Clerestory

Rich gothic interior light: a shaft through a high window against cold stone, deep saturated darks that hold, and a setting that survives the sun going behind a cloud without being touched. Named for the row of high windows that throws light down a nave. Derivation in the notes.

Derived from cinematographer Steve Yedlin's account of a 2025 mystery film, reported by PetaPixel. Yedlin calls it "more of a lighting movie than a camera movie" and describes a "rich, gothic" look built around churches and crypts. The governing idea came from director Rian Johnson, who wanted "the light changing a lot, where we feel the clouds coming in front of the sun and all these different changes within scenes", drawn from growing up in Colorado where "suddenly it would be like God turned the lights out and just things go from very sunny to very dark". His stated aim was "getting very theatrical with natural light shifts that you don't see very often in movies".

The brief is a range problem disguised as a mood

Most looks in this bank describe a colour and a contrast. This one describes light that moves: sun in and out of cloud, a shaft appearing behind a character mid-conversation, the room going from very sunny to very dark while the camera runs.

For a stills photographer that translates into a demand almost nothing else here makes. The recipe has to work in both states without being touched, because by the time you have changed a setting the light has changed back. So it is built for the widest ambient swing rather than for one lighting condition, and the settings are chosen to survive rather than to flatter.

• DR400 for the two stops. A shaft of sun through a high window against stone shadow is close to the widest range in interior photography, and the sun going in takes three or four stops out of the scene without warning.
• Highlight −1.5 for the window itself. This bank's standing note on this base is that −1 is usually enough because the highlight end is already soft, and that −2 goes limp. A leaded window with the sun behind it is the case that justifies the value in between.
• Exposure at −1/3 to 0, metering for the stone and letting the window go. Protecting the bright end is what keeps the frame usable when the sun reappears.

Shadow +1, which undoes what DR400 just did

DR400 buys range by underexposing and then lifting everything below the highlights. That is exactly what the window needs and exactly what the crypt does not: two stops of lift turns gothic darkness into a grey basement.

So the tone curve puts it back. Classic Chrome's shadow end is already the hard one, and this bank's own tone table records that the base "needs a Shadow lift just to reach normal shadow detail". Declining that lift gives deep. Going a notch positive on top of DR400's lift gives gothic, and it lands roughly where the base would sit unaided.

That is the opposite of the rule North Light is built on, where range comes from Dynamic Range and the tone curve stays out of the shadows. The difference is what the darkness is for. There, a lifted shadow is the amateur tell. Here it is the subject.

Classic Chrome, with the saturation put back

The requirement is unusual: a hard shadow end for density, under a soft highlight end for the window, with rich colour surviving both. Fujifilm's published tonality gives the first two to one simulation, which is second only to Velvia in overall contrast while staying soft at the top.

What it does not give is richness. It is the most subdued base in the lineup, with suppressed magenta and cool shadows by design.

Both of those turn out to be assets here.

• The cool shadows are wanted. Stone, slate, lead and cold north light are the palette, and this is the rare recipe that takes the base's documented coolness as delivered rather than correcting it. White Balance moves one notch of Red and no more, purely to keep skin alive inside it.
• Suppressed magenta drops reds toward terracotta and brick, which is the gothic register: oxblood, dark wood, brass, dried blood on a stone floor. A pure red would be wrong.
• The richness comes back through the documented route. This bank's note on the base is explicit that recipes wanting a true red must add it with Color and the Chrome effects rather than expecting the base to deliver it. Color +2 with Color Chrome Effect Strong is that instruction followed.

Color Chrome Effect Strong is doing precise work rather than general enrichment. It gives tonal separation to colours that are already saturated, and in a gothic interior those are stained glass, brass, candle flame and dark polished wood, each of which spans an enormous tonal range within a single object. Without it a stained-glass window becomes flat blocks of colour, which is the single most recognisable way this subject is photographed badly.

