0:00 / 0:00

Comment “AI” and I’ll send you a guide + links 🚀 It’s really hard to believe that realism that you can now achieve using Kling 3.0. I’ve been playing around with it for the past two days and I’m blown away. Things are moving way faster than expected and soon we will be able to create an entire movies with AI and a fraction of the normal cost while keeping consistency, realism impact and storytelling intact. @higgsfield.ai has an amazing offer right now on new plans so comment “AI” and I’ll send you the link 🤩 #contentcreation #aicontent #kling3.0 #kling

Simone Ferretti

@sferro21

INSTAGRAM · 2026-02-06Source

602likes

619comments

Remix This

Recreate with Wan2.6

Make your own AI viral video

Prompt

GLOBAL LOCK: A vertical 9:16 creator-economy tutorial reel that alternates between one male presenter speaking directly to camera and rounded-corner cinematic demo clips or dark-mode screen recordings above him. The presenter is a light-skinned man in his 20s or early 30s with side-parted brown hair, clean-shaven face, slim build, expressive hands, and a friendly but high-energy delivery style. He wears a cream textured overshirt or knit jacket over a black crew-neck shirt and speaks into a black podcast microphone positioned centrally in front of him. The base environment is a dark charcoal studio with soft frontal key light, warm amber background glow, crisp digital sharpness, and social-first edit pacing. The insert window above him cycles through realistic AI film shots, portrait references, and Higgsfield/Kling 3.0 interface screens. Speech should feel like an enthusiastic tutorial and sales-demo hybrid: one speaker, close-mic audio, clean articulation, medium-fast cadence, excited emphasis on realism, workflow ease, and the CTA to comment for the guide.

[00:00-00:07] Open on a dark vertical layout with bold white headline text reading “100% Made with AI” across the top. In the upper rounded insert window, show moody green-and-gold cinematic scenes with shallow depth of field, including a dim interior and an extreme close-up of a burning match or cigarette ember touching the floor. In the lower rounded talking-head panel, the creator points upward and speaks directly into the microphone with animated eyebrows and raised finger, introducing how realistic the AI results now look. Keep the lighting warm on his face and the lip-sync fairly tight.

[00:07-00:14] Accelerate into a realism montage in the upper insert: a boxing-ring close-up with a glove pushing into lens, a sharply lit city-street action shot of a man smashing glass with a bat, and a vintage car interior with a suited man driving through daylight streets. In the lower panel the same presenter keeps talking continuously, hands moving in small punches that match edit accents. Preserve clean, close podcast audio and energetic tutorial cadence.

[00:14-00:20] Cut to a portrait-reference stage. In the upper portion, show a full-body male character standing barefoot in a Japanese-style tatami room under a paper lantern, with the word “PORTRAIT” visible above. The man has dark hair, a dark hoodie, and light sweatpants, arms folded, used as the identity anchor for later generations. The presenter below explains this is the starting character image or reference needed for consistent output. Lighting in the reference image is neutral indoor daylight with soft warm wood trim.

[00:20-00:26] Transition to a dark-mode Higgsfield interface screen recording. The cursor scrolls past model cards where “Kling AI 3.0” is clearly visible, along with other video-generation options. The creator remains in the lower panel, still speaking in a persuasive, teacher-like tone about using the newest model and current offer. UI motion is smooth and cursor-driven; edits land on emphasized words.

[00:26-00:35] Move deeper into the workflow. Show upload panels, prompt fields, and example cinematic stills in the upper insert while the creator explains how to set up the generation. One prompt card references a character smoking and another visible text prompt describes the person getting frustrated while drawing, tearing up the page, and throwing it away. Keep the interface dark, minimal, and product-demo realistic. The presenter below gestures with one hand while staying centered in the lower frame.

