Sources: Imaging Resource Definitive Guide (Aug 2020) · Zen & The Art of Fujifilm Film Simulations · Pro 400H vs Pro Neg Hi comparison
The Core Problem: Film Is Multi-Dimensional, Sliders Are Not
When you adjust Hue/Saturation or Selective Color in Photoshop, you apply a single scalar transformation to a region of the color wheel. It does not know whether a pixel is bright or dark, saturated or dull, every pixel with that hue gets the same treatment.
Real photographic film does not work this way. Fujifilm's simulations are a mathematical characterisation of actual emulsions, where color transforms vary simultaneously across three independent axes:
- Hue, where the color sits on the color wheel
- Saturation. How saturated the original —. Luminance. How bright the color is
No slider operates across all three axes at once. Hue/Saturation, Selective Color and the HSL panel are all per-band, single-stage tools.
⚠Correction (2026-07-27): the old "structural impossibility" claim was wrong
This section used to say that a transform varying across hue, saturation and luminance simultaneously is "a structural impossibility" in post. That is false, and this repo disproves it two folders away: a 3D LUT is precisely a mapping from every (R,G,B) triple to another, so it varies across all three axes by construction, and
Knowledge/luts/ships two 33³.cubeLUTs built here. Photoshop has applied 3D LUTs since CS6 (Color Lookup); Resolve, Premiere and Capture One all do.A per-pixel colour transform is not what makes a film simulation unreproducible. The argument below is rebuilt on the parts that actually hold. Overclaiming here is worse than sloppy. It is the kind of error a knowledgeable buyer catches, and it would put the rest of this bank's credibility in doubt.
What genuinely cannot be reproduced, in four parts, the first three are structural, the fourth is practical:
- The demosaic is different before anything else happens. Every third-party raw converter. Lightroom, ACR, Capture One, RawTherapee, darktable. Reconstructs X-Trans photosites with Frank Markesteijn's algorithm, not Fujifilm's, which has never been published (sensor-architecture §7d). So a post-hoc grade does not start from the same image. No colour transform, however sophisticated, can correct for a different reconstruction of the underlying photosites.
- Several recipe components are not per-pixel transforms at all, so no LUT can hold them. A 3D LUT is memoryless, one pixel in, one pixel out, no knowledge of neighbours. But Grain is spatial and stochastic (and on ACROS is modelled, not overlaid), Clarity is a local-contrast operator, Sharpness and NR are spatial, and Color Chrome Effect / FX Blue act selectively on already-saturated regions, which requires knowing the neighbourhood rather than the pixel value alone. These are provably outside what any LUT or slider stack can express.
- The simulation runs upstream of the 8-bit JPEG. It operates on raw sensor data with the sensor's own spectral response known to it. Post work starts from output that is already transformed and quantised. Information the simulation used is gone by then.
- The characterisation data is not public. Even where a LUT could express the colour transform, you would need Fujifilm's spectral sensitivity and dye-density measurements to build the right one, and those have never been released. This is an availability limit, not a mathematical one, and it is worth stating as such rather than dressing it up.
Recipe implication (unchanged, and now on firmer ground): recipes built in-camera are the only mechanism that engages Fujifilm's own demosaic, grain model and Chrome operators. They are not stylistic presets. But the reason is the pipeline, not a mathematical impossibility about colour.
Fujifilm's 85-Year Film Heritage
Fujifilm has manufactured emulsions since 1934. Their digital simulations are built from:
- Spectral sensitivity curves of the actual film stocks (per-wavelength response of each emulsion layer)
- Dye density curves: how silver halide grains translate to dye clouds through development chemistry
- Frontier minilab calibration data: Fujifilm's professional photo printing system colour profiles
- Memory colour research: empirical data on which colour renditions humans perceive as most emotionally satisfying
This data does not exist outside Fujifilm. Third-party profiles (Lightroom, ACR, Capture One) are visual approximations, not physics reconstructions.
How Colour Film Actually Works (and why sliders can't touch it)
(For the black-and-white chemical foundation, silver halides, latent image, development, see film-chemistry-fundamentals.md.)
