Kodak Double-X 5222

Original (datasheet + characteristic validated). See the notes, the cinema README, and the Knowledge base.

The stock

Eastman Double-X Negative Film 5222/7222, in continuous production since 1959, making it one of the longest-lived film stocks in history. Kodak's datasheet (archived: kodak-motion/double-x_5222.pdf, pub. H-1-5222) describes a "high-speed, panchromatic material" with "good image-structure characteristics and excellent sharpness," suited to both studio and location work, including dim-light photography without added illumination.

Why it's on the "most notable and pleasing" list

This isn't a guess. Researched directly. Double-X's filmography is one of the strongest credibility signals of any stock in this bank: Raging Bull, Schindler's List, Manhattan, and it remains the reference point critics reach for whenever describing "the moody intensity of Old Hollywood and film noir." It is the stock people mean when they say a modern film "looks like classic black-and-white cinema."

Datasheet specifics used in the derivation

• Diffuse RMS granularity: 14: moderately heavy; heavier than any fine-grain still stock in this bank, in the same territory as the bank's dedicated push stocks (Delta 3200, T-Max P3200), though Double-X isn't itself a push stock. This is its native grain.
• Control gamma 0.65–0.70 (developed to this contrast target, measured via Status M blue densitometry), a moderate, versatile contrast, not an extreme one.
• Resolving power 32 lines/mm (1.6:1) to 100 lines/mm (1000:1): lower than still-photo B&W stocks, appropriate for cinema (motion, projection, and telecine don't demand still-print-level per-frame resolution).
• EI 250 daylight / 200 tungsten: a genuinely fast negative, usable in dim conditions.

The grade in one line

Classic cinema black & white. Moderate contrast with rich, deep blacks, moderately heavy grain, built for both studio control and available-light location work.

How the settings embody it

Acros carries real tonal richness under the grain. Grain —/— matches the RMS—eality. Heavier than fine-grain still stocks. — + — + Shadow —model the moderate-gamma, rich-shadow curve. Clarity —(a —, deliberate deviation from the 0 default) adds the tactile grit associated with its on-screen reputation, more restrained than Tri-X's +3 since Double-X's contrast target is genuinely gentler than Tri-X pushed.

A stylistic variant: period-drama warmth

The datasheet gives no colour cast (B&W has none), but films styled after 1940s–60s cinematography sometimes want a subtle warm toning rather than neutral grey. Try Monochromatic Color WC —/MG 0 for a gentle sepia-adjacent period lean, keeping everything else the same.

vs the bank's still-photography B&W family

Double-X lives in cinema/, not black-and-white/, deliberately. It's a different creature from Acros/HP5/Tri-X/Delta/Pan F even though all are monochrome. Those are still-photography stocks with still-photography grain/resolution targets; Double-X is built for projection, telecine, and motion, and its RMS-14 grain plus moderate gamma give it a specific "classic cinema," not "photojournalism," character.

Validation status

Datasheet-validated (RMS, gamma, resolving power, speed, archived PDF) + characteristic-validated against specific named productions (a stronger evidence class than generic reviews, since it's independently verifiable). No Commons frame-grabs exist (real film footage isn't Commons material). Scan-validation would need a user-supplied reference frame, e.g. a still from one of the cited films or your own Double-X test roll.

Original (research-backed). See the notes, the notes, the cinema README, and validation-methodology.

Status: datasheet-measured. Kodak's sensitometric curves have been traced and the published gamma independently reproduced. No real-footage reference frame, see Why a scan is structurally blocked.

---

This datasheet is the bank's calibration standard

Source: X-T5/_reference-sources/datasheets/kodak-motion/double-x_5222.pdf, EASTMAN DOUBLE-X Negative Film 5222 / 7222, Kodak Publication H-1-5222, revised 3-22. Archived in this repo.

Most datasheets plot a curve and leave you to measure it. This one plots five curves and prints the gamma of each, because the chart's purpose is to show how development time changes contrast:

12 min γ=1.05 · 9 min γ=0.84 · 6½ min γ=0.66 · 5 min γ=0.56 · 4 min γ=0.50 (Exposure: Daylight, 1/50 s intensity-scale sensitometer · Processing: KODAK Developer D-96 at 21 °C · Densitometry: Status M (Blue))

That makes it ground truth. Every other curve measurement in this bank is unverifiable against a published number. This one is verifiable against five. It was traced blind to those labels and then compared.

The result

Curve	Kodak's printed γ	Measured	Error
12 min	1.05	1.063	+0.013
9 min	0.84	0.823	−0.017
6½ min (recommended)	0.66	0.678	+0.018
5 min	0.56	0.578	+0.018
4 min	0.50	0.517	+0.017

Mean absolute error 0.016 · maximum 0.018 · correlation r = 0.9980 (n = 5)

⚠Corrected 2026-07-27: this was first published as r = 0.99908, which does not reproduce from the five pairs above. Recomputing gives 0.9980. More importantly, r is the wrong statistic here: it tests whether the points fall on a line, not the identity line, and stays ~0.99 even under a systematic scale error. The regression of measured on published gives slope 0.972 (95 % CI 0.86–1.09), intercept +0.030 (−0.055 to +0.114). Consistent with ideal, but n = 5 cannot exclude a ±12 % scale error. Mean absolute error carries the claim, and it is unchanged. See the notes §6.

The tracer reproduces published gamma to about 2 %, five times over. That is the strongest methodological result in this repo, and it retroactively supports the 2383 measurement, which had only an indirect system-gamma cross-check.

