30 Synthwave AI Video Prompts That Actually Work

Synthwave is the easiest genre to fail at with AI video. The aesthetic is so loaded with shorthand (neon, chrome, palm trees, retro grids) that almost every model will give you something that looks vaguely synthwave on a one-line prompt. The problem is that “vaguely synthwave” is the worst possible result. It reads as generated. It reads as cheap. It reads as the same image everyone is generating.

The way out is specificity. Real synthwave videos don’t just have neon, they have specific neon (magenta paired with cyan, not blue paired with pink). They don’t just have chrome, they have lacquered chrome with hard reflections from a single direction. The grammar of the genre is precise, and the more of it you encode in the prompt, the further you get from the generic look.

This piece is a library of thirty synthwave AI video prompts written with that grammar. They are organized by mood, with chase scenes, melancholy moments, neon cities, and retro tech. Each one specifies the shot, the lighting, the lens (where it matters), the motion, and a one-line mood. They have been tested across Runway, Pika, and Luma in 2026 and produce results that read as intentional rather than generic.

A note before the list. If you want a wider library across genres, the 100-prompt master library is the place to start. If you want to understand why these prompts are structured the way they are, the prompt-writing fundamentals piece breaks it down. If you are still picking which model to use, Runway vs Pika vs Luma is worth reading first.

Chase scenes

The defining synthwave shot. A car at speed at night under neon. Eight prompts, all variations on motion, framing, and what’s chasing what.

Low angle tracking shot of a black 1987 Lamborghini Countach driving on a wet asphalt road at night, magenta and cyan neon billboards flashing past, 35mm anamorphic, chrome reflections sharp on the bodywork, faint chromatic aberration. Camera move: side track at speed, 5 seconds. Mood: outrunning a 1986 nightmare.

Front-on shot of a chrome muscle car emerging from neon mist on a desert highway, 50mm, headlights cutting through the haze, cyan grid horizon behind, distant magenta sunset. Camera move: slow dolly back as it approaches, 6 seconds. Mood: arrival of the synth gods.

Over-the-shoulder of a driver in a leather jacket, dashboard lit cyan, road blurred ahead through the windshield, magenta city lights on the horizon, 50mm at f/1.8. Camera move: locked to the driver, world moves outside, 6 seconds. Mood: he knows where he is going.

Wide shot, an aerial behind a single car on an empty interstate at night, neon billboards cutting the dark on either side, 24mm equivalent on drone, light trails on the horizon. Camera move: drone push forward and slightly up, 7 seconds. Mood: alone at 100 mph.

Close up on a steering wheel and gloved hands, 35mm, dashboard glow magenta and cyan, occasional flash of red brake light from a car ahead, faint VHS scan lines. Camera move: slight handheld jitter, 5 seconds. Mood: the chase is real.

Side tracking shot of a motorcycle weaving through neon traffic, 35mm, taillights smearing into long magenta streaks, rain on the lens, helmet visor reflecting cyan. Camera move: side track matching speed, 6 seconds. Mood: cutting through 1987.

Rear three-quarter shot of a car launching down a tunnel, ceiling lights strobing, 50mm, chrome rear bumper catching cyan and magenta in alternation, faint lens flare. Camera move: dolly back as the car accelerates away, 5 seconds. Mood: leaving without looking back.

Top down on a car driving across a glowing grid, 35mm equivalent, the grid extending in all directions, sun on the horizon half magenta half cyan. Camera move: drone follow, 7 seconds. Mood: the road is the world.

Melancholy moments

Synthwave has a sadness that pop synth doesn’t. Use these for the bridges, the breakdowns, the verses that slow down before the chase resumes.

A woman in a leather jacket smoking on a fire escape, 35mm at f/2, magenta neon sign flashing across her face, distant city below in cyan haze, faint film grain. Camera move: slow push in, 6 seconds. Mood: the night before a decision.

A neon-lit hotel hallway at 3am, 24mm, magenta carpet, cyan ceiling light strips, a single figure walking away from camera in the distance. Camera move: locked off, 7 seconds. Mood: the door at the end is the only one open.

Close up on a phone receiver in a payphone, 50mm macro, cyan booth light, magenta neon visible through the booth glass, the phone is off the hook and dangling. Camera move: locked off, 5 seconds. Mood: the call he didn't take.

Wide shot of an empty diner at 2am, 35mm, neon OPEN sign casting magenta on the booth, single figure at the counter, one cup of coffee, cyan rain outside the window. Camera move: very slow dolly forward, 7 seconds. Mood: nobody is coming back.

A woman alone in a parked car under a flickering streetlight, 50mm at f/1.8, single magenta source from above, her face in soft shadow, faint condensation on the glass. Camera move: locked off, 6 seconds. Mood: not yet ready to drive home.

A figure standing on a rooftop at night looking out over a neon city, 24mm, back to camera, magenta and cyan towers below, low fog on the streets. Camera move: very slow pull back, 8 seconds. Mood: too high to hear it.

