this video is 100% AI.... heres exactly how i made it: > use the sora 2 prompting guide + claude > put the prompt into sora 2 pro > upscaled in topaz there's a lot of noise about new AI video tools every single day but sora 2 has remained one of the best models for AI ads and https://t.co/Vx2e9ocKBw

Miko

@Mho_23 · creator

TWITTER · 2026-03-05Source

358likes

26comments

Remix This

Prompt

GLOBAL LOCK:
Subject: Young East Asian woman, mid-20s, slim build, black hair with straight-cut bangs and loose strands framing the face, warm skin tone with natural texture.
Wardrobe: Simple black spaghetti-strap tank top.
Environment: Dimly lit bedroom, warm indoor lighting from a bedside lamp. Background includes a red flag/tapestry on the wall, a corkboard with pinned photos, and a bed with neutral-colored sheets.
Camera: Handheld iPhone-style movement, slight natural shake, Medium Close-Up (MCU) framing, 35mm lens feel with soft background bokeh.
Lighting: Warm, low-key, motivated by a lamp to the right, creating soft highlights on the face and deep shadows in the room.
Speech/Audio: Single speaker, female, youthful and casual tone, UGC confessional style, close-mic proximity with slight room ambience.

[00:00–00:03]
Subject is centered, holding a small wired lavalier microphone close to her mouth with her right hand. She looks directly into the lens with a serious, slightly tired expression. She begins speaking: "Okay, I literally thought I was just anxious." Her lips move in high-sync with the words. Subtle eye blinks. Camera has a very slight handheld sway.

[00:03–00:06]
Subject tilts her head slightly to the left, eyes looking momentarily up as if reflecting. "Like, I accepted that." Her expression is one of resignation. The lighting catches the gloss on her lips. Hand holding the mic remains steady.

[00:06–00:10]
Subject's expression shifts to realization. She looks back at the camera. "Turns out I was just super magnesium deficient. I started taking these magnesium gummies before bed..." Her eyebrows raise slightly on "super." Natural head movements and shoulder shifts.

[00:10–00:13]
Subject breaks into a wide, genuine smile, showing teeth. She looks down and away briefly, then back at the camera. "...and my body finally shut up." The smile is warm and reaches her eyes (Duchenne smile). High motion intensity in the facial muscles.

[00:13–00:15]
Subject gives a final small nod and a lingering smile. "Yeah, that was it." She looks relaxed. The video ends on this positive, casual note. The handheld camera shake remains consistent until the final frame.

NEGATIVE PROMPT:
Visual: Robotic movements, frozen facial features, inconsistent hair strands, flickering background lights, distorted fingers on the microphone, plastic-looking skin, over-saturated colors, studio lighting, tripod-steady camera.
Speech: Monotone voice, robotic cadence, mismatched lip-sync, harsh 'S' sounds, loud background noise, sudden cuts in audio, unnatural pauses between words.

SPEECH PACK:
Transcript: "Okay, I literally thought I was just anxious. Like, I accepted that. Turns out I was just super magnesium deficient. I started taking these magnesium gummies before bed and my body finally shut up. Yeah, that was it."

TAKE_A (Natural/Casual): [00:00] Okay... [pause] I literally thought I was just anxious. [00:03] Like... I accepted that. [00:06] Turns out I was just *super* magnesium deficient. [00:08] I started taking these magnesium gummies before bed [00:11] and my body finally... [laugh] shut up. [00:13] Yeah, that was it.
TAKE_B (Fast/Energetic): [00:00] Okay I literally thought I was just anxious like I accepted that. [00:05] Turns out I was just super magnesium deficient! [00:07] Started taking these gummies before bed and my body finally shut up. [00:12] Yeah that was it!
TAKE_C (Confessional/Soft): [00:00] [breath] Okay... I thought I was just anxious. [00:03] I just accepted it. [00:05] But I was just magnesium deficient. [00:08] Started taking gummies... [00:10] and my body finally shut up. [00:13] Yeah. That was it.

