AI Design just got a whole lot easier with lovarts design agent 🙌 Are you using Banana correctly? @Lovart_ai is the ultimate answer Subscribe to Lovart’s plan before Sep 14 (PST) and enjoy unlimited Nano Banana model usage throughout your membership Want to try it yourself? Comment “AI” for a link! #LovartAI #CreateWithLovart #NanoBanana #AIDesignAgent #AIImageGenerator #GenerativeAI #CreativeAI #AIForDesigners #AIForCreators #AIContentCreation #PosterDesignAI #LogoDesignAI #BrandingWithAI #EcommerceAI #SmartDesign #NextGenCreativity #InnovativeAI #DigitalCreators #VisualStorytelling #FutureOfDesign #CreatorEconomy

Rourke Sefton-Minns

@rourke · creator

INSTAGRAM · 2025-09-09Source

24.4Klikes

8.6Kcomments

Remix This

Prompt

GOAL
Maximize visual + motion + speech similarity to the reference video, prioritizing the composite nature of the shot (talking head overlaid on screen recording), the specific UI elements shown, and the energetic, fast-paced tutorial delivery.

WORKFLOW
A) MISE EN PLACE (prep)
- Invariants: The composite layout (split-screen style with the subject keyed out in the bottom center). The subject's appearance (Caucasian male, 30s, short beard, beige "Vans Off The Wall" cap, plain white t-shirt). The background environment (a crisp screen recording of a web browser showing the Lovart.ai interface). The lighting on the subject (soft, even, frontal). The audio signature (close-mic, dry, energetic podcast style).
- Variables: The specific content shown on the background screen recording (which changes rapidly), the subject's hand gestures and facial expressions, the text overlays.

B) SHOTLIST (blueprint)
- shot_id: 1, timecode_start: 00:00, timecode_end: 00:54, duration: 54s
- framing: Static MCU (Medium Close-Up) for the subject in the bottom center. The background is a full-screen capture.
- lens: 35mm equivalent for the subject, sharp focus. Infinite depth of field for the screen recording.
- camera movement: Static camera for the subject. The background screen recording features digital zooms and pans to highlight UI elements.
- subject: Caucasian male, 30s, short beard, beige "Vans" cap, white t-shirt. Energetic, using open-palm gestures, pointing upwards.
- environment: Background is a digital screen recording of a UI.
- lighting: Subject has soft, even ring-light style illumination.
- color grade: Natural, warm skin tones for the subject. High contrast, vibrant colors for the UI elements (especially the red handbag and blue Dior bottle).
- motion cues: Subject's hands moving rapidly, screen recording UI elements changing instantly without loading times.

C) STYLE BIBLE (global)
- visual_style: UGC tech tutorial composite.
- camera_signature: Locked-off camera for the subject, dynamic digital screen capture for the background.
- lighting_signature: High-key, flat lighting on the subject to ensure a clean chroma key look.
- grade_signature: Clean, sharp, high-definition digital look.
- pacing_signature: Extremely fast-paced, cutting out all dead air and loading times.
- speech_style: Energetic, authoritative, direct-to-camera tech tutorial voiceover.
- speaker_profile: Male, mid-30s, enthusiastic, fast talker, clear enunciation.
- mic_mix_profile: Dry, close-mic, heavily compressed for maximum clarity and presence, no room reverb.

D) PROMPT SYNTHESIS

MASTER PROMPT
GLOBAL LOCK: A continuous composite video. The background is a crisp, high-resolution screen recording of a web browser navigating a design tool interface. In the bottom center of the frame, overlaid on top of the screen recording, is a talking-head subject: a Caucasian male in his 30s, medium-length brown hair, a short beard, wearing a beige "Vans Off The Wall" baseball cap and a plain white crew-neck t-shirt. The subject is well-lit with soft, even frontal lighting, casting no harsh shadows, and has a clean cutout (green screen effect). The camera angle for the subject is a static medium close-up (chest up). The overall visual style is a polished UGC tech tutorial. The speech is energetic, fast-paced, and authoritative, recorded with a close-mic, dry podcast-style audio signature.

