Grade reconstruction report · 2014
Eight frames from The Grand Budapest Hotel, where pink stucco, elevator red, lobby-boy purple and 1968 tangerine are painted, dyed and built rather than graded in, and the grade's job is to carry saturations up to C* 48 without clipping, keep faces alive inside coloured rooms, and let miniatures, night and snow sit in the same storybook as the sets.
The Grand Budapest Hotel is a strongly chromatic and markedly warm film: across its 1,024 photographed shots the mean C*ab is 16.45, individual shots reach C* 63, and the warm/cool balance averages 0.78. Its colours are also the least ambiguous about where they come from. The elevator is red because it was painted red; the uniforms are purple because they were dyed purple; the 1968 lobby is orange because the set was dressed in orange, yellow and tan. Very little of the palette is the kind of global wash a grade lays on, though the grade did isolate and weight it area by area.
That reverses the usual reconstruction question. Instead of asking which grade produces the look, the studies ask what the grade must do so that colour built in front of the camera survives the pipeline: how to hold a wall at C* 48 without clipping a channel, how to keep a face readable when the light on it has bounced off red lacquer, how to let a miniature and a painted night sky belong to the same picture as a live set, and how to make snow blue or pink when the story wants it.
The film changes aspect ratio with its time frames, and the slide lab crops every shot to its own geometry: 1.37:1 for the 1930s and 2.30:1 scope for 1968. The eight studies cover a miniature hotel in daylight and at night, the 1968 concierge desk, the red elevator, Madame D. in her coffin, the prison visiting room, two men running across blue snow, and a sled run that dissolves into pink-white.
The film-level construction
Elevator red, uniform purple, Mendl's pink and the 1968 lobby's tangerine are paint, dye and set dressing. In the red elevator 49% of the frame is one red cluster; at the 1968 concierge desk 64% of pixels are high-chroma. A grade that tried to produce those colours from neutral footage would have to key every surface. The grade here carries them.
Mean C* reaches 48 in the elevator and 44 at the 1968 desk. At those levels the practical ceiling is not an aesthetic choice but the point where the red or blue channel clips and the hue shifts toward orange or magenta. The look depends on colours held just inside that edge, and on skin that has not been dragged along with them.
The 1932 story lives in pink, red and purple with cream and gold; 1968 in orange, ochre and avocado under hard fluorescent-and-tungsten light; night is deep blue with gold windows; snow can be blue or pink-white depending on the scene. The warm/cool balance swings from 0.98 in the pink façade and the 1968 lobby to 0.0 on the blue snowfield.
Miniature exteriors, painted skies and process shots are part of the film's grammar, not hidden effects. The grade's contribution is to put them on the same palette, softness and grain as the live-action sets, so that a model hotel reads as the same place a character walks into.
Resolve blueprint
Normalize camera originals into DaVinci Wide Gamut / DaVinci Intermediate with Resolve Colour Management or a documented input CST, and give miniature, backing and painted elements their own input transforms. Place exposure from faces and from the brightest painted surface that must keep its colour — a pink façade, a white habit, snow — before any palette work.
The film's palette is art-directed, so the graph is built to protect rather than to create: separate branches keep painted surfaces, skin and practical light at independent density, and the most important global operation is a gamut-aware saturation limit that lets reds and purples be as strong as the set made them without clipping a channel. Suggested settings are starting ranges for suitable log or raw footage, never recovered originals.
Normalize camera, miniature and backing sources correctly before judging painted colour.
Place exposure from faces and from the brightest painted surface that must keep its colour, and balance from real neutrals rather than from the dominant set colour.
Keep contrast moderate and the shoulder soft, so pastel and saturated surfaces read as painted objects rather than lit ones.
Let painted colour be as strong as the set made it while keeping every channel inside its range, so reds stay red and purples stay purple.
Hold each era's family — pink, red and purple in 1932; orange, ochre and avocado in 1968 — at its own density.
Keep faces readable and believable when the light on them has bounced off red, pink or orange surfaces.
Hold lamps, lit windows and neon as warm sources against blue night or cool fluorescent interiors.
Place snow as a field whose colour carries the scene — blue at dusk, pink-white in the chase — at a density that keeps texture.
Guide attention with soft windows through symmetrical, busy frames: lift the faces that lead, quiet large bright surfaces.
Keep colour believable at the ends of the scale: rich colour in the mids, slightly less in deep shadow and near-white.
Put miniatures, painted backings and process plates on the same palette, softness and grain as the live sets.
