Print film emulation — construction

ExpertDuration ~12 min video + build sessionTools DaVinci Resolve (free)

Why this lesson

In 2.10 you learned to apply a print-film-emulation LUT with discipline — normalize first, place it late, add halation and grain separately. This lesson is the level up: build the chain instead of applying a sealed one, so you understand what a print emulation actually models and can adjust any part of it. The single deep free artifact on this — Juan Melara rebuilding the Kodak 2383 node by node — is your text. Around it, the explainer gives you the two concepts a LUT hides: status-M density (how film response is measured) and printer lights (how a print is balanced). Own those and a “film look” stops being a download and becomes something you can construct and reason about.

Watch

Watch for: The order of the rebuild is the lesson. First the main curve: he desaturates to see it as pure grayscale, switches to a smooth-gradient image, grabs a reference still, and matches an editable spline to it — contrast first, color later. Then the tint: in Lattice he shows the LUT's R, G and B curves converging at the midpoint (no tint in the mid, blue building below, warmth above), and rebuilds only the deviation from the main curve on a half-gain node for precision. Then hue-vs-sat and hue-vs-hue matching to a chart, a white-point node with luma-mix at zero, and finally how he uses it: a CST from log into Rec.709 (the correct normalize), a balance node before the look, and — the whole argument — swapping the 'too contrasty' curve out because it's just a node.

The explainer

Film is a density system, and that’s the key that unlocks everything. When film is exposed and developed, the image is stored as density — how much dye sits in each layer, blocking light. Measure that density and you get film’s fingerprint. The standard measurement is status-M density: a densitometer reads the red, green and blue dye layers of a negative through defined filters, producing three characteristic curves (density vs log exposure). Those curves are what a print emulation is really reproducing — not “a look,” but the measured response of a specific stock. It’s why Melara can rebuild 2383 so faithfully: he’s matching curves that were originally read off film, which is also the physical construction Daniele describes in the FilmLight masterclass (densitometer readings → a print-LUT).

The pipeline: negative, then print. A print-emulation pipeline models the two-stage photochemical system from 2.10 as an ordered chain. Stage one is the negative response — the camera stock’s density curves, which give the toe-and-shoulder tonality. Stage two is the print — the negative projected through printer lights onto print stock like 2383, which adds the punchy contrast and the characteristic color cross-over. Built as nodes, that’s: a main tone curve (the combined density response, matched in grayscale), then a tint stage (the per-channel color deviation), then hue/saturation shaping, then a white-point step. Each is separable — the entire point.

Reading the tint the way Melara does. Load the 2383 into a LUT viewer and look at its R, G and B curves. They converge near the midpoint and diverge away from it: below mid-grey the blue rides highest (a cool shadow cast), above mid-grey red and green climb (warm highlights). Crucially, since your main curve already carries the contrast, the tint curves don’t need to repeat that S-shape — you only model their deviation from the main curve, which is why he drops a node to 50% gain first (halving sensitivity for precision) and draws only red and green offsets. That “model the deviation, not the whole thing” move is the craft.

Printer lights, done properly. In a real film lab, the print is balanced with printer lights: three lamps — red, green, blue — whose intensity is set in discrete steps called printer points (a point is a small, standardized change in exposure of that channel, classically about 0.025 in log density, with a full range around 0–50). The colorist “times” a shot by calling values like “25-25-25” for neutral and nudging, say, the red light up to warm it. This is the original color-balance interface, and it survives in modern grading: the printer-lights control in Resolve, the per-channel offsets in your 3.4 DCTL, and the language DPs still use (“give me a couple of points of green out”). The important technical point: printer lights act in log density, not linear light — a point is an additive shift on a log curve, which is why they feel even across the whole range instead of blowing out the highlights. Your linear-gain DCTL from 3.4 approximates them; a true printer-light control does it in log.

Why construction beats the LUT. Everything above is invisible inside a baked 2383. Melara’s rebuild exposes it, and the payoff is exactly the complaint everyone has about film LUTs: “too contrasty.” Because the contrast is its own node, he swaps it for a gentler LogC→Rec.709 transform and keeps all the color. He can retune the tint for a client, drop the two luma-shaping nodes when they stress the image, or dial the whole look’s strength. A constructed print emulation is a film emulation you can negotiate with; a LUT is one you can only accept or reject.

