How to make an AI music video from a track

Create a character card from the photos: on the left, a 2×2 grid showing the face in four angles (front, profile, three-quarter and back of the head); on the right, the character full-length, standing straight, looking at the camera with arms along the body. Keep the same look across every angle. The character wears the outfit from the sixth photo. Clean white background, studio lighting, no overlays, graphics or text.

Use the attached character card and 3 location photos as the main references and build a 3×3 storyboard with the same character in those locations, as if these were frames from a music video. Across all 9 frames the character keeps the same look while staying dynamic and in expressive poses. Add motion blur, stop-motion and ghost blur for energy and rhythm. Vary the compositions: interesting angles including dutch angle, bird's-eye view and ground-level shots; in a few frames, interaction with the camera. The result is one stylish storyboard with no text, captions or graphic overlays.

Use @image1 as a full 9-panel storyboard grid, not as a single image. Each panel is a separate beat-locked starting frame, used in exact reading order from left to right, top to bottom. Use @image2 as the strict character identity reference: preserve the character's face, hairstyle, wet black hair, pale skin, body proportions, outfit, hoodie graphic, pants, accessories, posture language, and overall silhouette exactly. Do not beautify, age, change gender, change clothing, change facial structure, change hairstyle, or replace the character. Use @audio1 as the full soundtrack and edit the entire video to its rhythm: every cut, body hit, hand gesture, camera shake, fake lip-sync movement, and impact flash must land on the beat.

Create a hyperdynamic dark music video in a frozen black-and-blue arctic ruin landscape. Style: raw handycamera, amateur extreme music video, chaotic handheld filming, unstable framing, crooked camera, sudden zooms, whip pans, stop-motion body jumps, heavy motion blur, ghost blur trails, frozen long-exposure frames, flash-frame transitions, blown-out icy highlights, crushed shadows, wet lens smears, snow hitting the lens, shaky autofocus hunting, VHS-like compression, cold cinematic haze, violent wind, black volcanic ground, snow, towering rock pillars.

Use @image1 as the full storyboard grid and treat the 9 panels as one continuous dark hyperdynamic music-video story, moving strictly left-to-right, top-to-bottom. Each panel is the exact starting frame for its own consecutive beat-cut segment. Use @image2 as the exact character identity reference: preserve the same face, hair, body proportions, outfit, expression style, accessories, skin details, and overall silhouette with zero face morphing, zero outfit change, zero age change, zero body change. Use @audio1 as the full soundtrack and edit every movement, cut, camera hit, flash, body accent, head snap, hand gesture, and fake lip-sync motion to the rhythm of @audio1.

Visual style: dark abandoned tiled corridor, wet reflective floor, dirty walls, teal-green fluorescent lighting, grime, haze, puddle reflections, horror-rap / underground music-video energy. Shoot like a chaotic handheld handycam: amateur framing, Dutch angles, sudden zooms, unstable focus, whip pans, crash zooms, flash-frame transitions, stop-motion jitter, motion blur, ghost blur trails, VHS compression, chromatic aberration, blown fluorescent highlights.

Use @image1 as the complete storyboard grid and visual sequence, reading left to right, top row to bottom row. Each panel is the exact starting frame for its matching scene beat. Use @image2 as the strict character identity card: preserve the character's exact face, hairstyle, wet black hair, facial tattoos, piercings, body proportions, outfit, hoodie details, pants, shoes, skin tone, expression language, and overall silhouette. No face morphing, no outfit change, no age change, no body change, no hairstyle change. Use @audio1 as the full soundtrack and drive every cut, camera hit, body movement, fake lip-sync gesture, head snap, hand motion, step, blur, flash, and transition directly to the rhythm of @audio1.

The video is a hyperdynamic dark underground music video at a rainy train station, handheld like an amateur filming on a damaged handycam: crooked framing, unstable zooms, missed focus, harsh flash cuts, dirty lens water drops, stop-motion body hits, motion blur, ghost blur trails, whip-pans, rain streaks, wet reflections, cold blue-gray lighting, gritty cinematic darkness.

Use @image1 as a complete 3x3 storyboard grid: each frame is a separate starting keyframe for the next scene, read left-to-right, top-to-bottom. Use @image2 as the exact character identity card: lock the character's face, haircut, wet black hair, pale skin, body proportions, outfit, hoodie graphic, pants, accessories, posture language, and overall appearance. Do not change, beautify, age, morph, replace, redesign, or simplify the character. Use @audio1 as the full soundtrack and drive every cut, body hit, hand gesture, fake lip-sync movement, camera shake, flash, blur, and transition to the beat.

Style: hyperdynamic dark arctic music video, frozen black-rock wasteland, snow, fog, brutal wind, giant stone monoliths, cold blue-gray grade, raw handheld handycam energy, amateur unstable filming, crooked framing, sudden focus hunting, harsh flash bursts, stop-motion stutter, motion blur, ghost blur trails, speed ramps, whip-pans, crash zooms, icy particles hitting the camera, VHS-like compression, frozen breath.