Prosody Markup:
- "literally": emphasis
- "anxious": falling intonation
- "super": elongated vowel, high pitch
- "shut up": punchy delivery followed by a soft laugh.
- "Yeah...": trailing off with a smile.

How Mho_23 Made This Magnesium Deficiency AI Video

This case study examines a 15-second AI-generated "talking head" video that perfectly mimics the UGC (User-Generated Content) aesthetic common on TikTok and Instagram Reels. Featuring a young East Asian woman in a dimly lit bedroom, the video uses a cinematic yet raw editorial portrait style. The core keywords here are "AI realism," "health hack storytelling," and "iPhone bedroom aesthetic." Despite being 100% AI, the video achieves high engagement by tapping into the "anxiety vs. nutrition" discourse, a high-growth niche in the wellness space. The creator, @mho-23, leverages Sora 2 for generation and Topaz Video AI for upscaling, proving that high-fidelity AI characters are now indistinguishable from real creators in low-stakes, high-trust environments.

What You’re Seeing: A Detailed Breakdown

The video features a single subject: a young East Asian woman with dark hair and bangs, wearing a simple black tank top. She is positioned in a Medium Close-Up (MCU) shot, holding a small, wired lavalier-style microphone close to her mouth—a classic "ASMR/confessional" visual cue used by Gen Z creators to signal authenticity. The scene is set in a typical bedroom with warm, low-key lighting coming from a bedside lamp (off-camera right), creating soft shadows on her face. Background elements like a red flag, a corkboard with photos, and a messy bed add to the "lived-in" feel. The color grade is warm with a slight film grain, and the motion is remarkably fluid, including subtle micro-expressions, eye blinks, and natural hand movements that sync with her speech.

Shot-by-Shot Analysis

Time Range	Visual Content	Shot Language	Lighting & Tone	Viewer Intent
00:00–00:03	Subject looks into lens, holding mic, serious expression.	MCU, Static, Eye-level	Warm, low-key, soft shadows	Hook: Immediate eye contact & relatable opening line.
00:03–00:06	Slight head tilt, "I accepted that." Realization phase.	MCU, subtle sway	Consistent warm glow	Build Trust: Vulnerability through "confession."
00:06–00:10	Explaining the "fix" (magnesium). Looking slightly off-camera.	MCU, natural eye movement	Golden hour bedroom vibe	Value Drop: Providing the "secret" solution.
00:10–00:13	"Body finally shut up." Big smile, looking down then up.	MCU, high motion intensity	Highlights on cheekbones	Emotional Payoff: Relief and satisfaction.
00:13–00:15	Final nod, "Yeah, that was it." Smile lingers.	MCU, fading motion	Soft fall-off	Retention: Loop-friendly ending.

Why It Went Viral: The Mechanics of Relatability

The Power of the "Health Hack" Script

The script taps into a massive psychological pain point: anxiety. By framing a common mental health struggle as a simple nutritional deficiency (magnesium), the video triggers a "discovery" response in the viewer. It’s not just a video; it’s a potential solution to a problem millions face. The "confessional" style—holding the mic close—mimics the intimacy of a FaceTime call with a friend, which bypasses the viewer's "ad filter."

The "Uncanny Valley" as a Growth Engine

From a platform perspective, this video wins because it sits on the edge of the Uncanny Valley. Viewers often watch multiple times or check the comments to see if the person is real. This "is it AI?" curiosity significantly boosts watch time and comment density, two primary signals that tell the Instagram/TikTok algorithm to push the content to a wider audience. The creator's transparency ("this video is 100% AI") actually fuels the fire, turning a simple health tip into a tech demonstration.

