Dark Ground

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

Part 0: the source

Read from the design archive's look collection on 2026-08-25: Suzanne Jongmans's Renaissance-style portraits, published by Colossal in August 2026. The costumes are built from recycled packaging material, which is the work's subject and not its look.

One frame, and normally that would be too thin. It is enough here because the lighting tradition behind it is four centuries old, unambiguous and enormously documented, so a single well-made example is a sample from something already characterised rather than an isolated observation.

What the frame holds:

Part 1: the requirements, ranked

1. Dark that is not empty. The velvet occupies more of the frame than the face and it carries fold, nap direction and a tonal change across the shoulder. If the low end crushes, half the picture becomes one shape.
2. Luminous skin. Modelled by a single soft key with no fill, which means a long gradient from lit cheek to unlit jaw, and any harshness in that gradient reads as a lighting error.
3. A warm ground that stays a colour. Umber has to be umber across a huge tonal range, not brown in the light and black in the shade.
4. One small bright that stays distinct. The lace is two per cent of the frame and it anchors the whole thing.
5. A restricted palette. Everything not skin, ground or lace should sit down.

Requirement 1 is the hard one and it is the reason the base choice was not obvious.

Part 2: Reala Ace, on the third attempt

This bank has rejected Reala Ace twice and both times on the same clause. The standing note is that recipes on this base "shouldn't fight it with heavy Highlight softening", because the hard highlight end is half of what makes the simulation read true to life.

North Light lost it because a blown window is the interior photographer's failure mode and any interior recipe has to push the top end hard. Austurbane lost it because its largest frame area is whitespace that has to hold gradation, and because the compression would have flattened the one refracted spectrum that had to stay saturated.

Neither objection applies here, and that is the whole argument.

• Nothing fights the highlights. The brightest thing in frame is a collar occupying two per cent of the picture, and requirement 4 wants it to stay a distinct bright rather than roll away. A hard highlight end delivers that for free, so … and the bank's rule is obeyed rather than broken.
• The compression is wanted. "The more saturated the subject, the more muted the color" is requirement 5 stated as a base behaviour. A restricted palette arrives without a setting.
• And the shadow clause, which is the one that decides it. Fujifilm describes Reala Ace's tonality as "deep, soft shadows and hard highlights", the inverse of most simulations, engineered to recreate era-accurate photographic prints while avoiding digital muddiness.

Deep and soft in one phrase is requirement 1 exactly. Deep gives a ground that reads black at a glance. Soft gives the gradation that keeps the velvet readable. No other simulation in the lineup names both, and it is unusual enough that Fujifilm calls it an inversion.

That a base rejected twice on one clause wins the moment the clause stops applying is worth recording, because it is the difference between a rule and a reflex.

The rejections

Part 3: Shadow …, which is the recipe's real claim

The instinctive setting for a dark moody portrait is Shadow …. It produces a silhouette against a black rectangle, and it is worth understanding why that is a different picture rather than a stronger version of this one.

An Old Master portrait is not a picture of a lit face on a black background. It is a picture of a figure emerging from a ground, and the ground is a rendered surface with paint on it. The velvet cape in the reference holds more information than the face does. Crush the low end and the composition loses its largest element, and what remains reads as a modern studio portrait with a plain black backdrop behind it, which is a look this bank already covers with Fashion Editorial at Shadow ….

Reala Ace's shadows are already deep. What a negative Shadow value buys is not brightness, it is protection of the softness half of "deep, soft shadows", so the gradation survives into the darkest part of the frame.

The darkness itself comes from three places, none of which is the tone curve: one key with no fill, a genuinely dark ground in front of the camera, and exposure compensation pulled down. Asking the tone curve to supply darkness that the lighting did not is what turns a reference like this into an underexposed photograph.

… follows from the same logic. Fujifilm's dynamic-range modes underexpose and then lift everything below the highlights, which is the single operation this look cannot survive. One stop covers the lace. … would raise the ground and undo the entire premise, which makes this one of the few recipes here where the higher setting is actively wrong rather than merely unnecessary.

Part 4: why … on a deliberately restrained look

… on a look this restricted looks like an oversight. It is the opposite, and the reason is specific to this base.

This bank's note on Reala Ace is that the Color control behaves differently here than anywhere else: because the nonlinear compression has already tamed whatever is most saturated, a Color adjustment mostly moves low-saturation content. That is why fujifilm-negative reads natural rather than punchy at Color ….

Run that backwards. The most saturated thing in this frame is the umber ground, which the compression is already holding. The least saturated thing is skin. So a negative Color here would leave the ground more or less where it was and drain the face, which is the one outcome requirement 2 cannot tolerate.

The restricted palette therefore has to come from the base and from what is in front of the camera. Color has to stay out of it. This is a good example of why a bank that records how each base behaves produces different settings from one that records how each base looks.

Color Chrome Effect … is the counterpart and it is safe for the same structural reason. The ground is the one genuinely saturated region, it spans an enormous tonal range from lit fold to deep shadow, and the control exists to keep gradation in saturated reds and yellows instead of letting them block up. Applied to velvet, that is the requirement written as a setting.

Part 5: where it sits in the bank

It is the bank's second Reala Ace original after AP's Portra, and its first low-key portrait built around a ground rather than around a subject.

Part 6: honest limits

• No frames. Creative-design tier and nothing more.
• One image. The thinnest evidence base of any recipe in this batch. What carries it is that the lighting tradition is documented independently of this photographer, but the specific rendering here is one frame by one artist and the reading could be wrong about what is hers and what is the tradition's.
• Half of this is lighting and the recipe cannot supply it. One soft key, placed high and to the side, with no fill and a genuinely dark surface behind the subject. Every setting in the table assumes that light already exists. Shot under a normal room's ambient this recipe produces a slightly warm, slightly dark photograph and nothing more.
• The costume and the ground are production design. Jongmans builds hers. A recipe cannot hand anyone a velvet cape.
• The Color argument is inferred rather than measured. It follows from Fujifilm's statement about nonlinear compression and from this bank's reading of how that affects the Color control. Neither is a measurement, and the skin frame in the test list is the only thing that can confirm it.
• Exposure is doing more work than any setting. That is stated in three places in the notes because it is the failure mode, and it means the recipe is less portable than most here: it depends on the operator overriding what the camera wants.
• No video version. Movie mode drops Clarity and both Chrome effects, and Color Chrome Effect … on the ground is central here.

Last updated: 2026-08-25