[00:35-00:45] Display the generated sketching sequence in the upper insert: the same male character sits in a workshop or cluttered room with a cigarette in his mouth, sketching intensely on paper under greenish tungsten lighting. Follow with a close-up of the pencil drawing a car, then show a start-frame and end-frame layout above a bright yellow “Generate” button, making the interpolation workflow obvious. Speech continues as a single uninterrupted explanation about how to prompt scenes and transitions while preserving realism and identity.

[00:45-00:54] Finish with a rapid cinematic payoff montage. The upper insert cycles through fireworks reflecting in a man’s sunglasses, a pink balloon near an older man’s face, a fiery explosion in the sky, a plane-window travel shot, and finally a suited man by the airplane window. Over the top, bold CTA text appears: “Comment ‘AI’”. The presenter below raises his finger again and delivers the closing call to action for the guide and links. Audio remains one-speaker, close-mic, confident, slightly urgent, with no crowd noise and with the final CTA synced to the on-screen text.

NEGATIVE PROMPT: inconsistent face shape between shots, different hair color, extra fingers, broken glasses reflections, rubber skin, flat UI screenshots, unreadable prompt boxes, cheap green-screen compositing, low-detail backgrounds, jittery motion, robotic lips, muddy audio, crowd ambience, subtitles, watermarks, duplicated props, oversaturated neon color cast.

SHOT PROMPTS: dark studio creator tutorial; rounded-corner insert window; 100 percent made with AI hook; cinematic realism montage; boxing insert; glass-smash action shot; vintage car driver; portrait reference in tatami room; Higgsfield dark-mode UI; Kling 3.0 model card; upload-image workflow; prompt field; frustrated drawing prompt; cigarette sketching scene; start-frame end-frame generation; fireworks reflected in glasses; plane-window final montage; comment AI CTA.

SPEECH PACK: Single male speaker only. Tone should be excited, persuasive, and instructional, like a creator sharing a breakthrough workflow and an exclusive offer. Keep close-mic podcast texture, medium-fast pace, clear consonants, and strong emphasis on “Kling 3.0,” “realism,” and the final “comment AI” call to action.

How sferro21 Made This Kling 3 Realism AI Video — and How to Recreate It

This Instagram Reel by Simone Ferretti is a compact AI-creator sales funnel disguised as a realism demo. In less than a minute, it proves that AI video can look cinematic, shows the exact workflow interface, and closes with a comment CTA that turns curiosity into inbound leads.

Case Snapshot

Creator: Simone Ferretti. Platform: Instagram Reels. Engagement snapshot at capture time: 602 likes and 619 comments. Caption hook: viewers are told to comment “AI” to receive a guide and links, while the creator frames Kling 3.0 as a major leap in realism and consistency.

The structure matters because it combines three proven content roles in one asset: visual proof, workflow tutorial, and lead capture. The reel is not just showing cool outputs. It is teaching enough to create trust while holding back enough to make the CTA valuable.

What You’re Seeing

The presenter sits in a dark studio, speaking directly into a microphone from a stable talking-head setup. Above him, the edit continuously swaps in AI-generated scenes and product UI screens. The first seconds establish the headline claim with the text 100% Made with AI, then the video quickly moves through action-heavy cinematic inserts, a portrait reference sheet, Higgsfield’s interface, visible prompt fields, and final generated results.

Visually, the reel is designed to keep one foot in trust and one foot in aspiration. The creator footage feels personal and believable, while the inserts feel expensive, dramatic, and movie-like. That contrast is the whole point: “Here is me, here is the tool, here is the impossible-looking result.”

