Tell the video model exactly what should move.
Define the first frame, one achievable action, camera behaviour, continuity locks and final frame so the generated clip has a clear visual path.
Act as an expert AI video prompt designer, cinematographer, visual storyteller and text-to-video generation specialist. TASK: Turn the supplied idea into a clear, model-ready text-to-video prompt with strong scene control, camera direction, subject consistency, realistic motion and visual continuity. The final result should describe what the video should actually show, not just the theme or mood. VIDEO GOAL: [What should this video communicate or achieve?] PLATFORM: [Facebook Reel / Instagram Reel / TikTok / YouTube Short / YouTube / website / advertisement / presentation / other.] VIDEO TYPE: [Product video / food video / cinematic scene / business visual / educational clip / lifestyle / commercial / social content / other.] DURATION: [For example: 5 sec / 8 sec / 10 sec / 15 sec / other.] ASPECT RATIO: [9:16 / 16:9 / 1:1 / other.] MAIN SUBJECT: [Person / product / food / object / environment / process.] SUBJECT DETAILS: [Appearance, clothing, product characteristics, food appearance, materials, colours, etc.] ACTION: [What exactly happens during the clip?] STARTING FRAME: [Describe what is visible at the beginning.] ENDING FRAME: [Describe the desired final state.] ENVIRONMENT: [Location, set, room, outdoor scene, background.] TIME OF DAY: [Optional.] VISUAL STYLE: [Photorealistic / cinematic / premium commercial / documentary / lifestyle / food-commercial / editorial / other.] CAMERA SHOT: [Close-up / medium / wide / macro / overhead / tracking / other.] CAMERA ANGLE: [Eye level / low angle / high angle / top-down / three-quarter / other.] CAMERA MOVEMENT: [Static / slow push-in / dolly / pan / tilt / orbit / tracking / handheld / other.] LIGHTING: [Natural / studio / warm / dramatic / soft / high-key / moody / other.] DEPTH OF FIELD: [Shallow / moderate / deep / optional.] FRAME RATE / MOTION FEEL: [Real-time / smooth / slow motion / cinematic / natural handheld / other.] COLOUR / MOOD: [Optional.] REFERENCE REQUIREMENTS: [Describe any reference image, character or product that must remain consistent.] AUDIO: [No audio / ambient sound / dialogue / voice-over / music cue / optional.] VISIBLE TEXT: [Exact text or "No visible text".] ELEMENTS TO AVOID: [Optional.] SPECIAL REQUIREMENTS: [Any other constraints.] TEXT-TO-VIDEO REQUIREMENTS: 1. DEFINE ONE CLEAR VIDEO ACTION The clip should communicate one primary visual event. 2. AVOID OVERLOADING SHORT CLIPS Do not attempt to show too many actions within a few seconds. 3. MATCH ACTION TO DURATION A 5-second clip should contain less action than a 15-second clip. 4. DEFINE THE STARTING STATE Describe what already exists in the first frame. 5. DEFINE THE ENDING STATE Describe what should be true by the final frame. 6. CREATE A LOGICAL MOTION PATH The movement between the first and last frame should be physically understandable. 7. PRESERVE SUBJECT IDENTITY If a person, product or object is central, keep its defining characteristics consistent throughout the clip. 8. DO NOT MORPH THE SUBJECT Avoid uncontrolled: Face changes Body changes Product redesign Colour shifts Material changes Object replacement 9. CHARACTER CONSISTENCY For a person, preserve: Face shape Skin tone Hair Age appearance Clothing Accessories Body proportions 10. PRODUCT CONSISTENCY For a product, preserve: Shape Packaging Logo Colour Material Labels Key physical features 11. FOOD CONSISTENCY For food, preserve: Dish type Shape Layers Texture Toppings Colour Serving vessel 12. DO NOT INVENT PRODUCT FEATURES No imaginary buttons, accessories, screens, functions or packaging. 13. CONTROL OBJECT COUNT State how many important objects are present where necessary. 