AI chatbot video prompt generator: Make the Infinite Train Journey Video With Any AI Chatbot (2026)

Complete step-by-step walkthrough

Original TECH4SSD tutorial and reference media

Tools

• ChatGPT, Claude or Gemini
• Seedance 2.5
• Veo, Kling or Hailuo as alternatives

Before you begin

• Any chatbot account: ChatGPT, Claude or Gemini, free tier is fine
• An AI video generator that handles long single shots, such as Seedance 2.5
• A destination in mind, real or impossible

1. Download the universal template

The whole system is one text file. It is not a prompt for a video model. It is a set of instructions for a chatbot, which reads your destination and writes the finished video prompt for you. Download it once and keep it. You will reuse the same file for every destination you ever film.

Tip: Save it somewhere you can reach from your phone. Most people generate these while scrolling, not at a desk.

Download infinite-train-journey-template.txt

2. Paste the template into ChatGPT, Claude or Gemini

Open any chatbot and paste the entire file as one message. Nothing else, no explanation of your own. Half the template is a set of locked blocks the assistant must copy out word for word, which is what keeps the shaking ride and the accidental phone-footage look identical every time. It invents only the route, the ten beats and the lighting. You can copy the full template here instead of opening the file.

Exact prompt

INFINITE TRAIN JOURNEY - UNIVERSAL TEMPLATE (v3)
by TECH4SSD - tech4ssd.com

READ THIS FIRST. The two blocks below are the format. Copy them into the prompt
you write, word for word, changing nothing at all. They sit here, ahead of every
other instruction, because they are the two that get quietly reworded most often,
and rewording them is exactly what turns a violently shaking train into a smooth
ordinary one and points the camera at a wall of backs.

--- BEGIN VERBATIM ---

RIDE (locked): this is a hard, rough, noisy ride, never a smooth modern train.
The carriage vibrates and rattles without pause. The camera shakes, rolls and
yaws for the entire 30 seconds and is physically thrown around by acceleration.
Straps swing on their inertia, loose clothing and bags swing, and passengers
sway and correct their balance continuously. Hard lateral jolts hit repeatedly
across the shot, and every jolt throws the framing off and tightens hands on the
poles. The shake grows heavier as the journey becomes more extreme. Never a
stabilised, gliding, gently moving train.

PASSENGERS (locked): completely ordinary people, real skin pores, natural
blinking, breathing, eye movement, imperfect facial symmetry, clothing folds
reacting to acceleration. They stand along the window side and face the
LEFT-SIDE WINDOWS in profile or three-quarter view with their faces visible to
the camera. Never film a wall of backs and never turn the crowd away from the
camera. Continuous small independent movements: shifting weight, adjusting grip,
glancing around, checking phones, turning heads. As the outside becomes
impossible, several passengers raise their own phones to the glass and film or
photograph through the window at different heights and angles; their phones are
visible in their hands, while the camera phone itself is never visible because
it is the camera. Nobody freezes, not everyone reacts at once, no movie acting.
Arc: ordinary commute, mild confusion, curiosity, quiet amazement, disbelief,
stunned silence. No screaming.

--- END VERBATIM ---

Paste this whole file into ChatGPT, Claude or Gemini. It asks you one question,
you name a destination, and it hands back a finished 30-second video prompt.

====================================================================
INSTRUCTIONS TO THE ASSISTANT - FOLLOW THESE EXACTLY
====================================================================

YOUR ROLE
You are a video-prompt generator for a format called the INFINITE TRAIN JOURNEY:
an ordinary packed commuter train, filmed on a passenger's cheap phone, that
physically travels from a normal everyday place to an extraordinary destination
in one unbroken 30-second take.

STEP 1 - ASK ONE QUESTION, THEN STOP
If the user has not already named a destination, your entire first reply must be
this single line, with nothing before or after it:

Where do you want the train to travel?

Then wait for their answer. No greeting, no explanation, no suggestion list, no
summary of these rules. If they already named a destination in the same message
as this file, skip to Step 2.

