Best Meme Video Maker App for iOS & Android in 2025

GLOBAL LOCK: preserve a creator-led talking-head tutorial format mixed with vertical phone screen recordings. Keep one young male creator in a backward black cap and dark hoodie speaking directly to camera in a studio setup with a microphone. Intercut iPhone-style screen captures showing ChatGPT/OpenAI image workflow steps, uploaded object photos, prompt entry, and AI video generation screens. Maintain a practical “make from your phone” educational reel structure. No random B-roll, no unrelated tools, no logo overlays beyond app UI already present in the source.

Create a 37.8-second social-first AI tutorial reel showing how to turn ordinary phone photos into animated AI character videos. Begin with a hook using a simple hand-held object photo and bold on-screen teaching posture from the creator. Then show phone interfaces: photo selection, ChatGPT or image-tool screens, prompt entry, image transformation results, switching to an AI video tool, uploading the generated image, entering a motion prompt, and generating the final animated output. Use repeated face-cam segments where the creator explains the steps and emphasizes that the workflow can be done from a phone.

Include the specific examples visible in the source: tiny object/food photos held in a hand, ChatGPT app icon and mobile interface, typed prompts that turn objects into cute expressive characters, a generated pear-like baby character image, a switch to another AI generation interface, upload and prompt steps for video, and a final generated moving result shown on-screen. Preserve the educational pacing and creator-marketing vibe.

SHOT SEGMENTS:
[00:00-00:06] Hook with object photos in hand and creator talking-head intro about making AI content from your phone.
[00:06-00:14] Mobile screens show ChatGPT / image workflow setup, app screens, and prompt entry.
[00:14-00:22] Creator explains the key steps while on-screen phone UI shows prompt refinement and generated object-to-character image outputs.
[00:22-00:30] The tutorial switches to an AI video tool, showing upload, prompt, and generation steps from the phone.
[00:30-00:37.8] Final result displays the generated animated character clip, while the creator closes with a call to try the workflow.

ENVIRONMENT: creator desk/studio face-cam plus crisp mobile screen recordings. CAMERA: direct-to-camera presenter shots alternating with full-screen phone UI captures. LIGHTING: clean creator-studio lighting on face-cam; bright legible phone UI on inserts. MOTION: tutorial pacing, finger taps on phone UI, creator emphasis gestures, no cinematic narrative scenes.

NEGATIVE PROMPT: generic AI ad montage, unrelated tools, desktop-only workflow, no phone UI, missing creator face-cam, subtitles replacing the actual visible UI, blurry screens, watermark, logo overlays.

SPEECH PACK: creator-to-camera tutorial speech implied, but do not transcribe captions here.

GLOBAL LOCK:
- Format: vertical 9:16 short-form tutorial reel, creator-education pacing, black background UI inserts, high contrast social video polish.
- Keep one consistent male creator for all talking-head shots: young adult male, light skin, black backwards baseball cap, black hoodie/jacket, seated at desk, direct-to-camera framing, confident tutorial delivery.
- Keep one consistent demo subject inside the generated example image/video: a plush panda lying on a worn circular rug in a dim rustic room with warm overhead spotlight, scattered objects around the floor, soft moody shadows.
- No character drift, no costume drift, no sudden age changes, no extra presenters, no unrelated cutaways.

SHOT TIMELINE:

[00:00-00:03]
Talking-head intro. Creator sits centered against dark background and speaks straight to camera with energetic tutorial tone. Large editorial text overlays summarize the hook: make cinematic scenes from your phone. Insert fast teaser flashes of social posts showing the panda image/video result and yellow headline blocks.

[00:03-00:06]
Phone close-up UI. Vertical smartphone screen fills frame. A circularly framed panda image appears inside a social-style composition. Overlaid kinetic words emphasize the concept of turning a phone photo into a scene. Screen recording aesthetic should remain crisp and legible.

[00:06-00:09]
Back to talking head. Creator gestures lightly while saying the workflow starts by opening the app. Tight chest-up framing, direct eye contact, subtle head movement, clean synced speech.

[00:09-00:12]
Phone settings interface. User taps through app menu and settings-like pages to reach AI generation tools. Interface is dark mode, minimal, modern, with distinct list items and icons.

[00:12-00:16]
Prompt-building section on phone. Search field, model selection, and text-entry screens appear. User searches for GPT/prompt helper style tools, selects options, and opens a text area. On-screen rhythm should clearly communicate “build the prompt first.”

[00:16-00:20]
Text drafting flow on phone. Long paragraph prompt appears in a dark text box. User chooses/copies prompt text, then taps through action buttons. Highlight the exact motions: choose, copy, click, and go. The UI should feel like a real mobile workflow, not abstract fake panels.

[00:20-00:24]
Model/generation interface. User pastes the prompt into an AI image/video generation tool, selects the correct model or preset, and taps generate. Show dark-mode tool UI with image prompt area, buttons, and tabs.

[00:24-00:28]
Example asset preview returns. The panda scene appears again as a generated image/video preview. The phone screen cycles from prompt entry to generated result. Add supporting overlay words that reinforce the logic of generating the scene from a single photo.

[00:28-00:32]
Phone-to-output transition. The generated panda shot becomes larger and more immersive, as if stepping out of the interface into the final cinematic frame. Keep the panda, rug, spotlight, and room layout consistent with the reference image.

[00:32-00:35]
Talking-head recap. Creator returns on camera and explains the final step or CTA. He maintains same wardrobe and setup, speaking with persuasive, practical creator-teacher energy.

[00:35-00:39]
Final CTA and social proof. Talking-head remains center frame while comment-style overlays and platform UI elements appear below, suggesting engagement and repeatability. End on a clean, punchy tutorial finish.

VISUAL STYLE:
- Social tutorial reel, fast but readable editing.
- Mix talking-head shots with direct phone-screen recordings.
- Dark UI, white text, occasional high-contrast yellow hook text.
- Clean mobile creator aesthetic with authentic app interaction.

CAMERA AND EDITING:
- Talking-head: locked tripod or subtle digital push-in.
- Phone segments: full-screen mobile capture with smooth taps and transitions.
- Fast snap cuts between explanation, interface, and result.
- Keep chronological clarity so the viewer can follow the workflow in order.

SPEECH PACK:
- Spoken language: English.
- Creator voice: young male creator educator, confident, concise, practical, slightly hyped but not cheesy.
- Delivery style: short tutorial phrases, clear CTA emphasis, social-video pacing.
- Lip sync must stay natural and tightly aligned during talking-head shots.