The fit window is the dominant error term: and it is now calibrated

Gamma is the slope of the "straight-line" portion, but the straight portion has no exact boundaries, so the answer depends on how wide a window you fit. Tested against the five known values:

Fit window (log E)	Mean abs. error	Failure mode
0.5	0.022	Biases high — taking a maximum over many noisy slope estimates is upward-biased
0.7	0.019
0.8	0.016 ✅	Optimum
1.0	0.019
1.2	0.033	Biases low — window outruns the straight section, pulling in toe and shoulder
1.5	0.061	Severe; the γ=1.05 curve collapses to 0.80

Both failure modes are real and directional, and they explain the residual pattern exactly: a fixed window is too wide for high-gamma curves (whose straight section is short) and too narrow to average out noise on low-gamma ones. 0.8 log E is the bank's standard fit window, but see the 2026-07-27 self-audit: the 0.003 MAE gap between 0.8 and its 0.7/1.0 neighbours is about one-sixth of a single point's error, so 0.7–1.0 are statistically indistinguishable and only ≥1.2 is demonstrably worse. 0.8 is retained as a convention for reproducibility, not a demonstrated optimum.

Calibration and validation of the trace itself

Render	300 dpi
Log-exposure axis	170.0 px per decade (from tick detection)
Density axis	169.75 px per density unit
Agreement between axes	0.15 %
Columns with all five curves resolved	323, spanning x 308–746
Masked before tracing	the inset gamma vs development time plot, the exposure/processing text block, and the per-curve labels

---

The curve at the recommended development (6½ min, γ 0.66)

This is the one that matters. Kodak's own processing recommendation, and the gamma the recipe is built on.

Quantity	Measured
D-min (base + fog)	0.24
Gamma	0.678
Toe: slope reaches 90 % of γ by	log E −2.69
Straight-line span	log E −2.69 → −0.41
Straight-line latitude	≈ 7.6 stops
Density at the right edge of the plotted range	1.66

Two findings, and the second corrects this recipe's own notes

1. The stock is enormously forgiving, 7.6 stops of straight-line response.

Set against the bank's other measured stocks, the contrast between a camera negative and a print stock is now quantified end to end:

	Gamma	Usable scale
Double-X 5222 (camera negative)	0.68	≈ 7.6 stops, straight
Kodak 2383 (print stock)	3.44	≈ 4.4 stops, hard

That is the whole photochemical chain in two rows. Capture wide and flat, then throw most of it away through a short steep transfer. It also puts Double-X in the same latitude class as FP4 Plus, whose datasheet claims eight stops in words; here the same order of latitude is measured.

2. ⚠The toe is short, so the emulsion does NOT crush shadows, which means Shadow — is a look, not fidelity.

The slope reaches 90 % of gamma by log E −2.69, essentially at the left edge of the plotted range. A short toe means shadow separation is good and shadows render linearly, across the full 7.6 stops.

This recipe's notes say Shadow —gives "the deep, rich blacks associated with its classic war-drama/noir cinematography." That reasoning is sound but the attribution needs care: those blacks come from how those films were exposed, printed and graded, not from the negative's own curve, which is linear and open in the shadows. Exactly the same stock-versus-grade distinction established for 2383.

No setting is changed on this. The recipe openly targets the cinematography rather than the raw negative, which is a legitimate and clearly-stated design choice, and it is what makes the recipe look like the films people actually remember. What changes is that the notes no longer imply the emulsion crushes its own shadows. If you want the negative's real shadow behaviour rather than the noir grade, drop to — or —, that is now a documented, evidence-backed variant rather than a guess.

⚠One limit stated plainly: the chart's plotted range ends at log E −0.41 (D 1.66) and no shoulder appears within it. That does not mean Double-X has no shoulder, every emulsion eventually shoulders. It means Kodak did not chart one, so nothing is claimed about highlight roll-— beyond the plotted range, and the 7.6 stops is straight-line latitude, not total latitude.

---

What the datasheet says in words

Kept separate from the measurement, because these are Kodak's statements rather than numbers derived here:

• Exposure indexes (for development to gamma 0.65): Tungsten (3200 K) 200 · Daylight 250. Note the index is quoted against a gamma. Speed and contrast are not independent on this stock.
• Control gamma: "Develop to recommended control gamma of 0.65 to 0.70 calculated using Status M densitometry (blue)."
• Diffuse rms granularity 14: the basis for Grain —/—.
• Resolving power: 32 lines/mm at TOC 1.6:1, 100 lines/mm at TOC 1000:1. (The X-T5's Nyquist limit is 164.4 lp/mm, so the sensor out-resolves this film by 1.6–5.1×, see sensor-architecture §7.)
• Image structure figures were generated in D-96 at 21 °C to the recommended control gamma: so grain, MTF and resolving power all assume the 6½-minute curve, the same one measured above. The datasheet is internally consistent, and this recipe is built on the same point.
• In continuous production since 1959.

---

Why a scan is structurally blocked

Double-X is a motion-picture negative. A faithful reference would be a frame from a film shot on it, and real theatrical footage is not Commons material, while frame-grabs of released films carry the film's own DI/print grade, which is the exact variable this recipe needs to hold constant. The achievable evidence ceiling is therefore datasheet + measurement + characteristic, which this recipe now meets in full.

A user-supplied frame, your own B&W test footage, or a scan of a Double-X negative. Dropped into references/ is the only route further.

---

What would change this recipe

1. A real reference frame showing the settings miss in a specific, repeatable direction.
2. A decision on the noir-grade question above. Worth testing Shadow — against Shadow — on one real scene, since the measurement shows +2 is a stylistic addition rather than an emulsion property.

No setting is changed on the strength of this measurement. It confirms the gamma the recipe was already built on, and reclassifies one setting's justification without altering its value.

---

Measured 2026-07-26 from the archived datasheet. Axis calibration, the five-point gamma validation and the 0.8 log E fit-window calibration are all reproducible from the chart on p. 3 of the source PDF.

Last updated: 2026-07-26