Grok prompts that turn a still into a short clip
Paste a still into Grok Imagine and ask for a short clip: product spin, portrait motion or landscape weather. Name duration, camera, motion and ambient audio. Not grok.com chat. Not NSFW.
In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Drafted with AI writing assistance and checked against our editorial policy.
Product Spin
5 promptsTurntable Yaw of the Pack
1/30✨ What it does
A 6 second locked yaw of the uploaded pack on a named surface, label staying readable. Keep the take where type does not morph and the motor stays quiet.
Animate [PRODUCT PHOTO] in Grok Imagine as a 6 second image-to-video clip, not a grok.com chat reply. Locked 50mm on a seamless sweep, 16:9. The pack sits on [SURFACE MATERIAL] and completes a 180 degree yaw so the front panel, then the side seam, stay readable. Motion is a constant-speed turntable with no bounce and no zoom. Softbox from camera left, thin rim from the right. Audio: quiet platter motor, cloth studio hush, no sung hook. Keep printed type from the still; do not redraw the mark. No extra hands. Photoreal product film, safe for a store listing.
Pro tip: Name the surface. Bare infinity often melts the base of the pack.
Bottle Label Catch Light
2/30✨ What it does
An 8 second close rotation where a kicker slides across the bottle label. Keep the pass where the highlight travels and no extra glass appears.
Turn [PRODUCT PHOTO] into an 8 second Grok Imagine clip. 85mm, shallow depth, 4:5. A glass [BOTTLE TYPE] rotates 90 degrees under a traveling kicker so the label catch-light slides from left serif to right. Camera is locked; only the bottle and the highlight move. Duration 8 seconds, end on a three-quarter view. Audio: a faint glass tick at the stop, fridge-like room tone, no voiceover. Keep condensation if the still has it; do not add ice cubes. No bartender, no extra glass. Photoreal. Not NSFW. Grok Imagine video, not chat.
Pro tip: Ask for a 90 degree turn. A full 360 often smears small type.
Sneaker Sole to Upper Orbit
3/30✨ What it does
A 10 second low-to-three-quarter orbit of the uploaded sneaker on a named floor. Keep the take where laces hold and only one scuff is heard.
From [PRODUCT PHOTO], generate a 10 second Grok Imagine orbit. Start 24mm low on the outsole of [SHOE COLOR] sneakers, then arc to a three-quarter of the upper. 16:9, gimbal-smooth, one continuous move. Motion: the shoe stays planted on [FLOOR TYPE]; the camera travels. Audio: rubber scuff once, distant HVAC, no stadium crowd. Keep laces and tooling from the still. Do not invent a second shoe unless the still shows a pair. No feet, no celebrity athlete. Photoreal retail film. Skip spoken lines.
Pro tip: Plant the shoe. Floating sneakers read as a 3D render, not a photo coming alive.
Watch Face Second Sweep
4/30✨ What it does
A 7 second macro where only the second hand moves on the uploaded watch. Keep the take where the case and strap stay still and the tick is dry.
Animate [PRODUCT PHOTO] for 7 seconds in Grok Imagine. 100mm macro, 1:1. The watch sits on [WRIST OR BLOCK]; only the second hand should move, at real clock speed, while the case, [STRAP MATERIAL] strap, and reflections stay locked to the still. Camera: tiny 2cm push, no tilt. Lighting: north-window soft. Audio: a precise tick, then a leather creak if a wrist is present; no radio bed. Do not add a second watch or a famous face attached to the wrist. Photoreal. Duration 7 seconds, then hold. Not a grok chat caption.
Pro tip: If the still is a phone screenshot of a watch, say so, or the bezel will invent facets.
Box Lid Lift Then Settle
5/30✨ What it does
A 9 second locked carton whose lid lifts a little, shows the nest, then settles. Keep the pass where print holds and no hand appears.