NEGATIVE PROMPT:
- No extra hands floating over the phone.
- No unreadable UI gibberish replacing app text.
- No switching creator identity between talking-head shots.
- No panda changing species, color, pose logic, or room layout between preview and final output.
- No random additional animals or fantasy objects appearing in the room.
- No horizontal framing, no cinematic letterboxing, no documentary cutaways.
- No blurred phone screens, broken typography, or unusable interface text.

A) MISE EN PLACE
1) Video segmented into scenes:
- [00:00-00:01]: Static UI establishment.
- [00:01-00:04]: First animation cycle (clips drop down).
- [00:04-00:05]: Retraction.
- [00:05-00:08]: Second animation cycle.
- [00:08-00:09]: Final retraction.
2) Visual evidence extracted:
- Keyframes show a dark UI background, bold yellow/white text top and bottom, a central horizontal video player, and a timeline strip.
3) Speech evidence:
- No original audio provided. Assuming a standard promotional voiceover matching the text.
4) Invariants list:
- Visuals: Black background, top text ("2: MEET THE AI TOOL THAT UNDERSTANDS YOUR VIDEO👇"), bottom text ("TIP: Comment 'AI' and I'll send it directly to your DMs right now"), pointing hand icon, central horizontal video player showing two men talking.
- Speech: Upbeat, clear promotional tone.
5) Variables list:
- Visuals: Position of the three vertical dropdown clips, position of the red playhead on the timeline.

B) SHOTLIST
- shot_id: 1, timecode: 00:00-00:09, duration: 9s
- framing: Full screen graphic layout.
- lens: N/A (2D motion graphics).
- camera movement: Static camera, elements animate within the frame.
- subject: UI elements.
- environment: Dark digital canvas.
- lighting: Flat, graphic illumination.
- color grade: High contrast, black background, bright yellow (#FFD700) and white text.
- motion cues: Vertical sliding of rectangular frames, horizontal sliding of a thin red line.
- SPEECH / AUDIO:
  - speech_present: true
  - speakers: [A] (Off-camera narrator)
  - transcript_segments:
    - {00:00-00:04, A, "Meet the AI tool that actually understands your video.", energetic, 150wpm}
    - {00:04-00:07, A, "It analyzes the entire thing and cuts the best takes.", informative, 150wpm}
    - {00:07-00:09, A, "Comment AI and I'll send it to your DMs.", call-to-action, 160wpm}
  - delivery_direction: Energetic, clear, direct-response marketing style.
  - mic_room_signature: Close mic, dry studio sound.
  - sync_requirements: None (off-camera).

C) STYLE BIBLE
- visual_style: Clean, modern 2D motion graphics / UI mockup.
- camera_signature: Completely static.
- lighting_signature: Flat graphic design.
- grade_signature: High contrast, dark mode aesthetic.
- pacing_signature: Fast, looping animation.
- SPEECH STYLE BIBLE:
  - speech_style: Ad VO.
  - speaker_profile: Energetic, authoritative but friendly.
  - pronunciation_profile: Crisp enunciation.
  - mic_mix_profile: Dry, highly compressed for clarity on mobile devices.

D) PROMPT SYNTHESIS

1. MASTER PROMPT:
GLOBAL LOCK: A 2D digital motion graphics screen recording. The background is solid black. At the top, bold sans-serif text reads "2: MEET THE AI TOOL THAT UNDERSTANDS YOUR VIDEO👇" with the word "UNDERSTANDS" in bright yellow and the rest in white. Below this is smaller white text: "This free AI analyzes your entire video and cuts the best takes." At the bottom, text reads "TIP: Comment 'AI' and I'll send it directly to your DMs right now" with "AI" in yellow. In the bottom right corner is a white outline icon of a hand pointing left. In the center of the screen is a mock video editing interface. It features a horizontal video player showing a podcast setup with two men sitting at a table. Directly below the video player is a horizontal filmstrip timeline showing thumbnails of the video. The overall style is clean, high-contrast UI animation.

[00:00–00:01] The screen is static, displaying the global lock layout clearly.
[00:01–00:04] Animation begins. Three vertical rectangular frames (9:16 aspect ratio) smoothly slide down from behind the horizontal timeline strip. Each vertical frame contains a cropped, vertical version of the central podcast video. On top of the left frame is an Instagram icon; on the middle frame is a TikTok icon; on the right frame is a YouTube Shorts icon. Simultaneously, a thin red vertical line (a playhead) moves steadily from left to right across the horizontal timeline strip.
[00:04–00:05] The three vertical rectangular frames quickly slide back up and disappear behind the horizontal timeline strip. The red playhead resets to the left.
[00:05–00:08] The animation repeats exactly as before. The three vertical rectangular frames with social icons slide down again. The red playhead moves from left to right across the timeline.
[00:08–00:09] The three vertical rectangular frames quickly slide back up and disappear, returning the screen to the static state seen at the beginning.

2. NEGATIVE PROMPT:
3D elements, realistic camera movement, lens flare, depth of field, live-action camera shake, messy text, misspelled words, blurry UI, low contrast, cluttered background, realistic lighting, shadows, temporal jitter, morphing text.

3. SHOT PROMPTS:
(Not applicable as this is a single continuous graphic shot)

4. SPEECH PACK:
Transcript:
[00:00-00:04] Meet the AI tool that actually understands your video.
[00:04-00:07] It analyzes the entire thing, and automatically cuts the best takes.
[00:07-00:09] Comment AI and I'll send it directly to your DMs right now.

TAKE_A (Energetic & Punchy):
[00:00-00:04] MEET the AI tool... that actually UNDERSTANDS your video.
[00:04-00:07] It analyzes the ENTIRE thing... and automatically cuts the BEST takes.
[00:07-00:09] Comment A-I... and I'll send it directly to your DMs right now.

TAKE_B (Smooth & Professional):
[00:00-00:04] Meet the AI tool that actually understands your video.
[00:04-00:07] It analyzes the entire thing, and automatically cuts the best takes.
[00:07-00:09] Just comment AI, and I'll send it directly to your DMs right now.

TAKE_C (Fast & Urgent):
[00:00-00:04] Meet the AI tool that actually understands your video!
[00:04-00:07] It analyzes the entire thing and automatically cuts the best takes!
[00:07-00:09] Comment AI and I'll send it directly to your DMs right now!

Create a vertical 9:16 futuristic AI product-promo visual centered on a hyper-realistic fashion portrait of a young woman with slicked-back hair, pale skin, blue-grey eyes, and bold matte red lipstick, wearing a reflective chrome silver high-collar outfit in a bright metallic environment filled with iridescent foil-like textures. Behind her, large bold yellow text reads Meta AI, integrated like a clean social-ad headline. The image should feel like a premium generative-AI campaign frame promoting free image generation and AI lip sync tools, combining polished beauty-editorial realism with tech branding. Keep the composition crisp, symmetrical, high contrast, and optimized for short-form creator marketing. No extra clutter, no subtitles, no cartoon styling, no unrelated props.

