AP's Superia

Original design (field-test pending). See the notes and the Knowledge base.

---

Part 1: Why this recipe was almost not built

The AP set had a gap and an objection sitting on top of it.

The gap: AP's UltraMax covers the warm street, AP's Portra covers the person, AP's Kodachrome covers the documentary picture of India. All three are built for light you can see by. None of them is built for a Mumbai interior, and a large share of the city's daily life happens under a tube light.

The objection: this bank's own analysis rules Classic Negative out. The base comparison in AP's Portra is blunt about it.

That verdict stands, and this recipe does not overturn it. What it does is take the verdict seriously enough to work out the mechanism, which turned out to be correctable in a way the earlier note assumed it was not.

Classic Negative is also not optional here. Fujifilm built the simulation from Superia. A Superia recipe on any other base is a colour grade wearing the name, so either the objection gets solved on this base or the recipe should not exist.

Part 2: What the datasheet says, and the one line nobody has used

Fujifilm's Superia X-Tra 400 sheet (AF3-0217E, mapped in full in the faithful build's research notes) claims fine grain for a high-speed film, excellent sharpness, wide exposure latitude, excellent grey balance from the brightest highlights to the deepest shadows, and excellent skin colour reproduction.

Then there is the fourth colour layer.

Superia carries a fourth, cyan-sensitive layer added so the film renders correctly under fluorescent light, where ordinary three-layer colour negatives go green. It is the film's genuine engineering distinction, it is the reason Superia became the default roll for indoor snapshots for a generation, and no recipe in this bank has ever used it as a design input. The faithful build records the fact and then builds a daylight recipe, which is reasonable, because the faithful build's job is the emulsion rather than a place.

For this brief it is the most useful sentence on the sheet. Mumbai's shops, homes, stairwells, local trains and market interiors run on tube light. The one film engineered for that light is also the film this recipe emulates. That is the recipe's reason to exist, and it is why the fluorescent variant in the notes is the headline rather than a footnote.

The honest limit: the X-T5 has no fourth layer. The correction has to come from the white balance grid, which is a blunter instrument than an extra emulsion layer, and it corrects the light without reproducing what the layer did to hue separation. The variant is a translation of the film's intent, not a reproduction of its chemistry.

Part 3: The chromatic problem, which is not the tonal one

The other two AP recipes solve a tonal problem. Classic Chrome's shadow end is hard, darker skin sits lower on the curve, hardness there costs facial modelling, and the fix is to lift the shadow.

Classic Negative's problem has the same shape and a different cause. Fujifilm's stated signature is that colours are "biased toward cyan-green in the shadows and magenta in the highlights" (Fujifilm Official Design Notes — from the X Stories / product pages official design notes). The shadow region is tinted, so anything sitting in it inherits a green cast regardless of how much detail survives.

Stack the three facts:

1. Darker skin renders at roughly half the separation from mid-grey that lighter skin does, about +0.16 against +0.31 in Kodak's complexion-density tables. So the face sits low.
2. The base tints that region cyan-green.
3. Maharashtra skin skews warm with olive common, and olive is yellow plus green.

Add a tube light, which casts green, and the same hue is arriving from four directions at once. This is a better explanation of the "Classic Negative makes Indian skin look tired" complaint than the usual one, which blames saturation.

The correction follows from the mechanism rather than from taste. Shadow …aises the face out of the tinted region. White balance … opposes the remaining cast directly, since red sits opposite cyan on that axis. And crucially, moving the face up puts it in the mid-tones, which is where Fujifilm says this base is deliberately at its most colourful: "colorful even in the midtones." So the fix does not trade colour away to buy neutrality. It moves the skin into the zone the base treats best.

Where the cool went

Removing Superia's cool white balance leaves a debt. The stock's coolness is its signature, and a build that takes it out without putting it back somewhere is a warm consumer-negative recipe wearing a Fuji name.

The resolution is that Superia's cool is a blue-highlight phenomenon rather than a global chill. The datasheet claims excellent skin reproduction, so the film does not make people green. The cool lives in neutrals, skies and highlights. That means it can be carried by a control that acts selectively on saturated blues instead of by one that acts on everything.

Skin is not a saturated blue, so the control does not reach it. This is the same mechanism the whole AP set runs on, stated at the level of a single setting: put the effect where the skin is not.

Two honest qualifications. The bank's X-Trans V & Recipe Conversion Notes conversion rule drops Color Chrome FX Blue one notch when porting a X-Trans IV recipe to the X-T5, because X-Trans V renders blue more strongly. That rule governs ports and this recipe was designed natively on V, so it does not bind here, but it does mean Strong on this body is stronger than Strong on a IV body. Reggie's Superia reaches the same conclusion on the same base and sensor generation, which is as close to corroboration as this gets without frames. And the mechanism is an outdoor one: tube-lit interiors contain almost no saturated blue, so indoors the control idles and the recipe leans on white balance and the tone curve alone.

Why the Red shift stops at +1

Everything above argues for more Red, and the highlight bias argues against it. The magenta lean at the top end belongs to the simulation, so Red added to fix the shadows also deepens a pink cast in the highlights that was never a white balance error. Fujifilm's own tuning implication, recorded in the Fujifilm Official Design Notes — from the X Stories / product pages design notes, is that pink highlights on this base call for less Red rather than more Blue.