Colour film is a stack of 9 or more layers: multiple light-sensitive emulsion layers for red, green and blue light, with colour-filtering and interface layers between them. Two independent curve families define the look:
- Spectral sensitivity curves. How each light-sensitive layer responds across the wavelength spectrum (input side)
- Dye absorption/density curves. How the cyan, magenta and yellow dyes that form the final image block light per wavelength (output side)
Case study: PROVIA 100F vs Velvia 100F dye purity
✅ Verified against the source, 2026-07-27. This section's claims were checked line by line against the Imaging Resource article and all hold. Note the stocks are Provia 100F and Velvia 100F; the curves are not from Velvia 50, despite that being the simulation's name. The guide reproduces Fujifilm's actual curves for both, and the finding is counter-intuitive:
- Velvia's sensitivity layers overlap more than Provia's (its red and blue layers react more to greens), which would normally reduce saturation.
- But Velvia's dyes have much less spectral overlap than Provia's. Perfect dyes would block only their target colour; Provia's magenta and cyan dyes absorb a lot of blue light they shouldn't, while Velvia's dyes are far purer.
- Net result: Velvia's legendary saturation comes primarily from the output dyes, not the input sensitivity, a two-stage interaction no single-stage editing tool can model.
Also note Velvia's green sensitivity layer is simply more sensitive than Provia's. Part of why its foliage greens glow.
Why curves and sliders fail here. Stated precisely. Film's look is the product of input sensitivity curves × interlayer filtering × output dye curves, where each colour channel contaminates the others differently at each stage. Per-channel tone curves cannot represent that, because they treat R, G and B independently by definition: no setting of a red curve encodes "the red layer is partially sensitive to green, feeding a cyan dye that partially absorbs blue."
A 3D LUT can represent the end-to-end colour mapping (see the correction at the top). Cross-channel contamination is exactly what a 3D LUT expresses. What defeats reproduction is not this stage; it is the demosaic, the spatial operators, and the absence of Fujifilm's characterisation data.
History of the Film Simulations
| Year | Camera | Milestone |
|---|---|---|
| 2003 | FinePix F700 | First variant, called FinePix COLOR — F-Standard, F-Chrome, F-B&W (not tied to named films) |
| 2004 | S3 Pro DSLR | The name "Film Simulation" first appears |
| 2007 | S5 Pro | More film types added |
| 2009 | FinePix F200EXR | First use of classic film names: PROVIA, Velvia, ASTIA (+ generic B&W, SEPIA) |
| 2012 | X-Pro1 | PRO Neg.Hi & Std added; B&W renamed MONOCHROME; Ye/R/G filters added |
| 2014 | X30 | CLASSIC CHROME |
| 2016 | X-Pro2 | ACROS |
| 2018 | X-H1 | ETERNA |
| 2019 | X-Pro3 | CLASSIC Neg. |
| 2020 | X-T4 | ETERNA BLEACH BYPASS |
As of the guide's writing (Aug 2020) there were 17 simulations covering 10 distinct film types. The X-T5 generation later added Nostalgic Neg. and REALA ACE. Fujifilm also continually refines existing simulations as processors improve, the same simulation name can render slightly differently across camera generations (relevant when converting recipes between X-Trans versions; see x-trans-v-and-conversion.md).
Memory Colors: Why Fuji Optimises for Perception, Not Truth
Fujifilm's colour science philosophy is explicit: the goal is not colorimetric accuracy. It is perceptual satisfaction.
Human visual memory is systematically biased: - Skies are remembered as bluer and more saturated than measured - Foliage is remembered as greener and more vivid - Skin is remembered as warmer and more even
Simulations intentionally push toward memory colours, not toward neutral measurement. This is why Fuji images feel emotionally "right" even when a color chart would show them as —.