Keep the film's gentle, slightly soft texture without blurring the fine pattern of wallpaper, uniforms and props.
Render the scene-referred grade to the calibrated delivery while keeping display-specific trims out of the shared look.
8 reconstruction studies
Each study begins with a literal visible description and tests it against the active-picture measurements of the verified midpoint. Metrics describe the whole frame and cannot identify a wall, a uniform or a face without the visual inspection recorded in the prose. High-chroma share is the fraction of pixels above the lab's chroma threshold; in this film it reaches 64% of a frame.
Build the base with every optional node bypassed. Add only the named shot operations, check keys and windows at full resolution, and inspect moving footage before approving qualifiers, tracked windows, glow or element integration.
01 · Study I · 1932 · The hotel · shot 17 · 01:03:24:08
The hotel stands frontally in pink stucco with pale trim and blue-grey domes, on a stone terrace with three arches; a blue-roofed funicular station sits at right, and snowy mauve pines rise behind.
The frame is high-key and almost entirely pink: mean L* 61.2, 46% of the picture in highlights and none in shadow; the five palette clusters run from pale #EBB3AE through #CB939B and #B27176 to #633F4A, all rose. Mean a* is +22.1, b* only +5.8, and the warm/cool balance is 0.98. Even the snowy forest behind is mauve-pink rather than white or blue. The blue-grey domes and the funicular's pale blue roof are the only cool colours, and they are small.
Everything is sharp from front to back, the lighting is flat and even, and there is no atmospheric haze between the façade and the hillside. The windows are small and regular; a few glow warm. The retaining wall's stone and the snow in its arches are finely detailed.
The construction is a miniature: a scale model of the hotel photographed in front of a painted or modelled hillside. The pink is the model's paint. The grade's contribution is to hold that pink below clipping in a high-key frame, keep the domes blue, and give the model the same softness and palette as the live-action hotel interiors.
Expose for the façade's lightest pink and roll the shoulder so it keeps colour. Resist adding haze or depth; the flatness is the storybook.
Can regular footage get there?
Settling on a useful reference rather than a generally reproducible candidate. The grade is simple; the picture is a model and a painted environment that no grade can supply from ordinary footage of a building.
Composite and background read
Uniform depth of field across façade and hillside, absent atmospheric perspective, regular scale detail, a hillside of identical trees and a funicular structure rendered as a model all point to a constructed miniature. The film's hotel exteriors are widely understood to be miniatures.
This is a reading of construction, not a production credit; the split between model, painting and digital work cannot be determined from a still.
Shot-specific Resolve pass
Match the model's softness and grain to adjacent live-action shots.
Why: The model must read as the same place.
Watch: A miniature sharper than the live sets.
Roll the top so the façade keeps colour at L* 60–75.
Why: 46% of the frame is highlight.
Watch: Pink clipping to white on the lit trim.
Narrow Hue vs Hue to keep the façade, hillside and snow in one rose family.
Why: The frame is one colour at different values.
Watch: The hillside drifting mauve-blue.
Lum vs Sat check on the domes and funicular roof.
Why: The only cool accents must stay cool.
Watch: Blue roofs going grey.
02 · Study II · 1968 · The concierge desk · shot 32 · 01:04:32:13
Between ochre marble columns, a concierge in a purple jacket leans on a yellow-green desk marked RING FOR SERVICE beside a man in tweed with a pipe; behind them an orange wall of key slots frames a portrait.
This frame has the highest share of saturated pixels in the report: 64% high-chroma, mean C* 43.7, mean b* +34.2. Orange #B36218 is the largest cluster at 30%, then rust #7C2706, brown #84482D and a pale peach #D08B62; the warm/cool balance is 0.98. The purple jacket is the only cool object, and it is central.
Light is hard and even, from above — a fixture is visible at the top — with shadows under the counter and inside the key slots. The yellow-green counter is a third colour family, avocado, distinct from the orange wall. The two faces are warm but not orange; they sit lighter than the wall and less saturated.
A 1968 set dressed in orange, ochre and avocado, a purple costume and top light supply the colour. The grade holds the orange below the red channel's clip, keeps the avocado distinct from the orange, and protects the faces from the wall's bounce.
Set exposure from the faces and the RING FOR SERVICE sign's white letters. Limit saturation by channel, not globally; the purple jacket must remain strong.
Can regular footage get there?
A dressed 1968 set photographed with even top light reaches this frame with exposure, a gamut-aware saturation ceiling and skin protection.
