The standard every recipe must meet to be promoted to originals/ and sold. The goal: faithful reproduction, proven against real film, not a single pretty example, and not vibes.
⭐ CORE PRINCIPLE: always study BOTH datasheets and real images
For every film look or simulation you MUST, without exception: 1. Read the manufacturer datasheet. Characteristic curve (contrast/gamma), RMS granularity / grain index, spectral sensitivity, exposure latitude, balance. The objective physics. 2. Fetch and analyse real reference images, ≥5 across lighting/subjects. A datasheet tells you the physics; only real scans tell you how the film actually renders once shot, developed, and scanned.
Never build a recipe from memory, or from only one of the two. And keep finding new strategies to improve authenticity. New image sources (Reddit, Flickr, Commons, lab galleries, user drops), new datasheets, side-by-side comparisons. This is a continuous discipline, not a one-time check.
Image-fetching strategies (in rough order of reliability):
1. User-provided reference files (downloaded PDFs / scans placed in X-T5/_reference-sources/). Render with pdftoppm and view every page/image. Best when available.
2. Direct image download, Invoke-WebRequest/curl a real scan JPG to disk, then Read it. Works on open hosts (Wikimedia Commons upload.wikimedia.org, Flickr static, blog media dirs). Bypasses the browser.
3. Manufacturer / lab galleries. Datasheets (fetch PDF → pdftotext/pdftoppm) for curves + grain, plus lab film-index galleries for real scans.
4. Search → collect many, always gather ≥5 across lighting/subjects (see scan rules) so you are not fitting to an outlier. Before searching Commons, read COMMONS-SCAN-INDEX.md. It records which stocks have genuine scans, which have only box shots, and the mislabelling traps already caught.
5. When a source is bot-protected (Cloudflare), don't fight it. Switch hosts, or ask the user to download the page/scans (as was done for Gold 200).
6. Video references (YouTube): the player blocks browser screenshots in this environment, but YouTube serves real auto-generated frame stills as plain JPGs, https://i.ytimg.com/vi/<VIDEO_ID>/sd1.jpg, sd2.jpg, sd3.jpg (~25/50/75% points) and maxresdefault.jpg (hero frame). Four genuine frames across different scenes, fetchable with a plain browser-UA request. Caveat: heavy compression → trust hue relationships, treat absolute saturation loosely. (Proven on the Silent Atlas verification, 2026-07-16.)
Record which images were viewed (source, light, subject) in the recipe's the notes. Images are analysis-only, never redistributed in the product.
Blind evaluation: borrowed from Fujifilm's own process (added 2026-07-20)
Fujifilm's Image Design team describes how they validate a simulation during development: photographers shoot diverse scenes with deliberately unconventional settings, prints are then evaluated by neutral parties who were not involved in the shoot, and the cycle repeats "for every test." (X Stories, ep. 1)
The reason that matters to us: the person who built a recipe is the worst judge of it. Knowing the intent makes you see the intent. This is the same failure mode as fitting to a single reference image, one step further along.
Adopt for the bank:
- When judging test shots, look at the frames before re-reading the settings. Grade the image, not the intention.
- Evaluate a recipe against reference scans mixed in, unlabelled, rather than side-by-side with labels.
- Prefer a second opinion from someone who doesn't know the target for any recipe being promoted to originals/.
- Fujifilm evaluates on prints; screens flatter. Worth a print check before declaring a flagship recipe finished.
The three evidence tiers
A recipe's validation status is one of:
- Datasheet-validated. Grounded in the manufacturer's technical data (balance, ISO/EI, contrast, grain/RMS or grain index, latitude, spectral/skin notes). Establishes the objective baseline.
- Characteristic-validated. Corroborated by ≥3 independent, reputable reviews describing the look (cast, contrast, saturation, grain, per-hue behaviour). Establishes the consensus look.
- Scan-validated (pixel). Checked against multiple real film scans (see rules below). Establishes the actual rendering. Required for the top "validated original" tier.
An original ships with its tier stated. Gold 200 is the reference example (all three tiers → fully scan-validated).
Scan-validation rules (to avoid building on outliers)
- ≥5 scans per film, from different photographers/sources where possible.
- Spread across lighting: direct sun, flat/overcast, open shade, golden hour, and artificial/night where the film is used there.
- Spread across subjects: at least one each of skin/portrait, foliage/greens, sky/blues, neutral whites/greys, and a saturated accent (red/yellow).