Different simulations implement different memory-colour strategies:
| Simulation | Memory Colour Strategy |
|---|---|
| Velvia | Maximum exaggeration across the board |
| PROVIA | Moderate, naturalistic bias |
| ASTIA | Optimised for skin — less rosy, hides blemishes |
| Classic Chrome | Inverted for blues/skies — magenta deliberately removed |
| Pro Neg simulations | Prioritises skin-tone smoothness over vibrancy |
Why Adobe Camera Raw / Lightroom Don't Match
The author of the Imaging Resource guide spent hours converting shots in ACR, then saw the camera JPEG of the same frame, and described the result as "a lot different." Key reasons:
- ACR presets are visual approximations. Reverse-engineered from the look, not from spectral physics
- Tone curve handling diverges dramatically, especially highlights. ASTIA's soft rolloff cannot be reproduced in ACR's Profile system
- Per-hue non-linearity is irreproducible. Classic Chrome in ACR gets the general direction (low saturation, harder contrast) but cannot reproduce the specific per-hue shifts that vary simultaneously by saturation and brightness
- Capture One vs Adobe. Capture One displays the embedded camera-generated JPEG thumbnail; Adobe ignores it and substitutes its own profile. This means in Lightroom, your RAF files will look different from your JPEGs even for the same shot
The Only Authentic Simulation Workflow
In-camera JPEG, or X RAW STUDIO (which routes through the camera's own ISP). Any other RAW converter produces a different image, not a variant of the simulation, but a different colour science entirely.
This is the foundational reason why this project treats JPEG as the archival primary and RAW as a recovery fallback only.
Post-Processing Safety Boundaries
Some adjustments are safe after the simulation is applied; others break the per-hue colour model:
⚠Reasons corrected 2026-07-27. This table previously justified three rows with claims that do not survive checking. They are marked below. The safety ratings are unchanged; the reasoning behind them was wrong, which matters because the reasons are what a reader generalises from.
| Adjustment | Safety | Reason |
|---|---|---|
| Global exposure ±0.5 EV | ✅ Safe | Moves values along an already-applied curve; small shifts stay clear of the shoulder and the toe, where the sim's character lives. (Previously justified as "within the tone curve's linear region" — a baked JPEG has no meaningfully linear region; it is gamma-encoded with the sim's curve already applied.) |
| White balance correction | ⚠️ Caution | Downgraded. (Previously "✅ Safe — pre-demosaic is clean". That reason applies to RAW, but this table is about editing the baked output, where WB is applied after the sim's per-hue transforms and therefore fights them.) Correct WB in camera, or in X RAW Studio. Small corrections on a JPEG are recoverable; large ones re-hue everything the sim placed deliberately. |
| Clarity / local contrast | ⚠️ Caution | Downgraded. (Previously "✅ Safe — does not touch colour". It does — local-contrast operators change perceived and measured saturation as a side effect of the luminance change.) Modest amounts are fine; heavy Clarity shifts saturation in the mid-tones. |
| Vibrance | ⚠️ Caution | Can break the hue-saturation balance built into the sim |
| Saturation (global) | ❌ Avoid | Overrides the sim's per-hue saturation strategy |
| Selective hue shifts | ❌ Avoid | Fights the sim's own hue-shift curves |
| Changing film sim in ACR | ❌ Avoid | Produces a different image — not a variation (see the demosaic point above) |
Per-Simulation Colour Behaviour (from Imaging Resource's colour maps)
⚠Header corrected 2026-07-27. This was titled "Per-Simulation Spectral Behaviour (from Imaging Resource spectral data)" It is not spectral data. Verified against the source: the article publishes spectral sensitivity and dye-density curves for exactly two stocks. Provia 100F and Velvia 100F: and describes the simulations through colour maps (input-colour → output-colour vector diagrams) plus observation. The entries below are therefore qualitative colour behaviour, which is what they read as. Calling them spectral data implied a per-wavelength measurement that does not exist here, the same over-attribution pattern found on the 2383 datasheet.