… is where those two pressures balance. It is deliberately less than Mumbai's …, which is a street recipe with no obligation to protect a face, and the difference between the two is the clearest illustration in the bank of what "tuned for skin" actually changes.

Why the Blue shift stops at −1

This is where this recipe and Reggie's Superia disagree most, and it is worth being precise about, because Reggie's recipe is good and was built with care for exactly this class of problem.

Negative Blue is a yellow push. Reggie's runs −4. On skin with red-brown undertones, a yellow push adds warmth and richness and reads as depth. On golden-olive undertones, which already carry yellow and green, the same push adds the hue the skin has too much of, and the documented result is sallow. AP's UltraMax reached the same conclusion from the other direction and pulled the reference recipe's … back to … for the same reason.

So the two recipes are not competing answers to one question. They are correct answers to two different undertones, and neither generalises to the other.

Why depth was added with Color Chrome rather than Color

The obvious lever for a richer frame is Color, and on this base it is the wrong one. Fujifilm states that Classic Negative applies a mid-tone saturation emphasis, staying "colorful even in the midtones". Mid-tones are where skin lives, so raising Color here is a direct hit on complexion, and on warm skin it goes orange quickly.

Color Chrome Effect does the opposite job. It restores tonal depth inside already-saturated colour, which means foliage, painted walls, fabric and signage, and it has almost nothing to act on in low-saturation skin. Color stays at …, matching a medium-saturation stock, and the richness arrives through a door skin is not standing in.

This is also why the recipe runs Strong where the faithful build runs Weak. The faithful build has no obligation to protect a face, so it can let Color and the base do the work between them. This one has capped Color for skin's sake and needs the depth from somewhere.

Part 4: The adjudication, and what it changed

This recipe was drafted twice, independently, in two different sessions working from the same repo. Both drafts are recorded here because the disagreement was more useful than either draft alone, and because the bank has a precedent for settling that kind of conflict on evidence rather than authorship.

The two drafts already agreed on 11 of 14 settings, including both of the values that carry the argument: WB …/… and Shadow …. They also reached those two by the same route, from Fujifilm's cyan-green shadow statement and from the bank's own magenta-highlight tuning rule. Two independent derivations landing on the same load-bearing values from the same primary sources is the strongest corroboration available for a recipe with no frames behind it.

Three settings disagreed, and all three resolved against the first draft.

The first draft's reasoning for Highlight … was that it keeps the frame clear of the magenta-biased highlight region. That conflated two controls: the magenta bias is corrected on the Red axis, as the recipe's own tuning row now says, and the tone curve should be set from the light instead.

What survived from the first draft is the material the second one did not have: the fourth colour layer read as a design input, the tube-light variants that follow from it, the undertone comparison against Reggie's Superia, and the AP-set positioning below.

Part 5: How this sits against the closest prior art

The shared ground is worth stating: Reggie reached Shadow … independently, and it is the right direction for the same reason it is right here. That is the second time in this set that his instinct has landed on the correction darker skin needs, after the same value in Reggie's Portra. Credit where it is due. The derivation here is an argument for a value he had already found by eye.

Part 6: Position in the AP set

Four recipes, four different colour mechanisms, one city.

The fourth position is the one the set was missing. Mumbai's everyday frame is mostly mid-tones, a person against a painted wall in clothing that is neither loud nor grey, and Classic Negative is the only base here whose saturation emphasis is placed exactly there.

There is no the notes, which matches the rest of the AP set. Movie mode drops Grain, both Color Chrome controls and Clarity, and this recipe's identity sits in white balance and the tone curve, both of which survive the conversion. A video version is a short job if anyone wants it.

Validation status

Original design, field-test pending. Grounded in:

1. The Fujifilm Superia X-Tra 400 datasheet, AF3-0217E. Fine grain, excellent sharpness, wide latitude, grey balance across the range, excellent skin reproduction, and the fourth colour layer.
2. Fujifilm's published design statements for Classic Negative. The cyan-green shadow and magenta highlight bias, the mid-tone saturation emphasis, and the consumer-negative design intent.
3. Kodak's complexion-density tables, inherited through AP's Portra and AP's UltraMax.
4. The two real Superia scans analysed for the faithful build, which confirmed the cool palette, muted olive greens and medium contrast.
5. The Mumbai 11-image place study.

Two gaps, stated plainly. It has not been shot on Indian skin, which is the premise of the whole AP set. And it has not been shot under a tube light, which is this recipe's specific claim to a place in the bank. The fluorescent variant is reasoning from the datasheet and the base's documented bias, with no frame behind it yet. Drop frames into test-shots/, and real Superia scans of indoor subjects into references/. An indoor scan with faces in it would be the single most valuable file this recipe could receive, and the bank has none.

The explicitly labelled inference, same as elsewhere in this set: mapping Kodak's negative-exposure density targets onto a digital tone curve is reasoning from the data rather than a manufacturer claim, and the tables come from Kodak films rather than from Superia's own sheet, which publishes no complexion densities. Where skin falls on a tone curve is a property of skin and the curve rather than of the stock, so the logic carries, but it is one step from the source and should be read that way.