Composite and background read
Shadows in the key slots, the actors' contact with the counter and the columns' occlusion are coherent with a built set. Portrait and signage could be inserted, but nothing requires it.
Invisible cleanup or insert replacement cannot be ruled out from a still.
Shot-specific Resolve pass
Soft gamut compression on orange and rust; watch the red channel.
Why: 64% of pixels are high-chroma.
Watch: Orange flattening into one tone.
Narrow Hue vs Hue so the counter stays yellow-green, not orange.
Why: Three families, not two.
Watch: The counter drifting into the wall's orange.
Face keys; reduce bounce saturation 10–15%.
Why: The faces must sit lighter and less saturated than the wall.
Watch: Skin going orange with the wall.
Even windows on both columns if any are used.
Why: The composition is symmetrical.
Watch: One column darker than the other.
03 · Study III · 1932 · The elevator · shot 108 · 01:10:57:21
Four people fill a red lacquered elevator: an operator in purple with gold buttons, a lobby boy with a leather case, Madame D. in a red coat and fur hat, and M. Gustave in purple; a brass panel is at left.
Red #AE1B02 is 49% of the frame, the most saturated dominant cluster in the report; mean C* is 48.4, mean a* +37.0, and 60% of pixels are high-chroma. Purple uniforms and dark purple shadows make the second family, and there is a near-black cluster at 22% in the uniforms' folds. The walls are lacquered and the panels show soft reflections.
The four faces are the grade's problem. They are lit inside a red box and every face carries some of it; yet they read as skin rather than as red, lighter and less chromatic than the walls. Madame D.'s pale made-up face and red lips are the clearest case: the lips are more saturated than the skin, and both are inside the same red light.
A red lacquered set, purple costumes and a soft top key supply the construction. The grade's work is the ceiling — holding the red wall inside the red channel — and the faces, where some of the bounce is removed without making skin look daylight-lit.
Place exposure from the faces, cap the red, then key skin and reduce only its excess red. Leave the lips and coat alone.
Can regular footage get there?
A red set, purple costumes and soft top light reach this frame; the grade protects rather than creates.
Composite and background read
Reflections in the lacquer, occlusion between the four figures and consistent top light point to a practical set.
Invisible cleanup cannot be ruled out from a still.
Shot-specific Resolve pass
Soft gamut compression on red; watch the red channel on the parade.
Why: 49% of the frame is one saturated red.
Watch: The wall turning orange and flat.
Face keys; reduce saturation 15–20% and move hue a few degrees toward skin.
Why: Faces must read as skin inside red light.
Watch: Neutralized faces that look pasted in.
Narrow hue range on uniforms; prevent drift toward blue or red.
Why: Purple is the second family.
Watch: Purple shadows going blue.
Lum vs Sat in the toe.
Why: 22% of the frame is near-black purple.
Watch: Coloured noise in the uniforms.
04 · Study IV · 1932 · The hotel at night · shot 179 · 01:17:40:01
The hotel at night: its façade nearly black against a deep blue forest, dozens of windows glowing gold, the lettered sign on the roof lit, and small lamps along the terrace below.
The same façade as the daylight study, now nearly black: mean L* 5.8 with 96% of the picture in shadow and 65% in deep shadow. The blue forest behind, #111735 at 22%, carries the ambient; the windows, sign and lamps are the gold accents, #B66B26 and #562F1C, together about 5% of the frame. The warm/cool balance is 0.11 — blue, but not exclusively.
The windows are lit unevenly: some bright gold, some pink, some dark, as in a real building. The sign on the roof glows more intensely than any window. The forest's snowy trees remain readable as blue shapes, a step above the black façade.
The construction is again a miniature, now lit from inside with small practicals and against a blue night backing. The grade holds the gold windows below clipping, keeps the blue ambient dense rather than bright, and places the façade just above black so its architecture remains faintly visible.
Grade the night as a relationship between two colours: gold sources and blue ambient. Lift nothing that is not lit.
Can regular footage get there?
Settling on a useful reference rather than a generally reproducible candidate. The miniature, its internal lighting and the backing are the image; a grade can place night but cannot supply them.
Composite and background read
The same model hotel as in daylight, with uniformly scaled windows, identical trees and no atmospheric depth, lit by many small interior sources.
This is a reading of construction, not a production credit; the split between model, painting and digital work cannot be determined from a still.
Shot-specific Resolve pass
Key the windows and sign; lower core density, keep the gold.
Why: The windows are the scene's only light.
Watch: Windows clipping to white rectangles.