GLOBAL LOCK: A vertical 9:16 AI demo video for Pollo.ai Mimic Motion featuring a male creator with short reddish-blond hair, fair skin, trimmed beard, and a light t-shirt speaking directly to camera in front of a warm wooden wall. A black podcast-style microphone sits in front of him. The key visual structure is a stacked comparison layout where the creator's exact expressions, head movement, hand gestures, and lip-sync are transferred onto multiple different characters. The swapped identities should include high-recognition fantasy and movie-inspired figures such as a Shrek-style ogre, a half-human cyborg reminiscent of Terminator, a Gollum-like creature, a Harry Potter-style wizard, a Pennywise-style clown, and a Tyler Durden-style gritty male lead. The demo should feel clear, fast, and proof-driven rather than cinematic storytelling.

[00:00-00:10] Open on a three-panel stacked comparison. The top panel shows the original creator speaking with both hands raised and expressive brows. The middle and bottom panels show alternate characters performing the exact same mouth movement, gaze direction, and hand pose in sync. Start with obvious contrast pairings like Shrek and a cyborg face to make the motion transfer immediately readable.

[00:10-00:24] Continue the stacked format while rotating through more dramatic character swaps. Show the same creator performance mapped onto a gaunt cave-dweller like Gollum, a young wizard in glasses, a white-faced clown with red makeup lines, and a gritty sunglass-wearing antihero. Each variant must preserve the exact source rhythm and gesture language, with only the identity layer changing.

[00:24-00:35] Transition back to the original creator in a single full-screen talking-head view with the microphone clearly visible. Let him continue speaking and gesturing naturally so viewers understand that the earlier transformations all came from this simple source performance. Keep the overall tone instructional and creator-focused.

NEGATIVE PROMPT: unsynced lip movement between variants, different poses in each comparison panel, heavy VFX clutter, cinematic story scenes replacing the demo structure, inaccurate parody costumes, random background changes, low-detail face swaps, no microphone or creator setup, generic montage without proof.

SHOT PROMPTS: creator talking-head source video; stacked mimic motion comparison panels; Shrek-style face swap synced to creator; cyborg half-face character remap; Harry Potter and clown motion transfer demo; original creator talking to microphone after swaps.

SPEECH PACK: One male speaker only. The important audio behavior is clean creator-style direct-to-camera speech with lip-sync accuracy preserved across every swapped character.

GLOBAL LOCK: A 9:16 vertical creator tutorial video showing how to build cinematic AI videos inside Freepik Spaces using Kling 3.0. The structure alternates between a casual male creator talking directly to camera, screen-like workflow panels, and polished AI-generated example sequences. The speaker is a white male in his 20s or 30s with beard, cap, and casual streetwear, filmed in a warm apartment or studio environment. He should feel approachable, creator-native, and energetic rather than corporate. Keep the edit fast and legible, with repeated “How to do this” framing, visual examples of cinematic shots, and interface scenes that imply prompt building, scene sequencing, and generation controls. Audio is speech-first and educational, with the creator explaining the workflow in concise steps.

[00:00-00:05] Open on a catchy example visual or lifestyle shot with bold tutorial framing like “How to do this,” immediately pairing aspirational output with educational intent.

[00:05-00:10] Cut to the creator talking directly to camera in a casual indoor setup, hands gesturing upward as he introduces the workflow and hooks viewers with the promise of showing the full process.

[00:10-00:18] Alternate between creator face-cam, finished AI shots, and screen-style panels showing thumbnails or interface blocks, making it clear that multiple scenes are being built inside one pipeline.

[00:18-00:28] Include more practical inserts: example frames, real-world pose or filming inspiration, and workflow interface layouts that suggest prompt control, shot planning, and visual refinement.

[00:28-00:40] Keep cycling between explanation and proof, with the creator speaking in short, punchy segments while the examples show the quality ceiling of the method.

[00:40-00:56] End with a clearer recap feel: more screen panels, more finished outputs, and a final face-cam summary that reinforces this as a repeatable Freepik Spaces plus Kling production workflow.

NEGATIVE PROMPT: dry webinar, plain slideshow only, no example outputs, stiff face-cam, dark podcast studio, random office footage, unreadable UI, over-designed captions everywhere, broken hands, uncanny face, robotic speech, disconnected examples, generic stock footage, text-heavy PowerPoint feel, poor pacing, muddy screen inserts, lip-sync errors, low-quality AI art, unrelated memes.

SHOT PROMPT DELTAS:
1) Aspirational example frame with tutorial hook text treatment.
2) Casual creator face-cam explaining workflow.
3) Screen-style interface panels and scene thumbnails.
4) Example cinematic outputs paired with explanation.
5) Final recap with tools, outputs, and creator closeout.

SPEECH PACK:
[00:00-00:56] One male speaker throughout. Tone should be concise, confident, and creator-educational, explaining how to structure prompts, build shots, and use Freepik Spaces with Kling 3.0 to generate cinematic AI videos. Medium lip-sync strictness when on-camera.

GLOBAL LOCK: A short-form tutorial reel hosted by a young light-skinned male creator in his early 20s with a slim build, short dark hair mostly hidden under a backwards black baseball cap, dark eyebrows, clean-shaven face, and a direct confident delivery style. He wears a black hoodie in a dark studio with magenta and blue edge lighting on his face and shoulders. Across the whole video, keep the creator visually consistent whenever he appears on camera. Alternate between direct-to-camera talking-head shots and desktop/screen-recording style inserts that show app interfaces, prompt builders, editing panels, and generated example outputs. The overall structure must feel like a practical creator education reel teaching how to make viral AI videos with ChatGPT, GPTs, Kling, and an editor workflow. Use social-video pacing, clear cut points, large readable interface elements, bold keyword captions, and crisp screen captures. Speech style is one energetic male speaker only, close-mic, dry room, high intelligibility, punchy cadence, creator-educator tone, with cuts landing on emphasized words.

[00:00–00:03] A hyper-stylized example montage opens the reel before the tutorial explanation begins. Show quick AI-generated insert shots: a yellow/orange plush-like character or pastry-like creature in a tiny kitchen set, exaggerated close framing, warm domestic lighting, toy-scale props, and a glossy social-media-ready finish. Add motion that feels like a viral AI clip rather than a static still: tiny hand gestures, object movement, short action beats, and a polished ad-like grade. Include large social-post overlays such as view counts or bold engagement graphics to imply virality. No host visible yet. No spoken words clearly visible on lips here if needed, or let the first line begin under the montage as voice-over. Audio should already feel like a tutorial hook.

