keeps arriving long after a sensor would have clipped. Kodak quantifies its own version: VISION3 500T's Sub-Micron Technology "enables 2 stops of extended highlight latitude", and in post "up to two stops more image information can be extracted from scene highlights".

Two consequences follow for anyone emulating this.

You cannot emulate latitude you never captured. A soft shoulder applied to footage that already clipped gives you a soft shoulder on top of a flat line. That is the argument for shooting log, laid out in what is log footage and how log formats allocate code values.

The curve must be applied in the right space. A film response is defined against scene exposure, so it belongs between a correct input transform and the display transform — not slapped on top of an already display-referred image. Applying it in the wrong place is one of the failure modes described in why CST, ACES and LUTs look different.

The print is a second curve on top of the first

The negative is not what audiences saw. A camera negative is deliberately low in contrast, and the picture is made when it is printed onto a much steeper print stock — so a film look is the composition of two characteristic curves, not one.

Kodak's 2383 print film is the stock most people are unknowingly describing when they say "film". Its data sheet claims "rich blacks and neutral highlights" and an "excellent tonal scale", and it is set up against a patch printed "to a neutral gray of 1.0 visual density", Status A densities 1.09 red, 1.06 green, 1.03 blue (Kodak H-1-2383t, read 13 September 2026). Print-film emulation exists as a grading stage because that second curve is where the contrast and highlight character live.

A third layer gets forgotten: the lab. Every shot in a photochemical release went through a timing decision before it reached the print stock — see color timing in film. A stock emulation reproduces the chemistry, not those decisions.

Halation is optics, not a glow filter

Halation is a halo around bright areas, caused by light passing through the emulsion, reflecting off the back of the film base, and exposing the emulsion a second time from behind. Film is built to fight it: an anti-halation layer exists for "preventing any light from being reflected back through the emulsion from the rear surface" of the base (en.wikipedia.org, read 13 September 2026). What you see in a film frame is what that layer failed to stop.

Kodak's 2383 sheet is specific: its antihalation dyes "offer superior protection against exposure by light reflected back from the support surfaces", "minimizing color fringing in critical scenes like white titles and night scenes with automobile headlights". Kodak's newer camera-negative structure moves the function into "an Anti-Halation Undercoat layer, or 'AHU'" (kodak.com, read 13 September 2026).

Three properties an emulation should respect:

• It is spatial. Halation spreads light across neighbouring pixels. No 3D LUT can produce it, because a LUT maps a color to a color with no knowledge of its neighbours.
• It is not neutral. Wikipedia's diagram of the effect, citing Prodigium Pictures, describes light passing the blue, green and red layers and bouncing "back into the red layer", which is why the halo reads warm rather than white.
• It is threshold-driven. Real halation appears around genuinely bright sources — a window, a headlight, a practical — not around every light-toned object. An emulation that glows uniformly reads as a diffusion filter.

Grain is a structure, not an overlay

Film grain is "the random optical texture of processed photographic film", arising from silver or dye particles, and — the part almost every plugin gets wrong — "it is not a particle but an optical effect" (en.wikipedia.org, read 13 September 2026). It also trades against sensitivity: "large crystals will therefore give more sensitive film, for the price of being visibly grainier".

The behaviour that matters is that grain is not uniform across the frame. Kodak measures it as rms granularity "read with a microdensitometer, using a 48-micrometre aperture", and instructs you to read the value for a given density off the characteristic curve — magnitude is a function of where you sit on that curve. Kodak also notes graininess is "highly dependent on scene content, complexity, color, and density" (Kodak H-1-5219, read 13 September 2026).

So a flat noise layer at fixed amplitude is wrong three ways: it does not vary with density, it is usually monochrome where real granularity differs per color record, and it goes on at the wrong point. Grain belongs at delivery resolution, after the grade — grading over grain rescales it, and heavy compression afterwards invites banding after grading and export.

Color response: dyes, not primaries

Film's color does not come from red, green and blue primaries. It comes from three subtractive dyes whose absorption bands overlap, which Kodak publishes as spectral dye-density curves: they "depict the spectral absorption of the dyes formed when the film is processed".

That overlap is why film renders certain hues in ways a saturation control cannot imitate: it is cross-channel behaviour, not per-channel. It is also why a credible emulation needs a 3D mapping or a physical model rather than three curves and a matrix — a per-channel curve set can reproduce the tonality and miss the color entirely.