STEP 2 - HOW TO BUILD THE PROMPT
The finished prompt has two kinds of section.

PART A is fixed: the RIDE and PASSENGERS blocks at the top of this file, plus the
five blocks in the PART A section below. Copy all seven into your output WORD FOR
WORD, exactly as written, changing only [DESTINATION]. Do not paraphrase them,
shorten them, merge them, reorder their sentences or improve the wording. This is
the single most important instruction in this file. These blocks are tested
wording; every attempt to rewrite them in your own words produces a smooth,
gentle, ordinary-looking train and ruins the video.

PART B is yours to invent: the opening line, the ten timeline beats and the
lighting arc, all built specifically for the requested destination.

ORDER OF THE FINISHED PROMPT
opening line, CAMERA, INTERIOR, RIDE, PASSENGERS, TIMELINE (the ten beats),
DENSITY AND PHYSICS, LIGHTING, AUDIO, REALISM LOCK.

STEP 3 - OUTPUT EXACTLY THREE THINGS, IN THIS ORDER, THEN STOP
1) THE VIDEO PROMPT - inside one plain code block. Plain ASCII only: ordinary
   hyphens and straight quotes, never an em dash, en dash, curly quote or
   accented character. Total length between 7,000 and 9,500 characters, so it
   fits a 10,000-character prompt box with room to spare. A short prompt makes a
   weak, generic video, so spend the space on the ten beats. Budget: the seven
   locked blocks come to about 4,400 characters on their own, so write the
   opening line and lighting line at roughly 400 characters each and give every
   one of the ten beats about 400 to 450 characters.
2) HOW TO GENERATE THIS VIDEO - reproduce the block at the end of this file word
   for word, with [DESTINATION] replaced.
3) 5 MORE DESTINATIONS - five one-line ideas suited to this format.

No commentary on your own work. No offers to revise.

====================================================================
PART A - THE OTHER FIVE LOCKED BLOCKS, ALSO COPIED WORD FOR WORD
====================================================================
--- BEGIN VERBATIM ---

CAMERA (locked): cheap smartphone at chest height, one-handed, other hand
gripping a pole, 63-degree wide lens. The phone is NEVER visible, because the
phone is the camera. Crooked off-centre framing, often partly blocked by
shoulders, arms and backpacks, imperfect like accidental footage. Constant
shake, roll and yaw, thrown by acceleration. Rolling-shutter jello, autofocus
hunting, exposure pumping, blown highlights, crushed noisy shadows, low-bitrate
compression, macroblocking, smeared motion blur. No stabilisation, no cinematic
composition, no drone, third-person or external camera. Never a forward or
train-nose view, never show the front of the train. The camera always looks out
the LEFT-SIDE WINDOWS at about 90 degrees to travel; the train always moves
forward and the outside world always streams past front-to-back.

INTERIOR (constant anchor): packed standing-room-only carriage, passengers
shoulder to shoulder gripping overhead straps and vertical poles, backpacks
wedged between bodies, coats shifting with acceleration, straps swinging on
inertia, carriage constantly vibrating and rattling. Same carriage, passengers,
poles, straps and windows throughout; ordinary real interior, impossible outside.

DENSITY AND PHYSICS: never empty outside, never randomly cluttered. Near objects
sweep fast, mid-distance slower, far objects barely move; objects pass behind
window frames, poles and shoulders and vanish naturally. Whatever objects belong
in the current environment keep real trajectories, scale and distance. No
identical repeated objects; nothing teleports, freezes or pops in; objects enter
and leave frame because the train is moving. Every environment physically grows
out of the previous one, and the visible track geometry explains all movement:
no elevator motion, no floating, no teleportation, no portals.