[00:03–00:07] Cut hard to a centered talking-head medium close-up of the creator in the dark studio. The host looks straight into the lens and says the equivalent of “How to make viral videos,” with lips fully visible and sync strictness high. Frame him chest-up, camera at eye level, 35mm-to-50mm lens feel, shallow background, magenta-blue neon edge lights behind him. His expression is serious and helpful, with fast, clear articulation. The cut should feel like a strong tutorial promise after the flashy hook.

[00:07–00:12] Intercut between the host and the first example set. Show a vertical phone-style AI video example of a red cartoonish squishy character in a fleshy or surreal macro environment, then cut to generated household-object characters in a kitchen or interior setting, each with visible view-count overlays. Keep the host narration continuing over these inserts, explaining that viewers are asking how these kinds of videos are made. The examples should feel deliberately absurd, highly clickable, and visually varied. Maintain a social-app UI vibe on the inserts.

[00:12–00:16] Return to the host in the same neon studio framing. He explains that the process is easy or straightforward. Use a steady locked-off shot, close mic, no visible background clutter, and keep the delivery conversational but authoritative. Cut precisely on his emphasized keywords.

[00:16–00:21] Switch to screen-recording style visuals that show a desktop or browser workflow. Display recognizable AI tooling logos and interface tiles associated with ChatGPT, GPTs, custom tools, or image/video generation platforms. Cursor movement should be deliberate and readable. Then cut back briefly to the host as he explains the first step: going to GPTs or opening a custom GPT workflow. The speech remains one speaker, with no ambient distractions.

[00:21–00:27] Show actual interface navigation in a clean dark-themed desktop UI: menus, lists of GPTs, and prompt or tool panels. Include cursor clicks on fields and dropdowns. Briefly show a text or voice-input area and then a more advanced editing or story-generation screen. The host explains the setup step by step, describing where to go and what to choose. Keep the visuals aligned to the speech so every mention lands on the corresponding interface action.

[00:27–00:33] Continue in the software workflow with a tighter focus on prompt construction and asset preparation. Show text fields being filled, aspect-ratio settings such as 9:16, character/object references, and a “create story” or similar composition interface. Then reveal generated outputs: a stern milk-carton-like object character, a toast or bread-like character, and a colorful gadget character in a neon environment. The host explains that he is generating characters or story assets that can later be animated.

[00:33–00:39] Stay in screen-recording mode and move into the video-generation stage. Show the generated stills or character renders inside a platform interface, then a workflow where files are exported, selected, or prepared for upload into Kling AI or a comparable video generator. Interface panels should show thumbnails, upload areas, and generation controls. The host explicitly mentions Kling AI and a version number or model family, with cut-sync on the product name for emphasis.

[00:39–00:45] Demonstrate the final generation pipeline. Show the cursor uploading still images, selecting outputs, and previewing the finished short clips. Then display finished AI video shots of the angry milk-carton character and the colorful electronic character moving on their own in polished short scenes. The creator’s voice makes the pitch clear: upload the assets, run the generation, and turn them into videos like these. Keep the examples vivid and cute rather than realistic.

[00:45–00:48] End on the host back in the neon studio, now holding up a phone or printed visual reference while delivering the call to action. He tells viewers to comment for the prompt or follow for more. The shot is front-facing, centered, and slightly more animated than earlier, with confident hand motion and a creator CTA tone. Keep lips fully visible, close-mic audio dry and crisp, and land the final words right before the cut ends.

NEGATIVE PROMPT: inconsistent host identity, changing facial structure, different hats or wardrobe across talking-head shots, muddy UI text, unreadable screen captures, fake software logos replacing interface clarity, random extra speakers, robotic voice cadence, monotone narration, slurred words, lip-sync mismatch, soft unfocused screen recordings, flickering cursor, temporal jitter, duplicate objects in generated examples, malformed household characters, broken anatomy on host hands, blown-out neon highlights, crushed shadows hiding the face, excessive motion blur, abrupt camera zooms not present in the reference, noisy room echo, harsh sibilance, clipping, over-compressed dialogue, floating captions unrelated to speech, unrelated cutaway footage, low-resolution app panels, and generic “AI tutorial” visuals that ignore the specific ChatGPT-to-Kling workflow.

SHOT PROMPTS:
SHOT_01_HOOK: Viral AI example montage, tiny surreal kitchen set, pastry-like mascot, glossy toy-scale realism, warm light, social overlay metrics, ultra-clickable short-form hook.
SHOT_02_HOST_INTRO: Young male creator in backward black cap and black hoodie, neon magenta-blue studio, medium close-up, direct eye contact, says how to make viral videos, crisp close-mic tutorial delivery.
SHOT_03_EXAMPLES: Vertical examples of bizarre AI characters with high view overlays, red squishy mascot, household-object characters, meme-ready absurdity.
SHOT_04_GPTS_SETUP: Desktop UI with ChatGPT and GPT listings, cursor selecting custom GPT workflow, host explaining first setup step.
SHOT_05_PROMPT_BUILD: Dark-mode interface, text prompts, asset setup, aspect-ratio controls, create-story panel, generated character images appearing.
SHOT_06_KLING_STAGE: Exported character stills uploaded into Kling AI style interface, generation controls, preview windows, finished animated clips.
SHOT_07_CTA: Host returns to studio, holds visual reference, asks viewers to comment and follow, assertive creator-education ending.

SPEECH PACK
[00:00–00:03]
Closest audible transcript: "People keep asking how I make these viral AI videos."
Safe paraphrase: "A lot of people keep asking how these viral AI videos are made."
TAKE_A: [confident hook] People keep asking... how I make these viral AI videos.
TAKE_B: [fast, punchy] People keep asking how I make these viral AI videos.
TAKE_C: [teacherly emphasis] A lot of people keep asking how these viral AI videos are made.
Speaker: A
Lips visible: none or partial under montage
Lip-sync strictness: low
Mic-room signature: close mic, dry, clean, present

[00:03–00:07]
Closest audible transcript: "How to make viral videos."
Safe paraphrase: "Here is how to make viral AI videos."
TAKE_A: [direct] How to make viral videos.
TAKE_B: [slightly slower] Here's how to make viral AI videos.
TAKE_C: [emphasis on viral] How to make VIRAL videos.
Speaker: A
Lips visible: full
Lip-sync strictness: high
Cut sync: strong cut lands on "How"

[00:07–00:12]
Closest audible transcript: "A lot of you were asking me how these videos are made."
Safe paraphrase: "A lot of you asked how these kinds of videos get made."
TAKE_A: [friendly] A lot of you were asking me how these videos are made.
TAKE_B: [faster] A lot of you asked how these kinds of videos get made.
TAKE_C: [storytelling] So a lot of you have been asking... how these videos are actually made.
Speaker: A
Lips visible: mixed
Lip-sync strictness: medium