14. PREVENT DUPLICATION Avoid duplicated: Hands People Products Utensils Food items Props 15. CONTROL HAND INTERACTION When hands interact with something, require: Correct anatomy Natural finger placement Believable grip Proper contact Consistent hand position 16. CONTROL BODY MOTION Human movement should be anatomically plausible. 17. AVOID TELEPORTATION Objects should not suddenly appear, disappear or move between unrelated positions. 18. AVOID UNEXPLAINED TRANSFORMATIONS Do not let objects change shape unless transformation is explicitly part of the scene. 19. CONTROL CLOTHING Clothing should not randomly change during the clip. 20. CONTROL ACCESSORIES Jewellery, glasses, watches and similar elements should remain stable. 21. CONTROL HAIR Hair movement may react naturally to motion or wind, but hairstyle and length should remain consistent. 22. CONTROL BACKGROUND CONTINUITY The environment should not randomly rearrange itself. 23. CONTROL PROP CONTINUITY Props should remain in consistent positions unless intentionally moved. 24. USE PHYSICALLY PLAUSIBLE MOTION Respect: Gravity Weight Momentum Object contact Liquid behaviour Fabric movement Steam Smoke 25. LIQUID MOTION Pouring should have: Continuous flow Correct source Correct destination Natural gravity 26. FOOD MOTION Cutting, stirring, pouring, lifting or serving should preserve realistic texture and physical behaviour. 27. STEAM / SMOKE Use subtle realistic movement rather than excessive artificial clouds. 28. CAMERA MOVEMENT MUST HAVE PURPOSE Do not add motion simply to make the scene feel cinematic. 29. STATIC CAMERA MODE Use when subject action itself is sufficient. 30. PUSH-IN MODE Use a slow camera move toward the subject to emphasize detail. 31. TRACKING MODE Use when following moving subjects. 32. ORBIT MODE Use only when the environment and subject geometry can remain stable. 33. PAN MODE Use for controlled horizontal reveal. 34. TILT MODE Use for vertical reveal. 35. HANDHELD MODE Use subtle realistic movement, not uncontrolled shaking. 36. AVOID IMPOSSIBLE CAMERA MOVES Do not describe multiple conflicting movements at once. 37. LIMIT CAMERA DIRECTIONS Prefer one dominant camera movement per short clip. 38. DEFINE CAMERA SPEED Use terms such as: Very slow Gentle Smooth Controlled Fast only when needed. 39. MAINTAIN CAMERA AXIS Do not unintentionally jump between opposite angles. 40. PRESERVE SCREEN DIRECTION Moving subjects should not randomly reverse direction. 41. CONTROL FRAMING Specify whether the subject remains: Centered Left weighted Right weighted Tracked Foreground dominant 42. KEEP IMPORTANT DETAILS IN FRAME Do not accidentally crop: Face Hands Product Food hero Important action 43. VERTICAL VIDEO MODE Keep important subjects inside the central safe area. 44. HORIZONTAL VIDEO MODE Use wider environmental composition when appropriate. 45. CLOSE-UP MODE Prioritize detail and reduce unnecessary background movement. 46. MACRO MODE Use for textures, food, products and small details. 47. WIDE MODE Use only when environment is relevant. 48. DEPTH OF FIELD Use shallow focus when isolating a subject, but keep critical action readable. 49. CONTROL FOCUS Avoid random focus hunting. 50. USE FOCUS PULL PURPOSEFULLY Only when shifting attention from one subject to another. 51. LIGHTING MUST REMAIN CONSISTENT Do not allow random changes in: Light direction Brightness Colour temperature Shadow direction 52. NATURAL LIGHT MODE Use believable daylight behaviour. 53. STUDIO MODE Use controlled, repeatable lighting. 54. FOOD-COMMERCIAL MODE Use: Soft highlights Texture-rich lighting Natural gloss Warm appetizing colour Controlled steam 55. PRODUCT-COMMERCIAL MODE Use: Clean reflections Controlled highlights Premium surface detail Strong product separation 56. CINEMATIC MODE Use cinematic lighting and composition without making the scene unrealistically dark. 57. REALISTIC SKIN Avoid plastic or waxy skin. 58. REALISTIC EYES Avoid flicker, asymmetry or unnatural eye motion. 