The rest

• Clarity 0, which is the restraint that matters most here. A window mullion against blown glass is the archetypal high-contrast hard edge, and Clarity is a local-contrast operator that fringes on exactly that. The subject is full of them: tracery, columns, ribs, railings. This is the same argument North Light makes about rooflines and it applies with more force in a church.
• Sharpness 0. Carved stone and wood grain want detail, and the X-T5 has no optical low-pass filter, so 0 already resolves close to Nyquist. Hard geometric edges are where over-sharpening announces itself first.
• Color Chrome FX Blue Weak. Cold window light and blue-grey stone are the two large cool areas. Weak keeps gradation in them. Strong on this sensor would push a grey nave toward navy and turn gothic into fantasy.
• Grain Weak and Small. Enough texture to stop a large dark area rendering as a digital gradient, and no more. The reference is a modern digital production.
• ISO 500 to 12800. The floor is set by DR400. Church interiors between shafts are darker than they look and the ceiling needs to be high.

Variants

Validation tier: Creative design, field-test pending. Grounded in the cinematographer's and director's published account of the lighting brief, in Fujifilm's stated tonality and colour design for Classic Chrome, and in this bank's Dynamic Range & Tone Curve, the contrast controls tone and Color, Color Chrome Effect & FX Blue, the saturation/richness controls colour notes for that base. No frames have been shot with it.

Test priority: in order

1. A shaft of sun through a high window into a dark interior. The whole recipe. Does the window hold any structure, does the stone stay genuinely dark, and is there still colour in the middle?
2. The same frame with the sun behind cloud, untouched. The robustness test, and the reference's actual idea. If the second frame needs a different setting the recipe has failed its brief regardless of how the first one looks.
3. Stained glass at 100 per cent. The Color Chrome Effect Strong test. It should hold gradation within each colour rather than reading as flat blocks.
4. A mullion or a railing against a bright window. The Clarity 0 check. If there is a bright fringe along the edge, something else in the chain is adding local contrast.
5. Skin in shadow, at +1 Red. Classic Chrome's cool shadows are documented and deliberate here, and this settles whether one notch is enough to keep a face from going bloodless in them.

What would count as the recipe being wrong rather than needing a tweak: if the darks read grey rather than deep even at Shadow +1, then DR400 is lifting more than the curve can put back, and the rebuild drops to DR200 and gives up the range that the changing light was supposed to need.

ISO 500 is the sweet spot, not a compromise (measured 2026-07-27). DR400 pins the ISO floor at 500, and that is exactly where the X-T5's conversion-gain step sits, so it is the cleanest ISO in that region, not a tax you pay for using DR400. ISO 640 is the expensive next click: roughly 0.35 EV of dynamic range for one third-stop, against about 0.02 EV for the step from 400 to 500. In good light, hold ISO at 500 and control exposure with aperture and shutter. Measured from published read-noise and PDR data, see Dynamic Range & Tone Curve, the contrast controls.

Pro tip: This recipe is built to be left alone. Its brief is light that changes inside a scene, so the test that matters is not one beautiful frame but two frames three minutes apart with the sun in and out. If you find yourself adjusting between them, use the overcast variant as a separate preset rather than chasing the light.

Pro tip: DR400 and Shadow +1 are a pair, not a contradiction. Dynamic Range modes buy highlight room by lifting everything underneath, which is precisely what a crypt cannot afford. The positive Shadow value buys the darkness back and lands roughly where the base would sit unaided. Range from Dynamic Range, density from the curve.

Pro tip: Take the cool shadows as delivered. Almost every recipe in this bank on this base spends a white-balance shift correcting Classic Chrome's documented cool shadow cast. Stone, slate and lead want it, so this one corrects by a single notch of Red and only to keep skin alive. If your frames feel cold, that is the base doing its job.

Shoot JPEG or HEIF. This recipe is HEIF-safe, since Clarity sits at 0.

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

Part 0: the rerun, and a correction to the last audit

The PetaPixel cinematographer series was audited on 2026-08-25 by following the related-article links in the footer of the piece that produced Revival House. That was the wrong method and it produced two wrong conclusions.

Footer links surface related articles, not a series index, and PetaPixel's tag and search pages refuse an automated fetch with a 403. Enumerating the series properly through search turned up four articles the footers never linked, including the one behind this recipe and one about a show whose look is among the most discussed on television.

