Setup/workflow:

LTX 2.5 TXT2VID using wan2gp

720p output

10-second clips

About 3 minute per clip

Around 18 generated shots for the full 3-minute video, all were the first generation

Prompts were written as continuous cinematic shots rather than shot lists

I kept the same general factory description in every prompt to get some visual consistency

The Git command signs were not generated inside the video, I created the signs separately and added them later in DaVinci Resolve

Here are the prompts used:

A glossy industrial music-video shot follows a female android technician operating an oversized hydraulic pushing machine in a pristine retro-futuristic workshop. She has a polished chrome body, human feminine facial features, short black geometric hair, metallic silver workwear, black protective gloves, and expressionless professional focus. A large steel block representing a completed software change rests on a conveyor belt, and she slowly pulls a heavy lever; the hydraulic ram extends with enormous mechanical force and pushes the block through a glowing remote transfer portal. The camera begins in a low medium-wide angle beside the ram, tracks smoothly alongside the moving block, then finishes close on the android's calm face as the mechanism locks into place. Teal fluorescent lights reflect across brushed steel surfaces, with strong white highlights, subtle haze, and early-2000s industrial music-video grain. Heavy hydraulic impacts, servo motors, metallic sliding sounds, and a relentless minimal electro-house rhythm synchronize with every movement. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician stands at a massive pneumatic stamping press inside the same immaculate teal-lit industrial workshop. Her polished chrome skin, short black geometric hair, metallic silver uniform, and black gloves match the established visual design. She carefully places several thin metal plates from a glowing staging tray beneath the press, lowers both hands onto two physical safety controls, and the machine slams down with a violent precise impact, permanently compressing the assembled stack into a single solid block. The camera begins overhead on the separate plates, cranes downward as they enter the press, then pushes toward the newly formed block while the android lifts it and examines it without emotion. Hard fluorescent reflections, oily metal textures, soft haze, and analog video grain create a polished early-2000s industrial-commercial aesthetic. Pneumatic hisses, a deep mechanical slam, relay clicks, and the continuous electro beat fill the soundtrack. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android warehouse technician rapidly operates a robotic loading arm beside a brightly illuminated staging platform in a clean futuristic factory. She has polished chrome features, short black geometric hair, a fitted metallic silver utility uniform, and black gloves. Individual anonymous metal cases arrive on several small conveyor belts; she guides the articulated arm to pick selected cases one by one and place them neatly onto the staging platform, while other cases continue past untouched. The camera starts with a close-up of the mechanical claw grabbing one case, pulls back into a medium tracking shot revealing the growing staged group, then arcs around the technician as she makes one final precise selection. The environment is dark teal and brushed aluminum with hard fluorescent lighting, soft atmospheric haze, and subtle analog grain. Robotic arm motors, metallic clicks, conveyor rattles, and rhythmic electronic percussion remain synchronized with the action. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician braces herself beside a gigantic industrial cable winch in the same retro-futuristic workshop. Her chrome body and silver work uniform gleam beneath cold teal fluorescent lights as she grips a long steel handle and activates the machine. Thick bundled cables run through a luminous opening suggesting a distant network connection, and the winch steadily pulls a series of sealed metal containers from the remote side toward her workstation; when they arrive, a mechanical routing mechanism feeds them directly into her local assembly line. The camera begins behind the rotating winch drum, tracks along the tightening cable toward the android, then circles into a frontal medium shot as the retrieved containers arrive beside her. Metallic surfaces, deep shadows, restrained fog, and slight film grain create a polished early-2000s industrial-commercial aesthetic. Cable tension creaks, electric motors, locking mechanisms, and a pulsing electro-house rhythm dominate the sound. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician controls a long telescopic robotic retrieval arm inside a vast dark teal industrial warehouse. She wears the same metallic silver uniform and black gloves, with polished chrome skin and a precise expressionless face. The arm reaches through a luminous remote gateway, collects several anonymous metal containers from distant storage racks, and deposits them neatly onto a separate receiving shelf beside her without touching the active assembly line in front of her. She calmly inspects the newly retrieved containers while the main working conveyor continues completely unchanged. The camera starts in a wide symmetrical composition, follows the telescopic arm outward with a smooth lateral track, then returns to a medium close-up showing the untouched workbench and newly filled receiving shelf side by side. Fluorescent cyan light, brushed steel, subtle haze, and analog grain maintain the industrial electro aesthetic. Servo motors, distant electrical hum, mechanical catches, and the ongoing beat remain crisp and rhythmic. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A striking female android technician enters a large industrial duplication machine in a polished futuristic workshop. She has chrome skin, short black geometric hair, black gloves, and a fitted silver utility uniform. She steps onto a circular platform while articulated scanning arms sweep over her body; a second chamber opens beside her and an identical android technician emerges along with a complete duplicate workstation, identical tool set, storage rack, and matching metal containers. Both androids turn toward each other at exactly the same moment and perform the same small head tilt before beginning identical movements. The camera begins as a centered full-body shot, slowly dollies forward during the scanning process, then slides sideways to reveal both identical technicians and both complete work areas in symmetrical composition. Cold teal lighting, reflective metal surfaces, restrained fog, and subtle VHS-era grain create a polished retro-tech commercial mood. Scanner pulses, mechanical shutters, servo movements, and synthetic electro percussion accompany the duplication. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician operates an elegant industrial rail-switching machine in a dark teal factory. Chrome skin, geometric black hair, metallic silver workwear, and black gloves give her the same precise mechanical appearance as the other technicians. A single conveyor carrying a sequence of identical steel blocks approaches a massive junction; she pulls a polished lever and a second parallel conveyor mechanically unfolds from the original line, allowing future blocks to travel independently while the original conveyor remains intact. She watches both paths operate simultaneously and calmly adjusts the new line's controls. The camera begins facing straight down the original conveyor, tracks backward as the junction separates, then rises into a slightly overhead wide shot revealing the two distinct parallel paths. Cyan fluorescent tubes, brushed aluminum, oily mechanical details, and faint analog grain reinforce the early-2000s industrial music-video aesthetic. Gear shifts, track locks, conveyor motors, and a steady dirty-electro beat synchronize tightly with the branching motion. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician stands between two active industrial conveyor lines carrying different sequences of steel components through the same futuristic workshop. Wearing a fitted silver work uniform and black gloves over polished chrome hands, she activates a heavy mechanical junction that gradually guides both conveyor paths toward a single shared line. The components approach the intersection in alternating order, mechanical gates carefully coordinate their entry, and the two streams successfully become one continuous sequence while the android monitors alignment with absolute concentration. The camera begins wide enough to show both parallel lines, glides toward the junction as they converge, then rotates smoothly around the android and ends looking down the newly unified conveyor. Teal fluorescent light reflects off worn steel, polished chrome, and transparent safety shields with slight analog grain. Mechanical gate impacts, synchronized conveyor sounds, hydraulic locks, and pulsing electronic bass build as the lines combine. