Vertigo (1958) — bell tower stairwell dolly zoom
See our character’s iconic-scene parody →A hero who is afraid of heights climbs the bell tower stairs and looks down. At that moment the stairwell stretches and seems to suck him in, which makes you dizzy. Borrow it when you need that one "gasp" shot, where a character is overwhelmed by fear or shock.
The moment Scottie looks down over the railing, the bottom of the stairwell seems to be sucked away. According to Wikipedia, a full-size set made this hard to shoot, so they built a model of the stairwell, laid it on its side, and shot the dolly zoom horizontally.
To recreate it: Give a single start frame of an empty stairwell, instruct only 'dolly zoom, vertigo effect', and keep it short, 2–3 seconds.
The dolly zoom shot is what Scottie sees. Wikipedia explains that Hitchcock used this effect to emphasize the height of the tower and Scottie's dizziness.
To recreate it: Start the prompt with 'POV looking straight down' and keep the character's body out of the frame.
The lines of the square stairwell converge on the floor at the center of the frame, so the composition drops the eye to a single point.
To recreate it: Write the position of the vanishing point, as in 'vanishing point in the center', so the perspective change is clear when you zoom.
Scottie's face looking down is followed by the stairwell he is looking at, so the audience follows his gaze exactly. In the clip, from 4:40 to 5:20, the face and the stairwell seem to alternate.
To recreate it: In the face shot, write that the gaze points downward, and make the next shot always a top-down angle.
After seeing the stairwell, we return to Scottie's frightened face.
To recreate it: For the reaction shot, lock the camera and instruct just one change of expression; the face distorts less.
Inside the tower it is dark except for the light from a small window, so the bottom of the stairs looks even deeper.
To recreate it: Write 'dim, single small window light, deep shadows', or the AI will fill the frame with bright light.
According to Wikipedia, the film was shot on VistaVision cameras, and Technicolor handled the processing and printing.
To recreate it: Add the same texture phrase, such as 'VistaVision, 1950s color film look', to every shot to match the tone.
Cinematic multi-shot sequence, 10 seconds total. Shot 1 (3s): full shot of a middle-aged man in a gray suit climbing a narrow wooden staircase inside an old stone bell tower, camera tilting up as he passes; Shot 2 (2s): locked-off close-up of him gripping the railing, freezing and looking down, eyes widening; Shot 3 (3s): first-person high-angle POV down the deep square stairwell with a dolly zoom, the bottom staying the same size while the walls stretch away; Shot 4 (2s): slow push-in on his frightened eyes as he presses against the stone wall. Consistent man and tower, dim window light, deep shadows, 35mm film grain, muted 1950s color, fear of heights.
If the camera moves back while zooming in (or the opposite), the subject stays the same size but the perspective of the background changes. The eye sees the subject as "not moving", but the space is moving, so the audience physically feels the dizziness of the hero looking down.
A middle-aged man in a gray suit climbs a narrow wooden staircase inside an old stone bell tower, gripping the handrail and breathing hard; the camera tilts up slowly as he passes; dim window light, dust in the air, 35mm film grain, muted 1950s color.
Close-up of the same man in a gray suit gripping the wooden railing; he freezes and looks straight down over the edge, eyes widening, sweat on his brow; locked-off camera, 50mm lens, hard side light from a small window, deep shadows, 35mm film grain.
POV looking straight down a deep square stairwell inside an old bell tower, wooden steps zigzagging down to a stone floor far below; dolly zoom, vertigo effect, the bottom stays the same size while the walls stretch away; dim window light, 35mm film grain.
Extreme close-up of the man's frightened eyes as he pulls back from the railing and presses himself against the stone wall; slow push-in, harsh side light from a small window, shallow depth of field, sweat on his skin, 35mm film grain, muted 1950s color.
Hard-cut in this order, with eyeline matches: the face looking down (Shot 2), the stairwell he sees (Shot 3), then the face again (Shot 4). Use the dolly zoom shot for only 2–3 seconds, and lay a low, heartbeat-like sound effect under it as a J-cut so it is heard just before Shot 3.
Leave only the dolly zoom shot to the AI (the Higgsfield 'Dolly Zoom' preset, or the phrase 'dolly zoom, vertigo effect'). Even the original was hard to shoot on a full-size set, so they laid a model of the stairwell on its side and filmed it horizontally. First make a start-frame image of a deep stairwell with no people, then confirm in a 480p draft that the perspective stretches before you commit.