[00:00–00:10] The background screen shows a collage of high-end design posters, including a glossy red patent leather handbag and a dark blue Dior Sauvage perfume bottle. The subject in the bottom center gestures enthusiastically with both hands, palms open, explaining the concept. A text overlay "Nano Banana x Design" appears at the top.
[00:10–00:20] The background screen transitions to show the process of generating a Dior Sauvage ad. The screen zooms slightly on the generated images featuring a model holding the bottle. The subject continues to talk rapidly, using small hand chops to emphasize points. A red circle graphic appears around the text "AI Design Agent" on the screen.
[00:20–00:35] The background screen shows the Lovart.ai interface. A red handbag image is uploaded. The screen shows text being typed: "create me 2 posters of this handbag, one elegant, and one bold". The subject points up at the screen with his index finger.
[00:35–00:45] The background screen displays the generated handbag posters. The UI shows the user selecting one, and then editing it to add a female model in a black suit holding the bag. The subject brings his hands together in a clarifying gesture.
[00:45–00:54] The background screen shows a new prompt being entered to place the poster in Times Square. The screen instantly updates to show the red handbag poster displayed on a massive glowing billboard in a neon-lit Times Square at night. The subject points directly at the camera. A large text overlay "Comment 'AI'" appears at the top of the screen.

NEGATIVE PROMPT
Visual artifacts to avoid: green spill on the subject, messy chroma key edges, blurry screen recording text, UI elements that look like AI hallucinations (keep the UI looking like a real website), inconsistent clothing on the subject, lighting changes on the subject.
Speech negatives: robotic cadence, unnatural emphasis, slurred words, harsh sibilance, plosives, clipping, pumping compression, over-denoise artifacts, lip-sync mismatch, long pauses, breathing sounds.

SPEECH PACK
Speaker: Male, 30s, energetic, fast-paced tech tutorial voice.
Transcript:
[00:00-00:10] "Google's Nano Banana is insane when it comes to design work. You can take a product, clothes, and location and put it together. But when you use this for design work, it's wild."
[00:10-00:20] "You can upload a photo of a product into Lovart, for example, and it will create all these custom posters using Google Nano Banana. I'm going to show you how you can do this because this is the efficiency revolution for design."
[00:20-00:35] "You can create a new project on Lovart, upload a photo of the product you want in the poster. You can write in a prompt like 'create me 2 different poster designs'. Then select the image model, select Nano Banana, and the agent will create different styles."
[00:35-00:45] "Pick the ones you want. Now we've got these two images, you can actually prompt and edit anything you want about that poster. So I now have this model holding the product, we can remove things and change the size."
[00:45-00:54] "Then we can even reprompt this to put this on Times Square for example. Boom. If you want to try this out for yourself, type AI in the comments and I'll send you a link."

How rourke Created a Lovart AI Design Agent Video Tutorial

A fast-paced, 54-second UGC-style software tutorial featuring a casually dressed creator (Vans cap, white tee) overlaid on a dynamic screen recording, demonstrating the rapid generation of high-end product photography (a glossy red handbag and a Dior Sauvage bottle) using the Lovart.ai design agent, culminating in a highly effective "Comment AI" automated engagement funnel.

2. What You're Seeing

Visual & Audio Elements

The video employs a classic "talking head over screen recording" composite format. The subject is a Caucasian male in his 30s with a short beard, wearing a beige "Vans Off The Wall" baseball cap and a plain white crew-neck t-shirt. He is keyed out (green screen effect) and positioned in the bottom center of the frame, taking up about 25% of the screen height. He uses expressive, open-palm hand gestures to emphasize points. The background is a crisp, high-resolution screen recording of the Lovart.ai web interface. The lighting on the subject is soft, even, and frontal, with no harsh shadows, suggesting a ring light or softbox setup. The color tone of the creator is natural and warm, contrasting sharply with the high-contrast, cinematic, and glossy outputs shown on the screen (like the deep red patent leather bag and the moody blue Dior bottle). Text overlays are minimal, using clean, sans-serif fonts with red highlights (e.g., circling "AI Design Agent"). The audio is a close-mic, dry, energetic voiceover, typical of fast-paced tech tutorials, with no noticeable background music, focusing entirely on the creator's authoritative delivery.