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician attempts to combine two industrial conveyor lines, but two incompatible steel components arrive at the same junction simultaneously and jam the mechanism. She has polished chrome features, short geometric black hair, metallic silver workwear, and black gloves, remaining eerily calm as red warning lamps begin flashing around the otherwise teal-lit factory. She stops the machinery, manually lifts both conflicting parts, compares their incompatible shapes, removes one small section with a precision rotary tool, and fits the corrected pieces together before restarting the conveyor. The camera starts as a close tracking shot following the two incoming components, stops abruptly when they collide, then moves around the technician's hands as she resolves the mechanical conflict and ends on the smoothly moving unified line. Metallic sparks remain brief and controlled, with hard cyan and red reflections and subtle analog grain. Alarm pulses, metal impacts, rotary-tool whine, then a satisfying mechanical lock synchronize with the continuous electro track. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician stands beside a long elevated conveyor containing several sequential steel blocks in a dramatic retro-futuristic workshop. Her polished chrome body and metallic silver uniform reflect cold teal fluorescent light as she activates a large mechanical lifting platform beneath a group of recent blocks. The machine raises those blocks away from their original foundation, slides a newer foundation section underneath them, then lowers the sequence onto the new base one piece at a time. Each block mechanically reconnects to the previous block as the sequence is rebuilt in the new position. The camera begins low beside the original foundation, tilts upward with the lifted sequence, then tracks sideways as the blocks are reapplied onto the new base and finishes in a perfectly aligned profile view. Brushed steel, deep shadows, cyan highlights, light haze, and restrained film grain maintain the polished industrial-advertisement style. Hydraulic lifts, magnetic clamps, precise locking clicks, and a hypnotic electro rhythm accompany the process. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician rides a narrow industrial work platform toward a junction where several parallel production lines extend through a futuristic teal factory. She wears polished chrome skin, short black geometric hair, a fitted silver uniform, and black protective gloves. Without stopping the factory, she reaches for a compact mechanical selector lever, moves it decisively to another position, and the entire platform smoothly shifts sideways onto a different active conveyor path while all surrounding machinery continues operating normally. The camera tracks alongside her in a medium profile shot, briefly moves ahead to reveal the approaching junction, then arcs around as the platform changes tracks and settles into the new route. Clean cyan fluorescence, reflective metal, controlled haze, and subtle analog grain create a stylish mechanical music-video atmosphere. Track switches, electric motors, relay clicks, and an unwavering electro-house pulse synchronize to the lateral movement. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician works at a cluttered metal bench when a warning lamp announces an urgent interruption inside the polished teal workshop. Her chrome face remains calm as she gathers several unfinished components, loose tools, and partially assembled mechanisms from the bench and rapidly places them inside a heavy secure steel locker hidden beneath the workstation. She closes the locker, turns a mechanical key, and the workbench instantly becomes completely clean; after handling a brief alternate mechanical task, she unlocks the locker and restores the saved unfinished pieces to exactly their previous positions. The camera starts in a close overhead view of the messy bench, follows her hands downward toward the hidden locker, then pulls back to show the spotless surface before moving in again as the saved work returns. Cyan fluorescence, polished steel, soft haze, and mild analog grain continue the visual identity. Locker mechanisms, tool clinks, servo movement, and tight electronic percussion remain rhythmically synchronized. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician stands before an intimidating industrial control console dominated by one oversized mechanical reset lever. She has polished chrome features, short black geometric hair, metallic silver workwear, and black gloves, framed by cold teal fluorescent lighting and restrained red warning lamps. Several unfinished components and recent mechanical modifications cover the active workstation; she studies them briefly, grips the lever with both hands, and pulls it fully downward. Powerful mechanical actuators immediately return the machine, storage trays, and working surface to an earlier pristine configuration while every recent unfinished modification is physically swept away into a disposal chute. The camera begins in a slow push toward the lever, drops to a low angle as she pulls it, then rapidly tracks across the workstation as the recent changes disappear before ending on her expressionless face. Deep hydraulic impacts, mechanical resets, falling metal parts, and a sudden bass hit accompany the action. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android technician examines a previously completed steel assembly moving through a mature production line in the same dark teal factory. Instead of dismantling or rewinding the existing sequence, she selects a new inverse-shaped component and uses a precision torque tool to attach it to the assembly, visibly neutralizing the effect of an earlier modification while leaving every previous part physically present in the production chain. She then attaches a fresh mechanical component to the end of the sequence and releases the complete assembly back onto the moving conveyor. The camera begins in close-up on the existing finished part, pulls back to show the intact sequence behind it, then circles around her hands as the compensating component is installed and finishes on the additional new piece at the end. Cool cyan fluorescence, brushed steel, chrome reflections, and analog texture preserve the established aesthetic. Ratchet clicks, magnetic locks, conveyor hum, and measured electro percussion fill the soundtrack. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android quality-control technician operates a large illuminated comparison table inside the retro-futuristic factory. With polished chrome skin, geometric black hair, metallic silver workwear, and black gloves, she places two nearly identical mechanical panels side by side beneath a scanning gantry. Thin beams sweep across both surfaces and illuminate only the physical components that differ between them, while matching sections remain dim and unchanged. She points a precision inspection tool toward one altered section, rotates the panels slightly, and studies the differences with methodical concentration. The camera begins directly overhead for a symmetrical comparison view, lowers into an oblique close-up as the scanner reveals each difference, then slowly pushes toward one changed component. Cyan and pale green inspection light contrasts with dark teal steel surfaces, subtle haze, and gentle analog grain. Scanner tones, electronic ticks, small mechanical adjustments, and a restrained electro groove provide synchronized sound. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android supervisor stands motionless at the center of a pristine industrial workstation while several categories of physical components surround her in carefully separated illuminated trays. Her polished chrome features, short black geometric hair, silver utility uniform, and black gloves appear sharply defined under cold teal fluorescent light. Without altering anything, she activates a diagnostic scanning mechanism that sweeps across the entire workspace; one tray contains completed prepared components, another contains modified unfinished parts, and a third contains loose untouched pieces. She simply observes the scan, turns her head from one category to another, and inspects each group while every physical object remains exactly where it was. The camera begins wide and symmetrical, slowly pushes toward the workstation as the diagnostic light passes, then settles into a calm medium close-up on the android supervisor. Soft scanner sweeps, electrical ambience, relay clicks, and minimal electronic rhythm fill the audio. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android archivist walks slowly beside an enormous mechanical history wall containing a chronological sequence of interconnected illuminated steel blocks stretching deep into a dark teal industrial archive. Her chrome face and metallic silver uniform catch narrow fluorescent highlights as she passes each block, and every block briefly illuminates when her handheld scanner points toward it. She stops at several entries, inspects their physical connections, then continues backward through the history while the connected chain remains completely intact. The camera begins behind her in a long tracking shot emphasizing the seemingly endless sequence, moves alongside her profile as she scans individual entries, then rises slightly to reveal branching historical paths converging behind her. Cyan edge lighting, deep shadows, dusty atmospheric particles, worn steel textures, and subtle analog grain create a dramatic archival mood. Scanner beeps, distant mechanical ambience, footsteps, and a persistent electronic bass pulse accompany the movement. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