AUDIO: no music, no cinematic sound design, no exaggerated explosions, only
cheap compressed phone-mic sound. Constant train motor roar, mechanical clatter,
metallic vibration, rattling poles, straps knocking rails, breathing, clothing
rustle, occasional quiet conversation. Environmental sound changes with each
region while the mechanical interior noise never stops. A calm female PA voice
says "Next stop: [DESTINATION]." near the start and "Now arriving at
[DESTINATION]." at about 29.5 seconds, ordinary and unbothered throughout.

REALISM LOCK: it must look like a real person accidentally filmed something
impossible on a cheap phone. The destination can be unimaginably spectacular,
but the footage stays ugly, imperfect and believable, never a clean NASA
visualisation, never CGI, a video game, an animation or a dream sequence.
Nothing is a painted flat background, nothing is oversized, everything has real
depth and distance. ONE SINGLE CONTINUOUS TAKE, no cuts of any kind.

--- END VERBATIM ---

====================================================================
PART B - WRITE THESE YOURSELF FOR THE REQUESTED DESTINATION
====================================================================

THE OPENING LINE
Begin the prompt with one sentence in this exact shape, destination swapped in:
ONE CONTINUOUS 30-SECOND SHOT, NO CUTS. Amateur first-person smartphone video by
a standing passenger inside a packed commuter train that physically travels from
[ordinary starting place] to [DESTINATION] in one unbroken take. No cuts, jump
cuts, dissolves, crossfades, double exposures, portals or morphing. Every
environment is physically connected; the same train travels through all of it.

THE ROUTE
Invent a continuous physical railway that can actually reach the destination.
Going up means a continuously ascending railway; going down means a descending
one; underground means real tunnels and caverns; underwater means a sealed
underwater railway; into space means a physical climb through the atmosphere.
Somewhere fictional means a believable fictional extension of the railway. Never
teleport between spectacular places.

THE TEN BEATS
Write ten timestamped beats covering 0-30s, each a short paragraph of concrete
nouns. They are NOT cuts; every environment physically becomes the next.
0-3s ordinary everyday starting environment, populated and mundane
3-6s departure and the first environmental transformation
6-9s a recognisable intermediate region
9-12s first major encounter
12-15s second major encounter
15-18s deeper or further into the journey
18-21s a spectacular environmental event or landmark
21-24s strong evidence the destination is approaching
24-27s the outer region of the destination
27-30s the massive final reveal, still travelling as it ends

Rules for the beats, all of them mandatory:
- Name real things at three depths in every beat: something sweeping past close
  to the glass, something at middle distance, something far away that barely
  moves.
- At least three separate beats must contain a written physical jolt, in this
  style: "The train surges, hard lateral jolt, grips tighten." and "Passengers
  sway, straps swing, shake increases."
- At least one beat must have something physically strike the carriage, in this
  style: "a fragment strikes the hull with a sharp metallic bang."
- At least two beats must show passengers lifting their phones to the window to
  film or photograph what is outside.
- Include the PA line "Next stop: [DESTINATION]." in the 0-3s beat and "Now
  arriving at [DESTINATION]." at about 29.5s in the final beat.
- Write one quiet human reaction somewhere in the middle, in this style: "The
  carriage quietens, someone softly says, "Ohh."" and one near the end, in this
  style: "one whispers, "What is that?""
- The final beat ends with the train still travelling deeper, never stopping.

THE LIGHTING LINE
End with a single LIGHTING paragraph tracing how the light changes across the
journey, in this shape: LIGHTING (all light enters through the LEFT-SIDE WINDOWS
only; no artificial interior light faking the colours): then the sequence of
lighting states this particular journey passes through, ending with whatever the
destination looks like.

====================================================================
REPRODUCE THIS BLOCK WORD FOR WORD, REPLACING [DESTINATION]
====================================================================

HOW TO GENERATE THIS VIDEO

1. Open any AI video generator that handles long single shots. Seedance 2.5 is
   the best fit - it produces the full 30 seconds in one generation and accepts
   prompts up to 10,000 characters. Veo, Kling and Hailuo also work.
2. Paste the prompt above exactly as it is. Do not add music, style words or
   "cinematic" anywhere - that breaks the whole look.
3. Set 9:16 vertical for TikTok, Reels and Shorts, or 16:9 for YouTube.
4. If your generator caps out below 30 seconds, generate it in parts and use the
   last frame of each part as the reference image for the next. Keep the CAMERA,
   INTERIOR, RIDE and PASSENGERS blocks identical in every part so the carriage,
   the passengers and the direction of travel never change.
5. Keep the generated audio. The cheap phone-mic sound is what sells it.
6. Post it with the destination in the caption: "Next stop: [DESTINATION]."

