New: get 25 ready-to-use AI prompts, free. Get the pack

Free AI Prompt Library

This free AI prompt library has copy-paste prompts for Gemini, ChatGPT, Claude, Veo, Kling, Seedance and Pollo AI. Browse prompts for photo edits, product photography, AI videos and Instagram captions, then customise the [bracketed] parts with your own details. New prompts are added regularly, and every prompt has its own page with tips.

🪔 New: Diwali AI Prompts 2026 for couple photos →
Coffee AI video trend before and after: two dogs making coffee replaced by a woman in a red kurta and a man in a light blue kurtaVideoFree

Trending coffee AI video prompt: turn pets into people

START Create a photorealistic kitchen comedy featuring EXACTLY TWO ADULT HUMANS: [Person A, e.g. the reference woman in a red kurta] and [Person B, e.g. the reference man in a pale blue-white kurta]. COMPLETELY REPLACE BOTH SOURCE DOGS [Person A] replaces the larger foreground dog. [Person B] replaces the smaller background-left dog. The two source dogs become these two humans; they must NOT remain as additional pets. Remove every dog, animal body, animal head, paw, ear, tail, fur patch and animal reflection from every frame. IDENTITY AND WARDROBE Preserve [Person A]'s recognizable face, [hairstyle], [accessories, e.g. small bindi and earrings] and [outfit]. Preserve [Person B]'s recognizable face, [hairstyle], [facial hair] and [outfit]. Use realistic adult human proportions. [Person B] is naturally [taller / shorter] than [Person A]. Keep their identities and clothing consistent during movement and head turns. Use the portraits for facial identity, NOT for a permanent smiling expression. Their expressions must change with the action. STORY AND COMIC REACTIONS Follow the original coffee-making actions and cup movements in their original order and timing. Make the cause and effect unmistakable: One person is aiming a spoon or pouring toward a red cup. The other person unexpectedly moves that cup out of position. The person preparing the drink notices the moved cup and resulting spill, then reacts directly to the person responsible. The frustrated reaction must happen immediately AFTER the cup moves: 1. Eyes track the moving cup. 2. The interrupted hand pauses naturally. 3. Brows draw together and lips press tight. 4. A sharp sideways look toward the other person. 5. An incredulous "seriously?" expression, followed by a restrained palm-up gesture or small head shake. For [Person A]'s foreground performance, show clear mounting annoyance: concentrated coffee preparation, sudden disbelief at the moved cup, then an irritated stare at [Person B]. Their mouth and face must not remain pleasantly smiling. [Person B] notices the reaction and responds with a brief guilty freeze, raised eyebrows and an awkward apologetic look. Keep the exchange playful and exaggerated enough to read as comedy, while remaining believable. Let reactions build with each mistake. Use short pauses and eye contact for comic timing. Do not invent extra spills, aggressive behaviour or unrelated actions. HUMAN HANDS AND PROP CONTACT Translate the source paw actions into anatomically natural human hands. Each person has two arms and two hands, with five fingers per hand. Use believable grips on the kettle, spoon, coffee jar and mug handles. Keep objects attached to the correct hand during movement. When a cup is moved, show a hand physically moving it across the counter; it must not teleport. Maintain the original number of mugs and utensils. Coffee granules and liquid spill only in response to visible actions. Keep spills on the counter consistently after they happen. SCENE AND CAMERA Preserve the source kitchen, wooden cabinets, countertop, lighting, framing and camera behaviour. Keep both faces readable during the reaction exchange. Maintain realistic human scale behind the counter, natural shadows and correct occlusion. Do not introduce either reference-photo background. AUDIO Preserve the original soundtrack and timing. Do not add dialogue, narration, barking or new music. Communicate the frustration through facial expressions, eye contact and body language. OUTPUT Approximately 15 seconds, vertical 9:16, 720p. Exactly two humans and zero animals throughout. Natural movement, recognizable faces and a clear, funny cup-moving misunderstanding. END

Pollo Recast
View full prompt →
AI lip-sync video prompt result: a man in a black turban and gold sunglasses on a New York subway platform, made with Pollo RecastVideoFree