A female android investigator stands alone in a dim industrial inspection room while a malfunctioning mechanical assembly rests under a focused white examination lamp. She has polished chrome skin, short black geometric hair, a metallic silver uniform, and black gloves, maintaining complete professional composure. She activates a narrow handheld forensic scanner and slowly passes it across individual components; each scanned piece triggers a small beam of light that traces physically backward through a network of mechanical connections until one specific component is linked directly to her own workstation. She pauses, lowers the scanner, looks toward the identified component, realizes the connection, then slowly turns her head and stares directly into the camera without moving any other part of her body. The camera starts over her shoulder on the damaged assembly, pushes closer as the scanner traces its origin, then finishes in an uncomfortable centered close-up of her expressionless face. The electro beat abruptly disappears during the final stare, leaving only scanner beeps, electrical room tone, and one dry mechanical click. No visible text, letters, numbers, logos, labels, screens, subtitles, or signage.

I made another full AI music video locally, this time using MiniMax Ref2Video instead of text-to-video.

The whole thing was generated on a single RTX 3090 24GB using Wan2GP, with Spectrum step skipping enabled.

The video is built from 20 separate 15-second clips, generated at 960x544, then edited together in DaVinci Resolve.

Current system setup:

• RTX 3090 24GB
• 96GB system RAM
• NVIDIA driver 610
• CUDA 13.2
• MiniMax Ref2Video
• Wan2GP
• Spectrum step skipping
• 960x544
• 15 seconds per clip
• 20 clips total

The model is not audio driven, so the motion, pacing, dancing, camera movement, and transitions were all handled through prompting and editing rather than being synchronized directly to the song.

For continuity, I used reference images for the recurring character of taxi driver of Total Recall and a waymo vehicle, then tried to keep the same cyberpunk city, lighting, costume, car, and overall 80s cinema aesthetic across the individual generations.

The biggest improvement over pure T2V was having a stronger visual anchor for each scene. Ref2Video made it much easier to keep recurring subjects recognizable, although continuity across 20 separate generations is still very much a negotiation rather than a guarantee.

After generation, I assembled everything in DaVinci Resolve, trimmed the usable parts from each 15-second clip, added transitions, and finished it with scanlines and some additional grading to push the 80s look.

Here are the prompts:

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable vehicle identity, white exterior body, compact proportions, rounded front shape, dark windshield and side glass, wheel design and placement, front lighting arrangement, roof-mounted autonomous-driving sensor assembly, and overall silhouette. Ignore the daylight street, palm trees, buildings, and environment visible in <Picture 2>; only the vehicle itself is referenced.

summary:
[reference generation] Place <Subject 1> inside <Subject 2>, cruising through the same neon cyberpunk downtown at night during one continuous fifteen-second establishing shot, ending in a clean side-tracking composition that naturally leads into the next scene.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - identity, recognizable face, facial proportions, hair, body proportions, pale blue uniform, cap, black tie, metallic robotic hands, footwear, and characteristic expression remain unchanged; only nighttime lighting, seating position, and subtle rhythmic movement are new.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - exact body design, white exterior, dark glass, wheels, lighting geometry, roof-mounted autonomous sensor equipment, proportions, and recognizable silhouette remain unchanged; only the environment, reflections, movement, and nighttime lighting are new.

detailed_description:
The target video is a fifteen-second photorealistic cinematic single take in a neon cyberpunk future photographed like an expensive 1987 science-fiction movie. The entire music video occurs during the same night in the same downtown district: wet black asphalt, massive brutalist concrete towers, practical cyan neon tubes, restrained magenta accent lights, deep blue-black shadows, chrome reflections, thin drifting steam, atmospheric haze, soft diffusion, subtle 35mm film grain, horizontal anamorphic lens flares, and believable physical lighting. Avoid a modern glossy CGI look.

[Shot 1] Begin from a low wide camera position approximately one meter above the wet boulevard. <Subject 2> appears far down the street and approaches smoothly through the neon city. The camera begins tracking backward at approximately the same speed, maintaining a stable front three-quarter view as the vehicle gradually becomes larger in frame. Cyan architectural lights and small magenta highlights travel naturally across the exact white body and dark windows of <Subject 2>.

As the vehicle approaches, reveal <Subject 1> clearly through the windshield, seated calmly in the front cabin. Cyan dashboard light softly illuminates their recognizable face, pale blue uniform, cap, black tie, and metallic robotic hands. <Subject 1> looks calmly forward with the same cheerful uncanny expression from <Picture 1>. Their right metallic hand rests naturally while two fingers gently tap an implied Italo-disco rhythm.

Keep identity, hands, seating position, car geometry, reflections, and camera movement physically stable.

During the final five seconds, the camera smoothly arcs from the front three-quarter position toward the left side of <Subject 2> without cutting and without changing speed.

END STATE / TRANSITION: finish on a stable medium side-profile tracking composition of <Subject 2> traveling from left to right, with <Subject 1> clearly visible through the side window. The next clip begins from this exact motion direction, framing, city block, and lighting state.

No redesign of either subject, no different vehicle, no costume change, no daytime environment, no palm trees, no added text, no subtitles, no generated logos.

overall_soundscape:
None required. Visual generation only; final song and sound design will be added during editing.

non_diegetic_music:
None generated. The final Italo-disco track will be added separately.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable vehicle identity, white exterior body, compact proportions, rounded front shape, dark windshield and side glass, wheel design and placement, front lighting arrangement, roof-mounted autonomous-driving sensor assembly, and overall silhouette. Ignore the daylight street, palm trees, buildings, and environment visible in <Picture 2>; only the vehicle itself is referenced.

summary:
[reference generation] Continue <Subject 1> riding inside <Subject 2> along the same neon boulevard during one uninterrupted fifteen-second side-tracking shot, emphasizing autonomous driving and restrained rhythmic character movement before approaching a cyan-lit intersection.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - exact face, identity, body proportions, pale blue uniform, cap, black tie, metallic hands, footwear, and expression remain recognizable and unchanged; only head direction and small seated dance gestures change.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - same exact vehicle body, white exterior, glass, wheels, roof sensor assembly, lighting arrangement, and proportions remain unchanged; only motion and neon reflections change.

detailed_description:
The target video is a fifteen-second photorealistic cinematic single take continuing directly from the previous scene. Same neon cyberpunk downtown district, same night, same wet boulevard, same cyan-dominant practical lighting, restrained magenta accents, brutalist architecture, chrome reflections, atmospheric steam, deep blue shadows, anamorphic horizontal flares, soft diffusion, subtle 35mm grain, authentic 1980s science-fiction cinematography.