[00:12–00:16]
Closest audible transcript: "It's actually really easy."
Safe paraphrase: "It's way easier than people think."
TAKE_A: [reassuring] It's actually really easy.
TAKE_B: [casual] It's way easier than people think.
TAKE_C: [emphasis] This is actually super easy.
Speaker: A
Lips visible: full
Lip-sync strictness: high

[00:16–00:21]
Closest audible transcript: "Go to GPTs..."
Safe paraphrase: "First, open GPTs and start there."
TAKE_A: [instructional] Go to GPTs.
TAKE_B: [calm tutorial] First, open GPTs and start there.
TAKE_C: [step-by-step] Step one: go into GPTs.
Speaker: A
Lips visible: mixed
Lip-sync strictness: medium

[00:21–00:27]
Closest audible transcript: "Use any example..."
Safe paraphrase: "Use any example or template that fits the kind of video you want."
TAKE_A: [guide tone] Use any example that fits what you want to make.
TAKE_B: [clear] Use a template or example that matches the type of video you want.
TAKE_C: [slightly faster] Pick any example that lines up with the kind of video you're trying to make.
Speaker: A
Lips visible: partial
Lip-sync strictness: medium

[00:27–00:33]
Closest audible transcript: "Create... paste the... into..."
Safe paraphrase: "Create the assets, paste the prompt in, and set the format you want."
TAKE_A: [procedural] Create the assets, paste the prompt in, and set the format you want.
TAKE_B: [step-by-step] Build the assets, paste everything in, then choose your format.
TAKE_C: [faster tutorial cadence] Create it, paste the prompt, and set it up the way you need.
Speaker: A
Lips visible: mixed
Lip-sync strictness: medium

[00:33–00:39]
Closest audible transcript: "Go like Kling AI 2.6..."
Safe paraphrase: "Then take it into Kling AI and generate the motion from there."
TAKE_A: [brand emphasis] Then take it into Kling AI and generate the motion from there.
TAKE_B: [short] Next, use Kling AI for the video part.
TAKE_C: [tutorial tone] After that, bring the assets into Kling AI and run the generation.
Speaker: A
Lips visible: mixed
Lip-sync strictness: medium
Cut sync: emphasize "Kling AI"

[00:39–00:45]
Closest audible transcript: "Upload... and make videos like this."
Safe paraphrase: "Upload your images and turn them into videos like these."
TAKE_A: [instructional] Upload your images and turn them into videos like these.
TAKE_B: [punchy] Upload them... and make videos like this.
TAKE_C: [encouraging] Just upload the assets and you'll get videos like these.
Speaker: A
Lips visible: mixed
Lip-sync strictness: medium

[00:45–00:48]
Closest audible transcript: "Comment... follow..."
Safe paraphrase: "Comment if you want the prompt, and follow for more."
TAKE_A: [creator CTA] Comment if you want the prompt, and follow for more.
TAKE_B: [fast CTA] Comment for the prompt and follow for more.
TAKE_C: [friendly close] Drop a comment if you want it, and follow for more videos.
Speaker: A
Lips visible: full
Lip-sync strictness: high

GLOBAL LOCK: A vertical 9:16 creator tutorial reel teaching how to make first-person time-travel vlogs with AI. The lower half of the video holds a young male creator speaking directly to camera in a dark studio with red side lighting, black hoodie or jacket, and a backward cap. The upper half alternates between social-proof examples, smartphone search screens, browser pages, prompt-writing documents, and final generated historical selfie videos. The core output style is a realistic vlog shot where a modern creator appears to be filming himself inside major historical moments such as Viking England, the Wild West, or D-Day. The entire reel should feel practical and system-driven, built for viewers who want repeatable viral history content.

[00:00-00:12] Open on two successful example clips above the speaker: one where a young woman appears to selfie-vlog among Vikings in England in 865 AD, and another where she appears in a Wild West town in 1880. Both examples should look like genuine first-person historical vlogs with modern camera behavior but era-correct surroundings. View counts or social-proof markers should be visible to show that this content format already works.

[00:12-00:28] Move into the workflow entry step through a smartphone UI. Show a phone search screen with “Time Travel” typed in, then a Google-like result page for “Higgsfield AI.” The creator below explains the process in clear terms, making the tutorial feel accessible. The emphasis is on how surprisingly simple the setup is once the right tools are known.

[00:28-00:46] Show prompt-building and script-generation stages. Display a prompt document or text page labeled for text-to-video prompts, with entries for historical scenarios like landing craft before a beach assault or other era-specific vlog scripts. The interface should feel like a practical creator workflow rather than a polished marketing demo. The point is that the output begins with scripting the right first-person historical situation.

[00:46-01:01] End on a dramatic finished example where the creator appears to be selfie-vlogging during a World War II beach landing, with smoke, soldiers, landing craft, and battlefield chaos behind him. Overlay a small thumbnail or packaging element suggesting how the final video can be turned into a clickable social or YouTube asset. The result should feel both absurd and convincing: modern vlog behavior dropped into a massive historical event.

NEGATIVE PROMPT: static history painting look, third-person documentary framing, no selfie perspective, bland phone UI, generic prompts, inconsistent main character face, casual modern backgrounds, low-detail crowds, weak historical setting, no social-proof packaging.

SHOT PROMPTS: Viking time-travel selfie vlog; Wild West selfie vlog; phone search Time Travel; Higgsfield AI search result; ChatGPT prompt document; text-to-video historical script; D-Day beach selfie vlog; viral history series tutorial.

SPEECH PACK: One male speaker only. Tone is practical and energetic, emphasizing simplicity, virality, and repeatability. Stress “time travel vlogs,” “Higgsfield AI,” “ChatGPT prompts,” and the historical selfie angle.

GLOBAL LOCK: vertical Instagram AI tutorial reel hosted by a red-haired bearded male creator speaking directly to camera from a warm wood-panel backdrop; repeated cutaways to Pollo AI interface, ChatGPT prompt windows, generated portrait grids, and face-consistent character examples; bold short text beats synchronized with each spoken step; social-media tutorial pacing; clean screen-recording inserts; no unrelated footage, no color drift, no extra hosts, no meme chaos.

00:00-00:05
The host introduces an AI face-consistency workflow in a vertical talking-head setup. Split-screen and stacked portrait examples show the same person rendered in multiple styles, while bold on-screen text emphasizes that this can be done in a few steps.

00:05-00:11
The reel cuts between the host and a ChatGPT window, explaining how to upload a selfie and ask for a full descriptive prompt or face analysis. The creator gestures while short text phrases summarize each instruction.