PROVIA / Standard
- Baseline reference. Moderate saturation lift, slight red and green exaggeration
- Sky: moderate blue saturation boost with a hint of magenta retained
- Oranges shift slightly toward yellow, without this shift, yellows and oranges look muddy and hard to separate visually
- Light cyan-blues shift toward pure blue, an industry-wide digital convention for pleasing skies, and Provia applies it conservatively
- Fujifilm's marketing calls it "the ace of spades — the strongest card of all"; it is the factory default on every camera
VELVIA 50
- Based on FUJICHROME Velvia, introduced 1991. Professional reversal film beloved by landscape photographers
- Extreme saturation lift, especially greens and magentas; dark greens, dark blues and pure reds most extreme
- Blues deliberately skew toward magenta. Producing the "royal blue" skies Velvia is famous for
- Can clip sRGB gamut for bright reds and pinks (flowers, sunsets). Visible loss of detail in the most saturated areas
- Hard S-curve shadow contrast; bright yellows also rendered with higher contrast than other hues
- Designed to represent scenes "as photographers remember them". Memory colour pushed to maximum
- Choose ASTIA when you want colour pop but need to preserve tonal gradation in bright subjects
ASTIA / Soft
- Highlights: substantially softer rolloff than PROVIA
- Shadows: darker than PROVIA. Overall contrast stays high despite soft highlights
- Skin: less rosy, lower saturation, and something other than a "softer PROVIA". Caucasian skin lands almost dead-on its true colour; darker skin tones get increased saturation, producing richer tones for darker complexions
- Hides skin blemishes and blotchiness better than any other simulation
- Colour map reads like a blend of Provia and Velvia: dark greens more saturated than Provia, yellows/oranges less; darker blues noticeably more saturated
- Fujifilm also recommends it for adapted vintage lenses, its harder shadows compensate for older optics' lower contrast
CLASSIC CHROME
- No single-film lineage. Designed to evoke Kodachrome-era reversal film and printed-magazine imagery
- Lowest saturation of any colour simulation (until Eterna Bleach Bypass)
- Shadow contrast is harder; highlight contrast is softer (opposite of ASTIA)
- Sky: magenta component intentionally and completely removed: documented Fujifilm design decision
- Cyan/blue hues behave non-monotonically: bright cyan stays near true colour; medium blue desaturates and shifts toward cyan; dark blue shifts differently again
- Explicitly stated by Imaging Resource: "cannot be approximated in Photoshop — the sorts of changes just aren't possible through any manual adjustment"
- Target use: street and documentary; evokes Life/Look magazine aesthetic of the 1970s
PRO Neg.Std
- Based on Fujifilm NPS 160 / NS 160. Professional studio portrait negative film
- The actual NPS 160 emulsion contained a fourth, cyan-sensitive layer that gave Fujifilm's chemists superior skin-tone control
- Very flat tone curve + muted, direction-accurate (not hue-shifted) colours
- Designed for studio use where lighting creates contrast; looks intentionally flat outdoors
- Hides blemishes and wrinkles through contrast reduction, not hue manipulation
PRO Neg.Hi
- "Pro Neg.Std Lite". Same colour direction, less extreme
- Slightly more saturation and contrast than Std
- Preferred for outdoor, wedding, and casual portrait situations
CLASSIC Neg.