Close up on hands holding a vinyl record, 50mm, single magenta neon catching the lacquer, the rest of the room dark cyan. Camera move: slight tilt down to the record, 5 seconds. Mood: the song that broke them.

Neon cities

Establishing shots. The wide world your synthwave video lives in. Use these to open and to bridge between scenes.

Wide shot, a neon-lit Tokyo arcade alley at 2am, slow dolly forward through rows of cabinets, magenta and cyan light spilling onto the wet floor, a lone figure at the back, 35mm film grain, shallow depth of field, faint smoke. Camera move: dolly in, 5 seconds. Mood: arcade ghost.

Aerial wide of a 1980s Los Angeles at night, 24mm equivalent on drone, palm tree silhouettes against magenta sunset, cyan city lights below, distant grid horizon. Camera move: drone push forward, 7 seconds. Mood: the city dreaming itself.

Street level, a neon-soaked back alley with steam from grates, 35mm, a chrome car parked at the end, magenta sign overhead, cyan light from a window above, faint chromatic aberration. Camera move: slow dolly forward, 6 seconds. Mood: somebody lives here who shouldn't.

Wide shot of a synthwave skyline reflected in a still pool of water at dusk, 24mm, magenta sky, cyan towers, the water perfectly still, faint grain. Camera move: very slow pull back, 8 seconds. Mood: the city as memory.

A neon-lit parking garage at night, 35mm, fluorescent ceiling strips alternating magenta and cyan, a single car parked at the far end, faint smoke. Camera move: locked off, 6 seconds. Mood: nobody was supposed to find it.

Wide shot, a 1986 shopping mall food court at night, lights still on, no people, 24mm, neon signage of long-gone restaurants, faint VHS scan lines. Camera move: slow dolly through the empty tables, 7 seconds. Mood: closed for thirty years.

Aerial above a neon highway with chrome cars in long light trails, 35mm equivalent, magenta and cyan trails crossing, distant city on the horizon. Camera move: drone descent toward the road, 6 seconds. Mood: the night is moving.

Retro tech

Inserts. CRT screens, cassette tapes, dial-up modems. Cut these in on beats for texture.

Close up on a CRT screen displaying a wireframe grid horizon with a magenta sun, 50mm macro, scan lines visible, faint glow on the surrounding bezel, dust on the screen. Camera move: very slow push in, 5 seconds. Mood: the future the past wanted.

A cassette tape ejecting from a tape deck, 50mm macro, single warm interior bulb, magenta neon visible through a window beyond, faint chromatic aberration. Camera move: locked off, 4 seconds. Mood: side B about to start.

A 1986 boombox on a windowsill, neon city visible through the glass, 35mm at f/2, single magenta source from outside, cyan from a kitchen lamp behind, the cassette spinning. Camera move: slow tilt up to the city, 6 seconds. Mood: signal in.

Close up on a rotary phone with a coiled cord, 50mm macro, magenta neon source from off frame right, the phone polished black, faint film grain. Camera move: locked off, 5 seconds. Mood: a number that no longer rings.

A wall of CRT televisions all displaying static, 24mm, magenta and cyan flicker patterns across the wall, a single figure in silhouette in front of them. Camera move: slow dolly back to reveal the figure, 6 seconds. Mood: too many channels of the same nothing.

Macro on a pinball machine plunger being pulled, 50mm, magenta and cyan playfield lights flashing, the ball about to launch, chrome reflections everywhere. Camera move: locked off, 4 seconds. Mood: the game starts.

A neon "OPEN" sign in a video rental store window at night, 35mm, magenta sign foreground, cyan store interior with VHS tapes on the wall, faint scan lines on the whole frame. Camera move: very slow push in to the sign, 5 seconds. Mood: still renting.

How to Adapt These

The mistake people make adapting synthwave prompts is escalating the synthwave grammar until it breaks. More neon, more chrome, more grid horizon, more VHS. At some point the model gives up trying to compose a coherent frame and just renders a wash of magenta. The skill is restraint. Pick two synthwave grammar elements per shot, not five. Magenta and cyan, plus chrome. Or grid horizon plus a single neon sign. Or VHS scan lines plus a chase. The constraint produces the look.

When you swap subjects, keep the lighting logic. If the original prompt was “magenta source from camera right, cyan source from camera left, single chrome subject in the middle,” the swap should preserve that geometry. Change the subject to whatever your video needs but keep the lights where they were. The lighting is the synthwave, not the noun.

A last note. Cuts on the beat are non-optional in this genre. Synthwave audio is rhythmically aggressive and any visual that sits past the beat reads as wrong. Plan your cuts when you write the prompt list. Mark which prompts are five seconds, which are seven, and assemble the order so the lengths fit your bars. Doing this at the prompt-writing stage saves you an afternoon in the timeline later.

If you are ready to take these from prompts to a finished video, open Melodex, drop in a synthwave track, and feed the prompts in scene by scene. The grammar is doing the work. The model is the executor.