0:00 / 0:00

Erik Gen

@erik_gen_ai

INSTAGRAM · 2026-02-23Source

97likes

15comments

Remix This

Recreate with Veo3.1

Make your own AI viral video

Prompt

GLOBAL LOCK:
Subject is a middle-aged Caucasian man, approximately 40-45 years old, with short, slightly messy brown hair, a prominent nose, and expressive blue-grey eyes. He has a medium build and fair skin with warm undertones. His facial identity must remain 100% consistent across all scenes. The visual style is hyper-photorealistic cinematic film, 4k resolution, with high production value. Lighting is motivated by the environment, ranging from natural daylight to warm candlelight and harsh artificial space lighting. Camera language uses a mix of handheld urgency, smooth tracking shots, and complex rotating movements. Speech is direct-to-camera or reactive dialogue with high lip-sync strictness when on-camera.

[00:00–00:05]
Subject: Man in a grey t-shirt and blue denim jacket.
Environment: Busy New York City street, yellow taxis in background.
Action: Man walks away from camera, then stops and looks back over his shoulder with a curious expression.
Camera: Medium shot, tracking behind him.
Lighting: Overcast daylight, soft shadows.
Speech: "Really interesting action shot." (Warm, conversational tone).

[00:05–00:14]
Subject: Same man, same clothes.
Environment: The city street begins to crack and split open. Buildings in the background start to crumble and collapse into dust.
Action: Man turns and starts running away from the camera as the ground behind him falls away. He looks terrified.
Camera: Wide shot, shaky handheld movement to simulate an earthquake.
Lighting: Dusty, hazy atmosphere with light filtering through debris.
Motion: Intense ground fracturing, falling debris, volumetric dust.

[00:14–00:28]
Subject: Man transitions from denim jacket to a green outdoor hiking jacket and a large olive-green tactical backpack.
Environment: Seamless transition from city to a dense, lush pine forest with mossy rocks.
Action: Man stops, looks around in wonder, then starts running again as a large grizzly bear emerges from the trees and chases him.
Camera: Tracking shot, medium-wide.
Lighting: Dappled sunlight through the forest canopy.
Speech: "Thanks man. How about appropriate outfit for this occasion?" (Relieved then urgent).

[00:28–00:43]
Subject: Man transitions into a full black and grey scuba diving suit with a mask and regulator.
Environment: Sudden transition from forest to deep blue ocean water.
Action: Man is underwater, initially confused, then sees a massive Great White Shark swimming towards him. He swims frantically.
Camera: Underwater wide shot, slow-motion feel.
Lighting: Caustic light rays from the surface, deep blue hues.
Speech: Muffled underwater vocalizations.

[00:43–01:12]
Subject: Man transitions to just blue swim trunks, shirtless.
Environment: Transition to a vast, orange-sand desert.
Action: Man is running across sand dunes. A rusted, post-apocalyptic car with two crazed men firing flamethrowers chases him.
Camera: High-speed tracking shot, low angle.
Lighting: Harsh, direct desert sun, high contrast.
Speech: "A little help here! I can't run like this!" (Panting, desperate).

[01:12–01:58]
Subject: Man transitions into a sharp navy blue suit with a white shirt and tie.
Environment: Transition to a high-end, candlelit restaurant.
Action: Man is sitting at a table, half-naked initially, then the suit appears. Suddenly, gunmen in black masks burst in. He dives under the table and crawls through the restaurant as sparks and bullets fly.
Camera: Close-ups on face, then wide shots of the chaos.
Lighting: Warm, moody candlelight with bright muzzle flashes.
Speech: "Do you mind?" followed by "Are you okay?" "I'm fine... this is embarrassing."

[01:58–02:47]
Subject: Man in the same navy suit, now disheveled and dusty.
Environment: A luxury hotel room with a view of the city at night.
Action: Man enters the room. The entire room begins to rotate 360 degrees. Furniture slides around. He falls against the window, it shatters, and he falls out.
Camera: Rotating camera movement matching the room's rotation.
Lighting: Warm interior lamps, cool blue city lights outside.
Motion: Pillows, chairs, and lamps flying through the air.

[02:47–03:15]
Subject: Man in suit, clinging to the wing of a small white propeller plane.
Environment: High altitude over a tropical jungle river.
Action: He loses his grip and falls through the air, landing in a river. He crawls onto the bank and is hit in the neck by a tribal arrow. He is then shown tied to a tree at night.
Camera: Wide aerial shots, then tight close-ups on the face.
Lighting: Sunset over jungle, then flickering orange firelight.
Speech: "I didn't mean on the plane like this!"

[03:15–03:20]
Subject: Man in striped pajamas, sweating and breathing heavily.
Environment: A dark, comfortable bedroom.
Action: Man sits up abruptly in bed, realizing it was a dream. He wipes his face and sighs.
Camera: Close-up on face.
Lighting: Dim, low-key moonlight.
Speech: "Nice." (Relieved, quiet).

