0:00 / 0:00

Comment “prompt” & I’ll send full guide👇 How to relight a light photo with AI and Nanobanana Pro. #relight #relightphoto #nanobananapro #nanobananaprompts

Gunhild Johanne Reumert

@shedoesai · Digital creator

INSTAGRAM · 2026-01-11Source

44likes

86comments

Remix This

Recreate with Kling 3

Make your own AI viral video

Prompt

GLOBAL LOCK:
Subject is a young blonde woman, light skin with warm undertones, approximately 20-25 years old. She has long, wavy blonde hair with curtain bangs. Her facial features are reminiscent of a pop star aesthetic (soft features, full lips). Wardrobe includes a pink camisole with strawberry prints and a white floral sundress. The environment for the subject is a soft-focus indoor bedroom or studio. The creator (narrator) is a blonde woman in her late 20s, wearing a black graphic t-shirt, holding a black Rode microphone. Creator's background is a dark studio with bookshelves and warm orange/red practical globe lights. Lighting for the subject transitions from flat/amateur to cinematic backlighting. Color grade is warm, saturated, and editorial.

[00:00–00:01]
Subject: Young blonde woman in pink strawberry-print camisole, sitting on a black stool.
Environment: Plain white wall background.
Action: Subject looks at the camera with a neutral, slightly pouty expression.
Framing: Medium shot, eye level.
Lighting: Flat, direct flash-style lighting, creating a slight shadow on the wall.
Motion: Static shot with a slight digital zoom-in.

[00:01–00:02]
Subject: Same woman, now in a white sundress with small red floral patterns.
Environment: Soft-focus indoor wall.
Action: Subject poses with a slight head tilt, looking into the camera.
Framing: Medium close-up.
Lighting: Soft, diffused side lighting.
Motion: Quick cut transition.

[00:02–00:03]
Subject: Same woman, close-up on face.
Action: She winks with her right eye and holds up a peace sign with her hand near her face.
Framing: Close-up (CU).
Lighting: Natural, low-contrast indoor light.
Motion: Quick cut transition.

[00:03–00:10]
Visual: iPhone screen recording showing the Google Gemini app interface. A circular overlay in the bottom center shows the creator speaking into a Rode microphone.
Action: On screen, the user selects "Create Image," uploads the winking photo, and pastes a long text prompt. The creator in the overlay is speaking and gesturing with her free hand.
Speech (Creator): "First head to Gemini and hit create image. Upload the photo, paste the prompt, hit generate, and just wait for the new image to be generated."
Cadence: Fast, instructional, energetic.
Mic Signature: Close-up, crisp, studio quality.

[00:10–00:14]
Subject: The creator (narrator) full screen.
Environment: Studio with bookshelves and warm orange/red glowing bulbs.
Action: Creator speaks directly to the camera, holding the Rode mic close to her mouth. She gestures towards the bottom of the screen.
Framing: Medium close-up (MCU).
Lighting: Dramatic studio lighting with warm rim lights.
Speech (Creator): "If you want the prompt I used, just comment 'prompt' and I'll send it over."
Cadence: Warm, inviting, clear call to action.
Lip Sync: High strictness; mouth movements must match the words "prompt" and "send it over."

NEGATIVE PROMPT:
Visual: distorted facial features, inconsistent hair color, blurry hands, extra fingers, flickering lights in background, low-resolution screen recording, robotic body movements, harsh shadows on face in the final "relit" version.
Speech: muffled audio, background hiss, robotic voice, misaligned lip-sync, long pauses between sentences, monotone delivery.

SPEECH PACK:
[00:03–00:10]
Transcript: "First head to Gemini and hit create image. Upload the photo, paste the prompt, hit generate, and just wait for the new image to be generated."
TAKE_A: (Fast, upbeat) "First head to Gemini and hit create image! Upload the photo, paste the prompt, hit generate... and just wait for the new image to be generated."
TAKE_B: (Instructional, steady) "First, head to Gemini and hit 'create image'. Upload your photo, paste the prompt, hit generate, and wait for the new image."
TAKE_C: (Casual, conversational) "So, first go to Gemini and click create image. You're gonna upload the photo, paste that prompt, hit generate, and then just wait for the magic."

[00:10–00:14]
Transcript: "If you want the prompt I used, just comment 'prompt' and I'll send it over."
TAKE_A: (Direct, friendly) "If you want the prompt I used, just comment 'prompt' and I'll send it over!"
TAKE_B: (Helpful, soft) "If you'd like the prompt I used here, just comment 'prompt' below and I'll send it right over to you."
TAKE_C: (Punchy, CTA-focused) "Want the prompt? Comment 'PROMPT' and I'll send it over now."

Why shedoesai's Relight Photo with AI Nanobanana Pro Went Viral — and the Formula Behind It

This viral tutorial by @shedoesai demonstrates a high-efficiency AI workflow for "relighting" flat, amateur photos into cinematic editorial portraits using Google Gemini. The video leverages a Sabrina Carpenter-esque aesthetic—featuring a blonde subject in soft, bedroom-style lighting—to capture immediate visual interest. By combining a screen-recorded software walkthrough with a high-quality "talking head" studio setup, the creator bridges the gap between complex AI technology and everyday social media utility. The core hook, "How to Light a Photo without Lights," addresses a common pain point for creators: poor lighting conditions. The video effectively uses a ManyChat-style engagement loop, promising a "full guide" in exchange for a specific comment, which significantly boosts the post's reach through platform signals.