Two corrections to that audit:

1. The cadence claim was wrong. It was recorded as roughly one article every six months, from a sample of five. With ten known the real spacing is closer to one every two or three months: April 2025, December 2025, May 2026, July 2026, August 2026, plus four across 2024.
2. The yield is better than reported, at three recipes from ten articles rather than two from five, though the ratio is similar and the shape of the finding stands.

What survives unchanged is the useful part. Read these for lighting design and colour palette. When an article is about camera movement, framing, capture format or lens choice, it will not produce a recipe however good the show is.

Part 1: what the source gives

Steve Yedlin, shooting for Rian Johnson on a 2025 mystery set around a rural church and its crypt, with the third instalment of a series they had already made twice together.

That last row and the two Johnson quotes are the brief. Everything in the settings table follows from them.

Part 2: the requirements, ranked

1. Survive a swing. The scene goes from very sunny to very dark and back while the camera runs. A stills recipe cannot be re-tuned between the two states, so it has to hold both. This is the requirement almost nothing else in this bank has.
2. Hold a window. A leaded window with sun behind it against interior stone is close to the widest range in interior photography, and it is the frame the look is built around.
3. Darks that stay dark. Crypt, nave, stone, shadow. A grey basement is the failure.
4. Rich rather than desaturated. Gothic here means stained glass, brass, oxblood and dark polished wood. This is what separates it from noir.
5. Cold. Stone, slate and lead. Warmth would make it candlelit, which is a different genre.

Requirements 2 and 3 are the pair that everything hinges on, and they pull directly against each other: the only way to hold a window is to buy range, and every way of buying range lifts the shadows.

Part 3: the settlement

The resolution is to buy the range with one control and take the darkness back with another, and it is worth stating as a rule because this bank normally forbids it.

• DR400 buys two stops. Fujifilm's dynamic-range modes underexpose and then lift everything below the highlights, which is precisely what requirement 2 needs and precisely what requirement 3 cannot afford.
• Shadow +1 returns the density. Classic Chrome's shadow end is already the hard one, and this bank's tone table records that the base "needs a Shadow lift just to reach normal shadow detail". Adding a notch on top of DR400's two-stop lift lands roughly where the base would sit unaided, which is deep.
• Highlight −1.5 covers the window. The standing note here is that −1 is usually enough on this base because the highlight end is already soft, and that −2 goes limp. A sunlit leaded window is the case that earns the half step between them.

North Light states the opposite rule and is right to: take range from Dynamic Range and leave the shadow end alone, because a lifted shadow is the tell that marks amateur interior work. The difference is what the darkness is doing. There it is a defect. Here it is the subject, and the recipe is allowed to reach for it as long as the reach is deliberate and recorded.

Exposure at −1/3 to 0 completes it. Metering for the stone and letting the window go is what keeps a frame usable when the sun comes back, and requirement 1 means the frame has to survive the state you did not meter for.

Part 4: the base, and the two rejections worth writing down

The shape required is a hard shadow end under a soft highlight end with colour retained. Fujifilm publishes exactly that shape for exactly one simulation: Classic Chrome, hard at the shadow end with soft tonality at the highlight end, and second only to Velvia in overall contrast.

The base's two documented weaknesses both turn into assets on this subject.

Its shadows are natively cool, which almost every recipe in this bank corrects with a positive Red shift. Stone, slate, lead and north light want cool shadows, so this is the rare case where the correction is a single notch and exists only to keep skin from going bloodless inside them. Requirement 5 arrives free.

Its magenta is suppressed, which drops reds toward terracotta and brick rather than letting them stay pure. That is the gothic register precisely: oxblood, dark wood, brass, old velvet. A pure red in a crypt would read as a costume.

What it lacks is richness, and this bank documents the fix rather than leaving it to taste: recipes wanting a true red on this base must add it with Color and the Chrome effects instead of expecting the base to deliver it. Color +2 with Color Chrome Effect Strong is that instruction followed, and the Chrome effect is doing something specific rather than general. Stained glass, brass and candle flame are each a single object spanning an enormous tonal range, and without gradation recovery a window becomes flat blocks of colour, which is how this subject is most often photographed badly.

