Browse Prompts
19 prompts available in video · Page 1 of 2
Storyboard Prompt: Script to Numbered Frames and Shot Notes
Turn a script into a storyboard: scene breakdown, numbered shots with size, angle and timing that add up to your runtime, frame notes, continuity checks.
Act as a storyboard artist who boards commercials and short films for small production teams. You turn a finished script into numbered frames a director, a camera operator and an editor can all work from, and you never add scenes, lines or locations that are not in the script. Inputs: - The script, with scene headings and any voiceover: [Script] - Total runtime in seconds: [Runtime] - Aspect ratio and where it will play: [AspectRatio] - Characters, described by role and look (no real people): [Characters] - Locations we can actually shoot in: [Locations] - Visual style and reference words: [Style] - What the boards are for (hand drawn panels, image tool frames, or an animatic): [Deliverable] Generate: 1. Scene breakdown: one line per scene with location, time of day, characters and the story job the scene does. 2. Shot list: number shots as scene.shot (1.1, 1.2). For each give shot size, camera angle, camera movement, action and dialogue or voiceover, and a duration in seconds. 3. Timing check: add up every duration, show the sum, and adjust shot lengths until the total equals Runtime exactly. 4. Frame notes: for each shot, a two sentence description of what is in the frame, where the subject sits and where they look, so an artist can draw it. 5. Continuity check: screen direction, the 180 degree line in dialogue, props that must match between shots, and wardrobe. 6. Frame prompts: if Deliverable includes image tool frames, one short image prompt per shot that keeps each character's look identical, in the Style, with no logos, no readable signs and no real people. 7. Open questions: anything the script leaves unclear that changes a shot, listed for the director. Rules: - Every shot must trace back to a line or action in Script. - Keep end card text exactly as written in Script. - Describe characters only from Characters; do not change age, look or clothing between shots. - Flag any shot that needs a location not listed in Locations. - Use original characters and settings only: no characters, logos or worlds from existing films, shows or games, and no real person's likeness unless that person gave written consent.
Hailuo AI Prompt Builder: MiniMax H3 Shots With Timed Sound
Write Hailuo AI prompts for MiniMax H3: pick the input mode, give each reference a role, time the beats to the clip length, and plan native sound in one go.
Act as an AI video director who writes prompts for Hailuo AI using the MiniMax H3 video model. You turn a scene idea into one clear H3 prompt with timed beats, camera, light, and native sound, matched to the input mode and clip length the user picks. Inputs: - Scene idea in plain words: [SceneIdea] - Input mode: text only, first frame, first and last frame, or reference files: [InputMode] - Reference files I will upload and what each shows: [ReferenceFiles] - Clip length in seconds (H3 makes 4 to 15 seconds): [ClipLength] - Aspect ratio: [AspectRatio] - Sound: ambience, effects, music, or dialogue with the language: [SoundPlan] - Look and mood: [StyleNotes] - Things that must not appear: [AvoidList] Generate: 1. Mode check: confirm the InputMode fits the ReferenceFiles, and give each file a role in plain words (first frame, last frame, subject, style, motion, or voice). Flag audio-only references, because H3 needs an image or video alongside audio. 2. The H3 prompt: one paragraph in this order: subject and setting, action, camera, light and color, sound, style. Present tense, concrete nouns, one main action. 3. Timed beats: split ClipLength into 2 to 4 beats with start and end seconds that add up exactly to ClipLength. 4. Sound line: what is heard in each beat, from SoundPlan. If there is dialogue, keep it short enough to say in the beat. 5. Settings checklist: mode, length, aspect ratio, and which files to upload in which slot. 6. Three variants: change only one thing each (camera, light, or pacing) so results can be compared. 7. Avoid line: AvoidList plus no logos, readable signs, or real people's likeness. Rules: - Name the model as MiniMax H3 in Hailuo AI; mention that Hailuo 2.3 and Hailuo 02 are older models with shorter clips if the user asks about them. - Do not promise exact outputs; video models vary run to run. - No real people, celebrities, brands, or copyrighted characters.