[Shot 1] Begin immediately in the exact side-profile tracking composition established previously: <Subject 2> moves smoothly from left to right while the camera travels perfectly parallel at the same speed and distance.

Keep the full recognizable side profile of <Subject 2> visible. Long cyan reflections and occasional magenta highlights slide naturally across its white body and black glass without altering its physical design.

Through the side window, <Subject 1> is clearly visible in the front cabin. Maintain the exact recognizable face and outfit from <Picture 1>. <Subject 1> initially looks forward, then slowly turns their head slightly toward camera.

<Subject 1> deliberately lifts both metallic robotic hands completely away from the vehicle controls, showing that <Subject 2> is operating autonomously. Without exaggeration, <Subject 1> performs a restrained seated Italo-disco movement: two small shoulder pulses, one subtle head nod, and one metallic index finger briefly pointing upward before relaxing again.

Sparse pedestrians and one cyclist may move through the distant background, but never obscure either referenced subject.

During the final four seconds, the camera smoothly advances from the pure side view into a front-left three-quarter tracking position as <Subject 2> approaches a large intersection illuminated by cyan traffic lights.

END STATE / TRANSITION: finish with <Subject 2> entering the intersection in a stable front-left three-quarter composition, still moving forward at controlled city speed, with <Subject 1> clearly visible through the windshield. The next scene begins from this exact position and direction.

No camera cuts, no high-speed driving, no new neighborhood, no vehicle redesign, no wardrobe change, no daytime, no readable text or generated logos.

overall_soundscape:
None required. Visual generation only.

non_diegetic_music:
None generated. Final song added in post-production.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable vehicle identity, white exterior body, compact proportions, rounded front shape, dark windshield and side glass, wheel design and placement, front lighting arrangement, roof-mounted autonomous-driving sensor assembly, and overall silhouette. Ignore the daylight environment visible in <Picture 2>; use only the vehicle as reference.

summary:
[reference generation] Continue <Subject 2> through the same cyan-lit city intersection while pedestrians and cyclists cross safely, with <Subject 1> calmly acknowledging a cyclist before the vehicle approaches the familiar nightlife curb.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - identity, face, outfit, proportions, metallic hands, cap, footwear, and expression remain unchanged; only a small two-finger gesture is introduced.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - exact vehicle identity, white exterior, body geometry, windows, wheels, sensors, lights, and proportions remain unchanged; only speed adjusts smoothly to surrounding traffic.

detailed_description:
The target is a fifteen-second photorealistic single-take continuation in the exact same cyberpunk downtown district during the same night. Authentic 1987 science-fiction cinema aesthetic: practical cyan neon, limited magenta accents, wet reflective road surface, heavy concrete buildings, atmospheric haze, thin steam, chrome highlights, deep blue-black shadows, subtle film grain, soft diffusion and horizontal anamorphic flares.

[Shot 1] Begin with <Subject 2> already entering the wide cyan-lit intersection in the same front-left three-quarter tracking composition established in the previous scene. Camera continues moving backward smoothly ahead of the vehicle.

Several pedestrians begin crossing far enough ahead to remain safe and visually clear. Two cyclists travel through a protected bicycle lane from right to left. Their motion is calm and natural, creating an elegant coordinated urban flow rather than danger.

<Subject 2> gently reduces speed without abrupt braking, maintaining perfectly stable geometry and orientation.

<Subject 1> remains clearly visible through the windshield. Preserve the exact recognizable face and blue uniform. As one cyclist passes, <Subject 1> lifts a metallic hand and gives a small friendly two-finger salute, then lowers it naturally. The distinctive cheerful expression remains unchanged.

After the crossing clears, <Subject 2> resumes smooth movement.

The camera slowly arcs toward the vehicle's right-front side while keeping both <Subject 1> and the recognizable front geometry of <Subject 2> visible.

Ahead, reveal the same nightlife block under a long cyan neon canopy, located immediately beyond the intersection.

During the final seconds, <Subject 2> moves gently toward the curb beneath that canopy.

END STATE / TRANSITION: finish with a stable low front-side composition of <Subject 2> approaching the cyan-lit curb and beginning to slow, with <Subject 1> still visible inside. The next scene begins at this exact curb approach.

No collision, no abrupt maneuver, no crowd chaos, no new vehicle, no character alteration, no new district, no text or subtitles.

overall_soundscape:
None required. Visual generation only.

non_diegetic_music:
None generated. Final music added separately.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable vehicle identity, white exterior body, compact proportions, rounded front, dark glass, wheels, front lights, roof-mounted autonomous-driving sensor assembly, and overall silhouette. Ignore the daylight environment of <Picture 2>.

summary:
[reference generation] Continue <Subject 2> stopping beneath the cyan canopy, then have <Subject 1> step out and perform a restrained Italo-disco gesture beside the exact same car in one continuous fifteen-second shot.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - recognizable face, identity, clothing, cap, tie, proportions, robotic hands, shoes, and expression remain unchanged; only posture changes from seated to standing and a simple dance gesture is added.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - exact vehicle body, glass, sensor equipment, wheels, lighting and proportions remain unchanged and clearly visible beside the character.

detailed_description:
The target video is a fifteen-second photorealistic cinematic single take in the same cyan-lit cyberpunk nightlife block, same night and same 1980s visual language: wet pavement, brutalist concrete facades, practical cyan canopy lighting, minimal magenta accents, chrome reflections, drifting steam, blue-black shadows, subtle grain, soft diffusion and anamorphic flares.

[Shot 1] Begin exactly with <Subject 2> approaching the familiar curb beneath the cyan canopy.

The camera tracks slowly beside the vehicle at low chest height.

<Subject 2> gently pulls into position and comes to a controlled stop. Hold long enough to establish that the exact reference vehicle remains visually stable.

Through the window, <Subject 1> is clearly visible.

The vehicle door opens naturally.

<Subject 1> steps out onto the wet pavement, one metallic hand briefly touching the door frame for physical stability. Preserve the exact face, proportions, blue suit, round cap, tie, robotic hands and shoes from <Picture 1>.

The camera gradually pulls backward while staying low enough to keep <Subject 1> and most of <Subject 2> together in frame.

<Subject 1> adjusts the front of the pale blue suit using both metallic hands, then performs a deliberately simple Italo-disco phrase: one side step, second side step, two restrained shoulder pulses, then one metallic finger points directly toward camera.

No complex dance choreography.

During the final three seconds, <Subject 1> relaxes the pose, turns slightly and leans casually against the front side of <Subject 2>.

END STATE / TRANSITION: medium hero composition with <Subject 1> leaning beside <Subject 2> under the cyan canopy, both identities clearly readable and physically stable. The next clip begins from this exact arrangement.