59. REALISTIC MOUTH Avoid deformation when no speech is required. 60. DIALOGUE MODE If dialogue is required, keep spoken content appropriate to the clip duration. 61. DO NOT ADD DIALOGUE WHEN NOT REQUESTED Silent visual clips should remain visually focused. 62. LIP MOVEMENT Only request visible speech when necessary. 63. AUDIO AND VISUAL TIMING If an audio cue is included, align it to visible action. 64. NO RANDOM TEXT Do not generate unintended captions, signs, logos or interface text. 65. EXACT TEXT If visible text is necessary, specify the exact wording. 66. KEEP TEXT SHORT AI video models may struggle with long text. 67. DO NOT INVENT LOGOS Use only supplied branding. 68. DO NOT REDESIGN LOGOS Preserve logo proportions and colours. 69. AVOID RANDOM WATERMARKS Explicitly prohibit them. 70. REALISTIC MATERIALS Maintain believable behaviour for: Glass Metal Plastic Wood Fabric Paper Ceramic Food Liquid 71. REFLECTIONS Ensure reflections match object position and lighting. 72. SHADOWS Shadows should remain attached and directionally consistent. 73. CONTACT SHADOWS Objects should feel physically grounded. 74. PERSPECTIVE Maintain consistent object scale and perspective as the camera moves. 75. DO NOT WARP STRAIGHT EDGES Protect packaging, furniture, screens and architecture. 76. ARCHITECTURAL CONTINUITY Walls, counters, shelves and doors should remain structurally stable. 77. NATURAL SPEED Human and object motion should not feel unintentionally sped up or slowed down. 78. SLOW MOTION Only use when explicitly requested and visually suitable. 79. MICRO-MOTION Subtle motion can include: Breathing Steam Hair movement Fabric movement Small hand adjustments 80. DO NOT MAKE EVERYTHING MOVE Stable elements improve realism. 81. CREATE VISUAL PRIORITY The viewer should immediately know what to watch. 82. REDUCE BACKGROUND DISTRACTION Background activity should not compete with the subject. 83. USE NEGATIVE SPACE WHEN USEFUL Allow room around the hero subject. 84. AVOID GENERIC AI EFFECTS Do not automatically add: Holograms Glowing circuits Floating particles Neon interfaces Sci-fi effects 85. USE EFFECTS ONLY WHEN STORY-RELEVANT Visual effects should support the requested concept. 86. SOCIAL VIDEO MODE Create immediate readability in the first second. 87. COMMERCIAL MODE Make the product or benefit visually obvious. 88. EDUCATIONAL MODE Show the concept or process clearly rather than prioritizing cinematic complexity. 89. LIFESTYLE MODE Use natural human behaviour and believable environments. 90. FOOD VIDEO MODE Prioritize appetite appeal, texture and realistic handling. 91. BUSINESS VIDEO MODE Use professional environments and restrained visual language. 92. DO NOT INVENT RESULTS Never visualize unsupported: Revenue Growth Weight loss Performance improvement Before/after outcomes 93. DO NOT INVENT SOCIAL PROOF No fake: Ratings Testimonials Followers Awards Customer numbers 94. DO NOT CREATE FALSE URGENCY Avoid unsupported countdowns or scarcity cues. 95. KEEP PROMPT LANGUAGE VISUAL Describe what the model should see and animate. 96. AVOID ABSTRACT MARKETING LANGUAGE Replace words like: Innovative Amazing Powerful Next-generation with visible scene details. 97. CREATE POSITIVE INSTRUCTIONS FIRST Clearly define what should happen before listing exclusions. 98. CREATE TARGETED NEGATIVE CONSTRAINTS Use only relevant protections such as: No face morphing No extra fingers No duplicate objects No random text No product redesign No camera jump No background warping No watermark 99. CREATE A CONTINUITY CHECK Verify: Subject Clothing Props Lighting Background Camera direction Object position remain coherent. 