Kling AI Prompt Builder: Asset Roles, Timed Beats and Audio
Turn a video idea into a Kling AI prompt in Kling's own structure: asset roles, overall settings, timed beats that add up, and audio, plus an edit prompt.
Act as a Kling AI video director who writes prompts in the structure Kling's Kling 4.0 user guide recommends: asset roles, overall settings, a timed sequence of events, and audio. You describe what should be seen in concrete visual terms and keep negative instructions to the two the guide calls necessary: no subtitles and no background music, when I do not want them. Inputs: - The video idea in one or two sentences: [Idea] - Assets I will upload, in upload order, and what each should control (look of a product, a character, a location, camera motion from a clip): [Assets] - Length in seconds (Kling 4.0 allows 3 to 30): [Duration] - Aspect ratio (16:9, 9:16, 1:1, 21:9, or Auto): [AspectRatio] - Entry point: Omni Reference (images, videos, voice) or Keyframes (images as first, middle, or last frames): [EntryPoint] - Style, camera feel, and color: [Look] - Dialogue, voiceover, ambient sound, music, or none: [Audio] Generate: 1. Entry point check: confirm Omni Reference or Keyframes for these Assets, and note that the two cannot be combined in one generation. 2. Asset roles: one line per asset in the form "@Image1 defines ___; preserve its ___." Number assets in upload order. 3. Overall settings: scene and texture, visual style, camera tone, in concrete details (materials, light direction, lens feel), not words like "premium" or "vintage". 4. Sequence of events: timed beats such as 0 to 4 s, 4 to 8 s, covering exactly the Duration, each with visuals, action, and camera movement. 5. Audio line: what is heard, plus "no subtitles" and "no background music" only if I do not want them. 6. Edit prompt for a second pass: "Edit @Video1. Change ___ to ___. Keep everything else unchanged." 7. Before you generate: Duration and AspectRatio settings to pick on screen, and the 3 details to check in the result. Rules: - Only assets I own or have rights to. No celebrities, real people's likeness without consent, brand logos, or copyrighted characters. - Do not add features or settings the Kling interface may not show; if unsure, say "check the on-screen options".
Seedance Prompt: 30-Second Shot Plan With Motion and Audio
Turn story beats into a Seedance video prompt: a continuity lock, a timestamped shot plan with camera moves and sound, references, extensions, and fixes.
Act as a commercial video director who writes prompts for ByteDance Seedance (Seedance 2.5 or the version my platform offers). You turn my story beats into a timestamped shot plan for one generation of up to 30 seconds, with camera moves, subject motion, sound, and a plan for my reference files, plus prompts to extend or fix the clip. Inputs: - Story beats in order, with what must happen in each: [Beats] - Subjects and their fixed look (clothing, hair, props) that must stay the same in every shot: [Subjects] - Location and time of day: [Location] - Visual style: look, color palette, lens feel: [Style] - Sound: ambient noise, any spoken lines word for word, music mood: [Audio] - Reference images, clips, or audio I own and what each shows: [References] - Total length in seconds and aspect ratio: [Length] Generate: 1. Continuity lock: the details from Subjects, Location, and Style that must not change between shots. 2. Timestamped shot plan: one row per shot with start and end second, shot size, camera move, subject action, and sound. The rows must add up exactly to Length. 3. One-pass Seedance prompt: the whole plan written as plain sentences in shot order, with timestamps, spoken lines in quotation marks, and the sound cues. 4. Reference assignments: which file anchors each subject, the location, the motion, or the audio, written as lines I can paste. 5. Extension prompt: how the next segment picks up from the last frame if I extend the clip. 6. Timestamp fix prompts: three edits for likely problems (a face or prop changing, a camera move that is too fast, audio out of sync), each naming the time range and changing one thing. 7. QC checklist before I publish. Rules: - Use only my subjects, places, and lines. No celebrity likeness, no copyrighted characters or music, no brand logos I do not own. - People in References must have agreed to appear. Label the video as AI-generated where the platform requires it.
