▶

～連鎖する転回～カオスの中で断片が結び直され、転回は次への合図となる。終わりに見えた地点は通過点へ変わり、同じ構造をなぞりながらも形を変え、進行は何度も更新されていく。～ Chain of Turns ～ Within chaos, fragments reconnect, and every turn becomes a signal to move on. What once seemed like an ending shifts into a passage, repeating familiar patterns while constantly transforming, as momentum renews itself again and again. 　　　　　　　　　　　　　　　 #KakuDrop #カクウドロップ #midjourney #klingai #nanobanana

Kaku Drop 架空飴

@kakudrop · creator

INSTAGRAM · 2025-12-28Source

14.5Klikes

177comments

Remix This

Prompt

GLOBAL LOCK: A young Japanese woman in her early 20s with a sharp black bob haircut and straight bangs. She wears a distinctive oversized white and teal-blue varsity bomber jacket over a black crop top, baggy teal-blue track pants, and black headphones around her neck. The visual style is cinematic photorealism with a heavy teal and orange color grade, 24fps film aesthetic, and significant motion blur during fast movements.

[00:00–00:02]
Interior of a modern Tokyo subway car. The subject stands holding a metal pole, looking directly into the camera with a calm, neutral expression. Fluorescent lights flicker slightly. The camera does a slow push-in. Teal shadows and cool tones dominate.

[00:02–00:03]
Extreme macro close-up of the subject's right eye. The iris is amber-orange with a glowing, intricate digital pattern. A reflection of a burning city is visible in the pupil. Sharp focus on the eyelashes and skin texture.

[00:03–00:04]
Surreal transition: The camera pulls back rapidly as the subject flies out of a giant, glowing open mouth in a dark, void-like space. High-speed motion.

[00:04–00:07]
Wide tracking shot of the subject skydiving through a pastel-colored futuristic city floating among thick white clouds. Neon signs in pink and teal are visible on the buildings. Wind aggressively blows her hair and jacket.

[00:07–00:10]
The subject is sitting cross-legged on the floor of a dark, circular room, facing a large round window. She is surrounded by a wall of stacked, glowing CRT monitors displaying her own face. Suddenly, the monitors explode outward in a burst of glass shards and orange sparks. She raises her hands to her head in shock.

[00:10–00:15]
Fast-paced driving sequence. The subject is behind the wheel of a white luxury sports car, wearing white-framed sunglasses. She is driving through a tunnel lined with glowing orange and blue light panels. Close-up on a glowing orange speedometer needle hitting the red line. Heavy motion blur on the tunnel walls.

[00:15–00:17]
A holographic 3D logo "KAKU DROP" in a glowing blue frame floats in the center of the tunnel as the car passes through it. Cinematic lens flare.

[00:17–00:19]
The subject is now riding a futuristic glowing hoverboard through a dark, industrial warehouse. Blue electrical arcs and lightning bolts crackle around the board. She leans into a turn, looking focused.

[00:19–00:20]
The camera is outside a moving subway train, looking through the window. The subject is seen briefly sitting inside as the train speeds past with heavy motion blur, transitioning back to the original subway environment to create a loop.

NEGATIVE PROMPT: Distorted facial features, inconsistent clothing colors, extra limbs, low resolution, blurry eyes, jittery frame transitions, cartoonish style, watermark, text overlays (except logo), flat lighting, dull colors, robotic movement.

SPEECH PACK:
(No spoken dialogue, but the visual rhythm follows a 128 BPM electronic track)
[00:00-00:20] Audio Intent: High-energy synth-wave with heavy bass drops on the transitions at 0:03, 0:08, and 0:17. Room tone: Subway hum (0:00), Wind rush (0:04), Glass shatter (0:09), Engine roar (0:12).

How kakudrop Made This Anime Transition AI Video and How to Recreate It

This high-octane video by @kakudrop is a masterclass in "Dream-Core" AI cinematography. It blends a grounded, relatable setting—a Tokyo subway—with escalating surrealist transitions that defy physics. The aesthetic is a perfect marriage of cinematic editorial portraiture and cyberpunk anime, characterized by a sharp teal-and-orange color grade, heavy motion blur, and impeccable character consistency. By utilizing a "loop" structure (starting and ending in the subway), the creator maximizes replayability, turning a 20-second clip into an immersive experience that feels like a high-budget movie trailer for an indie creator's brand.

What You’re Seeing

The video follows a consistent female protagonist through a series of increasingly chaotic and visually stunning environments. The wardrobe—a white and teal bomber jacket with matching baggy pants—serves as a visual anchor (the "Global Lock") that keeps the viewer grounded despite the shifting realities.

Shot-by-Shot Breakdown (Estimated)

Time Range	Visual Content	Shot Language	Lighting & Tone	Viewer Intent
00:00–00:02	Girl in subway car holding a pole.	Medium Close-Up (MCU), static.	Cool fluorescent, teal shadows.	Establish reality & character.
00:02–00:03	Macro of eye with fiery reflection.	Extreme Close-Up (ECU).	Warm orange glow in iris.	The "Inciting Incident" hook.
00:03–00:04	Flying out of a giant mouth.	Wide Shot, surreal morph.	High contrast, cinematic.	Shock value / Surrealism.
00:04–00:07	Skydiving through cloud city.	Tracking shot, high speed.	Pastel pinks and teals.	Sense of freedom/adventure.
00:08–00:10	Room of CRT monitors exploding.	Wide to MCU, center comp.	Warm amber practical lights.	Visual climax / "Glitch" vibe.
00:11–00:15	Driving sports car in neon tunnel.	Dutch angle, fast tracking.	Neon orange/blue streaks.	High energy / Speed.
00:16–00:20	Hoverboard to Subway return.	Low angle to profile.	Teal/Blue lightning.	Complete the loop.