Shot-by-Shot Breakdown

Time Range	Visual Content	Shot Language	Lighting & Color Tone	Viewer Intent
00:00 - 00:10	Creator overlaid on a collage of high-end design posters (red handbag, Dior bottle). Text "Nano Banana x Design" appears.	Static MCU (Medium Close-Up) of creator, dynamic screen recording background.	Warm, even lighting on creator; high-contrast, glossy colors on screen.	Hook the viewer immediately with premium visual outputs before explaining the tool.
00:10 - 00:20	Screen shows the process of generating a Dior Sauvage ad from a prompt. Creator gestures enthusiastically.	Static MCU of creator, screen recording zooms slightly on the generated images.	Consistent lighting on creator; moody, cinematic blues and blacks on screen.	Demonstrate the tool's capability to produce luxury-tier assets quickly.
00:20 - 00:35	Screen shows starting a new project on Lovart.ai, uploading a red handbag image, and typing a prompt for two posters.	Static MCU of creator, screen recording follows the UI flow.	Consistent lighting; clean, white UI background with the bright red bag.	Provide a step-by-step, "over-the-shoulder" tutorial to build trust and show ease of use.
00:35 - 00:45	Screen shows the generated handbag posters. The user edits one to add a model holding the bag and changes the size.	Static MCU of creator, screen recording highlights the editing tools.	Consistent lighting; high-contrast red, black, and white on the generated poster.	Showcase advanced features (inpainting/editing) to prove the tool is more than just a basic image generator.
00:45 - 00:54	Screen shows reprompting the poster to place it in Times Square. Final result is shown. Text "Comment 'AI'" appears.	Static MCU of creator pointing at the screen, screen recording shows the final dramatic transformation.	Consistent lighting; vibrant, neon colors of Times Square on screen.	Deliver a final "wow" moment and drive massive engagement through a clear, automated CTA.

3. Why It Went Viral (Mechanism Breakdown)

Topic Selection & Audience Psychology

This video taps directly into the "efficiency revolution" narrative that is currently dominating the creator and small business space. The core topic—using AI to replace expensive, time-consuming design work—is highly appealing to entrepreneurs, marketers, and freelance designers who lack the budget for high-end photoshoots. Psychologically, it offers a "magic button" solution. Seeing a simple, isolated image of a red handbag transformed into a Times Square billboard in seconds triggers a sense of instant gratification and empowerment. The video also leverages the contrast between the creator's casual, "skater" aesthetic (Vans cap, plain tee) and the ultra-premium, luxury outputs (Dior, glossy patent leather). This cognitive friction makes the tool feel accessible; if this casual guy can make luxury ads, the viewer feels they can too.

Platform Signals & Algorithm Triggers

From a platform perspective, this video is perfectly engineered for Instagram Reels. The 0-3 second hook doesn't waste time with an introduction; it immediately flashes the highest-quality outputs (the red bag and Dior posters) to stop the scroll. The pacing is relentless—the screen recording changes or zooms every 3-4 seconds, preventing viewer drop-off and ensuring high retention rates. However, the most significant algorithmic trigger is the "Comment 'AI'" call-to-action at the end. By trading a valuable resource (the link to the tool) for a specific comment, the creator generated 8,601 comments. This massive influx of engagement signals to the algorithm that the content is highly valuable, pushing it to a much broader audience and creating a viral loop.

5 Testable Viral Hypotheses

The "End-Result First" Hook: Evidence: The video starts with the polished red bag poster and Dior ad before showing the UI. Mechanism: Proves the value immediately, reducing the bounce rate of skeptical viewers. Replicate: Always show your best generated asset in the first 2 seconds of your tutorial.
High-Low Visual Contrast: Evidence: The creator wears a casual Vans cap while demonstrating ultra-luxury Dior ads. Mechanism: Creates cognitive friction and makes the complex tool feel accessible to "regular" people. Replicate: Don't overdress for tech tutorials; let the tool's output be the star of the show.
Rapid UI Progression: Evidence: The screen recording jumps from uploading to prompting to generating in seconds, skipping loading times. Mechanism: Compresses time, making the tool seem faster and easier than it actually is, maintaining viewer dopamine levels. Replicate: Edit out all loading bars, typing time, and dead air in your screen recordings.
The "God Mode" Edit: Evidence: Changing the background of the poster to Times Square with a single prompt at the end. Mechanism: Taps into the viewer's desire for effortless power and scale, providing a final "wow" moment before the CTA. Replicate: Show a dramatic transformation or scale-up as the final step of your tutorial.
Automated Engagement Funnel: Evidence: The "Comment 'AI'" text overlay and the resulting 8,601 comments. Mechanism: Trades a valuable resource (the link) for algorithmic juice (comments), creating a self-sustaining viral loop. Replicate: Use tools like ManyChat to auto-DM links in exchange for a specific keyword comment.

4. How to Recreate (Replication Tutorial: From 0 to 1)

Step 1: Topic Selection & Positioning

Identify a high-value AI tool that produces visually striking results quickly (e.g., Midjourney, Lovart.ai, Canva Magic Studio). This format suits accounts focused on marketing tips, AI tools, design hacks, or small business advice.