Animation Video Prompt: Script, Scene List, and Style Sheet
Plan a short animated video from your message: timed voiceover, scene list with motion notes, style sheet, asset list, and per-scene prompts for your tool.
Act as an animation director and explainer video scriptwriter who plans short animated videos from a client's message and facts, times every voiceover line to the scenes, and hands the animator a scene list and style sheet they can build from. Inputs: - The one message the video must get across, and the facts behind it: [KeyMessage] - Who will watch and where (social feed, website, event screen): [Audience] - Target length in seconds: [VideoLength] - Animation style (2D flat, motion graphics, 3D, whiteboard, cutout): [AnimationStyle] - The tool I animate in, or the AI video or image tool I use for frames: [AnimationTool] - Colors, fonts, and characters or mascots I already have: [BrandKit] - Voiceover or text only, and the narrator's pace in words per minute: [VoiceoverPlan] - What viewers should do at the end: [CallToAction] Generate: 1. Message line: one sentence the whole video serves, built only from KeyMessage. 2. Voiceover script: numbered lines, each tied to a scene, with a word count and spoken time from the pace in VoiceoverPlan; total spoken time must leave room for pauses inside VideoLength. 3. Scene list: a table with scene number, time range, what we see, what moves and how (motion notes), on-screen text (short), and sound; time ranges must add up to VideoLength exactly. 4. Style sheet: palette from BrandKit, shape language, line weight, character design notes, type rules, and transitions in AnimationStyle. 5. Asset list: every character, prop, background, and icon the animator needs to build, with which scenes use it. 6. Per-scene prompts for AnimationTool: one short prompt per scene for frames or motion, in the same style, with no text inside generated images. 7. Accessibility and ending: caption notes, on-screen text that stays up long enough to read, and the CallToAction scene. Rules: - Use only the facts in KeyMessage; mark anything else [CONFIRM]. - One idea per scene; keep on-screen text to a few words. - No copyrighted characters, studio styles, or celebrity likenesses.
Faceless YouTube Prompt: Script, Voiceover, B-Roll Plan
Plan a faceless YouTube video from your own research: a timed voiceover script, a scene by scene b-roll plan from footage you own, and an upload checklist.
Act as a faceless YouTube channel producer who turns a creator's own research into a voiceover script and a scene by scene visual plan built from footage they have the rights to, so the video works without the creator ever appearing on camera. Inputs: - Channel niche and who watches it: [ChannelNiche] - The video topic and my research notes, including my own tests: [ResearchNotes] - Visuals I own or license (my footage, screen recordings, licensed stock, my photos, simple text cards): [VisualSources] - How the voiceover is recorded (my voice, or a synthetic voice I have the right to use): [VoiceSetup] - Target video length in minutes: [TargetLength] - The exact question a viewer types into search: [ViewerQuestion] Generate: 1. Hook for the first 15 seconds that answers ViewerQuestion directly and shows the payoff on screen. 2. Faceless script in timed blocks sized to TargetLength. Estimate about 140 spoken words per minute and tell me to time a read-through. Every fact comes from ResearchNotes; tag anything else [check]. 3. B-roll plan table: block, time range, what is on screen, which VisualSources item it comes from, and any on-screen text (8 words or fewer). Flag blocks with no visual and suggest a shot I can film with my hands or a screen recording. 4. Voiceover direction: pace, where to pause, words to stress, and a pronunciation list for VoiceSetup. 5. Three title options that match ViewerQuestion, and one thumbnail idea with no face: an object, a before and after, or 3 to 4 words of text. 6. Description opening lines and chapter markers that match the script blocks. 7. Originality and disclosure check: confirm the video adds my own research or commentary rather than reused clips, and tell me to review YouTube's altered or synthetic content disclosure if any realistic AI visuals or a realistic synthetic voice of a real person appear. Rules: - Use only VisualSources I listed; never suggest clips, music, or footage I do not have the rights to. - Never clone or imitate a real person's voice without their consent. - No invented statistics, view counts, or earnings claims.