Shot-by-Shot Breakdown

Time Range	Visual Beat	Function	Why It Matters
0:00-0:07	100% Made with AI headline with moody cinematic inserts and the creator pointing upward.	Immediate proof hook.	The viewer understands the promise before any explanation starts.
0:07-0:14	Rapid montage of boxing, smashed glass, driving, and smoking scenes.	Escalation of realism.	The edit demonstrates range, not just one lucky shot.
0:14-0:20	A full-body portrait reference in a tatami-room setup labeled PORTRAIT.	Identity-lock explanation.	This tells advanced viewers where consistency starts.
0:20-0:26	Higgsfield dark-mode interface with Kling AI 3.0 visible.	Tool credibility.	The reel moves from abstract claim to concrete workflow.
0:26-0:35	Upload steps and prompt-writing screens, including a frustration-and-drawing scenario.	Tutorial value.	Showing prompts increases saves because viewers think they can replicate it.
0:35-0:45	Generated sketching scene, drawing close-up, and start-frame/end-frame layout.	Process proof.	The reel shows not only outputs but also transition logic.
0:45-0:54	Fireworks, balloon, explosion, plane-window shots, then Comment “AI”.	Final payoff plus conversion.	The strongest visuals are saved for the moment the CTA appears.

Why It Works

1. The hook is specific, not vague

“Made with AI” is already a familiar claim. Adding 100% and proving it instantly makes the statement feel testable. The audience starts looking for evidence rather than scrolling away.

2. The creator stays on screen as the trust anchor

The talking-head panel keeps a human guide present throughout the reel. That matters because raw AI outputs alone can feel fake, stitched, or cherry-picked. The presenter makes the workflow feel reachable.

3. The edit alternates between aspiration and utility

Every impressive cinematic insert is paired with enough interface footage to turn wonder into intent. This is a strong format for small creators because it sells possibility while still being save-worthy.

4. The CTA is aligned with the viewer’s state of mind

By the end of the reel, viewers are not just entertained. They want the prompts, the settings, or the deal. Asking them to comment AI is a low-friction next step that converts attention into engagement.

5 Testable Viral Hypotheses

A proof-first opener beats a talking-first opener for AI-tool content because the audience is still skeptical about realism.
Keeping the creator visible in a lower panel increases trust more than cutting to full-screen UI for the entire tutorial.
Showing one or two readable prompt fragments is enough to trigger saves without giving away the whole workflow.
Final-moment CTA placement works better when paired with the most cinematic montage shots rather than with the driest tutorial section.
Comment bait performs especially well when the promised resource is framed as a guide plus links, not just “more info.”

How to Recreate This Format

Film one clean talking-head anchor. Use a dark background, soft face light, and a visible mic so the creator footage already feels intentional and premium.
Open with the boldest claim you can prove. Place it on-screen immediately and support it with one or two cinematic inserts in the first three seconds.
Show the identity source. A portrait sheet or character reference is essential if your message is about realism and consistency.
Let the audience see the actual tool. Screen recordings of Higgsfield, Kling 3.0, or prompt panels create credibility that pure montage cannot.
Reveal one readable prompt fragment. This gives the content educational weight and raises the chance of saves and shares.
End with the strongest montage, not the weakest. Save the most dramatic visual payoff for the same moment you ask for the comment.

Growth Playbook

This format is useful because it can be repeated for every new AI model release. The repeatable pattern is simple: claim, proof, workflow, result, CTA.

Use the first sentence to frame a clear before-and-after belief change, such as realism, consistency, or speed.
Keep the creator on screen long enough to retain trust, even when most of the excitement comes from inserts.
Make the UI recognizable so search-driven viewers know exactly which tool they are learning.
Offer a concrete keyword CTA like AI, PROMPT, or GUIDE rather than a generic “link in bio.”
Turn each viral reel into an evergreen page targeting long-tail searches around the model, workflow, and tutorial intent.

FAQ

Why does the creator stay visible while the UI is on screen?

Because the human presenter is the trust layer. It prevents the tutorial from feeling like anonymous product footage.

Why show the portrait reference sheet?

It signals how character consistency is being controlled, which is one of the main pain points in AI video.

Why is the comment CTA effective here?

The viewer finishes the reel wanting prompts, links, or settings. Commenting is the fastest path to getting that value.

What should another creator copy from this example?

Lead with proof, show enough of the tool to earn trust, and save the flashiest result montage for the CTA moment.