5 Testable Viral Hypotheses

The Intimacy Hook: Using a handheld mic in a bedroom setting increases retention by 30% compared to professional studio setups because it signals "authentic UGC."
The "Problem-Pivot" Script: Starting with a negative emotion (anxiety) and pivoting to a physical solution (magnesium) creates a high "save" rate for future reference.
The Micro-Expression Loop: Including a genuine-looking laugh or smile at the end of a short clip encourages re-watching to catch the "human" detail.
The Tech Disclosure: Explicitly stating "This is AI" in the caption of a realistic video generates 2x more comments than not mentioning it.
The Low-Light Aesthetic: Warm, "messy" bedroom lighting performs better for personal storytelling than bright, clean lighting which feels "corporate."

How to Recreate: From 0 to 1

Step 1: Topic Selection & Scripting

Choose a "relatable struggle" (e.g., "I thought I was lazy, turns out I just needed Vitamin D"). Use Claude or ChatGPT to write a 15-second script in a "Gen Z confessional" tone. Keep sentences short and punchy.

Step 2: Character & Environment Consistency

Define your character: "East Asian woman, mid-20s, casual wardrobe." Keep the environment consistent by using keywords like "dimly lit bedroom," "corkboard background," and "warm lamp light."

Step 3: AI Video Generation (Sora 2 / Kling / Luma)

Input your prompt (see the Prompt section below). Focus on "Medium Close-Up" and "Direct to camera" to ensure the AI focuses on facial expressions and lip-syncing.

Step 4: Lip-Sync & Audio

If your video generator doesn't handle audio perfectly, use a tool like ElevenLabs for the voice and SyncLabs or Hedra to align the lips to the new audio track.

Step 5: Upscaling for Realism

Run the final render through Topaz Video AI. Use the "Iris" or "Proteus" model to enhance skin texture and remove AI "smudging," making it look like high-quality iPhone footage.

Step 6: Editing & Overlays

Add "dynamic captions" using CapCut. Use a font like 'The Bold Font' or 'Montserrat' and highlight key words (e.g., "ANXIOUS," "MAGNESIUM") in a different color.

Step 7: The "Hook" Title

Create a cover image with a text overlay: "Anxiety or just a deficiency?" This forces the user to stop scrolling.

Step 8: Publishing Strategy

Post as a Reel/TikTok. In the first 30 minutes, reply to every comment asking about the AI tool to boost the engagement rate.

Growth Playbook: Distribution & Scaling

3 Opening Hook Lines

"I literally thought I was losing my mind, but..."
"Stop scrolling if you feel 'anxious' for no reason."
"The one supplement that actually fixed my sleep."

4 Caption Templates

The Tech Reveal: "This video is 100% AI. 🤖 Can you tell? I used Sora 2 + Topaz to create this. The future of content is here. What do you think?"
The Value Drop: "I spent years thinking I had [Problem]. Turns out it was just [Solution]. 💊 Full breakdown in the comments. Have you tried this?"
The Curiosity Gap: "The secret to [Result] isn't what you think. 🤫 Watch until the end. Tag a friend who needs to hear this."
The Short & Sweet: "Magnesium gummies > Anxiety. ✌️ That's the post. Who else relates?"

Hashtag Strategy

Broad (Reach): #AI #Wellness #HealthTips #TechTrends
Mid-Tier (Niche): #MagnesiumDeficiency #AIVideo #ContentCreatorTips #Biohacking
Long-Tail (Community): #SoraAI #AIInfluencer #AnxietyReliefTips #UGCStyle

Frequently Asked Questions

What tools make it look the most similar?

Sora 2 for generation, ElevenLabs for voice, and Topaz Video AI for the final high-res texture.

What are the 3 most important words in the prompt?

"UGC," "Medium Close-Up," and "Micro-expressions."

Why does the generated face look inconsistent?

Usually due to high motion; use a lower "motion bucket" setting or shorter clips (3-5s) and stitch them.

How can I avoid making it look like AI?

Add a slight film grain overlay and ensure the lighting is "motivated" (comes from a visible source).

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

TikTok favors the "tech reveal" and "is it real?" debate more than Instagram.