Step 2: Screen Recording the Workflow

Record your screen using OBS or QuickTime while performing the task. Ensure your browser is clean (no messy bookmarks). Perform the actions smoothly: upload an image (like the red handbag), type a clear prompt, and generate the result. Do not worry about speed while recording; you will speed it up in post.

Step 3: Scripting the Voiceover

Write a script that matches the key actions on screen. Keep it punchy. Use phrases like "efficiency revolution" or "wild" to inject energy. Ensure the script builds up to a final, impressive feature (like the Times Square edit).

Step 4: Filming the A-Roll (Talking Head)

Film yourself against a green screen or a clean, contrasting background. Wear casual, relatable clothing (like the Vans cap and white tee) to contrast with the professional tool. Use a ring light for even, shadowless lighting. Deliver the script with high energy and use hand gestures to point "up" at where the screen recording will be.

Step 5: Keying and Compositing

In your editing software (CapCut, Premiere), remove your background using a chroma key or AI background removal tool. Place your talking head in the bottom center of the frame, taking up about 25-30% of the screen. Place the screen recording on the layer below.

Step 6: Pacing and Zooming

Edit the screen recording to match your voiceover. Cut out all typing time, loading screens, and mistakes. Add subtle digital zooms to the screen recording to highlight specific UI elements (like the "Nano Banana" model selection) when you mention them.

Step 7: Adding Visual Hooks and Subtitles

Add clean, sans-serif subtitles. Highlight key words in a contrasting color (like red). Add simple graphic overlays, such as a red circle around important buttons or text like "Efficiency Revolution," to guide the viewer's eye.

Step 8: Setting up the Automation Funnel

Before publishing, set up an automation rule in ManyChat (or similar) for your Instagram account. Set the trigger word to "AI" and the action to send a direct message containing the link to the tool you demonstrated.

5. Growth Playbook (Distribution & Scaling Strategy)

3 Ready-to-Use Opening Hooks

"Stop paying thousands for product photography until you try this..."
"This AI design agent just replaced my entire creative team..."
"How to create luxury brand ads in under 60 seconds..."

4 Caption Templates

The Time-Saver: [Hook] AI Design just got a whole lot easier. [Value] You can now generate high-end posters from a single product photo in seconds. [Question] Are you still using traditional design tools? [CTA] Comment "AI" and I'll send you the link to try it for free.
The Cost-Cutter: [Hook] Why pay for a photoshoot when AI can do this? [Value] I took a basic photo of a bag and put it in Times Square with one prompt. [Question] How much did your last product shoot cost? [CTA] Comment "TOOL" for the exact software I used.
The Agency Secret: [Hook] The secret tool design agencies don't want you to know about. [Value] This agent handles composition, lighting, and typography automatically. [Question] Ready to upgrade your brand's look? [CTA] Comment "DESIGN" and I'll DM you the access link.
The Tutorial: [Hook] Step-by-step guide to creating luxury ads with AI. [Value] Watch how I transform this simple red bag into a Dior-level campaign. [Question] What product would you test this on? [CTA] Comment "AI" to get started today.

Hashtag Strategy

Broad (To reach the algorithm's general tech audience): #AItools, #ArtificialIntelligence, #TechHacks. (These cast a wide net to find users generally interested in new technology).
Mid-Tier (To target specific professional niches): #DesignHacks, #MarketingTips, #SmallBusinessTools. (These target the actual buyers—people looking for efficiency in their work).
Niche Long-Tail (To rank for specific searches): #LovartAI, #ProductPhotographyAI, #AIDesignAgent. (These capture high-intent users searching for the exact workflow or tool shown).

6. FAQ

What tools make it look the most similar?

You need a screen recording tool like OBS, an AI background removal tool like CapCut for the talking head, and an AI design platform like Lovart.ai to demonstrate the workflow.

How do I get the clean green-screen effect without a green screen?

Use CapCut's "Remove Background" feature or RunwayML's rotoscoping tools; ensure you film yourself against a plain, contrasting wall with good, even lighting.

Why does the generated face look inconsistent in my AI tools?

You need to use models that support character consistency or provide a reference image and use strict prompting for facial features, though this specific video focuses on product consistency rather than faces.

How can I avoid making the tutorial look boring?

Cut out all loading times and typing, use digital zooms to highlight the UI, and ensure your talking-head delivery is highly energetic and uses hand gestures.

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

Instagram Reels currently favors this format heavily due to the effectiveness of ManyChat comment automations, which drive massive engagement signals.

How should I properly disclose AI use for this type of content?

The video itself is a tutorial on an AI tool, so the use is inherently disclosed; however, adding a hashtag like #AIGenerated or a small watermark is best practice for the final output images.