- Note the pipeline: lab scan vs home scan, and scanner, shift results. Flag when known, since scanning changes colour as much as the film.
- Read per-channel behaviour explicitly: where do whites sit (warm/cool), how are greens shifted, how saturated are blues, how do highlights roll and shadows hold.
- Reject outliers: if one scan disagrees with the other four, weight the consensus, do not chase the odd one.
- ⚠B&W carries two extra confounders: DEVELOPER and FORMAT. (added 2026-07-21) A B&W scan description that proudly states its full pipeline is telling you the frame is not neutral evidence: Rodinal/Blazinal accentuates grain and acutance, D—/ID—smooths it, and stand development compresses highlights dramatically. Differences larger than some film-to-film differences. Separately, format dominates apparent grain: a 4×5 negative shows almost none regardless of stock, so a large-format frame cannot validate a 35mm grain setting. Consequence for this bank: our B&W recipes emulate a film and a development choice, not a film alone: state which combination each recipe targets.
- ⚠Verifying the stock is only half the job. Check what else is driving the image. A frame can be provably shot on the right film and still be useless evidence. Before counting any scan, check the lens (toy/plastic optics impose vignetting, softness and flare that read as film character), the light source (uncorrected sodium/mercury/mixed lighting swamps the emulsion's own colour), the process (cross-processing, heavy expiry, push/pull), and the scan pipeline. A frame failing any of these should be logged as seen and rejected, not silently skipped. Otherwise the next pass re-finds it. Worked example: the Portra 800 rejection.
From evidence → settings → proof
Each original records, in its the notes (or the notes):
1. The datasheet facts used.
2. The scans analysed (source, light, subject). Analysis only, never redistributed; source files kept locally in X-T5/_reference-sources/.
3. A trait → setting table (each observed trait mapped to the specific control producing it).
4. What changed because of the evidence (e.g. Gold 200: WB —→— and Clarity —→0 after scan analysis).
5. A faithfulness verdict and what's still open (missing conditions, pipeline caveats).
Quality rules that always apply
- — is the default, not a hard rule. Keep it at 0 whenever the look allows (avoids halos/artifacts, faster saves). But faithfulness wins: if the real film's rendering genuinely needs a non-zero Clarity (e.g. the gritty micro-contrast of Tri-X, or a soft dreamy stock), change it, and justify why in the validation. Quality and faithfulness above the preference.
- Faithful reproduction beats stylistic flourish. A recipe is judged by how close it lands, not how "nice" it looks in isolation.
- Prefer fixed WB for colour-critical films (repeatable); use Auto only when the look tolerates scene-to-scene drift.
Promotion: reference → original
A reference recipe (attributed capture) becomes a validated original only after it is re-derived from evidence and passes at least datasheet + characteristic validation (scan-validation for the top tier). Copying values is not derivation; the analysis and write-up must be our own. This is both a quality bar and the clean-IP basis for selling the bank.
Evidence state of the bank: at a glance (consolidated 2026-07-21)
| Evidence level | Count | Recipes |
|---|---|---|
| Scan-validated, multi-frame (3+ frames or full pass) | 5 | Gold 200 (~35 frames), Acros 100 (4), HP5 Plus (3), Vision3 50D (3), Tri-X 400 (2, two extreme conditions) |
| Scan-validated, 2 frames | 4 | Vision3 500T (both key balances), Vision3 250D, Delta 3200 (two photographers), Ektar 100 |
| Scan (partial — 1 frame, or 2 from one photographer) | 6 | Portra 400, Ektachrome E100, Superia 400, Pro 400H, UltraMax 400 (now a full original), Kodachrome 25 |
| Datasheet — curve measured | 2 | 2383 Print · Double-X 5222 — tone scale traced from Kodak's published charts. Double-X is the calibration standard: its datasheet prints five gammas and the tracer reproduced all five to ~2 %, which is what licenses the method everywhere else |
| Datasheet + characteristic, no scan | 7 | Ektachrome 100D, Pan F Plus 50, Delta 100, Delta 400, XP2 Super 400, Kodachrome 64, FP4 Plus 125 |
| Datasheet-validated, partial-fidelity ⚠️ | 2 | SFX 200, Ortho Plus 80 — spectral opposites; see the tier definition below |
| Characteristic only ⚠️ | 2 | Fortia SP · ColorPlus 200 — the bank's weakest tier, both flagged in-recipe. ColorPlus is differentially derived: no Kodak datasheet exists, so it changes only the three axes on which independent reviewers agree and inherits the rest from scan-validated Gold 200 |
| Original design / place study | 9 | Mumbai (11-image study), Mumbai Monsoon, creative looks, + the three AP recipes (field-test pending) + Clay Season · Sun Bleached · Cold Open (2026 trend-led video looks, stills + movie mode, field-test pending) |
The "partial-fidelity" tier (added 2026-07-25)
A recipe is datasheet-validated, partial-fidelity when its settings are fully datasheet-grounded but a specific documented behaviour of the film is provably unreachable on the X-T5, for a reason that is physical rather than a tuning shortfall. It is not a synonym for "not very accurate" and it is not a hedge. Using it requires naming the exact effect, the exact mechanism, and what would achieve it.
