The easiest workflow to recreate cheating fruit AI videos using ChatGPT for story and prompts, and AICUT for characters, frames, and video.
This is the tool I use for character generation, reference sheets, scene start frames, and final video generation.
Use my AICUT link here:
If you want the exact prompts and written guide I used, write fruit and I’ll send them to you.
Pick a cheating fruit AI video you want to recreate in your own version.
Save it so you can study the cuts, pacing, expressions, and scene order.
The script controls the dialogue, story beats, and later animation prompts.
Usually every major cut becomes one new scene.
Upload screenshots, paste the script, paste the scene breakdown, then paste the image-generating master prompt.
Ask ChatGPT to swap the fruits, lock the cast, lock the world, and write all scene image prompts.
Make each character alone first, then create a reference sheet for each one.
Use the scene prompts from ChatGPT and keep the format at 9:16.
Only do this after your start frames look right.
Use each approved start frame as the scene start frame and generate the clips one by one.
Bring everything into CapCut or your editor, stitch the scenes together, trim the pacing, and export.
1. Reference Material
2. Script
3. Scene Breakdown
4. Fruit Swap / Cast Direction
5. Special Notes
Original wife → Peachnita, peach-headed cheating wife
Original husband → Peachino, peach-headed betrayed husband
Original boss / third party → Cucumbro, cucumber-headed boss
Original doctor / witness → Dr. Kiwiwell, kiwi-headed doctor
What to do: upload screenshots, paste your script, paste your scene breakdown, then paste this into a new ChatGPT chat.
Analyze the reference, improve the story if needed, swap the fruits into a new cast, lock the world, and write the scene-by-scene image prompts only.
I want you to act like a premium cinematic AI showrunner, storyboard director, script doctor, visual development artist, fruit-character designer, and elite image-prompt engineer. Your job is to help me build a premium fruit-headed cinematic short story for vertical short-form content. IMPORTANT: This is not only an image prompt task. This is a full story-development task. You must think like a serious short-form drama creator. Your job is to: 1. analyze my reference images / reference video shots 2. analyze my idea if I provide one 3. analyze my script if I provide one 4. if the script is weak, rewrite and improve it 5. if the script is missing, create one from the idea and scene references 6. create a strong viral story structure 7. lock the visual style 8. lock the characters 9. lock the wardrobe 10. lock the locations 11. lock the palette logic 12. create scene-by-scene dialogue blueprint for later animation 13. generate scene-by-scene IMAGE PROMPTS ONLY IMPORTANT: Do NOT generate animation prompts yet unless I explicitly ask later. MAIN GOAL: We are building a premium fruit-headed cinematic short story for vertical short-form content. This project must feel: - premium - cinematic - luxurious - emotionally readable - elegant - dramatic - modern - consistent - socially believable inside a fruit world STORY ENGINE RULES: You are not only writing prompts. You are building the STORY ENGINE first. SCRIPT RULES: If I provide only an idea and reference scenes: - build the story from that If I provide a rough script: - rewrite it into a stronger cinematic short-form script If I provide a strong script: - tighten it for short-form retention DIALOGUE BLUEPRINT RULES: Before writing image prompts, create a DIALOGUE BLUEPRINT for later animation. For each scene, define: - who speaks - what the line is - what the emotional purpose of that line is - whether the scene should be silent STYLE TARGET: - vertical 9:16 - high-end cinematic 3D animated film still - premium studio-quality 3D - ultra-detailed fruit skin texture with soft natural gloss - realistic fruit pores, seeds, speckles, stem texture, crown detail, and natural surface variation - realistic subsurface scattering - realistic cloth simulation - global illumination - soft film grain - shallow depth of field - natural color grade - high dynamic range - clean edges - stable anatomy - no text - no subtitles - no logos - no watermark CHARACTER SYSTEM RULES: Create a full LOCKED CHARACTER SYSTEM before writing the scene prompts. For each major character, define: - name - fruit type - gender / age feel - skin / peel / seed / pore / texture detail - crown / leaf / stem / top detail - eye shape and eye energy - brow styling - mouth / lip styling - face shape - body build - wardrobe - shoes - jewelry / accessories - emotional energy - continuity rules COLOR PALETTE RULES: I want every major scene prompt to use a strict 60/30/10 palette when appropriate. LOCATION SYSTEM RULES: Create a LOCKED LOCATION SYSTEM. CAMERA RULES: Every scene prompt must include a strong camera description. SCENE WRITING RULES: Each scene prompt must capture: - the exact emotional second - the exact power dynamic - the exact physical arrangement - the exact implied story beat PRODUCT / BRAND / SHOP INTEGRATION RULES: If there is a brand, product, shop, logo, note, bouquet card, phone app, or visual identity involved, integrate it naturally into the story. NEGATIVE PROMPT RULES: Every scene prompt should include a disciplined negative prompt when useful. OUTPUT FORMAT: Please respond in this exact structure: 1. STORY STRATEGY 2. SCRIPT 3. DIALOGUE BLUEPRINT 4. VISUAL STYLE SUMMARY 5. LOCKED CHARACTER SYSTEM 6. LOCKED LOCATION SYSTEM 7. PALETTE SYSTEM 8. GLOBAL RENDER LOCK 9. SCENE-BY-SCENE IMAGE PROMPTS ONLY FRUIT SWAP / ORIGINALITY RULE: If the reference contains fruit characters I do not want to copy literally, preserve: - the role - the energy - the relationship dynamic - the fashion signal - the emotional function But change the fruit identity into a new premium fruit choice. REFERENCE INPUTS I WILL PROVIDE: - reference screenshots - reference scene images - script - rough idea - scene breakdown - style notes - fruit swap direction - logo / branding if relevant IMPORTANT QUALITY RULES: 1. The prompts must feel premium. 2. The prompts must feel cinematic. 3. The prompts must be very specific. 4. The prompts must lock the characters strongly. 5. The prompts must support visual consistency. 6. The prompts must support later animation. MATERIAL I AM GIVING YOU REFERENCE MATERIAL: [PASTE OR UPLOAD REFERENCE SCENES HERE] SCRIPT: [PASTE FULL SCRIPT HERE OR SAY “CREATE FROM IDEA”] SCENE BREAKDOWN: [PASTE SCENE-BY-SCENE BREAKDOWN HERE OR SAY “BUILD FROM STORY”] FRUIT SWAP / CAST DIRECTION: [PASTE ANY CHARACTER SWAP RULES HERE] BRAND / PRODUCT / SHOP NOTES: [PASTE ANY BRANDING NOTES HERE] SPECIAL NOTES: [PASTE ANY EXTRA NOTES HERE] IMPORTANT: Start with: - story strategy - script - dialogue blueprint - visual style summary - locked character system - locked location system - palette system - global render lock - scene-by-scene image prompts only Then stop and wait for me to ask for animation prompts.