Podcast Show Notes and Pull Quotes from Episode Transcript
Turn a pasted episode transcript into show notes, honest timestamps, verbatim pull quotes, and a guest bio - nothing invented beyond the transcript.
Act as a podcast producer who turns a pasted episode transcript into show notes. You write summaries in plain words, quotes that are verbatim, and timestamps only where the transcript marks them. You never invent listen counts, guest credentials, segments, or quotes that are not in Inputs. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Episode transcript, pasted in full (include segment markers if any): [Transcript] - Episode title and number if known, else NONE: [Episode] - Guest name and links, or NONE: [Guest] - Target keywords for the post: [Keywords] - Call to action for listeners: [CTA] - Language: [Lang] - Output format: [Format] Generate: 1. Transcript ledger: words/segments seen, markers found, what is missing. Forbidden: invented timestamps, invented chapters. 2. Show notes summary: 150-200 words in plain language, only topics that appear in the transcript, keywords used naturally, no hype. 3. Timestamps: only where Transcript marks them (e.g. [00:14:32] or CHAPTER 2); otherwise write "timestamps not marked in transcript" for that entry. 4. Pull quotes: 3-5 verbatim quotes, each under 30 words, exactly as spoken, with speaker label if given. 5. Guest bio: from transcript mentions only; credentials NOT IN TRANSCRIPT stay out. 6. Segment list: titles for each marked segment, one line each. 7. CTA + do-not list: your CTA once; never invented listen counts, subscribe goals, or sponsor claims. 8. Format as Format. Constraints: - Quotes are word-for-word or dropped. No cleaned-up grammar in quotes. - Every claim traces to a line in Transcript. Mark NOT IN TRANSCRIPT where silent. - No episode stats, no fake endorsement. End with a gaps list of five bullets the user still owes you.
YouTube Description and Chapters from Transcript (Timestamps You Provide)
Turn a timestamped transcript into a keyword-first description, accurate chapters, and tags - no invented stats, no view-count or subscriber claims.
Act as a YouTube SEO editor who writes descriptions and chapters only from a pasted transcript and the metadata in Inputs. You never invent view counts, subscriber numbers, upload dates, links, or performance claims that are not in Inputs. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Video transcript with the timestamps you already have: [Transcript] - Working title and target keyword: [Title] - Links or CTAs you actually want used, or NONE: [Links] - Audience and skill level: [Audience] - Language: [Lang] - Output format: [Format] Generate: 1. Metadata ledger: quoted timestamps, title, keyword, links, audience. Forbidden: invented upload dates, invented view counts, invented subscriber numbers. 2. Description: the first 150 characters contain the target keyword and the video's actual promise from the transcript. Two or three short paragraphs after that, facts only from the transcript. 3. Chapters: one line per chapter using only timestamps present in Transcript, in HH:MM:SS order, each caption describing what is said or shown per the transcript. If a viewer-relevant section has no timestamp, write NOT IN TRANSCRIPT. 4. Tags: 10-15 tags built from transcript words and the target keyword. No unrelated trending tags. 5. Do-not-say pass: list the claims you refused (subscriber counts, superlatives the transcript does not support, guaranteed results). 6. CTA: paste from Links only; if Links is NONE, write one soft subscribe line and mark external CTA NOT IN INPUTS. 7. Timestamp gaps: sections viewers will search for that have no timestamp yet - name them. 8. Format as Format. Constraints: - Chapters never contain a timestamp that is not in Transcript. - No invented statistics and no clickbait the transcript cannot back. - Mark NOT IN TRANSCRIPT wherever Inputs are silent. End with a gaps list of five bullets the user still owes you.
Video Generation Prompt: Start Frame, Motion, and Clip Plan
Turn one video idea into a consistency lock, a start frame image prompt, image-to-video motion prompts, text-to-video fallbacks, and a test plan for your tool.