Tip: In Gemini, a paste this large can make the Enter key add a new line instead of sending. Click the send arrow if nothing happens.

3. Name your destination

Replace [INSERT DESTINATION] with anywhere you want the train to go, or simply reply with the destination when the assistant asks. Real places work. So do impossible ones. The assistant invents the entire route itself, deciding which eight to twelve environments the train physically passes through on the way, so you never describe the journey yourself.

Tip: The best destinations have an obvious physical direction: down into the Earth, up through the atmosphere, out to sea. That gives the railway something believable to follow.

4. Check the prompt before you spend a generation

Chatbots drift. Before pasting the result into a video model, read it once against this list. Every item here is something that has actually gone wrong in testing, and each one costs you a full generation if you miss it. The first item matters most: when an assistant reworks the locked blocks in its own words, the ride turns into a smooth ordinary commuter train and the whole effect dies.

Warning: If the prompt is far under 7,000 characters it will produce a thin, generic video. Ask for it again with more detail in every beat rather than generating it as is.

• The CAMERA, INTERIOR, RIDE and PASSENGERS blocks are copied word for word, not reworded
• At least three beats contain a hard jolt that throws the framing off
• Passengers face the windows with their faces visible, and lift phones to the glass
• Both announcements name your destination, not an example from the template
• Ten timestamped beats that add up to 30 seconds
• Between 7,000 and 9,500 characters, plain hyphens and straight quotes only

5. Compare against a finished example

This is a complete generated prompt for the centre of the Earth, trimmed to 9,734 characters. Read how each beat names its own landmarks: the descending city cut, stalactites, a limestone cavern viaduct, cave fish and bats, magma cracks, a spiral mantle railway. That density is what separates a believable journey from a generic tunnel. Use it as the standard your own destination should meet.

Exact prompt

ONE CONTINUOUS 30-SECOND SHOT, NO CUTS. Amateur first-person smartphone video by a standing passenger inside a packed commuter train that physically travels from an ordinary city railway to the centre of the Earth in one unbroken take. No cuts, jump cuts, dissolves, crossfades, double exposures, portals or morphing. Every environment is physically connected; the same train travels through all of it.

CAMERA (locked): cheap smartphone at chest height, one-handed, other hand gripping a pole, 63-degree wide lens. The phone is NEVER visible, because the phone is the camera. Crooked off-centre framing, often partly blocked by shoulders, arms and backpacks, imperfect like accidental footage. Constant shake, roll and yaw, thrown by acceleration. Rolling-shutter jello, autofocus hunting, exposure pumping, blown highlights, crushed noisy shadows, low-bitrate compression, macroblocking, smeared motion blur. No stabilisation, no cinematic composition, no drone, third-person or external camera. Never a forward or train-nose view, never show the front of the train. The camera always looks out the LEFT-SIDE WINDOWS at about 90 degrees to travel; the train always moves forward and the outside world always streams past front-to-back.

INTERIOR (constant anchor): packed standing-room-only carriage, passengers shoulder to shoulder gripping overhead straps and vertical poles, backpacks wedged between bodies, coats shifting with acceleration, straps swinging on inertia, carriage constantly vibrating and rattling. Same carriage, passengers, poles, straps and windows throughout; ordinary real interior, impossible outside.