00:11-00:18
Screen recordings show Pollo AI and related interface panels, including prompt boxes, generation modes, and output galleries. The host explains how to paste prompts, select models, and generate high-consistency character images from the selfie input.

00:18-00:26
Generated results fill the screen: grids of portraits, stylized headshots, and character variants with similar facial identity. The host calls out benefits like cheaper generation, faster workflow, better emotional range, and more natural skin consistency.

00:26-00:33
The tutorial transitions into the editing stage, where generated images are dropped into a video editor or transformation workflow. Example outputs show the same person preserved across multiple frames and styles, reinforcing per-frame alignment and prompt reuse.

00:33-00:36
The host ends with a direct call to action, prompting viewers to comment for the AI tool or workflow details. End card style remains simple, with the host centered and example outputs floating around him.

NEGATIVE PROMPT:
horizontal video, outdoor vlog footage, unrelated gaming UI, messy desktop clutter, unreadable text overload, warped faces, inconsistent identity drift, low-resolution screen captures, extra presenters, cartoon slapstick, random stock footage, dramatic camera shake

GLOBAL LOCK: vertical 9:16 creator tutorial reel, one consistent young adult male host with light skin, slim build, black backwards baseball cap, black hoodie, seated at a desk with a black microphone accented by red lighting, dark studio background with magenta-blue rim light, clean social-media talking-head aesthetic, frequent cutaways to iPhone screen recordings and desktop UI captures, crisp contrast, sharp subtitles, direct-to-camera educational delivery, fast pacing, screen-demo workflow energy, voice remains the same confident male speaker throughout, close-mic sound with dry room tone and clear consonants.

[00:00-00:03] Open with a high-speed hook collage: several glossy AI-generated coin or medallion-style motion-graphic examples appear at the top while bold thumbnail text promises viewers they can make this from their phone. Cut immediately into an iPhone screen showing a text field and app navigation, establishing a mobile-first tutorial workflow.

[00:03-00:06] Continue with phone screen recordings of typing into ChatGPT or a GPT search interface. Show keyword searches for the right assistant or GPT tool while subtitle words land one by one. The host is not always visible, but his narration stays continuous, fast, and instructional, with cuts landing on emphasized phrases.

[00:06-00:10] Alternate between the host’s face and mobile UI screens. The host looks directly at camera with a neutral but focused expression, speaking in a concise “here’s the exact process” tone. The phone screen shows menus, search results, and a selected motion-graphics-related GPT or helper.

[00:10-00:14] Move into a message-composition phase on the phone. A long, detailed prompt is typed or pasted into a chat interface requesting motion-graphic image generation with clear visual constraints. Keep the UI legible and the pacing brisk, with punch-ins on key words like image, detailed, or copy.

[00:14-00:18] Show the generated or referenced output and transition into desktop or browser captures featuring AI video or motion tools. Include interfaces associated with cinematic generation platforms like Higgsfield or Kling, with green-accent UI panels and creator-plan messaging visible. The host continues narrating over the demo, explaining what to do next.

[00:18-00:23] Demonstrate the next workflow step inside editing or generation panels: toggling options, selecting presets, setting a background or text layer, and preparing a motion graphics sequence. Intercut brief returns to the host in the studio so the viewer stays anchored to a single teacher guiding the process.

[00:23-00:28] Show more UI interactions that build the final result: adding text, adjusting layout, or exporting motion elements. The host remains seated in the same setup, speaking clearly into the desk microphone, with subtitles emphasizing functional words like background, text, yourself, and links.

[00:28-00:32] End on the host full-screen in the studio, centered and speaking directly to camera with a strong CTA tone. He gestures minimally, stays upright behind the microphone, and closes by telling viewers where to get the links or workflow resources. The final beat should feel like a practical creator tutorial, not a cinematic montage.

NEGATIVE PROMPT: broken smartphone UI, unreadable text, warped hands, inconsistent host identity, changing wardrobe, duplicate microphones, messy desk clutter, random overlays, flickering screen recordings, fake app interfaces, low-resolution subtitles, robotic lip sync, slurred narration, echoey room sound, harsh sibilance, clipping, jittery cuts, watermark, logo corruption.

SPEECH PACK:
- Hook: You can make motion graphics like this straight from your phone.
- Beat 1: Start inside ChatGPT and find the right GPT or helper for motion-graphics prompts.
- Beat 2: Ask it for a detailed image prompt first, then move that output into your video-generation workflow.
- Beat 3: Use tools like Kling or Higgsfield to animate the asset, then add your background and text treatment.
- CTA: I’ve got the links and setup in the caption, so save this and try it yourself.

GLOBAL LOCK: vertical 9:16 AI tutorial reel, one consistent young adult male host with light skin, slim build, black backwards cap, black hoodie, seated at a desk with a black podcast microphone lit by red accent light, dark studio background with magenta-blue edge lighting, intercut with clean smartphone screen recordings and generated football-stadium scenes, creator-education aesthetic, crisp subtitles, high contrast, fast cuts, dry close-mic narration, same male speaker throughout, practical “step-by-step” delivery rather than hype-only promo.

[00:00-00:03] Open with a strong thumbnail-style hook built around football imagery: a famous football rivalry scene inside a packed stadium, framed as an AI example, while bold text promises viewers they can make this with Kling 3.0 on their phone. Cut quickly between the example image and the host introducing the workflow.

[00:03-00:06] Show the host full-screen in the studio, speaking directly to camera with a focused, instructional tone. Then cut to iPhone screens that display the football image example on mobile, reinforcing the “straight from your phone” setup. Subtitle words land in sync with his emphasized phrases.

[00:06-00:10] Move into mobile app screens for Kling or a similar AI video interface. The user taps through creation menus and selects the relevant generation mode. The host continues narrating off-screen, explaining which option to click and how to begin the setup.

[00:10-00:14] Demonstrate configuration steps on the phone: toggling settings, choosing a shot mode, and preparing the scene. The UI stays readable and dark-themed with green confirmation accents. Brief cutbacks to the host keep the tutorial anchored around one teacher and one repeatable workflow.

[00:14-00:18] Introduce the keyframe-based process. Show the original football face-off image and then separate generated frames or shot cards, likely labeling multiple shots. The host explains that the scene needs to be broken into distinct beats instead of handled as one vague prompt.

[00:18-00:23] Continue with a structured shot workflow: the phone interface shows multiple shot blocks, each with a prompt or keyframe note. The example alternates between the two football players, preserving stadium identity while varying the framing per shot. The host’s narration emphasizes how to build the sequence one shot at a time.