No cuts, no additional performers, no costume change, no vehicle redesign, no daylight, no generated signage or subtitles.

overall_soundscape:
None required. Visual generation only.

non_diegetic_music:
None generated. Final Italo-disco music added in edit.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable identity, white exterior, compact rounded body, dark glass, wheels, lighting, roof-mounted autonomous sensor equipment, dimensions and silhouette. Ignore the original daylight surroundings.

summary:
[reference generation] Keep <Subject 1> dancing beside the parked <Subject 2> beneath the same cyan canopy in a single restrained 1980s performance shot, ending with <Subject 1> standing at the vehicle door ready to enter.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - exact identity, facial appearance, costume, proportions, metallic hands, cap and shoes remain unchanged during simple controlled choreography.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - remains stationary and visually identical to <Picture 2>, with only environmental neon reflections added.

detailed_description:
Fifteen-second photorealistic single-take performance in the exact same nightlife curb location. Authentic 1980s cyberpunk film appearance: practical cyan neon canopy, tiny magenta accents, wet pavement, dark brutalist architecture, chrome reflections, steam, blue-black shadows, soft optical bloom, anamorphic flare and subtle 35mm grain.

[Shot 1] Begin with <Subject 1> leaning naturally against the front side of <Subject 2>, exactly matching the previous ending.

The camera begins a slow clockwise orbit around the character and vehicle together. Keep both reference subjects visible for almost the entire shot.

<Subject 1> gently pushes away from the vehicle and begins a restrained, repeatable Italo-disco dance phrase designed to preserve identity: two lateral steps, one controlled shoulder roll, metallic right hand sweeps horizontally across the chest, left metallic index finger points upward, a small pivot, then two measured steps backward.

Keep limb proportions stable and movements humanly achievable. Preserve the exact uncanny friendly facial expression.

<Subject 2> remains parked in precisely the same position. Cyan reflections travel naturally over its white panels and dark glass, but the vehicle shape and sensor equipment never change.

The wet ground produces soft reflections of both subjects.

As the orbit approaches completion, <Subject 1> stops dancing, turns toward <Subject 2>, walks the short distance to the door and reaches for the opening.

END STATE / TRANSITION: <Subject 1> stands immediately beside the open door of <Subject 2>, one metallic hand resting on the door frame, body oriented toward the cabin and ready to sit. The next scene begins here.

No new vehicle, no background change, no additional dancers, no body deformation, no wardrobe change, no readable text.

overall_soundscape:
None required.

non_diegetic_music:
None generated. Music added separately.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable identity, white exterior, body shape, proportions, dark glass, wheels, front and rear lighting, roof-mounted sensor equipment and overall silhouette. Ignore the sunny location visible in <Picture 2>.

summary:
[reference generation] Continue <Subject 1> entering <Subject 2>, closing the door and smoothly departing the same cyan curb during one continuous fifteen-second tracking shot.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - face, identity, clothing, accessories, proportions and robotic hands remain unchanged while transitioning naturally from standing to seated.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - vehicle identity and geometry remain constant through stationary and moving states.

detailed_description:
The target is a fifteen-second photorealistic continuous shot maintaining the exact same downtown curb, same cyan lighting, wet street, brutalist architecture, subtle magenta accents, haze, steam, anamorphic bloom, soft diffusion and 1980s film grain.

[Shot 1] Begin with <Subject 1> standing beside the already open door of <Subject 2>, one metallic hand on the upper door frame.

Without cutting, <Subject 1> smoothly lowers into the front cabin. Maintain realistic limb articulation and stable body proportions. Both metallic hands move naturally inside, followed by the legs and black shoes.

The door closes.

The camera remains outside and begins gliding parallel along the side window as <Subject 2> gently pulls away from the curb.

Through the glass, keep <Subject 1> clearly recognizable under cyan dashboard illumination. <Subject 1> looks forward and taps one metallic hand lightly against the upper leg in a restrained rhythmic pattern.

<Subject 2> smoothly merges back into the exact same wet boulevard.

Camera continues beside the car for several seconds without changing distance abruptly.

During the final four seconds, camera gradually reduces speed while <Subject 2> maintains forward motion. The car naturally moves ahead until camera settles into a rear-left three-quarter view.

END STATE / TRANSITION: stable rear-left tracking view of <Subject 2> traveling away along the familiar cyan-lit boulevard. The next clip begins directly from behind this moving vehicle.

No cuts, no teleportation, no different vehicle, no character change, no new architecture, no text.

overall_soundscape:
None required.

non_diegetic_music:
None generated.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable identity, white exterior, compact proportions, rounded geometry, dark windows, wheels, lighting arrangement, roof-mounted autonomous sensor system and silhouette. Ignore the daytime background from <Picture 2>.

summary:
[reference generation] Follow <Subject 2> through the familiar neon downtown while <Subject 1> rides inside, ending with the vehicle entering a long cyan tunnel during one uninterrupted fifteen-second night-driving shot.

retention_analysis:
<Subject 1> (appears through vehicle glass): fully_preserved - same identity, facial appearance, costume, proportions, cap and robotic hands remain consistent.
<Subject 2> (primary visual subject): fully_preserved - exact shape, white body, dark glazing, wheels, sensor equipment and proportions remain unchanged during the entire drive.

detailed_description:
Fifteen-second photorealistic continuous tracking shot in the same cyberpunk downtown, same night and same 1980s film aesthetic: practical cyan architectural lights, minimal magenta accents, wet asphalt, concrete towers, thin steam, deep blue shadows, chrome reflections, soft diffusion, anamorphic streaks and subtle grain.

[Shot 1] Begin directly behind and slightly left of <Subject 2>, matching the rear-left three-quarter ending of the previous clip.

Camera travels at approximately the same speed and maintains a consistent following distance.

<Subject 2> drives calmly through the established downtown boulevard. Wet pavement reflects the white vehicle and repeating cyan architecture.

Sparse pedestrians remain safely on the sidewalks. One cyclist travels in a separated lane.

The road gradually curves to the right. Camera follows the same smooth arc and slowly moves closer toward the left side of the vehicle.

Through the dark side glass, briefly reveal <Subject 1> seated comfortably in the front cabin, still wearing the exact pale blue uniform and cap. <Subject 1> gives one gentle head nod and one small shoulder movement while looking forward.

Do not make <Subject 1> dominate this shot; the drive itself is the focus.

Ahead, reveal a long rectangular road tunnel built into the same downtown architecture. The tunnel entrance is illuminated by repeating cyan rectangular lights.

<Subject 2> aligns smoothly with the tunnel entrance.

END STATE / TRANSITION: centered rear view of <Subject 2> just beginning to cross into the cyan tunnel, with the repeating light geometry visible ahead. Next clip begins inside this exact tunnel.