Viral subway run AI lip-sync video prompt

Edit Video 1. Replace the man in the burgundy Renaissance costume with the person in Image 1, and make them lip-sync the song in Video 1. Image 1 supplies the complete character, including identity, face, body, anatomy, silhouette and clothing, with nothing kept from the source figure. Keep the subway train, commuters, platform, signage and all other environment details from Video 1. Video 1 controls the original performance, blocking, interactions, camera movement, framing, cuts, timing and audio; everything outside the listed replacement keeps its look and sound from Video 1. REFERENCES Image 1: [describe yourself: height and build, hairstyle or headwear, facial hair, glasses, top, trousers or skirt, shoes and accessories, e.g. "a tall man with short black hair, full beard, round glasses, a denim jacket over a white T-shirt, black jeans, white sneakers and a silver watch"]. Use their complete face, height, body proportions, clothing and accessories. AUDIO AND LIP-SYNC Keep the original song from Video 1 exactly: same vocals, lyrics, rhythm, volume and timing, with no new music, voices or sound effects. The person sings every line the source performer sings, with mouth shapes matched syllable by syllable: lip closures on M, B and P sounds, open mouth on long vowels, natural jaw movement and visible breathing. Their lips stay in sync while they walk, turn and sprint, with no lag, drift or frozen mouth. Any facial hair moves naturally with their jaw. SOURCE VIDEO Video 1 shows the performer rushing to a stationary subway car, pressing both palms to its window, then after a hard cut singing and reacting to camera before accelerating into a sprint along the platform and exiting as the camera whip-pans to a tiled wall. Keep the hard cut at 2.6s and the final whip-pan exactly in place. SHOTS Shot 1 (0.0–2.6s, continuous): Medium full handheld tracking shot from behind and side. The person moves from center toward mid-right beside the train on the left, taking long strides along the yellow tactile strip, keeping their height and proportions from Image 1. At 1.2s they reach the window, face it and press both open palms flat to the glass, then tap it in dramatic distress, lip-syncing wherever the source performer sings. [Their accessories, e.g. glasses or a cap] stay securely in place. Hard cut. Shot 2 (2.6–5.2s, hard cut): Medium close-up, head and upper torso fill the frame, no full legs. The person faces camera at center, turns from the train, puts one hand to their chest and sings the line toward camera with a wide, expressive open-mouthed delivery, matching the source performer's expression and mouth shapes exactly. Keep their face, hair or headwear, facial hair, clothing and accessories identical to Image 1; they begin moving forward. Shot 3 (5.2–11.6s, continuous): Handheld medium shot tracking backward and beside them. The person accelerates, then sprints left to right down the platform past maroon columns and the staircase, with long strides, arms pumping and clothing moving naturally, singing mid-run with breathless, energetic delivery and turning their head back toward camera. Their hair or headwear and accessories remain firmly in place. The train stays deep in the background on the left from 6.8s onward, partly hidden by columns. Shot 4 (11.6–12.2s, continuous): The person exits right/forward. The camera pans across the empty white tiled wall and ends in a fast, blurred whip-pan. PROPS [List any accessories from Image 1, e.g. "rigid metal sunglasses with dark lenses" or "a silver watch"]. Each accessory keeps its exact shape and stays correctly placed in every frame, including during the chest touch and sprint. SETTING Keep the original nighttime New York subway platform from Video 1: stainless steel train, yellow tactile edge strip, maroon steel columns, white tiled walls, staircase railings, commuters and cool fluorescent overhead light. NEGATIVE No extra people, no duplicated or cloned characters, no hybrid of source and target bodies, no source Renaissance costume or wig left over, no missing or changed features from Image 1 (hair, headwear, facial hair, glasses or accessories), no lip-sync lag, no mouth drift, no frozen or blurred mouth, no changed song or added voices, no morphing or melting objects, no gibberish text, no extra cuts, no new camera moves, no slow motion, no subtitles, no source person in the last shot, clean frame edge to edge.