The first and so far only case is SFX 200: Ilford's sheet claims a red filter renders skies near-black and green foliage near-white. The first is reproduced by ACROS+R; the second is inverted, because it depends on the vegetation red edge above ~700 nm and every unconverted digital body has an IR-cut filter over the sensor that removes that band before any processing occurs. The route to it is an infrared conversion or the actual film, not a different Grain, Shadow or WB value.
Use this tier sparingly. Most shortfalls are tuning problems and should be fixed, not labelled. Reach for it only when the missing information was never captured.
Nine of the film-based originals now carry real pixel evidence, up from three at the start of the July validation push. Every cinema recipe except 2383, Double-X and 100D is scan-backed.
Which remaining gaps are actually closable
Honest triage, so effort goes where it can succeed:
| Gap | Closable? | Why |
|---|---|---|
| More daylight frames for partials | ✅ Yes, easily | Commons is rich in clean daylight film photography |
| 2383 Print / Double-X / Ektachrome 100D | ❌ Structurally blocked | Motion-picture stocks: real footage isn't Commons material, and frame-grabs of released films are neither free nor faithful (they carry the DI grade) |
| Pan F Plus 50, T-Max | ❌ Blocked on Commons | Only box shots and software simulations. Needs Flickr or a user drop |
| Fortia SP | ❌ Effectively closed | Japan-only limited run, 2004–07. No datasheet, no scans anywhere open |
| Night / tungsten / interior colour | ⚠️ Blocked by contamination, not scarcity | Frames are abundant but the light source dominates them — see scan index. B&W is exempt |
| Colour skin tones / portraits | ⚠️ Only via user drops | The bank's largest gap. Commons portrait searching also carries a real risk of surfacing inappropriate imagery of identifiable people — a earlier attempt did, and was discarded |
The two highest-value things a human can contribute, in order: a colour portrait scan for Pro 400H or Portra 400, and any frame shot with a grey card in shot under tungsten or mixed light.
Current validation ledger
2026-07-15 datasheet audit: all originals below re-verified against locally archived manufacturer PDFs (
X-T5/_reference-sources/datasheets/), see X-T5/VALIDATION-AUDIT.md. The datasheet tier for every film-based original is now archive-backed rather than fetch-based.