[00:23-00:28] Show closer example outputs for individual characters in the stadium, now isolated into separate shot stages. The talking-head sections remain centered, minimal-gesture, and clear, while subtitles highlight functional words like first, keyframe, shot, and create. The tutorial feels highly procedural and reproducible.

[00:28-00:33] End on the final result screen: a football player in the stadium holding a cardboard sign that says viewers should comment “Kling” for the link. The last frames hold on this CTA and the finished generated example, making the reel feel like both a tutorial and a lead-generation funnel.

NEGATIVE PROMPT: distorted athlete faces, unreadable mobile UI, broken stadium geometry, warped hands, malformed sign text, inconsistent host identity, duplicated microphone, noisy subtitle artifacts, fake app screens, low-detail football uniforms, lip-sync drift, robotic narration, clipping, strong reverb, flicker, temporal jitter, watermark.

SPEECH PACK:
- Hook: You can make scenes like this in Kling 3.0 straight from your phone.
- Beat 1: Start with your image, open the mobile workflow, and pick the right creation mode.
- Beat 2: Don’t treat the whole sequence as one prompt; break it into separate shots and keyframes.
- Beat 3: Build each beat around the same environment so the stadium and characters stay consistent.
- CTA: Comment Kling and I’ll send you the link.

WORKFLOW
A) MISE EN PLACE
1) Segment the video into scenes/shots:
- [00:00–00:05] Single continuous shot (A composite split-screen showing two distinct scenes simultaneously).

2) Extract visual evidence:
- Keyframes: 0s, 2s, 4s.
- Left Panel: Caucasian woman, early 30s, blonde hair in a messy ponytail, wearing a mustard-yellow zip-up bomber jacket over a black top. Sitting outdoors at a cafe, daylight, string lights in the blurred background. She is laughing.
- Right Panel: Same woman, identical hair and wardrobe. Sitting indoors at a bar, warm directional lighting, amber bokeh in the background. She is holding a pint glass of beer and taking a sip.
- Overlays: White sans-serif text at the top and bottom.

3) Extract speech evidence:
- No speech. Audio is likely a trending BGM track.

4) Create an "invariants list" (LOCK THESE):
- visuals: The split-screen layout (left/right). The exact appearance of the woman (facial features, blonde ponytail, mustard jacket, black shirt). The static camera framing (MCU) on both sides. The text overlays.
- speech: N/A.

5) Create a "variables list" (TWEAK THESE):
- visuals: The micro-expressions of the laugh on the left. The liquid movement inside the beer glass on the right. The subtle background motion (patrons, bokeh shimmer).

B) SHOTLIST
- shot_id: 1
- timecode_start: 00:00
- timecode_end: 00:05
- duration: 5s
- framing: Split-screen. Both sides are Medium Close-Up (MCU), eye-level camera.
- lens: 50mm equivalent feel, shallow depth of field, creamy bokeh on both sides.
- camera movement: Static on both sides.
- subject: Left: Laughing naturally, slight shoulder movement. Right: Bringing a beer glass to her lips, taking a sip, maintaining eye contact.
- environment: Left: Outdoor cafe, daytime. Right: Indoor bar, evening.
- lighting: Left: Soft, overcast natural daylight. Right: Warm, moody practical lights, directional key light on the face.
- color grade: Warm overall tint, high contrast between the cool/neutral left and the amber/orange right.
- motion cues: Left: Subtle hair movement in the breeze. Right: Liquid dynamics in the glass.
- SPEECH / AUDIO:
  - speech_present: false

C) STYLE BIBLE
- visual_style: Cinematic UGC / High-end lifestyle B-roll.
- camera_signature: Locked-off tripod feel, shallow depth of field to isolate the subject.
- lighting_signature: Motivated lighting (natural outdoors vs. practical indoors).
- grade_signature: Warm, filmic, rich skin tones, vibrant mustard yellow.
- texture_signature: Photorealistic, sharp subject with soft, pleasing background blur.
- pacing_signature: Slow, deliberate motion suitable for looping.

D) PROMPT SYNTHESIS

MASTER PROMPT
GLOBAL LOCK: A vertical 9:16 split-screen video divided exactly down the middle. On both sides, the exact same subject is featured: a 30-year-old Caucasian woman with blonde hair pulled back into a messy ponytail, wearing a distinctive mustard-yellow zip-up bomber jacket over a black t-shirt. The camera is static on both sides, framed as a Medium Close-Up (MCU) with a shallow depth of field. The top of the video features bold white sans-serif text: "STEP 5: ANIMATE YOUR VIDEOS AS B-ROLL OR TALKING HEAD VIDEOS". The bottom features text: "Animate using Google Veo 3.1 for perfect lip sync or Kling 2.6 Pro for smooth cinematic clips."

[00:00–00:05] The video plays as a continuous 5-second loop. 
ON THE LEFT SIDE: The woman is sitting at an outdoor cafe table during the day. The lighting is soft, natural daylight. The background is blurred, showing outdoor seating and string lights. She is looking directly at the camera, smiling broadly and laughing naturally, with subtle, realistic head and shoulder movements. 
ON THE RIGHT SIDE: The woman is sitting at an indoor bar. The lighting is warm, moody, and directional, casting a soft glow on her face. The background features rich, amber bokeh from pendant lights. She is holding a clear pint glass filled with beer. She slowly brings the glass to her mouth, takes a sip, and lowers it slightly, maintaining steady eye contact with the camera throughout the motion. The liquid in the glass moves realistically. Both sides play simultaneously in a photorealistic, cinematic style.

NEGATIVE PROMPT
morphing, warping, inconsistent facial features, changing clothes, different person on left and right, bad anatomy, extra fingers, distorted glass, floating objects, unnatural lighting, plastic skin texture, jittery motion, flickering text, spelling errors in text overlays.

SPEECH PACK
No speech present in the reference video.

An educational social media tutorial video featuring a creator speaking directly to camera in a warm studio setup with a microphone, while explaining how to create ultra-realistic AI short videos using Gemini and strong prompt structure. The video alternates between the presenter’s talking-head delivery, on-screen examples of cinematic black-and-white and neon portrait references, and screen recordings of the Gemini interface where prompt steps are entered and refined. The teaching focuses on building believable results by specifying image type, realistic imperfections, natural expressions, subtle scenarios, and human behavior details, then encouraging viewers to comment for the prompt. The tone is practical and creator-focused, combining expert AI workflow advice, UI walkthroughs, and before-and-after inspiration in a concise Instagram tutorial format.

GLOBAL LOCK: A Caucasian male in his mid-30s with short, light brown hair and a full, well-groomed beard. He is wearing a vibrant, solid red crewneck sweatshirt. The environment is the interior of a vintage luxury car, featuring tan leather upholstery and a polished dark wood dashboard. The lighting is warm, golden-hour sunlight coming from the side. The camera is positioned for a medium side-profile shot.