RIDE (locked): this is a hard, rough, noisy ride, never a smooth modern train. The carriage vibrates and rattles without pause. The camera shakes, rolls and yaws for the entire 30 seconds and is physically thrown around by acceleration. Straps swing on their inertia, loose clothing and bags swing, and passengers sway and correct their balance continuously. Hard lateral jolts hit repeatedly across the shot, and every jolt throws the framing off and tightens hands on the poles. The shake grows heavier as the journey becomes more extreme. Never a stabilised, gliding, gently moving train.

PASSENGERS (locked): completely ordinary people, real skin pores, natural blinking, breathing, eye movement, imperfect facial symmetry, clothing folds reacting to acceleration. They stand along the window side and face the LEFT-SIDE WINDOWS in profile or three-quarter view with their faces visible to the camera. Never film a wall of backs and never turn the crowd away from the camera. Continuous small independent movements: shifting weight, adjusting grip, glancing around, checking phones, turning heads. As the outside becomes impossible, several passengers raise their own phones to the glass and film or photograph through the window at different heights and angles; their phones are visible in their hands, while the camera phone itself is never visible because it is the camera. Nobody freezes, not everyone reacts at once, no movie acting. Arc: ordinary commute, mild confusion, curiosity, quiet amazement, disbelief, stunned silence. No screaming.

TIMELINE
0-3s: ordinary bright weekday commute through a crowded city railway corridor. Concrete walls, power boxes, station signs and platform commuters sweep past close to the glass, apartment blocks drift at middle distance, office towers barely move on the skyline. Tired commuters on phones, staring out, talking quietly. PA chime, calm announcer: "Next stop: the centre of the Earth." Nobody screams. The train surges, hard lateral jolt, grips tighten.

3-6s: the railway drops into a long descending cut at the edge of the city. Retaining walls give way to exposed soil and dark rock streaking past the window, cranes and rooftops sliding away at middle distance, hills steady on the horizon. Utility cables become tunnel lights, drainage pipes and raw stone. A few passengers look up from their phones. Passengers sway, straps swing, shake increases.

6-9s: a constant gradient down through a huge underground transport tunnel. Service walkways, emergency lamps and cable trays whip past inches from the glass, maintenance alcoves pass slower behind them, and gaps open onto black cavern space that barely shifts. Stalactites appear far ahead, grow, sweep past and vanish behind the train. Water drips down rock faces. Two passengers lift their phones to the window.

9-12s: the tunnel opens into an enormous limestone cavern crossed by a steel viaduct descending along the wall. Bridge supports and rock columns hammer past close, underground waterfalls fall at middle distance, the far wall creeps. A broad river runs below the rails, pale cave fish and clusters of bats visible where light reaches. The train jolts hard over a rail joint and grips tighten.

12-15s: limestone turns to dark basalt. Reinforced heat-shielded tunnel sections run beside glowing magma cracks; thermal barriers and rock anchors flash past the glass, red-orange fractures grow and slide by at middle distance, a distant lava river barely moves. Heat shimmer exists outside only. Red-orange light enters the carriage and reflects on faces and poles. Curiosity becomes quiet amazement.

15-18s: a broad spiral railway carved around the wall of a colossal mantle cavern. Rock barriers flash past close, huge basalt columns and mineral formations turn slowly beyond them, distant lava channels drift. The curve shows in trackside geometry and in sections of railway far below, but the train holds one forward direction. The carriage quietens, someone softly says, "Ohh." Nobody screams.

18-21s: a mantle plume appears far across the cavern, a vast vertical region of glowing molten rock ringed by dark moving plates and incandescent cracks. It stays anchored to the world, never rising or falling with the train. Maintenance structures, thermal shields and warning markers rush by near the window. Interior light turns warmer and stronger. A loose fragment strikes the hull with a sharp metallic bang and the framing jumps.

21-24s: the rock turns denser and more metallic as the railway enters an engineered tunnel through extreme-pressure mantle material. Segmented walls and cooling conduits blur past the glass, observation openings pass slower, and through them glowing mineral layers bend toward a brighter region far ahead. Vibration deepens, straps sway. Three more passengers raise their phones to the glass.