| Tier | Notes |
|---|---|
| Datasheet + Characteristic + Scan (full) | ~35 scans, all light conditions; WB & Clarity refined from evidence |
| Datasheet (E-4050 archived: PGI 37@4×6, latitude, skin aims) + Characteristic (strong) + Scan (partial) | 1 real scan 2026-07-20 (sunny daylight) — restrained saturation, open shadows, fine grain confirmed |
| Datasheet (E-4000 archived: RMS 8, curves, low-contrast tone scale) + Scan (partial) | 2 confirmed scans 2026-07-20 (blue hour + mixed light, same photographer) — subject-following saturation and dual-end retention confirmed; daylight/skin still needed |
| Datasheet + Characteristic + Scan (partial) | 2 scans, one photographer — cool palette/tone confirmed; grain level provisional |
| Datasheet (RMS 7) + Characteristic + Scan (4 frames) | 2 more frames 2026-07-21 (flat hazy daylight + low raking sun) — official curve visibly confirmed: crisp unclipped highlights, full detail inside deep shadows, invisible grain. One photographer/location; portrait + high-ISO still wanted |
| Datasheet + Characteristic + Scan (3 frames) | 2026-07-21: 1989 Prague reportage frame, HP5/ID-11/EI 400 — neutral development, so clean evidence. Best skin-tone evidence in the bank; grain visibly smoother than Tri-X. 3rd frame = night street (same pipeline): blooming lamps + textured black coats in one exposure. Latitude is the defining quality |
| Datasheet (F-4017) + Characteristic + Scan (2 frames) | 2026-07-21: Leica reportage frames confirm Grain —/—, deep blacks, the tactile "bite" justifying Clarity —, and wide latitude. Refinement: the grit shows on normally-processed Tri-X, without pushing — recipe is more broadly faithful than its notes claimed |
| Datasheet + Characteristic + Scan (2 real-stock frames) | 2026-07-21: daylight vs artificial-light pair proves the "500T look" is the stock used out of balance — neutral/accurate under tungsten, teal-orange only in uncorrected daylight. First skin-tone evidence for any cinema recipe. Supersedes the earlier CineStill proxy |
| Datasheet (H-1-5213: "image structure of a 100-speed film" claim, EI 200 tungsten / 125 daylight w/ 85 filter, full 7-source exposure-index table, ±200 K colour-balance tolerance, reciprocity flat from 1/1000 s to 1 s, the "two stops more highlight information" latitude claim — archived PDF) + Characteristic | completes the Vision3 line (50D · 200T · 250D · 500T); derived 2026-07-25; no scan. The only stock in the family sold on a speed/grain mismatch. Granularity is a curve, not a number, so no RMS is claimed and the "100-speed structure" claim is treated as prose — it lands on — (a notch above its speed peers) rather than on Grain, which is already at the floor. Flagged as interpretation, not direct derivation |
| Datasheet + Characteristic + Scan (2 frames) | 2026-07-21: a full snowfield in flat overcast held with no clipping — decisive proof of — + Highlight —. Neutral, restrained, open shadows |
| Datasheet — curve traced and measured (H-1-2383, archived) + Characteristic | 2026-07-26: the only recipe in the bank whose tone scale is measured rather than described. Kodak's sensitometric chart traced at 300 dpi; ordering validated 93/93, and cross-checked from an independent direction — negative γ 0.55–0.60 × measured print γ 3.44 = system γ 1.89–2.07, bracketing the classic ~1.8–2.0 dark-surround target. γ ≈ 3.44 · D-max 4.09 · usable scale ≈ 4.4 stops (vs ~13 for the negative) — the short hard window is why —, not —. ⚠️ A per-channel claim made here on 2026-07-26 (red ≈15 % steeper → teal-orange intrinsic) has been RETRACTED — the channel ordering flips under a different estimator, so it was an artefact (why). D-max, D-min, the ~4.4-stop scale and the order-of-magnitude gamma all survive, since none depends on inter-channel structure. Note also that γ ≈3.5 sits 3–4× outside the method's calibrated range, so the 2383 gamma is quoted with its estimator named. Corrected the previous ledger wording: LAD was never extracted (the datasheet gives no LAD numbers, only a pointer to Pub. H-61). See validation.md. Scan structurally blocked |
| Datasheet (finest-grain claim) + Characteristic + Scan (3 frames) | 3rd frame 2026-07-21 (Vienna, Nikon F100 + 50/1.4G — best glass in the reference set): grain very nearly absent, very low contrast, subject-following saturation |