What You’re Seeing: Visual Breakdown

The video is a hybrid of UGC (User Generated Content) aesthetics and professional studio production. It features a young blonde woman as the subject of the AI transformation, wearing a strawberry-print camisole and a floral dress, evoking a "cozy-chic" vibe. The creator herself appears as a narrator, positioned in a well-lit studio with warm practical lights (orange and red bulbs) and a professional Rode microphone, which establishes immediate authority. The color palette transitions from the flat, slightly overexposed "before" photos to the rich, high-contrast "after" images characterized by ethereal rim lighting and golden hour warmth. The editing is fast-paced, utilizing a screen recording of an iPhone interface to show the Gemini AI app in action, making the "magic" feel tangible and accessible.

Shot-by-Shot Analysis

Time Range	Visual Content	Shot Language	Lighting & Tone	Viewer Intent
00:00–00:02	Rapid cuts of blonde subject in strawberry PJs and floral dress.	Close-up (CU) / Medium Shot (MS)	Flat, amateur flash lighting.	Hook: Establish the "before" state and aesthetic appeal.
00:02–00:03	Subject winking with a peace sign.	Close-up (CU)	Soft, low-contrast natural light.	Reinforce persona and relatability.
00:03–00:10	iPhone screen recording of Gemini app with creator in a circular overlay.	Screen Capture + Overlay	UI-focused; Creator in warm studio light.	Tutorial: Show the "how-to" process step-by-step.
00:10–00:14	Creator full-screen, holding a Rode mic, speaking to camera.	Medium Close-up (MCU)	Cinematic studio lighting (warm practicals).	CTA: Build trust and drive engagement (comments).

Why It Went Viral

The Power of the "Sabrina Aesthetic"

The choice of subject is no accident. The blonde, "cozy-girl" aesthetic heavily mirrors the current visual brand of pop stars like Sabrina Carpenter. This taps into biological triggers for beauty and cultural trends simultaneously. For the audience—largely Gen Z and Millennial creators—this look is highly aspirational. By showing how to achieve this "editorial" look with a simple AI prompt, the creator provides massive perceived value. The "before" photos look like something anyone could take on their phone, while the "after" looks like a professional photoshoot, creating a powerful contrast gap that keeps viewers watching.

Platform Signals & Engagement Loops

From a platform perspective, this video is a masterclass in retention and conversion. The first 3 seconds offer a clear "What's in it for me?" (lighting photos without lights). The middle section provides the "Aha!" moment by showing the AI actually working. Finally, the "Comment PROMPT" call-to-action is the engine of its virality. On Instagram, a high comment-to-view ratio signals to the algorithm that the content is highly engaging, pushing it to the Explore page. Furthermore, using an automated DM tool (like ManyChat) ensures that every comment leads to a direct interaction, further boosting the account's health.

5 Testable Viral Hypotheses

The Lookalike Effect: Using a subject that resembles a trending celebrity (Sabrina Carpenter) increases initial stop-rate by 40%.
The "Magic Wand" Fallacy: Showing a complex task (professional lighting) being solved by a single "paste" action creates a high save rate because it feels like a "cheat code."
The Authority Anchor: Holding a physical, professional microphone (Rode) while explaining AI tech increases viewer trust and watch time compared to using phone audio.
The UI Transparency: Showing the actual phone screen recording reduces "AI skepticism" because the viewer can see the exact buttons to press.
The Frictionless CTA: Asking for a one-word comment ("prompt") has a 5x higher conversion rate than asking a complex question.

How to Recreate (Step-by-Step)

Identify a Visual Pain Point: Choose a common problem (e.g., bad lighting, messy background, boring outfit) that AI can fix.
Source Your "Before" Image: Use a high-quality but "flat" photo. Using a subject that fits a current aesthetic (like the blonde/cozy look) helps.
Set Up Your Studio: Record your "talking head" segments with a clean background. Use warm practical lights (lamps) behind you to create depth.
Record the Workflow: Use your phone's screen recorder to capture the process in Gemini or your AI tool of choice. Keep it snappy—no long loading screens.
Draft the "Magic" Prompt: Create a prompt that focuses on "Backlighting," "Rim Light," and "Cinematic Glow." (See the Prompt section below).
Edit for Pacing: Use a "Hook -> Process -> Result -> CTA" structure. Ensure the transition from the screen recording to your face is seamless.
Set Up Automation: Use a tool like ManyChat to automatically send the "Full Guide" to anyone who comments your keyword.
Optimize for Reels/TikTok: Use trending, low-volume audio that complements a "tutorial" vibe—something rhythmic but not distracting.

Growth Playbook: Distribution & Scaling

3 Opening Hook Lines

"Stop throwing away photos with bad lighting. Do this instead."
"How to get the 'Sabrina Carpenter' glow without a professional crew."
"I found a secret AI button that fixes any flat photo in 5 seconds."

4 Caption Templates

Option 1: The Value Lead
Stop settling for flat photos! 📸 I used to spend hours in Lightroom, but now I just use this AI trick.
💡 Value: Professional rim lighting in one click.
👇 Comment "LIGHT" and I'll DM you the exact prompt I used!

Hashtag Strategy

Broad (Reach): #AI #PhotographyTips #PhotoEditing #ContentCreator
Mid-Tier (Niche): #GeminiAI #AITutorial #Relight #SocialMediaGrowth
Long-Tail (Specific): #NanobananaPro #ShedoesAI #SabrinaCarpenterAesthetic #AIRelighting

Frequently Asked Questions

What tools make it look the most similar?

Google Gemini (Advanced) or Midjourney's "Vary Region" tool are best for this.

What are the 3 most important words in the prompt?

"Backlighting," "Rim light," and "Ethereal glow."

Why does the generated face look inconsistent?

Ensure you are using a "Relight" or "Image-to-Image" function rather than a full regeneration.

How can I avoid making it look like AI?

Keep the "Denoising Strength" or "Image Influence" high so the original facial structure remains.

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

Instagram Reels currently favors "aesthetic tutorial" content with strong comment-based CTAs.