Why It Went Viral: The "Dream Logic" Hook

The core of this video's success lies in its pacing and transition logic. It uses "Dream Logic"—where one scene flows into another based on a visual cue (like the eye reflection) rather than narrative logic. This keeps the brain engaged because it cannot predict the next frame. The 0–3 second hook is particularly strong: it starts with a mundane, relatable scene (subway) and immediately breaks it with a macro eye shot, signaling to the viewer that "something big is coming."

From a platform perspective, the high information density is key. Every 2 seconds, the environment changes completely. This forces the viewer to re-watch the video to catch all the details (the monitors, the car dashboard, the hoverboard), which signals to the Instagram/TikTok algorithm that the content is highly engaging. The use of a "brand logo" (KAKU DROP) halfway through also positions the creator as a high-end studio rather than just a casual user, increasing "Save" rates for aesthetic inspiration.

5 Testable Viral Hypotheses

The "Macro-to-Micro" Hook: Starting with a wide shot and immediately cutting to an extreme macro (the eye) creates a psychological "zoom" that captures attention. Replicate by: Cutting from a landscape to a 1:1 detail of an object.
Color Contrast Narrative: Using Teal (Subway) vs. Orange (Explosions/Car) creates a subconscious "Conflict and Resolution" feel. Replicate by: Using complementary colors for scene transitions.
The "Impossible" Action: Flying out of a mouth or riding a hoverboard triggers the "How did they do that?" response. Replicate by: Using AI to generate physics-defying movements.
Seamless Loop: Ending the video exactly where it started (the subway) encourages infinite playback. Replicate by: Matching the first and last frame's lighting and subject position.
Audio-Visual Sync: The fast cuts are timed to the rhythmic beats of the BGM, creating a "satisfying" sensory experience. Replicate by: Editing on the beat in CapCut or Premiere.

How to Recreate: From 0 to 1

Define Your "Global Lock": Choose a character (e.g., "Asian girl, bob cut, teal jacket"). This description must be in every prompt to maintain consistency.
Generate the "Anchor" Scene: Use an AI video tool (Runway Gen-3, Luma, or Kling) to create the subway scene. This is your "reality" base.
Create the "Transition" Assets: Generate the macro eye shot. Use a prompt that specifies "fiery reflection in the pupil" to bridge to the next scene.
Build the "Surreal" Sequence: Generate 2-3 second clips of the skydiving, the monitor room, and the car. Keep the "Global Lock" character description identical.
Use Image-to-Video: For better consistency, generate high-quality AI images of the character in the different scenes first (using Midjourney), then animate them.
The "Logo" Reveal: Create a 3D text element or use a tool like Luma to animate a static logo into a 3D space.
Editing & Speed Ramping: Bring all clips into CapCut. Use "Speed Ramping" (fast-slow-fast) to match the high-energy rhythm of the video.
Post-Processing: Add "Motion Blur" and "Film Grain" overlays to hide AI artifacts and give it a cinematic texture.

Growth Playbook

Opening Hook Lines

"POV: You're stuck in a glitch in the matrix."
"This is what my dreams actually look like."
"Wait for the transition at 0:08... 🤯"

Caption Templates

Template 1 (Story-driven):
Lost in the loop. 🌀 Sometimes the only way out is to go deeper. Which scene was your favorite? 👇 #AIArt #Cinematic #DreamCore

Template 2 (Creator-focused):
Pushing the limits of AI cinematography. 🎬 Created this entire sequence using [Tool Name]. The future of filmmaking is here. Save this for your next edit inspo! #AIVideo #CreativeDirection #DigitalArt

Hashtag Strategy

Broad: #AI #DigitalArt #Cinematography #VFX (Reach wide audiences)
Mid-tier: #RunwayGen3 #LumaAI #CyberpunkAesthetic #Surrealism (Target tech/art enthusiasts)
Niche: #KakuDropStyle #DreamLogic #AIFilmFestival (Build a specific community)

FAQ

What tools make it look the most similar?

Runway Gen-3 Alpha or Luma Dream Machine for the video, and Midjourney for the initial character consistency images.

What are the 3 most important words in the prompt?

"Cinematic," "Motion Blur," and your specific "Character Description" (e.g., "Teal Jacket").

Why does the generated face look inconsistent?

You likely aren't using a strong enough "Global Lock" or a reference image (IP-Adapter) to fix the character's features.

How can I avoid making it look like AI?

Add real film grain, motion blur, and color grade it manually in DaVinci Resolve or CapCut.

Is it easier to go viral on Instagram or TikTok with this?

Instagram Reels currently favors high-aesthetic "eye candy" like this, while TikTok prefers a "How-to" hook at the start.