P—01 / COLLAPSING SEA-CATHEDRAL

Shot brief prompt for MiniMax H3 Max

Two thousand nine hundred characters, fifteen seconds, and not a single cut.

The longest prompt fal published for H3 Max, and the one that proves the point it was post-trained on. It is not a description of a scene, it is a shot brief: named subjects, a colour ratio, a shutter angle, and three timed movements inside one unbroken take. Nothing in it is decoration — every clause is either a constraint the render has to satisfy or a beat it has to arrive at on time.

Steps
One shot
Engine
MiniMax H3 Max
Mode
Text to Video
Duration
15s
Ratio
16:9
Resolution
768P
Audio
Foley + wind, no music
Cost
$1.20

Reference render — not generated on this site. Source: fal.ai

Output reference

A wind-skiff's rigging cuts across a mirrored tidal flat while a burning offshore rig collapses on the horizonVideo
The clip
2930 / 7000
  • Free clipThis run spends credits — the free clip is 5s · 480P · 16:9.

1 FREE CLIP · NO AUDIO

01 — Inside

One prompt, one workflow

The clip beside the verbatim source. Copy it into the console above, change what you need, generate.

Reference output

A wind-skiff's rigging cuts across a mirrored tidal flat while a burning offshore rig collapses on the horizonVideo
The clip
The promptthe clipMiniMax H3 Max · Text to video
2,930 chars

The whole take

Style, cinematography, colour, camera, physics, subjects, location, then three timed beats and a constraints list.

BLACK MIRROR TIDEFLATS / THE COLLAPSING SEA-CATHEDRAL Style: 8K cinematic. Photorealistic, no 3D render, no game engine, no game-cutscene aesthetic. Cinematography: naturalistic master cinematography. Lighting: high-contrast dramatic light. Cold storm-blue dusk across the flats vs. blinding sodium-orange furnace glow pouring from the toppling structure. The wet mirror surface doubles every light source into a second inverted world. Color: 60:30:10, dominant gunmetal blue and wet slate (the tideflats and sky) / secondary sodium amber and rust-red (burning floodlights, corroded steel) / accent magnesium white and cobalt blue (the failing reactor core and its arc-flashes). Camera: physical cine lens. 180 degree shutter motion blur. Continuous handheld operator shake, motivated by the ground concussions rolling through the shallow water. Physics: gravity and inertia respected. Steel trusses buckle and fold under their own mass; spray, ash and salt mist obey atmospheric drag. SUBJECTS: @pilot: a lone figure in a salt-crusted pressure suit crewing a skeletal wind-skiff, a low wheeled land-yacht with a torn carbon sail, hunched at the tiller, cracked gold-tinted visor reflecting the falling structure, throwing twin walls of white spray from its outriggers. @cathedral: a colossal derelict sea-rig, a kilometre of rusted iron lattice and floodlight towers, listing and buckling as it collapses into the shallows, shedding burning platform decks and cobalt arc-flash from its ruptured core. LOCATION: an endless tidal flat of black volcanic glass under two inches of standing water, a perfect mirror to the horizon, broken by rotting mooring pylons and drifting curtains of salt mist, the skyline dominated by the burning collapse. ACTION, ONE CONTINUOUS TAKE, single unbroken handheld tracking shot, NO cuts, 15 seconds. 0:00-0:05 Medium-Wide on @pilot: camera tracks alongside the skiff as it carves between rotting pylons that burst into splinters and spray behind the wheels; the horizon tears open as @cathedral enters frame, leaning, trailing burning deck plates. 0:05-0:10 continuous zoom-in / tilt up: camera glides past the skiff and tilts up through the pylons onto the collapse; a core vessel detonates, the shockwave ripping the standing water into a flat sheet and bending the flare into a horizontal streak across frame. 0:10-0:15 Extreme Close-Up on @cathedral: telephoto-compressed view of the disintegrating rig against the sky, steel plate peeling and glowing white-hot, raining burning debris down toward the camera. CONSTRAINTS: 16:9 anamorphic widescreen. ONE CONTINUOUS SHOT, absolutely NO hard cuts. Telephoto compression during the zoom. No slow-motion. Photoreal throughout. AUDIO (NO MUSIC): deep structural groan of failing steel, the hiss and slap of water under the hull, tearing-metal screech from above, the rigging-hum and canvas crack of the skiff at speed, wind roar across open flats.

4 levers

Make it yours

What is safe to change. Most libraries publish only this list, which is why so many copied prompts come back worse than the original.