Toe so the façade stays just above black with its architecture visible.
Why: The hotel must still be the hotel.
Watch: The façade disappearing into black.
Match black and grain between the miniature and the forest backing.
Why: They must read as one night.
Watch: A cleaner black in the backing than in the model.
Lum vs Sat in the toe.
Why: Blue noise in black is very visible.
Watch: Blue speckle in the façade.
05 · Study V · 1932 · The coffin · shot 304 · 01:24:12:12
From directly above, Madame D. lies in a coffin lined with ruched pink satin, white hair piled high, lips red, eyes closed, white flowers in her folded hands; greenery and white roses surround her in the dark.
A frame dark at its edges and warm at its centre: mean L* 13.5 with 73% in shadow and 46% in deep shadow, yet the face, hair, dress and satin form a lit pool in the middle. The palette runs from black #070304 through dark browns #311714 and #5A271D to a pale warm #BC7A56 at 6%, and mean a* +9.3, b* +8.2: the pool is warm-pink, not white.
The face is heavily made up — pale foundation, age spots, red lips — and lit softly from above and slightly in front, so every texture is visible. The white dress and hair are the brightest surfaces but stay below clipping; the white roses at the edges are a step darker, pulled into the wreath.
Makeup, costume, pink satin, a dark floral surround and a soft overhead key supply the image. The grade holds the pale face and white hair just below white with their texture, keeps the satin pink rather than peach, and keeps the red lips saturated without letting the face go red.
Place the face first. Let the surround fall into darkness; it frames the pool and carries no information the frame needs.
Can regular footage get there?
A soft overhead key on a made-up performer in a satin-lined coffin reaches this frame directly with exposure, a soft shoulder and skin work.
Composite and background read
Occlusion between flowers, hands and satin, and consistent falloff, point to a single practical overhead shot.
Invisible cleanup or beauty work cannot be ruled out from a still.
Shot-specific Resolve pass
Soft shoulder so white hair and lace keep texture.
Why: They are the brightest surfaces in the pool.
Watch: Hair clipping to a white shape.
Face key; keep foundation pale and age spots visible; protect lips.
Why: The makeup is the character.
Watch: Softening the skin texture.
Narrow hue range on the satin.
Why: The lining is part of the palette.
Watch: Satin drifting orange.
Soft vignette following the real falloff.
Why: The pool is the subject.
Watch: A visible vignette shape on the greenery.
06 · Study VI · 1932 · The prison · shot 401 · 01:36:58:21
M. Gustave, in a striped prison uniform and cap, drinks from a glass behind a steel-framed mesh screen; a guard stands before barred cells receding into cold light, and a warm wooden counter runs along the bottom.
The film turns cold here. Mean a* is −4.2 and b* −9.3, the warm/cool balance 0.12, and the palette is blue-grey: #527788 at 30%, deep navy #0B1E2E and #254358, a pale grey-green #B4BBAF at 18% (the uniform and the light in the cells) and a warm grey-brown #8A716C at 6% — the counter and the face. Mean L* 39.7, σL* 23.4: a frame with a full tonal range, unlike most of this report.
The mesh screen overlays everything in the foreground, a fine grid that must survive compression and sharpening. The cell block behind recedes into pale light from above. The face is lit warm-neutral and sits in front of the cool field, the same separation the red elevator makes in reverse.
A steel-and-concrete set with cool practical top light, a striped costume and a wooden counter supply the colour; the scene's cold is designed into the location and the light. The grade holds the blue-grey field, keeps the uniform's cream from going blue, and keeps the face warm relative to the room.
Place the face first, then let the room be cold around it. Protect the mesh from sharpening artefacts.
Can regular footage get there?
A steel set with cool top light and a warm face light reaches this frame with exposure, a cool ambient and skin protection.
Composite and background read
Mesh occlusion over the actor, the guard's position in the depth and the consistent top light are coherent with a built set. The upper cell block could be extended, but nothing here requires it.
Set extension beyond the visible tiers cannot be ruled out from a still.
Shot-specific Resolve pass
Neutral from the cream stripes; let the steel stay blue.
Why: The uniform is the frame's neutral reference.
Watch: Balancing from the steel and warming the whole room.
Face key; hold warmth against the blue field.
Why: The face must separate by temperature.
Watch: Blue spill on the face left uncorrected.
Key the cell block's light; keep it pale grey-green.
Why: The cold is the scene's mood.
Watch: Cyan saturation in the top light.
No sharpening; fine grain only.
Why: The mesh is a fine regular pattern.