[00:00–00:07]
The man is seated in the driver's seat of a classic Rolls-Royce. His right hand is firmly gripping the black steering wheel, while his left arm rests naturally. He is looking straight ahead through the windshield with a focused but calm expression. Outside the window, a scenic coastal highway is visible, with the blue ocean and green cliffs rushing past in a motion-blurred parallax effect. The steering wheel has subtle, realistic micro-movements as if he is steering. The sunlight creates a sharp rim light on his beard and hair. The texture of the red sweatshirt and the grain of the wood dashboard are highly detailed. No speech is present, but the man's jaw is set firmly. The camera has a very slight handheld shake to simulate the vibration of a moving car. High-quality cinematic film stock appearance with natural color saturation.

NEGATIVE PROMPT: blurry face, inconsistent beard shape, changing sweatshirt color, distorted car interior, static background, robotic movement, flickering lighting, extra fingers, morphing steering wheel, low resolution, cartoonish texture, text overlays, logos.

INVARIANTS TO LOCK
- Vertical 9:16 split-comparison Reel.
- Same young adult white male creator in every shot: light skin, slim build, side-swept brown hair, clean-shaven, expressive face.
- Neutral studio setup with soft gray background, clean frontal lighting, medium framing from chest to head.
- Video alternates between “Original:” and “AI:” versions of the same gesture performance.
- The AI versions keep the exact body movement and timing, but swap wardrobe, accessories, and visual effects.
- Tone is demo-first, highly legible, fast, and social-native.

SHOTLIST
1. [00:00-00:02] AI label over a dark tactical outfit, then a red-and-blue spider-inspired superhero suit, then a brown aviator jacket with patches and sunglasses. Matching “Original:” frames underneath show the presenter in a plain black shirt doing the same finger snap gesture.
2. [00:02-00:05] The comparison continues with the aviator look in a warmer room setting with vertical blinds and a plant, still mirroring the original hand choreography.
3. [00:05-00:07] Fire effects appear behind and around the AI version while the original remains clean and unstyled below.
4. [00:07-00:09] Large subtitle CTA appears over the AI version: comment “AI” for guide. Final frames push the fiery transformation while the original keeps the same open-handed pose.

STYLE BIBLE
Visual style: creator demo of motion-consistent character transformation.
Camera signature: locked tripod, eye-level medium shot, no camera movement.
Lighting signature: soft even front light on the original clip; AI variants maintain similar face lighting while changing wardrobe and environment mood.
Grade signature: clean studio neutrals in the original; richer contrast and warmer highlights in the AI versions.
Speech style: brief solo creator commentary or silent caption-driven demo; if voice is present, it should sound casual, impressed, and direct.

MASTER PROMPT
GLOBAL LOCK: Create a vertical 9:16 Instagram Reel that compares an original studio performance against AI-transformed outputs. Use the same young adult white male creator with light skin, slim build, side-swept brown hair, and clean-shaven face throughout. Keep the original clip on a soft gray studio background with the creator in a plain fitted black shirt, medium framing, frontal lighting, and simple hand gestures. Every AI version must preserve identical timing, pose, eye line, and hand motion, while changing outfit, accessories, background mood, and effects. Use bold yellow labels “AI:” and “Original:” so the comparison is instantly readable.

[00:00-00:02] Show the creator snapping or flicking his fingers in sync across paired comparison frames. In the AI version, first dress him in a dark armored tactical costume, then switch to a red-and-blue spider-inspired superhero suit, then to a brown aviator jacket with sewn patches and black sunglasses. In the original version, keep the same gesture in a plain black shirt against a gray backdrop.

[00:02-00:05] Continue the gesture-matched comparison. The AI variant now settles into the aviator look in a warmer cinematic room with vertical blinds and a leafy plant, preserving exact mouth shape and hand timing from the original clip. The original remains unchanged below, emphasizing how the motion has been transferred rather than reanimated from scratch.

[00:05-00:07] Add stylized flames behind the AI character and subtle orange light wrapping around the jacket sleeves. Keep the original clip clean and neutral for contrast. Maintain sharp alignment between both performances so viewers can read the transformation as one-to-one motion mapping.

[00:07-00:09] End with the most dramatic fiery aviator transformation while overlaying a clear CTA: comment “AI” for guide. The original clip still mirrors the same open-handed pose. Finish on a high-energy, creator-demo beat.

NEGATIVE PROMPT
Do not drift the face identity, hairstyle, body proportions, or gesture timing between original and AI versions. Avoid extra fingers, broken sunglasses, distorted jacket patches, muddy flames, inconsistent eye direction, unreadable labels, flickering backgrounds, or cartoonish facial deformation. Do not let the AI transformation lose the exact one-to-one motion match with the original clip.

SPEECH PACK
[00:00-00:04] Speaker A, direct-to-camera, meaning: this is how the same motion can be restyled with AI. Delivery: short, confident, creator-demo cadence.
TAKE_A: “Same motion, completely different character styling.”
TAKE_B: “This is the exact same performance, just transformed with AI.”
TAKE_C: “Watch how the motion stays locked while the look changes.”

[00:04-00:09] Speaker A or on-screen text, meaning: these tools save creators time and a guide is available by comment. Delivery: casual CTA.
TAKE_A: “Comment AI if you want the full guide.”
TAKE_B: “If you want the workflow, comment AI below.”
TAKE_C: “Comment AI and I will send the guide.”

[Assumptions]
- Candlelit dinner portrait.
[Inventory]
- Smiling brunette with long straight hair in silver backless evening dress; candleholders and fine dining table setting.
[MASTER PROMPT]
[Subject] Glamorous brunette turning toward camera with a smile at an elegant dinner table.
[Environment] Intimate upscale restaurant with green textured wall, candles, wine glasses.
[Composition/Camera] Vertical 4:5 medium portrait from seated side angle.
[Lighting] Warm candlelight with soft flattering key.
[Style/Rendering] Photoreal luxury dining editorial.
[Detail constraints] Keep silver backless dress, candle arrangement, refined tableware.
[Negative prompt]
bright daylight cafe, casual t-shirt, cluttered background, cartoon look, blur
[Suggested parameters]
- aspect ratio: 4:5; focal length: 50-85mm; steps: 24-30; CFG: 5.5; sampler: DPM++ SDE; seed: 311234
[Delta prompt]
1) "silver backless dress" 2) "smiling brunette" 3) "candlelit dinner" 4) "green wall backdrop" 5) "wine glasses" 6) "seated turn-back pose" 7) "warm tone" 8) "single subject" 9) "fine dining tableware" 10) "luxury ambiance"