No new environment, no speed racing, no vehicle mutation, no daylight, no captions.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable identity, white exterior, compact proportions, rounded front geometry, black glass, wheels, lights, roof-mounted autonomous sensor assembly, and overall silhouette.

summary:
[reference generation] Follow <Subject 2> through the cyan tunnel, gradually move alongside it and reveal <Subject 1> taking both robotic hands away from the controls for a small seated disco gesture before the car reaches the tunnel exit.

retention_analysis:
<Subject 1> (appears prominently in second half of [Shot 1]): fully_preserved - exact identity, face, blue clothing, cap, tie, metallic hands and body proportions remain unchanged; only restrained arm and shoulder movement is introduced.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - exact vehicle geometry, sensors, windows, body panels, wheels and color remain recognizable and constant under moving tunnel lights.

detailed_description:
The target is a fifteen-second photorealistic uninterrupted shot inside the same urban tunnel. Authentic 1980s science-fiction cinematography: repeating cyan practical light rectangles, dark concrete walls, wet pavement, occasional subtle magenta reflection, light atmospheric haze, anamorphic streaking, soft diffusion and tactile film grain.

[Shot 1] Begin directly behind <Subject 2> as it completes entry into the cyan tunnel.

Camera follows the vehicle at identical speed. Repeating cyan light bands travel rhythmically across the exact white body, dark windows and roof-mounted sensor system without changing their physical forms.

After several seconds, camera slowly moves from directly behind to the left side of <Subject 2>, arriving at a clean parallel tracking composition.

Through the side window, clearly reveal <Subject 1> in the front cabin.

Maintain exact facial identity and outfit. <Subject 1> calmly lifts both metallic hands completely away from the controls and brings them loosely to chest height.

Perform only a tiny seated Italo-disco gesture: two synchronized metallic fingertip taps in empty air, one subtle shoulder pulse and one relaxed head nod.

<Subject 2> continues perfectly straight without visible human control.

The effect should feel confident, cool and slightly humorous, never slapstick.

During the final four seconds, cyan and magenta city lights become visible beyond the tunnel exit. Camera gradually advances into a front-left side position.

END STATE / TRANSITION: front-left side tracking view of <Subject 2> precisely at the tunnel exit, with the familiar nighttime city visible immediately beyond. Next scene continues the same forward movement.

No visual transformation, no speed jump, no different vehicle, no costume changes, no text.

overall_soundscape:
None required.

non_diegetic_music:
None generated.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable vehicle identity, white body, dimensions, rounded design, dark glass, wheel geometry, front lighting, roof-mounted sensor equipment and silhouette. Ignore the original daylight environment.

summary:
[reference generation] Continue <Subject 2> exiting the cyan tunnel and traveling through the familiar downtown while <Subject 1> performs a slightly more energetic seated disco gesture, ending with the car stopped beneath the previously established cyan canopy.

retention_analysis:
<Subject 1> (appears prominently through windshield): fully_preserved - face, identity, body proportions, clothing, cap, tie and metallic hands remain exact; only controlled rhythmic motion changes.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - vehicle identity, body, windows, wheels, sensor system and lighting remain unchanged.

detailed_description:
Fifteen-second photorealistic continuous shot, same downtown, same night, same weather and exact same 1980s cyberpunk color grade: cyan practical lights, restrained magenta highlights, wet asphalt, brutalist facades, steam, deep shadows, analog diffusion, anamorphic lens streaks and 35mm grain.

[Shot 1] Begin as <Subject 2> exits the cyan tunnel from the front-left side composition established previously.

Camera smoothly transitions into a low front-left three-quarter tracking position while moving backward at identical speed.

The vehicle's white surface reflects long cyan lines and occasional magenta highlights from the same familiar architecture.

Through the windshield, <Subject 1> is clearly visible and slightly more animated than before while remaining physically stable.

<Subject 1> performs two gentle shoulder pulses, one head nod and then raises one metallic hand for a playful forward finger point.

The autonomous vehicle continues operating smoothly and safely.

Camera gradually gets closer to the windshield while maintaining enough visible vehicle body to preserve <Subject 2>'s identity.

<Subject 1> slowly turns toward the camera and shows the same recognizable friendly uncanny smile.

During the last four seconds, <Subject 2> slows and returns to the exact same curb beneath the cyan canopy used earlier.

END STATE / TRANSITION: <Subject 2> completely stopped beneath the familiar canopy, viewed from a stable front-side position, with <Subject 1> visible through the window looking toward camera. The next scene begins from this exact setup.

No alternate neighborhood, no new car, no daylight, no character redesign, no text.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable vehicle identity, white exterior, body design, proportions, windows, wheels, lighting and roof-mounted autonomous sensor assembly. Ignore the sunny environment in <Picture 2>.

summary:
[reference generation] Have <Subject 1> step out of the stopped <Subject 2> beneath the familiar cyan canopy, perform one final simple Italo-disco dance, then return to the vehicle in one coherent continuous fifteen-second performance shot.

retention_analysis:
<Subject 1> (appears throughout [Shot 1]): fully_preserved - exact facial identity, pale blue uniform, cap, tie, metallic hands, footwear and proportions remain unchanged during restrained choreography.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - remains parked and visually identical to the reference while serving as a stable visual anchor.

detailed_description:
The target is a fifteen-second photorealistic continuous performance shot in the exact same cyan-lit curb location established earlier. Same wet pavement, same concrete facades, same practical cyan lighting, same subtle magenta accents, atmospheric steam, deep blue-black shadows, anamorphic flares, diffusion and textured 1980s film grain.

[Shot 1] Begin with <Subject 2> stopped beneath the cyan canopy and <Subject 1> visible through the side window.

The vehicle door opens smoothly.

<Subject 1> steps out naturally and stands beside the exact car.

Camera begins slowly pulling backward as <Subject 1> walks two measured steps toward lens.

<Subject 2> must remain clearly visible behind <Subject 1> throughout the performance.

<Subject 1> performs the final restrained Italo-disco phrase: two side steps, one metallic right-hand finger point, one controlled shoulder roll, a small half-turn, one smooth backward glide, then both metallic hands briefly rise symmetrically at chest height.

Keep the choreography simple, physically believable and identity-preserving.

Cyan light reflects across the pale blue suit and chrome robotic hands while magenta remains only a secondary accent.

After the short dance, <Subject 1> stops, looks over the shoulder toward <Subject 2>, turns and calmly walks back to the open vehicle door.

<Subject 1> begins lowering into the seat.

END STATE / TRANSITION: <Subject 1> is halfway seated inside <Subject 2>, one black shoe still on the wet pavement, door open, cyan canopy overhead. Next scene begins from this exact physical pose.