| Datasheet — curve traced and measured (H-1-5222, archived) + Characteristic (Raging Bull, Schindler's List, Manhattan) | 2026-07-26: the bank's calibration standard. Its datasheet plots five curves and prints the gamma of each (1.05/0.84/0.66/0.56/0.50); traced blind and reproduced to mean abs. error 0.016, max 0.018, r = 0.9980 (n = 5) — published gamma matched to ~2 %, five times over. (r corrected 2026-07-27 from a non-reproducing 0.99908; MAE carries the claim.) Also calibrated the method's dominant error term: 0.8 log E is the bank's standard fit window — a convention, not a demonstrated optimum, since 0.7–1.0 are statistically indistinguishable at n = 5 (see film-curve-measurement.md §6). At the recommended development γ 0.678 · D-min 0.24 · ≈7.6 stops of straight-line latitude, and the toe is short — so the emulsion does not crush shadows and Shadow — is a noir-grade choice, not fidelity (setting unchanged; variant documented). See validation.md. Scan structurally blocked |
| Datasheet (saturation/sharpness language, archived) + Characteristic (Poor Things, 2023) | no Commons frame-grabs exist; derived 2026-07-16 |
| Datasheet (E-4046: PGI<25, saturation/sharpness) + Characteristic + Scan | 2 scans (saturated autumn street, night long-exposure) — derived 2026-07-16 |
| Datasheet (RMS 4) + video-comparison + Scan (partial) | first real scan 2026-07-20 (overcast, neutral rail + foliage) — cool-neutral whites, muted greens, low contrast, invisible grain all confirmed; no skin-tone frame yet |
| Datasheet (true ISO 1000, EI-3200-by-design) + Scan (2, different photographers) | 2nd scan 2026-07-20 — grain far finer than reputation; grit comes from pushing, not the base emulsion |
| Datasheet (extremely fine grain, "edge contrast") | a found "reference" was a synthetic software preset — correctly discarded, not used; genuine scans still needed; derived 2026-07-16 |
| Datasheet (EI 50–200 range, T-grain construction, reciprocity data — archived PDF) | closes a long-flagged gap (datasheet had been archived since 2026-07-15 but unread); derived 2026-07-25; no scan yet |
| Datasheet (EI 200–3200 development range, Gbar 0.62 contrast target, T-grain construction — archived PDF) | closes a long-flagged gap; derived 2026-07-25; no scan yet |
| Datasheet (EI 50–800 single-roll exposure range, the inverse exposure/grain relationship, "SHARP, FINE GRAIN" product identity, explicit no-push statement, C-41 chromogenic construction — archived PDF) | closes the second half of the long-flagged "archived but unread" queue; derived 2026-07-25; no scan yet. Breaks four bank conventions on datasheet authority — first 400-speed original at Weak/Small grain, first B&W original with a negative Shadow, lowest ISO ceiling of any 400 (800, because it cannot be pushed), most positive default EC. Highest-value scan: a frame with a genuinely wide subject brightness range |
| Datasheet (six-stops-over/two-under latitude specification, "exceptionally fine grain", Ta = Tm^1.26 reciprocity formula with a ½-second threshold, EI 50–200 quality range, 14-developer tables, giant-enlargement/copying/technical application claims — archived PDF) | closes the top of the promotion queue; derived 2026-07-25; no scan yet. The widest quantified latitude in the bank (8 stops) and asymmetric — 6 over vs 2 under — which is why — + Highlight —sit on the highlight end and —. Corrects four values in the attributed reference recipe, including a sign reversal on exposure compensation (the 2026——audit accepted negative EC on the EI 50–200 quality range; the latitude spec, now read, points the other way). Falsifiable open question: does a 2-stop underexposure hold, or is — too conservative? |
| Datasheet (orthochromatic blue-green sensitivity with no red response, ISO 80 daylight / ISO 40 tungsten, four-filter factor table inverting between light sources, Gbar 0.62–0.70 pictorial contrast, Ta = Tm^1.25 reciprocity, deep-red safelight, "high-resolution copy film" design statement — archived PDF) | ⚠️ Datasheet-validated, partial-fidelity — the second recipe at this tier, and the spectral opposite of SFX 200. Derived 2026-07-26; no scan. ACROS+G reproduces the red suppression but inverts the blue behaviour (ortho lifts blue, a green filter darkens it). Both partial-fidelity recipes fail for the same structural reason: a filter subtracts a band, an emulsion's sensitivity is a shape. Two numbers quantify the red-blindness directly — the 1-stop daylight/tungsten split and the inverting filter table. Grain —/— is an inference (no RMS published). This recipe makes the contested monochrome-WB experiment urgent: if WB acts as a channel mixer, −Red/+Blue on plain ACROS would suppress red and lift blue simultaneously, which is orthochromatic response exactly — a filter cannot do both |
