When you want to show everyday life on an ordinary commuter street being swallowed by a huge disaster in an instant, capturing both the size of the disaster and the people's fear. Use it for climate and safety campaigns, disaster-preparedness training, and blockbuster parodies.
A disaster film starts with the signs that build up before the disaster. An oddity in an everyday object, like ripples on the surface of coffee or a swinging traffic light, becomes the first warning.
To recreate it: Keep inserts short, 1 to 2 seconds, and use them with a very low rumble.
Disaster films mostly follow the disaster through the viewpoint of a specific character and family, so you have to show the threat from the eye level of someone standing in the street for the audience to stand there too.
To recreate it: Direct the tidal wave wide shot as 'eye-level from the street' too, so it doesn't become a picture looking down from above.
If you show the face looking up at the disaster before the disaster itself, the audience gets tense wondering 'what are they looking at'.
To recreate it: In the face shot, have the gaze go off-screen and upward, and reveal the disaster in the next shot.
The tidal wave in The Day After Tomorrow rushes between buildings inside a lidar-scanned 13-block Manhattan model, comparing the size of the whole city with the wave in one frame.
To recreate it: Always put something to compare the size with (a taxi, a person) in the foreground.
Layer a person running away in front and the approaching disaster behind, showing distance and scale in one frame.
To recreate it: Tell it to make the hero sharp and the crowd behind blurry, which shrinks the area where the AI can ruin faces.
When the camera shakes and runs along with them from the moment they start running, the audience too goes from onlooker to someone fleeing.
To recreate it: Keep the shots before the disaster is revealed locked-off, and switch only the shots after it to handheld to create contrast.
The Day After Tomorrow was split among 9 VFX companies, who made 416 VFX shots. Live action and CG made in several places have to be matched to the same sky color and water color to look like one scene.
To recreate it: AI clips also get a different sky color in each shot, so match them all at once at the end.
Cinematic multi-shot sequence, 15 seconds total. Shot 1 (3s): static wide shot of a busy city avenue on an ordinary gray morning, commuters crossing, a young woman in a red raincoat in the foreground. Shot 2 (2s): static insert of coffee trembling in rings inside a paper cup on a newsstand counter. Shot 3 (3s): slow push-in on the woman as she looks up at something enormous off-screen. Shot 4 (3s): static eye-level extreme wide shot of a colossal wall of seawater rising between skyscrapers at the end of the avenue. Shot 5 (4s): handheld shot backing away as she sprints toward camera through a panicked crowd. Consistent woman and red raincoat, cold overcast light, desaturated blue tones.
Disaster films usually run in the order of signs, disaster, and aftermath, and follow a huge event through the viewpoint of a particular character or family. That is why eye-level shots of people standing on the ground, and faces looking up at something, matter as much as the huge picture seen from the sky. For The Day After Tomorrow, they made New York by lidar-scanning a 3D model the size of 13 blocks of Manhattan and covering the buildings with textures from more than 50,000 photos. The water of the tidal wave was split between Digital Domain and Tweak Films depending on the shot.
Wide shot of a busy city avenue on an ordinary gray morning, commuters with coffee cups crossing at a crosswalk, yellow taxis waiting at the light, a young woman in a red raincoat in the foreground. Static camera, flat overcast light, natural colors, calm everyday bustle.
Insert shot of a paper coffee cup on a newsstand counter, the surface of the coffee trembling in concentric rings as a distant low rumble grows. Static camera, macro detail, flat overcast light, natural colors, uneasy quiet.
Close-up of the young woman in the red raincoat slowly lifting her eyes from her phone toward something enormous off-screen, color draining from her face, people behind her freezing in place. Slow push-in, cold overcast light, muted colors.
Eye-level extreme wide shot down the long avenue as a colossal gray wall of seawater rises between the skyscrapers at the far end, tiny figures and taxis in the foreground for scale. Static camera, stormy diffused light, cold desaturated blue tones.
Handheld shot moving backward in front of the young woman in the red raincoat as she sprints toward the camera through a panicked blurred crowd, the giant wave surging into the street far behind her. Shaky handheld, cold stormy light, spray in the air.
Fill the first wide shot with everyday sounds like horns and conversation, and from the cup insert on, let a low rumble push those sounds out. Show the looking-up reaction shot before the wave, then open the wave wide shot with an eyeline match. From the running shots on, move to fast cutting of about 1 second per shot.
AI easily makes the size and speed of huge physical events like tidal waves and earthquakes uneven from shot to shot, so set the scale first with a locked camera and a start frame for disaster shots. Shots of crowds running tend to melt faces and limbs, so keep just the hero clear and use them short, 3 to 4 seconds. Don't show people being hurt or swept away. Show the threat only through scale and sound.
Wikipedia — The Day After Tomorrow · Wikipedia — Disaster film