Pollo Recast
View full prompt →
Gangnam Style AI video prompt result: a still from an AI dance video made with Pollo RecastVideoFree

Trending Gangnam Style AI video editing prompt

Replace the three main performers in the source video with the three men in the uploaded plain-gray-background reference. Use that image only for each man's identity, body proportions and jersey outfit. The placement in the reference is NOT the desired group placement in the video. CRITICAL CAST MAPPING: VIDEO CENTER / LEAD: the bearded man at the RIGHT of the reference image wearing a RED jersey. Replace the original lead performer in the black/red sports top, in all close-ups and group shots. In the final three-person shot, the RED jersey man must be CENTER. VIDEO SCREEN-LEFT SUPPORT: the TALL man at the LEFT of the reference wearing BLUE, with swept-back hair and short moustache/goatee. Replace the source supporting performer appearing screen-left of the lead in the final group shot. VIDEO SCREEN-RIGHT SUPPORT: the SHORTER man in the MIDDLE of the reference wearing BLUE, with upright short hair and a narrow face and moustache. Replace the source supporting performer appearing screen-right of the lead in the final group shot. Final group order from viewer's left to right: tall BLUE man, RED man, shorter BLUE man. Track each identity across cuts and movement. Never swap, blend, duplicate or merge identities. In individual shots show only the corresponding replacement; preserve original framing. IDENTITY AND CLOTHING: Match each man's individual facial proportions, skin tone, age, hairstyle, facial hair, visible tattoos, build and jersey details from his own portion of the reference. Completely replace source costumes with the appropriate red or blue jersey. Preserve colors, collars, gold trim on blue jerseys, sleeve styling, sponsor markings, fabric and fit as closely as possible. Use matching sports trousers where not shown. Do not transfer the original women's hair or makeup. All three reference men have uncovered eyes: REMOVE the source sunglasses and DO NOT add glasses. Keep natural animated expressions, not frozen reference faces. SONG, LIP-SYNC AND MOVEMENT: Preserve the original audio track exactly, including the same song, voices, lyrics, rhythm, duration and timing. Transfer each source performer's own singing mouth movements to the corresponding replacement, matching syllables, lip closures, jaw motion, facial expression, pauses and head turns. Each replacement sings only when that source performer sings; retain simultaneous singing where present. No new dialogue, voices, lyrics or audio offset. Follow the original dance, arm gestures, hand actions, posture, gaze and body movement. SCENE: Keep the original nighttime location, architecture, lights, greenery, background people, camera movement, cuts, framing, shot sequence and pacing. Match scale, perspective, lighting, shadows, contact and occlusion. Do not force all three into single-person shots. Do not copy the gray studio background into the source location. Remove added source text overlays and subtitles. No new captions, graphic overlays or watermark. Avoid face drift, mixed identities, disappearing tattoos, eyewear, costume changes, duplicated people, extra limbs, malformed fingers, distorted mouths or teeth, lip-sync lag, flicker and background deformation.

Pollo Recast
View full prompt →

What’s in the free AI prompt library

  • Image prompts: photo edits, portraits, 3D figures and product photography for Gemini and ChatGPT.
  • Video prompts: complete AI video workflows for Pollo AI, Kling and Seedance, with source videos and step-by-step instructions.
  • ChatGPT prompts: Reel hooks, captions and scripts for creators.
  • Business prompts: product descriptions for Amazon, Meesho and your own store.

New to AI video? Start with my beginner’s guide on how to make realistic AI videos.

Prefer to write your own? Use the free AI prompt builder to turn your idea into a detailed prompt for Gemini, ChatGPT and more in seconds.

How to use this free AI prompt library

  1. Pick a prompt by category, or search for what you want to make.
  2. Click Copy prompt and paste it into the tool listed on the card.
  3. Replace the [bracketed] parts with your own subject, product or photo.
  4. Generate, then adjust the style, lighting or format until it looks right.

Frequently asked questions

Are these AI prompts free?

Yes. Every prompt in the library is free to copy and use.

Which AI tools do the prompts work with?

Each prompt card shows the tools it works best with. Image prompts are written for Gemini and ChatGPT, video prompts for Pollo AI, Kling and Seedance, and text prompts for ChatGPT and Claude. Most prompts also work in similar tools with small changes.

Can I use my own photo?

Yes. Prompts that mention an uploaded photo are made for your own picture. Only use photos you have permission to use.

How often are new prompts added?

Regularly. Follow @jasdeep.ai on Instagram for new prompts.

Can I use these AI prompts for commercial projects?

Yes. You can use the prompts from this free AI prompt library in your own personal and commercial projects, like client work, ads or social media posts. Please don’t resell or republish the prompt collection itself. Always follow each AI tool’s own terms, and don’t use prompts to copy real people or brands without permission.