// FREE PROMPT DROP

One photo. One prompt.
A storm-temple ninja.

The exact Seedance 2 image-to-video prompt behind the Ninja Assassin reel — and the simpler trick most people overshoot: no start/end keyframes. One reference photo of your face, one prompt, and the model fills all fifteen seconds. Watch it, copy it, drop your own photo in. No email, no gatekeeping.

Seedance 2 · 9:16 · one reference photo · made with Jeremy Knox AI

The exact prompt
NINJA ASSASSIN

SETTING
Abandoned mountain temple hidden deep within a stormy bamboo forest at night. Heavy rain crashes against ancient stone paths, shattered lanterns flicker violently in the wind, thick fog crawls low across the soaked ground. Massive hanging prayer bells sway slowly.

CHARACTER
@Image1 wears a black layered shinobi robe fused with modern tactical armor, dark wet fabric clinging to his body, reinforced gauntlets, utility straps, flowing scarf, and glowing crimson markings beneath the skin.

TIMELINE (15 seconds)

0-3s — EXTREME PUSH-IN
Violent handheld shot. Rapid push-in through pouring rain directly toward @Image1. His head slowly tilts upward. One glowing crimson eye ignites through the darkness. Thin red energy cracks pulse beneath his skin like molten veins. Rain droplets freeze momentarily around him.

3-8s — ORBITING COMBAT
Chaotic orbiting handheld camera. Enemy assassins emerge from the fog from every direction. @Image1 performs hyper-fast knife combat mixed with fluid kungfu counters. Every slash creates glowing red energy streaks and explosive water splashes. Bodies spin and crash through bamboo stalks.

8-12s — OVERHEAD SPINNING CRANE
Camera rises rapidly into a spinning aerial shot above the battlefield. @Image1 slams his palm into the wet ground. A gigantic crimson shockwave erupts outward in a rotating circular wave, lifting rainwater, rocks, bamboo debris, and enemies into the air. The storm clouds above begin swirling unnaturally.

12-15s — SLOW MOTION FINISH
Slow motion tracking shot. One final assassin lunges forward in slow motion. Time suddenly slows to near-complete stillness. @Image1 walks calmly through frozen rain particles. He appears behind the assassin instantly. A delayed energy slash suddenly erupts across the enemy's body. Time snaps back violently. The enemy is launched backward through the temple gate in an explosive impact.

SETTINGS
@Image1 = upload one clear, well-lit photo of your face as the reference image. Model: Seedance 2.0 (Higgsfield) · 9:16 · 720p · 15s · genre action · audio on. No start/end keyframes — one reference image, the prompt drives the rest.
#seedance2 #aivideo #imagetovideo #ninja #aifilmmaking
// THE TRICK

No keyframes. One reference image.

You don't need to bake two keyframes to keep your face. You give Seedance a single reference photo as the start image and let the prompt drive every beat — push-in, combat, shockwave, finish. The identity rides on that one image, so it stays you the whole way through instead of drifting into a stranger.

Reference input — one real photo of the subject's face

INPUT — one reference photo

Result — the crimson-eyed shinobi at the full-front hero finish, same face

OUTPUT — same face, 15s later

  1. 01Take one clear, well-lit photo of your face — front-on, no sunglasses. That single image is your reference (@Image1).
  2. 02Feed it into Seedance 2 image-to-video as the reference image with the prompt above. 9:16, 720p, 15s, genre: action, audio on.
  3. 03Upscale the keeper to 1080. That's the whole reel — one reference photo in, a 15-second storm-temple fight out, your face locked throughout.
The lever most people skip

My first cut was meh — stiff, over-directed, the energy gone. The fix wasn't a longer prompt. It was two settings: genre: action instead of cinematic for kinetic punch, and the full 15 seconds so the four beats breathe instead of cramming into twelve.

And the identity rule people get backwards: don't describe your face in words — anchor it with a reference image. A paragraph of "short curly hair, salt-and-pepper beard" drifts into a stranger by second eight. One real photo doesn't. Let the words drive the action; let the image hold the face.

// WANT THE WHOLE SYSTEM

This is one prompt. I teach the whole pipeline.

How I run AI as an operating system — from a single prompt to an unattended fleet that ships content, code, and trades. 350+ lessons in the Academy.

Explore the Academy
Grab more free drops