| Datasheet (740 nm extended red sensitivity, ISO 200/24°, EI 200/400/800 development tables, full filter-factor table, focus-shift caution — archived PDF) + published red-edge/hot-mirror physics (secondary, snippet-level — see note) | ⚠️ Datasheet-validated, partial-fidelity — a tier of its own, and the only recipe in the bank carrying it. Derived 2026-07-25; no scan yet. ACROS+R reproduces Ilford's near-black skies but inverts the documented pale-foliage effect, because the vegetation red edge (~680–750 nm) sits behind the X-T5's IR-cut filter. Not a tuning gap — unreachable on an unconverted body. Highest-value scan: a red-filtered frame containing both blue sky and green foliage |
| Datasheet (E-7023: Print Grain Index 46, T-GRAIN emulsion, complexion-density table, "accurate skin-tone reproduction" design claim, Gold print-compatibility, fluorescent/HID correction tables — archived PDF) + Scan (1 frame, inherited) | derived 2026-07-25 as the faithful counterpart to the attributed reference recipe and to AP's UltraMax. PGI 46 against a visibility threshold of 25 makes it the grainiest colour emulsion in the bank — settles a grain argument the bank had twice on prose alone. Scan is flat overcast only: no skin, sun or golden-hour evidence |
| Datasheet (E-55: diffuse RMS 10, "Extremely sharp", "Reproduces subtle color naturally", K—, daylight/tungsten EI tables, reciprocity limits, explicit no-push statement — archived PDF) + Characteristic (Fujifilm's published Classic Chrome tonality and colour-design data) | derived 2026-07-25 as the faithful counterpart to the reference recipe and AP's Kodachrome. Structural ceiling on validation, stated in-recipe: Classic Chrome is based on no analog film at all, so a scan can confirm resemblance but never derivation. Corroborated indirectly by the dated 1980 Kodachrome 25 frame. Deviates from both other builds on Clarity (—→ 0) and ISO ceiling (6400 → 800, because Kodak declines to recommend pushing this emulsion) |
| 11-image place study across 7 condition types | refined once from evidence (Highlight —) |
| Original design | field-test pending |
| Datasheet (E-7023: complexion density targets, T-GRAIN, skin-tone design claim) + inherits the 1 UltraMax scan + Mumbai 11-image study + Fujifilm curve data | ⚠️ field-test pending — never shot on Indian skin, which is the entire premise. Derived 2026-07-25. Open question: is Shadow —enough, or —? |
| Datasheet (E-4050: complexion density targets, PGI 37, "lighting cannot be controlled" brief) + E-4040 for the grain audit + Fujifilm tonality data (Reala Ace, Classic Chrome) + Mumbai 11-image study + audit of all 5 existing Portra recipes | ⚠️ field-test pending — never shot on Indian skin. Derived 2026-07-25. Base changed to Reala Ace (deep soft shadows) because Classic Chrome's documented hard shadow end compresses exactly where darker skin sits. Ships a Classic Chrome fallback. Open questions: does Highlight —hold Mumbai midday; is CCE — too restrained? |
| Datasheet (E-55: RMS 9/10, "extremely high sharpness", "reproduces subtle color naturally", K—transparency process) + Fujifilm tonality/colour-design data for Classic Chrome + Mumbai 11-image study + documented Singh/McCurry lineage. ⚠️ Complexion-density evidence is inherited from E-4050 (Portra) and E-7023 (UltraMax) — the Kodachrome sheet has no such table, because complexion targets are published for colour negatives, not transparencies | ⚠️ field-test pending — never shot on Indian skin. Derived 2026-07-25. Classic Chrome deliberately kept (unlike AP's Portra): contrast is Kodachrome's identity, so the fix is to place the skin above the crush (Shadow — vs the reference recipe's +0.5), not to escape the base. Clarity —→ 0. One inferential step further from source than the other two AP recipes — flagged in both files. Open questions: does Shadow —clear the face or does it want −1.5; does — hold Mumbai midday; do reds land or does Classic Chrome's suppressed magenta pull them to brick despite CCE Strong? |
| Aesthetic designs | tune against user-dropped references/ |
| Scan (1 frame) | 2026-07-21, flat daylight — warm lean, magenta sand, accents punchy without flooding (validates Color —) |
| Datasheet (E-55) + Scan (1 dated 1980 frame) | micro-detail, no visible grain, restrained earth tones |
| ❌ Scan attempted and rejected | 2 frames confirmed by rebate markings but toy fisheye + sodium-vapour light — recorded as seen-and-rejected, not counted |
| ⚠️ Characteristic only | No datasheet, no scans exist. Derivation rests on Fujifilm’s own Color Chrome Effect being a documented Fortia model. Weakest evidence in the bank — flagged in-recipe |
| Attributed captures | not yet re-derived |
Last updated: 2026-07-26
The settings live in the packs
This page explains how the controls behave. The validated recipes that put them to work, with every value and the datasheet evidence behind it, are in the packs.