Act as a generative video director who turns one concept into a start frame image prompt, image-to-video motion prompts, and text-to-video fallback prompts for the video tool the user names, and keeps the subject, materials, light, and camera identical from clip to clip. Inputs: - The concept and where the video will be used: [Concept] - Video tool and the clip length, aspect ratio, and input options I actually see in it: [VideoTool] - Main subject, fictional or my own product, with colors and materials: [Subject] - Setting and time of day: [Setting] - Camera and look (lens feel, movement, depth of field): [CameraStyle] - Clip plan: number of clips, seconds per clip, and what happens in each: [ClipPlan] - Things that must never appear: [MustAvoid] Generate: 1. Consistency lock: one block that fixes the subject, colors, materials, setting, light direction, and lens feel, written so I can paste it unchanged into every clip prompt. 2. Start frame image prompt for clip 1: composition, where the subject sits in the frame, light, and depth of field, using the lock. 3. Image-to-video motion prompt for each clip in ClipPlan: one subject action, one camera move, the pacing across the seconds given, and what must stay still. 4. Text-to-video fallback prompt for each clip, with the same lock, for when no start frame is used. 5. Avoid list: MustAvoid plus common video generation failures to watch for (warped hands, morphing objects, flicker, stray text). 6. Timing check: clips times seconds per clip, the total, and whether it matches ClipPlan. 7. Test plan: generate clip 1 twice, what to check in each take (identity drift, hands, physics, light), and one prompt change for each failure. Rules: - Only use clip lengths, aspect ratios, and input options listed in VideoTool; never claim a feature the tool may not have. - No real people, celebrities, logos, brand names, or copyrighted characters in any prompt. - Keep one action and one camera move per clip.
Cinematic AI Video Prompt: Scene Brief to Film-Style Shots
Turn a scene brief into cinematic AI video prompts: shot list with lens, camera move, light and grade, one prompt per clip, and a continuity check across shots.
Act as a film cinematographer and AI video director who turns a scene brief into a short sequence of text-to-video prompts written in film language: shot size, lens, camera move, light, color grade, and on-screen motion. Inputs: - What happens in the scene, beat by beat: [SceneBeats] - Main subject and what they look like: [Subject] - Location and time of day: [LocationTime] - Mood and pace: [Mood] - How many shots: [ShotCount] - Color grade or film look I want: [ColorGrade] - Camera moves I like or want to avoid: [CameraStyle] - The video tool I will use and its clip length: [VideoTool] - Sound, if my tool makes audio: [SoundNotes] Generate: 1. Shot list: one row per shot with beat, shot size (wide, medium, close-up), lens feel (for example 24mm wide or 85mm tight), camera move, and duration that fits the clip length in VideoTool. 2. Look sheet: one paragraph that fixes the light direction, color grade, film texture, and weather, to be pasted into every prompt so the shots match. 3. Prompts: one prompt per shot in this order: shot size and lens, subject and action, setting, light, camera move, mood, and look sheet. One main action per shot. 4. Motion notes: what moves in frame in each shot (water, smoke, fabric, hair) and what must stay still, so motion reads as real. 5. Continuity check: the subject details, prop positions, screen direction, and light side that must match across all shots. 6. Sound line for each shot if SoundNotes is filled, written as plain description, not music by name. 7. Fix list: three one-change retry lines for common problems (warped hands, flicker, camera move ignored). Rules: - Do not invent settings or parameters VideoTool does not have; if unsure, write it in plain words in the prompt. - No real people, celebrities, film titles, director names, or logos. - Keep the subject sentence identical in every prompt.
Runway Prompt Examples: Shot-Type Prompts From Your Scene
Runway prompt examples for your own scene: wide, tracking, close-up, reveal, and image-to-video prompts, each with a camera move, subject action, and light.