NEGATIVE PROMPT:
Visual: Character face morphing, inconsistent eye color, extra limbs, flickering clothing textures, blurry face, distorted background architecture, unnatural physics in non-rotating scenes, low resolution, watermarks, text logos.
Speech: Robotic voice, mismatched lip-sync, audio clipping, background hiss, unnatural pauses, monotone delivery.

SPEECH PACK:
[00:03] "Really interesting action shot."
TAKE_A: (Casual, impressed)
TAKE_B: (Slightly sarcastic)
TAKE_C: (Curious)

[00:36] "Thanks man. How about appropriate outfit for this occasion?"
TAKE_A: (Breathless, relieved)
TAKE_B: (Quick, urgent)
TAKE_C: (Grateful but rushed)

[01:46] "A little help here! I can't run like this!"
TAKE_A: (Screaming, high energy)
TAKE_B: (Strained, panting)
TAKE_C: (Desperate, cracking voice)

[02:05] "Do you mind?"
TAKE_A: (Annoyed, dry)
TAKE_B: (Embarrassed, quiet)
TAKE_C: (Polite but confused)

[03:22] "I didn't mean on the plane like this!"
TAKE_A: (Terrified scream)
TAKE_B: (Frantic, high pitch)
TAKE_C: (Gasping for air)

[04:58] "Nice."
TAKE_A: (Deep sigh of relief)
TAKE_B: (Whispered, exhausted)
TAKE_C: (Slightly amused)

How erik_gen_ai Made This Higgsfield AI Cinematic Storytelling Video — and How to Recreate It

This high-octane showcase for Higgsfield AI is a masterclass in cinematic AI storytelling and character consistency. It features a single protagonist navigating a rapidly shifting reality—from a New York earthquake to a deep-sea shark chase, a Mad Max-style desert pursuit, and even a space-bound alien encounter. The video uses a "meta-prompting" narrative device where the character’s reality is literally rewritten by on-screen text prompts. With its high-fidelity textures, complex physics (rotating rooms, crumbling buildings), and seamless costume swaps, it serves as a "North Star" for indie creators looking to push the boundaries of AI-generated video content.

What You’re Seeing: A Cinematic AI Odyssey

The video is a 300-second epic that blends multiple genres: disaster, thriller, action, and sci-fi. The subject is a middle-aged Caucasian man with brown hair, whose wardrobe and environment are constantly altered by "AI prompts." The lighting shifts from harsh daylight in the city to moody, warm candlelight in a restaurant, and finally the cold, high-contrast lighting of deep space. The editing is fast-paced, using "match cuts" and environmental transitions to keep the viewer disoriented yet engaged.

Shot-by-shot Breakdown

Time Range	Visual Content	Shot Language	Lighting & Color	Viewer Intent
00:00–00:14	City street splits open; buildings crumble in an earthquake.	Wide shot to MCU; shaky cam.	Natural daylight; dusty/grey.	Hook: Immediate high-stakes action.
00:14–00:43	Transition to forest; man gets hiker gear; chased by a bear.	Tracking shot; medium framing.	Lush greens; soft forest light.	Reinforce persona: The "Everyman" survivor.
00:43–01:12	Underwater transition; scuba gear; shark attack.	POV and close-ups; slow motion.	Deep blue; caustic light rays.	Curiosity: How far can the AI go?
01:12–01:58	Desert chase; flamethrowers; transition to a fancy restaurant.	High-speed tracking; wide shots.	Orange/Teal; warm desert sun.	Contrast: From chaos to "civilization."
01:58–02:47	Rotating hotel room; falling out a window to a plane wing.	Dutch angles; rotating camera.	Warm interior to night city.	Aesthetic value: Complex physics showcase.
02:47–03:20	Jungle capture; desert car jump; space alien fight.	Rapid cuts; ECU to Wide.	Firelight; harsh space contrast.	Climax: Peak intensity and genre-bending.

Why It Went Viral: The "God Mode" Narrative

The Power of the "Meta-Prompt"

The core of this video’s success is the "God Mode" fantasy. By showing the prompts on screen, the creator taps into the collective curiosity about AI's power. It’s not just a cool video; it’s a demonstration of control. The audience isn't just watching a man run from a bear; they are watching the process of creation in real-time. This "meta" layer reduces the explanation cost—you don't need a caption to tell you it's AI; the video proves it through its own narrative structure.

Psychological Hook: The Dream Logic

The video follows the erratic, high-stakes logic of a nightmare. One moment you're in a restaurant, the next you're falling through the sky. This triggers a biological response: hyper-vigilance. The viewer cannot look away because the environment might change at any second. The final "waking up" scene provides a satisfying emotional release, resolving the tension built over the previous five minutes.