Eterna, which had the best claim and lost anyway

Requirement 1 is a latitude problem, and Eterna has the most latitude in the lineup by a wide margin: Fujifilm publishes a 12-stop figure and calls it comparable to F-Log. This bank already leans on that fact to explain why the cinema recipes can carry Highlight −2 without producing mush.

It fails everything else at once. Fujifilm states openly that Eterna's muted palette and flat gradation make it look weak on isolated still images, by design, because it is built as an ungraded canvas. Its saturation suppression is described as "an art of omission" so that no single colour stands out. Its blues shift toward cyan.

Rich is the opposite of omission, and a look whose adjective is gothic cannot start from a base engineered to be weak as a still. Buying its latitude would mean spending Color, both Chrome effects and a white balance shift fighting the base's entire design intent, which is a worse position than adding saturation to a base that is already the right shape.

Velvia, briefly

Rich and deep, and hard at both ends. The hard highlight end clips a sunlit window immediately, which loses requirement 2 and with it the frame the whole look exists for.

Part 5: the two articles that gave notes rather than recipes

Both are worth recording, because a rerun that produces only one recipe should still bank what it learned.

Severance, Jessica Lee Gagné. The piece is largely about process: mood-board hallways, tests watched by every department together, a set of rules the team calls "that's so Severance". One idea is concrete and it corroborates another recipe here. The season-two flashback episode was shot on film while the rest of the show is digital, and her reason is a sentence that could sit inside Revival House: she wanted it "to feel like a flashback without actually putting some kind of filter on it because that cheapens it sometimes".

That is Sprenger's "not treated and gamey" said independently by a different cinematographer about a different show. Two practitioners arriving at the same position, that a period or a memory should be reached through capture character rather than through a look laid on afterward, is stronger support for Revival House's central argument than either statement alone. She also notes that some film shots came back soft and were kept, described as "painted soft", which is the same permission Revival House's Sharpness −2 asks for.

The Pitt, Johanna Coelho. Mostly logistics: no lighting on the floor, custom onboard camera lights, 3D-printed diffusion frames magnetted to the ceiling troffers, the crew in scrubs so accidental reflections read as background staff. One detail is genuinely valuable and has nothing to do with a camera setting. She and the production designer "tested over 50 shades of white to find the perfect one", and the reason given is that "it was important to find a white that would work with every skin tone because of the diversity of our cast".

That is a bounce-surface decision made for skin, which is the same problem AP's Portra and AP's Superia solve from the camera side. It belongs in a note rather than a recipe, but it is a reminder that some of what a recipe is asked to fix was decided by whoever chose the wall colour.

Part 6: where it sits in the bank

It is the bank's third recipe from a cinematographer's stated brief, and the only one anywhere here designed for light that changes while you are shooting it.

Part 7: honest limits

• No frames. Creative-design tier and nothing more.
• The brief is about motion and this is a stills recipe. Light changing within a scene is something a moving camera records and a still photograph cannot. What transfers is the requirement it creates, which is that one setting has to serve both states, and that is a real and testable property. It is not the same thing as the reference.
• The robustness claim is the central promise and the least tested. Frame two in the test list, the same view with the sun behind cloud and nothing touched, is the only thing that can confirm it.
• Shadow +1 on top of DR400 is arithmetic done by eye. Two stops of lift against one notch of curve is an approximation, and if the darks still read grey the honest fix is to give up the range rather than to keep adding Shadow.
• No measurement under any of the tonality claims. Hard and soft here are Fujifilm's descriptions, since no characteristic curve exists for a simulation.
• The subject supplies most of the look. A shaft of light through a high window into cold stone is a place and a hour before it is a setting, and the overcast variant exists because the recipe on its own cannot manufacture the shaft.
• No video version, which is again odd for a look read from a film. Movie mode drops both Chrome effects, and Color Chrome Effect Strong on stained glass is central here.

Last updated: 2026-08-25