No additional dancers, no crowd, no costume change, no vehicle variation, no text.

subject_definitions:
<Subject 1> is the person in <Picture 1>, preserving their exact identity, facial features, skin tone, hairstyle, facial proportions, body proportions, pale blue uniform suit, white shirt, black tie, matching round pale blue cap, metallic robotic hands, black footwear, and distinctive cheerful uncanny expression.
<Subject 2> is the autonomous vehicle in <Picture 2>, preserving its exact recognizable vehicle identity, white exterior, compact proportions, rounded body shape, dark windows, wheels, lighting, roof-mounted autonomous sensor equipment and overall silhouette. Ignore all daylight environmental information from <Picture 2>.

summary:
[reference generation] Complete the video with <Subject 1> entering <Subject 2>, giving one small final gesture through the window, and the exact vehicle driving away through the familiar neon boulevard during one continuous eleven-second closing shot.

retention_analysis:
<Subject 1> (appears during first half of [Shot 1]): fully_preserved - exact face, identity, clothing, cap, tie, metallic hands, footwear and proportions remain unchanged during the final seating and farewell gesture.
<Subject 2> (appears throughout [Shot 1]): fully_preserved - exact white autonomous vehicle, body design, windows, wheels, sensor equipment, lights and proportions remain stable until it disappears naturally into the city.

detailed_description:
The target is an eleven-second photorealistic cinematic final single take in the exact same cyberpunk downtown district on the same night. Maintain the established 1980s science-fiction film aesthetic: practical cyan neon, restrained magenta accents, wet black boulevard, brutalist concrete buildings, chrome reflections, drifting steam, deep blue-black shadows, soft optical diffusion, anamorphic horizontal flares and subtle 35mm grain.

[Shot 1] Begin exactly with <Subject 1> halfway seated inside <Subject 2> beneath the familiar cyan canopy, with one black shoe still outside.

<Subject 1> smoothly brings the remaining leg and metallic hands into the cabin, settles into the seat and closes the vehicle door.

Through the side window, <Subject 1> turns toward camera one final time and performs a tiny understated farewell: two metallic fingers rise briefly in a restrained disco gesture.

<Subject 2> gently begins moving away from the curb.

Camera remains stationary at street level at first, watching the exact vehicle move deeper down the same familiar wet boulevard.

After several seconds, camera begins a slow cinematic crane upward, revealing the same cyan-lit brutalist architecture already established throughout the video. Do not introduce any new landmark or district.

<Subject 2> becomes progressively smaller while its white body and roof-mounted sensors remain recognizable under the neon light.

Cyan reflections stretch along the wet road behind it.

During the final seconds, <Subject 2> reaches the same distant corner previously seen in the video and turns gently behind a building.

The vehicle disappears naturally from sight.

Hold very briefly on the empty wet boulevard, cyan neon reflecting across the pavement and a small cloud of steam drifting through frame.

Slow cinematic fade to black.

No new subjects, no new vehicles, no location change, no transformation, no text, no subtitles, no generated logos.

overall_soundscape:
None required. Visual generation only.

non_diegetic_music:
None generated. Final song continues underneath during editing and fades with the image.

The whole video was generated locally on a single RTX 3090 24GB using ComfyUI and the official MiniMax text-to-video node.

It’s made from 12 separate clips, 15 seconds each, so 3 minutes total. Not audio driven.

Generation settings were roughly:

• RTX 3090 24GB
• MiniMax T2V
• Official ComfyUI MiniMax node
• 0.5 MP resolution
• 20 steps
• 15 seconds per clip
• Around 35 minutes generation time per clip

So the full thing was roughly 7 hours of GPU generation. The edited with Davinci Resolve and added scanlines

Here are the prompts

A surreal psychedelic cinematic music-video shot. A single glossy electric-blue mouth floats in total darkness, emerging slowly from black as if waking up. Highly detailed wet luminous lips with ultraviolet highlights and liquid-cyan reflections.

The mouth starts almost still and then moves as if singing to an unheard electronic song. Simulate musical rhythm visually: at regular intervals the lips pulse open and closed and each pulse emits a circular cyan soundwave traveling outward through the darkness. Tiny glowing particles shake and react as the waves pass.

Small abstract wireless and Bluetooth-like symbols occasionally flicker around the mouth and dissolve into blue vapor.

Very slow cinematic macro push toward the lips, shallow depth of field, halation, bloom, subtle analog-video texture.

During the final seconds the mouth opens wider, revealing an intense tunnel of blue light inside. The camera accelerates forward and enters the glowing mouth.

End while the camera is traveling into the mouth so the next scene can continue from inside it.

Continue directly from inside the glowing electric-blue mouth. The camera flies forward through a surreal organic tunnel made from glossy lip textures, translucent crystal teeth, liquid membranes and rippling cyan light.

Simulate an electronic musical rhythm visually. Bright pulses travel down the tunnel at repeating intervals. Concentric soundwave rings, spirals and glowing frequency ribbons race along the walls. Tiny wireless symbols drift through the space like bioluminescent insects.

The tunnel stretches, contracts and breathes rhythmically.

The camera moves rapidly forward with smooth banking turns and gentle rolls, maintaining constant momentum.

Near the end the tunnel expands into a vast black void. A gigantic electric-blue mouth floats ahead.

As the camera approaches, that single mouth begins dividing into many identical mouths.

End at the moment the duplication begins.

Begin with the giant blue mouth from the previous scene dividing into dozens of identical glossy electric-blue mouths floating in deep black space like a constellation.

Each mouth behaves like a different musical instrument. Some pulse slowly like bass, others open and close rapidly. Each emits a different visible waveform: circular cyan rings, spirals, jagged crystalline waves, vibrating ribbons and translucent frequency walls.

Where the waves intersect they briefly create luminous geometric flowers, abstract faces and interference patterns.

The camera moves continuously through the constellation in zero gravity, weaving between mouths and passing extremely close to some of them before pulling rapidly into wide views.

Toward the end all mouths suddenly rotate toward the same distant point.

They open together and release one enormous synchronized beam of blue frequency light.

The camera accelerates along the beam.

End as the beam reaches a distant humanoid silhouette.

Begin with the cyan frequency beam from the previous scene reaching an androgynous humanoid figure standing inside a dark surreal nightclub.

The room contains reflective black floors, giant mirrors, ultraviolet fog and isolated cyan lights.

The figure initially has a perfectly smooth face with no mouth.

The incoming beam hits the face and a glossy electric-blue mouth forms there from liquid neon.

Immediately the new mouth begins moving rhythmically as if singing.

Every implied beat produces a powerful visible bass wave expanding through the room. Curtains ripple, mirrors bend elastically, reflections distort and pools of light pulse across the floor.

Camera continuously orbits around the figure, alternating between medium shots and extreme close-ups of the mouth.

The movement becomes increasingly energetic.

During the final seconds a huge soundwave hits a mirror behind the figure. The mirror liquefies into chrome water.

The camera follows the wave directly through the liquid mirror.

End while crossing through it.

Continue through the liquid mirror into an inverted psychedelic chrome-and-blue world.