Process variations sit in the same category. Skip-bleach retains silver that is normally removed, which "increases the contrast of the image and decreases the color saturation" (kodak.com, read 13 September 2026) — a chemical look with a real cost, not a preset.

What film emulation cannot give you

Worth being blunt, because this is where disappointment comes from.

Lenses. Flare, field curvature, focus falloff and the rendering of spherical or anamorphic glass are a large part of what reads as "film", and none of it is in the stock.

Lighting. Exposure placement, contrast ratios and practicals do more than any emulation stage.

Information you did not record. Clipped highlights stay clipped. An emulation redistributes what is there.

Sharpness behaviour. A stock's modulation transfer function describes how sharpness falls off with spatial frequency; digital footage arrives with a different one.

How a colorist actually uses it

The order is the same as any grade, with emulation late.

1. Normalise. Apply the correct input transform for the format, so the image is photographic before any look. A washed-out starting point is usually this step missing — see why S-Log3 footage looks washed out.
2. Correct. Exposure and white balance per shot until the scene agrees with itself.
3. Match shots. Emulation on shots that do not agree makes the disagreement more visible, not less.
4. Grade. Build the look you want with curves and palette.
5. Emulate, at reduced strength. Most stock emulations look right at well under full opacity. Start low.
6. Grain last, at delivery resolution.

Check on scopes, not by eye alone: the waveform monitor shows whether the emulated toe lifted the black point or crushed it, and the vectorscope shows whether the dye response moved faces off the skin line.

Three misconceptions

"A film emulation LUT is a film emulation." A LUT is a color-to-color mapping. It can carry the tonal curve and the dye response; it cannot carry halation or grain, both of which are spatial. Any product that is only a LUT is doing half the job by construction.

"Pick the stock the film was shot on and you get the look." The negative stock is one of at least four variables, alongside the print stock, the lab's timing and the lighting. Named-stock presets are a starting point, not a reconstruction.

"It will make small-format footage look like film." Emulation adds a response, not information. On heavily compressed 8-bit footage the added curve often exposes banding that was invisible before, and the grain stage then has to survive delivery encoding.

Where Leumos fits

Leumos is our product, so treat this as disclosure rather than a recommendation. It is a browser color workspace: import footage, it detects cuts, you select one shot and match it to a still reference image, review the result in motion, fine-tune with creative controls, and export the graded timeline in Rec.709 (leumos.ai/llms.txt, read 13 September 2026).

Because this article is about film looks, here is what it covers and what it does not. The Studio has a halation stage and a film-texture stage (grain, acutance, resolution and grain chroma) that sit after the color match, so two of the four components above are available as controls. What it does not have: named film-stock presets built from manufacturer sensitometry, and any LUT or grade-data export — the output is rendered video. Matching runs on one selected shot at a time and is reviewed before you move on. For a specific stock's published curve you want a dedicated emulation tool — the landscape is surveyed in best FilmConvert alternatives. Plans: pricing.

Frequently asked questions

What is film emulation in color grading?

Film emulation reproduces on digital footage the measurable behaviour of a film chain: the negative's density-versus-log-exposure curve, the steeper response of the print stock, halation inside the emulsion, and granular structure. It is four separate effects, not one filter.

Is film emulation just a LUT?

No. A LUT maps one color to another and can carry the tonal curve and the dye response, but halation and grain are spatial — they depend on neighbouring pixels — so no lookup table can produce them. A complete emulation needs additional stages.

Why does film emulation look wrong on my footage?

Most often one of three: no input transform, so the curve is applied to an image that is not scene-referred; the footage was clipped, so there is nothing for the shoulder to roll off; or the emulation went on before the shots were corrected and matched.

Should grain go on before or after the grade?

After, at delivery resolution. Grading on top of grain rescales and distorts it, and grain added before a resize is resampled into something that no longer looks like granularity.

Does film emulation need the right stock name?

Less than people expect. The print stock, the lighting and the lab's timing decisions contributed as much as the camera negative, so a named preset is a starting point rather than a reconstruction of a particular film's look.

Can film emulation fix flat log footage?

No. Flat log footage needs a correct input transform and a balance pass first; emulation is a later stage that shapes an already photographic image. The order is in how to color grade video.

Sources

Related reading

• Color timing in film — the lab decisions a stock emulation cannot carry
• Cinematic color grading — where texture sits in the order of operations
• What is log footage — why the shoulder needs something to roll off
• Why CST, ACES and LUTs look different — applying a response in the wrong space
• Banding after grading and export — what grain and delivery encoding do to each other