Seedance 2.0 · Kaze.ai · Cinematic AI Video

The Next
Opponent
Is You

A cinematic AI video about facing your future self — shot in darkness, silence, and light. Generated entirely with Seedance 2.0 on Kaze.ai.

Final Output

Final output — generated using Seedance 2.0 on Kaze.ai

The Concept

Two versions of the same person

“A lone figure climbs upward into a single beam of light — and finds himself already waiting.”

The creative direction was simple but precise: one spotlight, two figures, total darkness. No scenery, no props, no distractions. Just the weight of the space between who you are and who you’ll become.

The younger version rises from below the frame into the light — cautious, breathless, uncertain. The older version stands still in the far background, already arrived. They never speak. They never move toward each other. The tension lives entirely in the silence between them.

This is the opponent worth facing. Not an enemy. Not a rival. Your future self — already watching.

Behind the Scenes

Reference stills used in generation

These reference images were passed as @image1 and @image2 to anchor the character identities across all three shots.

Reference still — new version, older self in spotlight — Shot 1 — New Version

Reference still — low angle perspective of subject — Shot 2 — Low Angle

Reference still — wide shot showing both figures facing each other — Shot 3 — Final Wide

The Prompt

Full Seedance Prompt (Copy & Use)

Exact 3-shot prompt used to generate the video. Replace @image1 with your teen photo and @image2 with your adult photo.

SHOT 1 (0–5s):Static medium shot framed from chest to waist, never changes. One single diagonal spotlight beam from upper right corner cuts through the darkness. At the start @image1 is below the frame, completely hidden in darkness. He slowly climbs upward, rising from the darkness into frame, entering the diagonal light beam and becoming illuminated. Slim South Indian teenage boy, short dark black wavy hair, white shirt, dark trousers. Slight breathlessness, tense nervous posture. He looks around cautiously. Raises his eyes upward toward the light, squints, lifts one hand to shield his eyes. Pauses. Slowly lowers his hand and looks directly forward. Pure dark void background. Audio: faint breath, dead silence of an empty dark space.

SHOT 2 (5–10s):Static wide shot, framing never changes. Same single diagonal spotlight beam from upper right corner illuminating both figures. Camera faces forward. @image1 stands LEFT FOREGROUND with his BACK to the camera on a very low gentle mound, closer to camera. Slim South Indian teenage boy, short dark black wavy hair, white shirt, dark trousers. @image2 stands FAR BACKGROUND on the RIGHT on a slightly higher gentle mound, facing toward camera, completely still, hair moves very slightly. Slim South Indian male mid-20s, short dark black wavy hair swept back, thin neat mustache and short pointed beard, white formal dress shirt untucked hanging loose over dark grey trousers, dark charcoal striped tie. Vast darkness everywhere else. Audio: deep ambient low frequency hum, barely perceptible.

SHOT 3 (10–15s):Static wide side-angle shot, framing never changes. Same single diagonal spotlight beam from upper right corner. LEFT side: @image1 on lower gentle mound, seen from side, facing right, still. Slim South Indian teenage boy, short dark black wavy hair, white shirt, dark trousers. RIGHT side: @image2 on slightly higher gentle mound, seen from side, facing left, still. Slim South Indian male mid-20s, short dark black wavy hair swept back, thin neat mustache and short pointed beard, white formal dress shirt untucked hanging loose over dark grey trousers, dark charcoal striped tie. Both full body visible, large distance between them. Deep darkness surrounding both. Cinematic, moody, atmospheric. Audio: complete silence, dead quiet for final 3 seconds.

The Tool

Generated on Kaze.ai with Seedance 2.0

Kaze.ai is one of the cleanest interfaces for running Seedance 2.0 — the AI video model from ByteDance. It supports multi-image reference inputs, which is what makes the @image1 / @image2 technique possible.

No complex setup. No API keys. Upload your reference images, paste your prompt, and generate.

Try Kaze.ai Free →

Kaze.ai interface showing Seedance 2.0 video generation

What I Changed

Beyond the prompt

Beyond the prompt, I made some manual edits and refinements to get the final result. The AI output gave me a strong base — the lighting, the silhouettes, the stillness were mostly there. But getting the emotional weight right required a few passes: trimming frames where motion felt off, adjusting the pacing of the reveal in Shot 1, and making sure the final wide in Shot 3 held long enough to land the silence.

The prompt engineering itself took more iteration than the editing. Getting Seedance to hold a static camera while a figure moves within frame — without the camera following them — required being very explicit about what shouldn’t change, not just what should.

The NextOpponentIs You