  1. 01

    The two @subjects

    @pilot and @cathedral are declared once and referred to by handle afterwards. Swap what they are — a courier and a burning refinery, a diver and a capsizing rig — and the rest of the brief still holds, because every later clause points at the handle rather than at the description.

  2. 02

    The colour ratio

    60:30:10 with three named families is doing more work than any adjective in the prompt. Change the three families and you have graded the shot: swap gunmetal blue for sodium green and rust for bone white and the same collapse reads toxic rather than cold.

  3. 03

    The location

    The mirror flat is chosen so every light source appears twice. Any replacement has to keep that property — wet tarmac, a flooded quarry, a salt pan after rain. A dry location silently halves the light in the frame and the brief never mentions it again.

  4. 04

    The audio line

    AUDIO (NO MUSIC) names five sound sources and forbids a score. H3 Max writes sound in the same pass as the picture, so this line is the mix. Adding "low sustained brass under the collapse" gives you a score; deleting the line entirely gives you whatever it invents.

Three ways to break it

  • Leaving the timestamps in but shortening the run

    The brief schedules three beats across 0:00–0:15. Ask for five seconds and all fifteen seconds of action still have to happen, so the model fits them in by slowing everything down. Cut beats before you cut seconds.

  • Dropping "ONE CONTINUOUS SHOT, absolutely NO hard cuts"

    The three timed segments read as a shot list the moment the constraint is gone, and the model will cut between them. The constraint is what makes the timestamps camera moves instead of edits.

  • Using an @handle you never declared

    The SUBJECTS block is what binds a handle to an appearance. Referring to @skiff in the action without declaring it leaves the model to invent one, and it will invent a different one in each beat.

What it does

What the collapsing sea-cathedral template does

Most prompt libraries teach you to write more adjectives. This one teaches you to write a brief, and the difference is that a brief has sections that answer different questions rather than one paragraph that answers all of them badly.

Look at the order. Style, cinematography, lighting, colour, camera, physics come first — those are the standing rules for the whole clip. Then SUBJECTS declares who is in it, with a handle attached to each. Then LOCATION. Only then does ACTION begin, and by the time it does, every noun it uses has already been defined. That ordering is not stylistic; it is the reason the action section can be five clauses long and still unambiguous.

The physics line is the clause people skip and should not. "Gravity and inertia respected. Steel trusses buckle and fold under their own mass; spray, ash and salt mist obey atmospheric drag" does not describe anything you can see in a single frame. It describes how things are allowed to move, which is exactly the decision a video model makes badly when left to itself. Naming it once at the top governs every frame after.

The timed beats are the part worth stealing even if you never write anything this long. Three segments of five seconds, each with its own framing and its own camera behaviour, and each written as a continuation of the last rather than as a new shot. That is how you get a fifteen-second clip with structure without asking for cuts you do not want.

The CONSTRAINTS block at the end is a negative space list: no hard cuts, no slow-motion, photoreal throughout, telephoto compression during the zoom. Every one of those is a mistake the model would otherwise be free to make. A prompt this specific about what should happen also has to be specific about what should not, because the more you describe, the more places there are for the render to improvise.

One thing to notice about the sound: it is the last block and it is written as a list of sources rather than a mood. Deep structural groan, water under the hull, tearing metal, rigging hum, wind roar. H3 Max generates audio in the same pass as the picture, so a list of sources is a list of things that have to appear in the mix — and "no music" is the only way to stop a score from arriving uninvited.

Everything here runs in text to video, on MiniMax H3 Max.

3 questions

Collapsing sea-cathedral — common questions

  • 01

    Do I have to write prompts this long?

    No. Three of the prompts in this library are under 400 characters and came from the same showcase. Length buys control, and control is only worth paying for when the shot has beats that have to land in a particular order. A single continuous move does not need 2,900 characters.

  • 02

    Does the @handle syntax mean anything to the model?

    It is not special syntax. It works because the brief declares each handle once with a full description and then reuses the same token, which is exactly what any consistent naming would do. The value is in declaring subjects before you use them, not in the @ character.

  • 03

    Why does it forbid slow motion explicitly?

    Because fifteen seconds of scheduled action is close to the point where the model starts stretching to fit. Naming the constraint is cheap insurance. If a run still comes back slow, the fix is fewer beats rather than more words.

Copy it, change three things, run it.

Every character is on this page. 15s at 768P costs $1.20.

Generate free · queue

Written and maintained by the MiniMax H3 Max AI Video Generator editorial teamPublished Last updated