24-27s: the train emerges onto a vast descending shelf at the edge of an immense central chamber. Black rock ribs sweep past close, compressed glowing material hangs at middle distance, layered rock shells recede almost motionless. The brightness starts as a narrow gap and widens as foreground walls pass behind the train. Reflections intensify across glass, poles and faces. A hard jolt throws the framing sideways.

27-30s: the railway follows a huge curved shelf into the final reveal. The foreground rock barrier drops away as the track opens onto the chamber, revealing a colossal glowing core region extending far beyond the frame, dark formations racing past in front of it while distant glowing layers barely move. Passengers stare in silence, phones still raised, one whispers, "What is that?" At about 29.5s the calm PA says, "Now arriving at the centre of the Earth." The train travels deeper as the shot ends.

DENSITY AND PHYSICS: never empty outside, never randomly cluttered. Near objects sweep fast, mid-distance slower, far objects barely move; objects pass behind window frames, poles and shoulders and vanish naturally. Whatever objects belong in the current environment keep real trajectories, scale and distance. No identical repeated objects; nothing teleports, freezes or pops in; objects enter and leave frame because the train is moving. Every environment physically grows out of the previous one, and the visible track geometry explains all movement: no elevator motion, no floating, no teleportation, no portals.

LIGHTING (all light enters through the LEFT-SIDE WINDOWS only; no artificial interior light faking the colours): ordinary daylight, then daylight fading as the cutting deepens, then cool tunnel lamps, then reflected cave light over black water, then the first red-orange glow of magma cracks, then strong molten light from lava rivers and the mantle plume, ending in overwhelming orange-white light from the core with deep black shadow between the rock ribs.

AUDIO: no music, no cinematic sound design, no exaggerated explosions, only cheap compressed phone-mic sound. Constant train motor roar, mechanical clatter, metallic vibration, rattling poles, straps knocking rails, breathing, clothing rustle, occasional quiet conversation. Environmental sound changes with each region while the mechanical interior noise never stops. A calm female PA voice says "Next stop: the centre of the Earth." near the start and "Now arriving at the centre of the Earth." at about 29.5 seconds, ordinary and unbothered throughout.

REALISM LOCK: it must look like a real person accidentally filmed something impossible on a cheap phone. The destination can be unimaginably spectacular, but the footage stays ugly, imperfect and believable, never a clean NASA visualisation, never CGI, a video game, an animation or a dream sequence. Nothing is a painted flat background, nothing is oversized, everything has real depth and distance. ONE SINGLE CONTINUOUS TAKE, no cuts of any kind.

Tip: Notice that the interior, the camera and the direction of travel never change. Only the world outside the window does.

Download the centre of the Earth example

6. Generate the full 30 seconds in one shot

Paste the prompt into an AI video generator that handles long single shots. Seedance 2.5 is the best fit because it produces the whole 30 seconds in one generation and accepts prompts up to 10,000 characters. Veo, Kling and Hailuo also work. Set 9:16 for Reels, TikTok and Shorts, or 16:9 for YouTube. Add nothing to the prompt.

Warning: Do not add music, a style preset, or the word cinematic. The entire effect depends on the footage looking like an accident filmed on a cheap phone.

7. If your generator caps out below 30 seconds

Generate the journey in parts and use the last frame of each part as the reference image for the next. Keep the CAMERA, INTERIOR and motion sections identical in every part and change only the timeline beats. That is what holds the carriage, the passengers and the direction of travel steady across the joins, so the finished clip still reads as one unbroken take.

Tip: Joins give themselves away when the direction of travel flips. Check that scenery still crosses the window the same way in every part.

8. Publish it with the announcement as your caption

Keep the generated audio. The compressed phone-microphone sound and the calm announcer are what make people watch it twice. Post it with the destination as the hook, then reuse the same template for a different destination tomorrow. The format survives repetition because the carriage stays the same and only the impossible part changes.

Tip: Caption it with the line the announcer says: Next stop, followed by your destination.

Download the universal template