Watch: Moiré on the mesh.
07 · Study VII · 1932 · The escape · shot 699 · 02:02:45:21
Two small figures run across a wide snowfield from left to right under a heavy sky; a dark wood lines the horizon and a single bare tree stands at left, all in one deep blue.
Almost the whole frame is one colour: #184D7B is 74% of the picture, with darker blues making up the rest, mean b* −30.3 and a* +0.5, warm/cool balance 0.0. The frame is also almost flat: σL* only 5.9, mean L* 29.2, no highlights and hardly any deep shadow. Snow, sky and forest are distinguished only by small steps of density.
The snow is darker than a daylight snowfield and bluer than a dusk one; it reads as twilight or a stylized night. The two runners are soft with motion and dark, the only strong value contrast in the frame.
An overcast or twilight snowfield supplies the texture and scale; the depth of blue and the lowering of exposure are plausibly the grade's, in the tradition of day-for-dusk. The grade reduces exposure, compresses contrast and moves the whole scene into one blue, while keeping the runners dark and the horizon readable.
Lower exposure before adding blue; blue added to a bright snowfield reads as a filter. Keep contrast low and let the runners carry the value.
Can regular footage get there?
Achievable from overcast or twilight footage with low contrast and a snowfield without hard sunlight; a sunny capture with shadows on the snow cannot be flattened into this without losing texture.
Composite and background read
The runners' motion blur, their tracks and the soft horizon are coherent with location photography. Sky replacement or darkening is common in such scenes, but the sky here is featureless and consistent with the snow.
A replaced or darkened sky cannot be ruled out from a still.
Shot-specific Resolve pass
Offset down 0.5–1 stop from a daylight placement.
Why: Blue added to bright snow reads as a filter.
Watch: Blue noise in lifted shadows.
Key snow and sky; move hue toward the measured blue, raise saturation moderately.
Why: The scene is one colour.
Watch: Sky and snow separating into two blues.
Reduce contrast so σL* stays low; keep the runners dark.
Why: The runners carry the only value contrast.
Watch: Crushing the runners into black.
Lum vs Sat in the toe.
Why: Saturated blue in the dark areas turns noisy.
Watch: Blue blocking in the forest.
08 · Study VIII · 1932 · The sled · shot 862 · 02:18:04:02
Two men in white hooded coats and flat caps lie on a sled rushing toward the camera, their faces sharp and intent, while the snow around them streaks into pink-white blur.
Mean L* 67.6 with 71% of the picture in highlights, the brightest frame in the report; the palette's largest cluster is pale pink #EBC9C3 at 48%. It is warm but barely coloured: mean C* 10.5, a* +8.0, warm/cool balance 0.85. The snow is pink-white rather than white or blue, the opposite of the escape across the snowfield.
The two faces are sharp and exposed normally while everything else streaks into blur, a separation between foreground and background that is typical of a process shot. The white coats sit just below the snow in value, and the caps and sled are the only darks.
The construction is plausibly a process or composite shot: actors on a sled rig in front of a moving background plate or screen, with the background blurred to sell speed. The grade holds the high-key field below white with a pink cast, keeps the faces readable and matches foreground and plate in contrast and grain.
Keep the snow pink-white, not blue. Place the faces first, then roll the shoulder so the background streaks still carry texture.
Can regular footage get there?
Reachable with a controlled foreground on a rig and a moving background plate or screen; a real sled run with a camera rig would give different blur and exposure behaviour. The grade itself is simple once the elements exist.
Composite and background read
Sharp, steady faces against a background of heavy streaking, flat even light on the actors and an absence of snow spray around the sled point to a foreground and background photographed separately. The chase is widely described as combining miniature and process work, though this frame alone is not decisive.
A still cannot distinguish rear projection, compositing or an in-camera rig; moving footage or production records would be needed.
Shot-specific Resolve pass
Match black level, contrast and grain between actors and background.
Why: They were photographed separately.
Watch: A foreground with deeper blacks than the plate.
Soft shoulder keeping texture in the streaks.
Why: 71% of the frame is highlight.
Watch: The background clipping flat white.
Hold the snow's warm cast.
Why: The chase's snow is warm by design.
Watch: Snow drifting toward blue or neutral.
Face keys; keep skin readable against the bright field.
Why: Faces must separate from a field brighter than they are.
Watch: Faces going grey against the snow.
May I count your visit? Google Analytics would tell me which pages get read. Nothing loads from Google unless you say yes, and you can change your mind from the footer.