100. FINAL QUALITY CHECK Before finalizing verify: - platform is defined - duration is realistic - aspect ratio is correct - one primary action is clear - starting frame is defined - ending frame is defined - subject identity is stable - product or food identity is stable - hands are plausible - props remain consistent - background remains stable - camera movement is achievable - focus behaviour is controlled - lighting is consistent - object motion follows physics - no random text appears - no fake branding appears - no unsupported results are shown - no unnecessary AI effects are added - video remains understandable without excessive complexity OUTPUT FORMAT: 1. VIDEO INTENT Summarize the purpose and primary action. 2. FIRST FRAME Describe exactly what should be visible at the start. 3. ACTION TIMELINE Break the clip into simple time-based movement. Example: 0–2 sec 2–5 sec 5–8 sec Adapt timing to the requested duration. 4. CAMERA PLAN Specify: Shot size Angle Movement Speed Focus behaviour 5. SUBJECT & CONTINUITY LOCK List characteristics that must remain unchanged. 6. FULL TEXT-TO-VIDEO PROMPT Provide one polished model-ready prompt written as a coherent visual description. 7. NEGATIVE / PROTECTION INSTRUCTIONS List only likely generation failures. 8. FINAL FRAME Describe the desired ending composition. 9. CONTINUITY CHECK Confirm: Subject Objects Hands Background Lighting Camera Motion IMPORTANT: - Keep the requested action achievable within the duration. - Use one dominant camera movement. - Preserve subject identity and physical continuity. - Describe visible actions rather than abstract concepts. - Do not invent product features, results or branding. - Avoid random text, duplicated objects, morphing and impossible physics. - Optimize specifically for the selected platform and aspect ratio.
Use the prompt effectively.
Define one achievable action
Start with a simple event the viewer can understand within the requested clip duration instead of packing several unrelated actions into one generation.
Lock the subject and environment
Describe the person's identity, product, food, clothing, props and background details that must remain stable while the action occurs.
Control the camera
Choose one clear shot, angle and dominant camera movement so the model is not trying to execute several conflicting cinematic directions.
Describe the ending
Specify the final composition so the model has a clear destination for the motion instead of allowing the scene to drift unpredictably.
Turn one simple food action into an 8-second video prompt.
Platform: Facebook Reel.
Duration: 8 seconds.
Aspect ratio: 9:16.
Subject: Freshly baked chocolate cake on a wooden countertop.
Action: A hand slowly cuts one slice and gently pulls it forward to reveal the moist interior.
Camera: Fixed close-up with a very subtle push-in.
Style: Premium photorealistic food-commercial.
FIRST FRAME: Whole chocolate cake centered on a wooden countertop, knife already positioned above the cutting line, warm kitchen background softly blurred.
ACTION: The hand makes one clean downward cut, completes the slice, then gently pulls the piece forward to reveal the moist crumb.
CAMERA: Close-up, front three-quarter angle, fixed framing with only a very subtle controlled push-in.
CONTINUITY: Same cake, plate, hand, knife, countertop, lighting and background throughout; no extra slices or utensils appear.
PROTECTION: No distorted fingers, no melting cake structure, no duplicated knife, no camera jump, no random text or excessive steam.
Reduce generation errors before adding more cinematic detail.
Short clips need simple choreography
A single well-described action usually generates more reliably than asking the model to perform several cuts, camera changes and scene transitions in a few seconds.
Describe what moves and what stays still
Separating the moving subject from stable objects, background and camera elements gives the model a clearer continuity target.
The last frame matters
Defining the final visual state can significantly reduce drifting, unwanted object changes and unfinished-looking motion.