- Based on Fujifilm Superia 100 consumer negative film (introduced 1998)
- High contrast, slight red-magenta tint, lower saturation
- Colour gamut behaviour: squashed away from bright yellows, greens, purples and blues. Ballooned out in the reds. The most complex colour map of any simulation
- Tone curve varies between colour channels more than any other profile, some channels deliberately flatter than others to increase inter-colour contrast
- "Old family shoebox print" aesthetic. Evokes colour-negative prints of the 1960s—s more strongly than any other simulation
ETERNA Cinema
- Based on Fujifilm's motion picture negative stocks, never sold to consumers
- ~12-stop dynamic range with softened shadows and highlights. Extended latitude without going to full log-gamma
- Lifted shadows (milky blacks, never pure black); extreme highlight protection
- Colour palette squashed like Classic Neg (greens, magentas, purples desaturated most) but far gentler, and per Fujifilm with deliberately low hue shift. Colour accuracy was a design goal, unlike most simulations
- Requires colour grading to reach a finished look, not a standalone aesthetic; it is a neutral starting point
- Correct workflow: ETERNA in-camera → grade in Resolve/FCPX
ETERNA Bleach Bypass
- Introduced Feb 2020 with the X-T4, the newest simulation in the guide
- Simulates skipping the bleaching step of colour-negative processing, which normally removes the image silver and leaves only dye. Retained silver = a contrasty B&W image overlaid on a faint colour one
- Lowest saturation of all colour simulations: but the desaturation is uniform across the spectrum with almost no hue shifts; its hue map matches ETERNA almost exactly, just with saturation floored and contrast maxed
- Washed-out highlights + very dark shadows are the hallmark. On real film, shots destined for bleach bypass were exposed ~1 stop under; for in-camera JPEGs consider dialling exposure back similarly
- Signature use: urban-grunge, gritty editorial; also striking on normally colourful subjects
ACROS
- Emulates the Neopan ACROS series black-and-white film (orthopanchromatic; the modern stock is Acros 100 II)
- Tone curve (per Fujifilm's own plot): harder/higher-contrast from lower midtones to middle highlights, much softer through lower midtones and shadows. Richer shadow detail
- Distinctive grain structure, not equivalent to adding digital noise (see grain section below)
- More contrasty than the Monochrome simulation
- Preferred B&W simulation for most subjects
- B&W films vary widely in spectral response (panchromatic vs orthochromatic etc.), which is why a Photoshop RGB→Grayscale conversion, which just discards colour information, produces bland results by comparison
- The +Ye / +R / +G filter variants: per Fujifilm's manual, these "deepen shades of gray corresponding to hues complementary to the selected color." Read that in the darkening direction, not the lightening one, +R darkens greens and cyans (it does not lighten foliage), +G darkens magentas and reds, +Ye mildly darkens blues. This trips people up when trying to model films with unusual spectral response; see the hard case immediately below.
The spectral ceiling, the cleanest demonstration of this whole document (added 2026-07-25)
The argument on this page is usually made about colour dye layers, where it is true but hard to show. Black-and-white gives a case that reduces to a single number.
Ilford SFX 200 is a B&W film with extended red sensitivity up to 740 nm. Its datasheet claims two effects when shot through a deep red filter: skies render "almost black" and green vegetation renders "almost white."
- The sky effect is ordinary visible-light filtration. ACROS+R reproduces it.
- The foliage effect is not. It comes from the vegetation red edge: chlorophyll absorbs strongly through the visible red and then becomes nearly transparent above ~700 nm, so leaf reflectance climbs steeply, by a factor of five or more. Across roughly 680–750 nm. (This is the same physics behind the Wood effect in true infrared photography.) SFX's 740 nm ceiling reaches the bottom of that ramp; that is the whole trick.
- The X-T5 never records that band. Like every unconverted digital camera, it sits behind an IR-cut / hot-mirror filter whose transmission falls away through the 600s. Infrared-conversion services exist specifically to remove it, for Fujifilm X-mount among others.
- So ACROS+R works purely on visible red, where chlorophyll's absorption peak (~660 nm) means healthy leaves reflect little red. Per the filter rule above, +R therefore renders foliage dark. The recipe fails to reproduce the effect and produces the opposite one.
No setting recovers this. Not Grain, not Shadow, not WB, not Clarity, not a physical R72 filter on the lens, the information was removed by a piece of glass in front of the sensor before any processing happened. That is exactly the claim this page makes about colour, stated in a form that can be checked: a film simulation is a spectral-sensitivity-aware conversion of the raw sensor signal, and it cannot reconstruct what the sensor was never given.
Worked out in full, with the datasheet quotations, in the notes. The recipe carries a partial-fidelity validation tier because of it, see validation-methodology.md.
Sourcing: the Ilford figures are primary (archived datasheet). The red-edge and hot-mirror physics is well-established published science, cited here at web-search-snippet level. Full-page fetches returned HTTP 403 from this environment. Fujifilm does not publish the X-T5's IR-cut transmission curve, so the cutoff is described qualitatively rather than quoted.