Do it

1. Take a normalized, exposed log shot. Load Resolve’s Kodak 2383 LUT and, in a viewer like Lattice (or Resolve’s curve on a desaturated grab), study its R/G/B curves — find the midpoint convergence, the cool shadows, the warm highlights.
2. Rebuild the main curve first, desaturated: match an editable spline to the 2383’s grayscale contrast on a smooth-gradient reference.
3. Add a tint node at reduced gain and draw only the deviation — blue up in the shadows, red/green up in the highlights — matching the LUT’s cross-over.
4. Add a printer-lights node (Primaries → Log, or the Printer Lights control) and rebalance in points; feel how even the moves are because they’re log-density, not linear.
5. Prove the payoff: swap the main curve for a plain LogC→Rec.709 CST and note you’ve tamed the contrast while keeping the color — the thing a LUT can never let you do. Add halation and grain (from 2.10) on top.

Level 3 workbook — every Do it exercise, 3.1–3.10, plus the capstone (printable)level-3-workbook.pdf764 KBOriginal course material — free to use

Terms introduced

Terms introduced in this lesson

• Status-M density
• Print-emulation pipeline
• Printer lights

Each term links to the glossary, where it is defined by the way you just used it.

Check

Check yourself

1. When Melara rebuilds the 2383 LUT, why does he desaturate before matching the main curve?

   Because 2383 has no colorTo make the file smallerTo speed up renderingTo see the tone response as a pure grayscale curve, stripping away the color complexity so he can match contrast first, then handle tint separately

2. In the RGB tint curves of a print emulation, what does it mean that the three curves converge near the midpoint?

   The LUT only affects highlightsThere is little to no color tint in the mid-grey — the tint builds away from the midpoint: a cooler cast in the shadows, warmth in the highlightsThe image is brokenSaturation is at zero everywhere

3. What are printer lights?

   The red, green and blue timing lamps in a film printer, set in discrete "points," that expose the print stock — the original per-channel balance control, worked in log densityA type of LED panel for on-set lightingA Resolve scopeThe lamps on a colorist’s desk

4. Why build a print emulation as a node chain instead of using the baked 2383 LUT?

   The LUT is illegal to useThere is no differenceA chain renders fasterA chain gives you control over every element — you can swap the "too contrasty" curve, adjust the tint, or dial strength — none of which a sealed LUT allows

Check answers

You can move on when you can… explain a print emulation as negative-density → print → tint → white-point stages, describe what status-M density measures, define printer lights and why they work in log, and rebuild a print LUT’s main curve and tint as separable nodes you can adjust.

Mark this lesson complete

Go deeper

Cullen Kelly — “Look User,” auditioning print LUTs (from 2.10): the discipline half — why a print emulation must serve every shot, and why a sealed LUT leaves you nowhere to go, which is precisely the limitation your constructed chain removes.

Segment: 23:46–31:56 — auditioning 2383 / Fuji PowerGrade LUTs and their limitswatch full video

Watch for: The 2383-versus-Fuji audition, the LUT that breaks on the second shot, and the case for owning a look you can edit — the argument for building the chain you just built.

Read

• Steve Yedlin — Display Prep Demo & On Color Science (yedlin.net): the most rigorous public argument that a well-built display/print rendering — not a stack of stock LUTs — is what actually makes digital look like film. Self-hosted, so it can’t be embedded; open it directly. Watch the Display Prep Demo, then read the notes. This is the deepest free thinking on the subject that exists.
• Chris Brejon — CG Cinematography, ch.1 (chrisbrejon.com): the density-and-film-response section, from the color-management side.

The paid path: structured print-film-emulation construction (not application) is taught in Lowepost’s PFE training and threaded through colour.training. Juan Melara’s free rebuild is the one deep public artifact; those courses are where the systematic version lives. When you’re building film pipelines for clients, they’re the upgrade.

Next up: 3.6 · HDR grading — grading in a much bigger box, where the tone and gamut mapping of 3.2 meet nits, PQ and Rec.2100.