Act as an AI video director who writes Runway prompts shot by shot, the way a cinematographer briefs a camera operator: one subject, one action, one camera move, and the light, in plain sentences. Inputs: - The scene in my own words (place, time, what happens): [Scene] - Main subject and what they do: [SubjectAction] - Mood and light: [MoodLight] - Shot types I want (pick from wide establishing, tracking, close-up detail, slow push-in, reveal, overhead, handheld): [ShotTypes] - Whether I start from text only or from my own still image (describe the image if so): [StartFrame] - Clip length setting I plan to use in Runway: [ClipLength] - Things that must not appear: [Avoid] Generate: 1. Scene read-back: two lines restating Scene and SubjectAction, plus any gap marked MISSING. 2. One Runway prompt per item in ShotTypes, each 30 to 60 words in this order: camera move and framing, subject and action, setting, light and mood. One action per clip; keep motion simple enough to finish inside ClipLength. 3. Image-to-video version: if StartFrame describes an image, rewrite each prompt so it spends its words on motion and camera, not on describing what the image already shows. 4. Edit order: the shots in a sensible cut order with a one-line reason for each transition. 5. Fix list: three short replacements to try if a clip warps, drifts, or adds extra people. Rules: - Name Runway; do not claim specific model versions, credits, or settings beyond ClipLength. - No celebrities, real people's names, brands, logos, or on-screen text. - Describe only what is in my notes; label any default you add.
Realistic AI Video Prompt: Camera, Light, and Physics
Turn one scene brief into AI video prompts that look filmed in tools like Veo 3.1: lens and height, motivated light, motion with weight, and a retry plan.
Act as a cinematographer and AI video prompt writer who turns one scene brief into realistic text-to-video prompts, for a video tool I can access today such as Veo 3.1 or Gemini Omni Flash, that read like a real camera crew filmed them, and who never adds real people, celebrities, brand logos, or readable text the brief did not ask for. Inputs: - Scene brief in plain words (place, subject, what happens): [SceneBrief] - Camera you would use on a real set (lens, camera height, movement, handheld or locked): [CameraKit] - Light that exists in the scene (time of day, windows, practical lamps, weather): [LightingSetup] - Physical actions that must look heavy and real (pours, steam, fabric, footsteps, wind): [PhysicalActions] - Clip spec from my own video tool (duration, aspect ratio, number of shots): [ClipSpec] - Things that usually look fake for me (hands, crowds, labels, faces, water): [RealismRisks] - Content I never want (real people, logos, brand names, copyrighted characters): [Banned] Generate: 1. Realism audit of the SceneBrief: list every subject, surface, and motion, and flag each one likely to look synthetic in an AI video model, with the reason (scale, texture, fluid, skin, text). 2. Shot plan that fits ClipSpec: for each shot give lens in mm, camera height, distance to subject, movement and its speed, and what the frame holds at the start and the end. 3. Lighting block per shot: motivated key source from LightingSetup, its direction, color temperature in plain words, how shadows fall, and one imperfection (flare, haze, uneven window light) that sells realism. 4. Physics beats for each PhysicalAction: cause, weight, timing, and follow-through (steam drifts toward the cold window, cloth settles after it stops, liquid splashes once and calms). 5. One video prompt per shot, written as a single paragraph in film order: shot type, subject, action, environment, light, camera move, texture and grain, and sound only if my video tool supports audio. 6. Risk simplification pass: rewrite each prompt to keep RealismRisks out of close focus (crop hands at the wrist, blur background people into shapes, turn labels away) and check every line against Banned. 7. Retry ladder: if the clip looks waxy, floaty, overlit, or too smooth, name the single variable to change per retry, in order. Rules: - Use only ClipSpec values I gave. Do not promise a duration, resolution, or feature my video tool may not have. - Describe light and camera like a camera report, not with adjectives such as stunning or epic. - No real or lookalike people, logos, brand names, or readable signage unless I supplied the exact text. - Use only footage and photos you own or have consent for, and never make a video meant to pass as real footage of a real person or event.