REFERENCE MATERIAL: - office cheating screenshot - home dinner screenshot - hospital baby reveal screenshot - hospital hallway confrontation screenshot
SCRIPT: Peachino comes home happy and loving, but Peachnita feels guilty and emotionally distant. At work, Peachnita is too close to her cucumber boss, and the chemistry feels wrong immediately. Later in the hospital, Peachino is handed a newborn and realizes the baby is clearly a cucumber, not a peach. Peachnita admits it was a mistake. Peachino storms down the hallway holding the baby, determined to confront the boss.
SCENE BREAKDOWN: Scene 1: Peachino comes home to Peachnita. Warm home scene, but she looks nervous and emotionally distant. Scene 2: At work, Peachnita stands too close to her cucumber boss. The chemistry feels wrong immediately. Scene 3: In the hospital, the doctor presents the newborn. Peachino realizes the baby is a cucumber. Scene 4: Peachino confronts Peachnita in shock and rage. She admits it was a mistake. Scene 5: Peachino storms through the hospital hallway holding the baby, ready to confront the boss.
After your start frames or images are approved, paste this into the same ChatGPT chat.
Keep the same locked characters, same world, same palette, and write all the scene-by-scene animation prompts in order.
Now use the locked story strategy, final script, dialogue blueprint, visual style summary, locked character system, locked location system, palette system, global render lock, and approved scene image prompts we already created. Your task now is to generate the ANIMATION PROMPT for each scene. IMPORTANT: Do NOT redesign the story. Do NOT rewrite the character identities. Do NOT change the wardrobe unless the story timeline clearly requires it. Do NOT change the scene order. Do NOT regenerate image prompts. Only generate scene-by-scene ANIMATION PROMPTS. This is a premium fruit-headed cinematic short story for vertical short-form content. Each animation prompt must: - preserve the exact approved image scene - preserve the exact character design - preserve the exact wardrobe - preserve the exact fruit textures and colors - preserve the exact palette logic - preserve the exact emotional moment - preserve the exact location mood - use strong short-form dialogue when needed - stay readable, cinematic, and controlled IDENTITY PRESERVATION RULE: Every animation prompt must clearly state that identity must be preserved perfectly. PALETTE PRESERVATION RULE: The animation prompt must preserve the 60/30/10 palette logic from the approved image prompt whenever relevant. DIALOGUE RULE: Use the approved script and dialogue blueprint as the base. Dialogue rules: - keep lines short - keep lines sharp - keep lines emotionally clear - keep lines instantly understandable - do not write weak generic dialogue - if silence is stronger, keep the scene silent MOTION RULE: Motion must be: - realistic - restrained - elegant - cinematic - readable - emotionally motivated AUDIO RULE: - spoken dialogue only if needed - natural room tone only - no music - no narrator unless I explicitly ask LIP-SYNC RULE: - only the active speaker’s mouth moves - no mouth movement from silent characters - no accidental background talking CAMERA RULE: Camera must stay controlled and intentional. OUTPUT FORMAT: For each scene, respond in this exact structure: SCENE [NUMBER] — ANIMATION PROMPT [full animation prompt] Inside each animation prompt include these sections in this exact order: 1. CLIP LENGTH AND FORMAT 2. IDENTITY PRESERVATION 3. AUDIO LOCK 4. WORLD 5. VISIBLE CAST ONLY 6. MOTION 7. SPEAKER LOCK AND DIALOGUE TIMING 8. IMPORTANT VOICE RULE 9. LIP-SYNC RULE 10. SFX / AMBIENCE ONLY 11. END STATE 12. CAMERA IMPORTANT DO NOTS: Do NOT rewrite the entire story. Do NOT redesign the cast. Do NOT change fruit types. Do NOT change wardrobe for no reason. Do NOT overanimate the scene. Do NOT add music. Do NOT weaken the image prompt quality. Generate the scene-by-scene ANIMATION PROMPTS for the approved scenes in order. Start from Scene 1 and continue in order. Only output animation prompts.
Use the cast details from ChatGPT and create each character by itself first.
Upload each character and make a reusable sheet so the look stays consistent.
Use the scene prompts from ChatGPT. Keep the format at 9:16.
Paste the animation prompt for that scene and generate the clips one by one.
Keep your characters, sheets, frames, and clips organized in folders.
Opening / normal moment.
Suspicion / something feels wrong.
Discovery / reveal.
Confrontation.
Payoff / hallway / final reaction.
Write fruit and I’ll send you the exact prompts and written guide I used for this workflow.