Platform Perspective: Retention & Re-watchability

From a platform perspective (Instagram/TikTok), this video is a retention monster. Because the scene changes every 10–15 seconds, the "boredom threshold" is never reached. Each new prompt acts as a "micro-hook" that resets the viewer's attention span. Furthermore, the sheer density of detail—the way the suit fits, the physics of the rotating room, the alien's texture—encourages re-watching to catch things missed the first time.

5 Testable Viral Hypotheses

The "Prompt-as-Action" Hypothesis: Including the text prompt inside the video frame increases watch time by 20% because it creates a "cause-and-effect" loop.
The "Genre-Mashup" Hypothesis: Rapidly switching between 3+ distinct genres (Horror, Action, Sci-Fi) prevents "scroll-past" behavior by constantly subverting expectations.
The "Character Consistency" Flex: Using the exact same face in wildly different lighting (Space vs. Candlelight) proves technical mastery, leading to higher "Saves" for reference.

The "Physics Porn" Hypothesis:

The "Relatable Nightmare" Hook: Ending a high-intensity sequence with the "it was all a dream" trope increases "Shares" because it resonates with a universal human experience.

How to Recreate: From Prompt to Production

Step 1: Define Your "Anchor" Character

To achieve the consistency seen in the video, you need a strong character reference. Use a tool like Midjourney to create a "Character Sheet" (front, side, 3/4 views). This will be your --cref (Character Reference) for all subsequent generations.

Step 2: Storyboard the "Reality Shifts"

Don't just prompt randomly. Plan your transitions. Use "Match Cuts"—if the character is running in the forest, have them be in the same running pose when they transition to the desert. This maintains the visual flow.

Step 3: Generate Keyframes for Each Scene

Generate high-quality stills for the start and end of each major sequence.

Scene A: Man in city, looking back.
Scene B: Man in forest, wearing a backpack.

Step 4: Use Video-to-Video or Image-to-Video

Tools like Kling AI, Luma Dream Machine, or Runway Gen-3 are ideal. Upload your keyframe and use the "Motion Brush" or specific motion prompts (e.g., "ground splitting open, camera shaking") to animate the scene.

Step 5: Implement the "Prompt Overlay"

In your editing software (CapCut/Premiere), add a text box that mimics a terminal or a search bar. Use a "typewriter" effect to show the prompt being entered right before the visual change occurs.

Step 6: Sound Design is 50% of the Magic

The video relies heavily on foley. You need the sound of crumbling stone, the roar of a bear, the splash of water, and the hum of a spaceship. Use an AI audio tool like ElevenLabs for the character's voice and Udio or Suno for the cinematic score.

Step 7: Color Grading for Consistency

Even though the scenes change, apply a global "film grain" or a slight "teal and orange" grade to the entire video to make it feel like a single cohesive movie rather than a collection of clips.

Step 8: The "Loop" or "Resolution"

Always end with a scene that brings the viewer back to reality. The "waking up" trope works, but you could also have the character "exit" the AI simulation.

Growth Playbook: Distribution & Scaling

3 Opening Hook Lines

"I let AI rewrite my reality for 24 hours... it got weird."
"This is why you should never give an AI 'God Mode' permissions."
"The most realistic AI video I've ever seen. Watch the transitions."

4 Caption Templates

The Tech Enthusiast: "AI video is officially at 'Inception' levels. 🤯 I used [Tool Name] to create this reality-bending sequence. Which transition was your favorite? 👇"
The Storyteller: "What if your thoughts became reality instantly? That’s the prompt I gave the AI. The result was a literal nightmare. 😱"
The Tutorial Tease: "Character consistency in AI is finally solved. 🧵 I’m breaking down exactly how I kept the same face across 10 different worlds in my bio link!"
The Short & Punchy: "Prompt: Everything, everywhere, all at once. AI delivered. 🎥✨"

Hashtag Strategy

Broad: #AI #VFX #Filmmaking #DigitalArt #FutureTech
Mid-tier: #AIVideo #HiggsfieldAI #CinematicAI #GenerativeAI #CreativeTech
Niche: #CharacterConsistency #AIStorytelling #IndieCreator #AIAnimation #PromptEngineering

Frequently Asked Questions

What tools make it look the most similar?

Higgsfield AI (as seen in the video), Kling AI, and Runway Gen-3 Alpha are currently the leaders in this level of motion and consistency.

How do I keep the face the same in every scene?

Use a "Character Reference" (Cref) image and keep your base prompt describing the man's features identical in every generation.

Why does my AI video look "jittery"?

Jitter usually happens due to low frame rates or conflicting motion prompts; try using a higher "Motion" setting and consistent seed numbers.

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

Instagram currently favors high-production "aesthetic" AI, while TikTok favors the "meta" storytelling and tutorial aspects.

How should I disclose AI use?

Use the platform's "AI Generated" tag and mention the tools used in the caption to build trust with your audience.