Monochrome
- Effectively PROVIA with the colour information removed (originally named "B&W" before the X-Pro1)
- Flatter than ACROS, and its grain option is a simpler random pattern. Preferable for high-key, dreamy, or extremely contrasty subjects where ACROS would clip
SEPIA
- Historical note: sepia toning replaced silver particles with silver sulfide, which is chemically more resistant to environmental sulfur from industrial coal combustion
- Practical archival function in the 19th century; now purely aesthetic. Full chemistry in film-chemistry-fundamentals.md
The ACROS Grain Simulation: Why It Isn't a Noise Overlay
Film grain is silver-halide crystals that clumped into visible particles during development (chemistry in film-chemistry-fundamentals.md). Fujifilm's ACROS simulation models that entire process algorithmically rather than scattering random dots:
- The simulated grain reproduces the structure of real ACROS grain, including how the texture changes between highlight and shadow regions of the same frame. Real grain clumps differently at different densities
- The base grain processing is always identical, but it combines with natural sensor noise, so images at higher ISO look grainier, just like film. Practical trick: if you want a grainy look, deliberately shoot at higher ISO even in bright light
- The separate Grain Effect menu (—/—, plus Size on newer bodies) layers on top of this
- Adobe's grain sliders (amount/size/roughness) can match the overall visual weight but have visibly less structure, the gap widens the harder the effect is pushed
- Photoshop's demosaicing can produce serpentine noise patterns that superficially resemble ACROS grain, but that is a demosaicing artifact seen across many cameras, not grain modelling
- Only ACROS gets the structured grain model. The Monochrome simulation's grain option is a simpler random digital pattern
Ye / R / G Filters and B&W Adjust
Both B&W simulations (ACROS and Monochrome) can apply a simulated colour filter before the monochrome conversion, replicating the classic screw-on filters of the film era:
| Filter | Effect |
|---|---|
| Yellow | Darkens blue skies slightly so clouds pop; minimal effect on foliage — the classic all-purpose landscape filter |
| Red | Yellow "only more so" — skies go very dark and dramatic; foliage lightens slightly |
| Green | Slightly darkens sky; mainly softens contrast within green foliage; flatters skin in portraits |
Cameras since the X-T3/X-T30 also offer B&W Adjust: ±9 steps of warm/cool toning on the monochrome image, mimicking the distinctive paper tones of different B&W printing papers (warm for portraits, cool for snowscapes).
Fujifilm's own words for this argument (added 2026-07-20)
The strongest first-party statement of the thesis on this page comes from the Image Design team:
"Image design is a work of converting the raw signal while understanding the difference of the spectral sensitivity." X Stories: Film Simulation, Revolution by Continuous Evolution
That single sentence is the whole case. A film simulation is not a colour grade applied to a finished picture. It is a spectral-sensitivity-aware conversion of the raw sensor signal, performed at the point where the sensor's response is reconciled with how the eye actually sees. Once a file has been demosaiced and handed to Lightroom, that step has already happened and cannot be re-run. This is why the in-camera JPEG is the archival primary in this bank.
Two supporting statements from the same source: - The target never moved, only the machinery: "The vision of ideal colour reproduction for FUJIFILM remains the same. It's just that devices have evolved, processor has evolved, and the algorithm has evolved.", and, candidly, "there are still hundreds of things to do to reach the ideal vision." - Newer processors extract more from the same signal, "it picks up signals that were not previously picked up" Giving more detail "without supersaturation", and granting every colour simulation greater "toughness against colour supersaturation." This is a generational trait, and part of why X-Trans IV→V ports tolerate Color values that would have looked garish on older bodies.
A third, easily-missed constraint: the EVF must render the simulation live, "If the film simulation is set to Velvia, then we have to show the world of Velvia in real time. EVF has to be that what you see is what you get." Practically, you are previewing the finished JPEG before the shutter fires; recipes should be judged in the viewfinder in the actual light, not guessed at from a settings table.
Full extraction: fujifilm-official-design-notes.md.
Last updated: 2026-07-27
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.