Two giant glossy electric-blue mouths float facing one another from opposite sides of the frame.

They move rapidly toward each other, trailing long ribbons of cyan waveform light.

They stop only millimeters apart without touching.

Between them the air vibrates violently with visible frequency patterns.

The mouths pulse and move as if harmonizing to an unheard electronic song. Every implied beat compresses the light between them further.

A brilliant sphere of cyan energy gradually forms between the lips, made from waveform lines, liquid light and tiny wireless symbols.

The camera performs a fast continuous orbit around the mouths while moving closer.

At the climax the sphere bursts outward into thousands of miniature glowing blue mouths and fragments.

The camera immediately chooses one tiny falling mouth and dives downward after it.

End while rapidly following the falling mouth.

Follow the miniature blue mouth falling directly out of the previous scene.

It drops through darkness and suddenly enters a surreal nighttime city.

Thousands of tiny glowing electric-blue mouths are falling from the sky like rain.

The camera descends rapidly toward street level and begins flying forward between buildings.

Each mouth opens and closes rhythmically while falling. Whenever one strikes a window, car, rooftop or puddle it generates a glowing circular soundwave.

Streetlights flash rhythmically. Building reflections ripple. Neon signs distort. Puddles produce synchronized frequency patterns.

The camera races low over the wet street, occasionally passing through clouds of falling mouths and narrowly avoiding chrome objects.

As the sequence progresses, the pavement begins turning into electric-blue liquid.

Buildings stretch downward into their own reflections.

The entire city melts into one enormous blue ocean.

End with the camera racing just above the newly formed liquid surface.

Begin directly above the infinite electric-blue liquid ocean created from the melting city.

The camera flies extremely low over the surface at high speed.

Huge glossy blue mouths rise explosively from beneath the water like surreal islands.

Each mouth opens on an implied beat and sends enormous circular waves across the ocean.

The camera banks around the waves, dives between giant lips and skims across the liquid surface while glowing droplets explode upward.

A chrome humanoid figure suddenly appears running across the water.

Every footstep generates luminous frequency rings.

The camera follows beside and behind the figure with aggressive tracking movement.

Ahead, an enormous mouth rises from the ocean and opens.

The water begins rotating into a giant whirlpool shaped like an abstract Bluetooth symbol.

The chrome figure accelerates toward it and is pulled into the vortex.

The camera dives in immediately behind the figure.

End while falling into the whirlpool.

Continue falling through the blue whirlpool.

The chrome figure suddenly splits into two mirrored humanoid bodies suspended in an enormous electric-blue void.

Each body has a glowing blue mouth embedded in the chest where the heart would be.

The bodies orbit each other rapidly.

Their chest-mouths repeatedly open and fire thin cyan waveform lines toward one another.

The connection fails several times, causing sharp visual glitches, duplicate frame echoes and bursts of distorted space.

Simulate rhythmic music visually through repeated connection attempts, pulsing light and synchronized body movement.

The camera revolves around both figures while constantly changing distance, moving from extreme close-ups of the heart-mouths to wide shots of the orbiting bodies.

Finally one bright continuous waveform successfully connects both mouths.

Their movements instantly synchronize.

The cyan connection becomes brighter and thicker until it fills the center of the frame.

The camera accelerates directly into the connecting waveform.

End inside the bright line.

The bright connecting waveform from the previous scene expands around the camera and transforms into an extreme macro view of two electric-blue mouths approaching one another.

The lips are constructed from glossy liquid, pixels, tiny waveform fragments and floating droplets.

The camera moves rapidly around and between them while maintaining extreme macro detail.

As the mouths approach, streams of luminous data begin moving between them.

Tiny landscapes, abstract memories, faces, wireless symbols and pulses of cyan light flow from one mouth into the other.

The mouths never physically collide. Instead they exchange increasingly intense streams of information through the narrow gap between them.

Each implied musical beat causes the lips to pulse, the data stream to surge and the camera to change speed.

The two mouths gradually lose their solid form and dissolve into one enormous shared waveform.

The waveform twists violently into a spiral.

As the camera follows it, the spiral begins forming the petals of a giant electric-blue flower.

End as the flower starts opening.

Continue from the waveform flower opening in the previous scene.

Reveal an enormous psychedelic flower floating in black space, constructed entirely from hundreds of glossy electric-blue mouths.

Every petal is a mouth.

The flower rotates rapidly while different rings of mouths open sequentially, creating visible waves traveling around its circumference.

Each pulse generates cyan frequency ribbons, glowing pollen and expanding geometric mandalas.

The camera spirals aggressively through the flower, diving between petals, rotating around the center and constantly shifting between huge wide shots and extreme macro mouth close-ups.

The rotational speed steadily increases.

The mouth petals stretch into liquid ribbons and the entire flower transforms into an enormous rotating blue mandala.

At the climax every mouth closes simultaneously.

The entire structure collapses rapidly inward.

Everything disappears except one single isolated electric-blue mouth floating in darkness.

The camera brakes suddenly and stops close to it.

End on the lonely mouth.

Begin with the isolated electric-blue mouth floating alone in nearly total darkness.

The atmosphere is quieter but still constantly moving.

The camera slowly circles the mouth while drifting closer and farther away.

The mouth attempts to sing.

Each time it opens, a visible cyan waveform travels outward but quickly breaks apart into digital particles.

The mouth tries repeatedly.

Every failed signal creates analog glitches, transparent duplicates and brief distortions of the surrounding darkness.

The pulses become weaker.

Then a tiny blue signal appears extremely far away.

The camera suddenly changes direction and moves toward the distant light while keeping the mouth visible behind.

The isolated mouth reacts and fires a stronger waveform.

The distant signal answers with another wave.

Both waves race toward each other through the void.

The camera accelerates alongside them.

End one instant before the two signals collide.

Begin exactly before the two cyan signals collide.

They connect and immediately produce an enormous radiant blue shockwave.

The camera is thrown backward at extreme speed.

The entire psychedelic universe from the previous scenes suddenly appears around the camera: hundreds of electric-blue mouths, liquid oceans, chrome bodies, waveform flowers, falling mouth rain, rotating wireless symbols and glowing frequency ribbons.

Everything moves in synchronized rhythmic pulses.

The camera races through the environment, passing between giant mouths, diving through waveform rings, rolling around chrome bodies and skimming above liquid-blue surfaces.

Every implied beat triggers another transformation.

The movement reaches maximum intensity.

Then the camera begins an extremely fast continuous pull backward.

As it moves away, the entire impossible universe becomes smaller.

Eventually reveal that everything we have been seeing exists only as a reflection on the surface of one enormous pair of glossy electric-blue lips floating in black space.

The camera continues pulling back.

The giant lips slowly close.

One final cyan soundwave explodes outward toward the camera, expanding until it fills the entire frame.

The light fades rapidly to pure black.

Hold black for the final second.