Use [PRODUCT PHOTO] as the hero carton. 9 second Grok Imagine clip, 35mm, 16:9, locked. A lid on [BOX COLOR] packaging lifts 20 degrees as if a hidden magnet opened it, reveals the product nest, then settles back without a slam. Motion is mechanical and slow. Audio: cardboard fiber, a soft magnet catch, studio air. No hands entering frame. Keep print and dielines from the still; do not invent a new slogan. Photoreal pack shot. Safe, not NSFW. Grok Imagine, not grok.com chat.
Pro tip: Cap the lift at 20 degrees. A full open often fabricates a new inner tray.
Portrait Motion
5 promptsBlink and a Quiet Breath
6/30✨ What it does
A 5 second locked portrait with one blink and a small breath, wardrobe unchanged. Keep the take where the face matches the upload and no one else appears.
Start from [YOUR PHOTO]. Make a 5 second Grok Imagine clip, 85mm, eye level, 4:5. The adult blinks once, then the chest rises for a small breath; hair, [OUTFIT], and background stay as photographed. Camera locked on a tripod. Lighting unchanged from the still. Audio: a soft inhale, room tone of [ROOM TYPE], no words. Keep identity locked: same age, same face, no beauty filter, no extra person over a shoulder. Mouths stay closed. Photoreal. Not NSFW. This is Imagine video from a still, not a grok chat portrait.
Pro tip: One blink. A second blink often warps eyelashes.
Hair Line in a Side Breeze
7/30✨ What it does
An 8 second side breeze that moves hair and a collar while the uploaded face holds. Keep the pass where the glance returns and no friend is invented.
Animate [YOUR PHOTO] for 8 seconds. 50mm, 16:9, slow lateral 10cm. A side breeze moves only [HAIR LENGTH] hair and a collar edge; the face holds, eyes may glance with the wind then return. Setting stays [SETTING]. Audio: fabric rustle, distant leaves or HVAC matching the setting, no song. Identity lock on the uploaded adult; do not youth-ify, do not add a friend. Photoreal daylight as in the still. Grok Imagine clip, not chat. Safe wardrobe, not NSFW.
Pro tip: Name hair length. Vague wind often turns short hair into a wig.
Look Off, Then Back
8/30✨ What it does
A 6 second locked head turn toward a named target, then back to camera. Keep the take where shoulders stay put and the face still matches.
From [YOUR PHOTO], produce a 6 second Grok Imagine take. 35mm, 3:2. Head turns 20 degrees toward [LOOK TARGET], holds two beats, returns to camera. Shoulders stay put. Camera locked. Audio: chair creak if seated, otherwise a faint city or office bed from [LOCATION HINT], mouths shut. Keep the same adult; no celebrity likeness, no second sitter. Photoreal. Duration 6 seconds. Not a grok.com chat rewrite of the face.
Pro tip: Cap the turn at 20 degrees so the far ear does not invent jewelry.
Smile That Stops Short of Teeth
9/30✨ What it does
A 7 second closed-lip smile that holds, eyes crinkling only as the still allows. Keep the take where teeth stay hidden if they were hidden.
Animate [YOUR PHOTO] into a 7 second Grok Imagine clip. 85mm, 4:5, locked. The mouth moves from the still expression to a small closed-lip smile, then holds. Eyes crinkle only as much as the still lighting allows. Outfit is [OUTFIT]. Audio: a quiet exhale, cafe cups far off if [SETTING] is public, no spoken hello. Do not add teeth if the still showed none. Do not add a second diner. Photoreal, identity locked, not NSFW. Imagine video, not chat.
Pro tip: Say closed-lip. Open smiles often change tooth shape.
Coat Collar Lifts in a Doorway
10/30✨ What it does
A 9 second vertical of a coat collar lifting once in a doorway while the person stays planted. Keep the take where the gust is single and the face holds.
Use [YOUR PHOTO] in a 9 second Grok Imagine doorway clip. 24mm, 9:16, slight low angle, locked. Wind from behind the door lifts [COAT TYPE] collar once, then it drops. The adult does not walk; only fabric and a loose hair move. Background stays the photographed [DOORWAY LOOK]. Audio: door weatherstrip, one gust, street hush. No extra pedestrian crossing. Face locked, adult, not a famous person. Photoreal. Not grok chat. Not NSFW.
Pro tip: Keep the body still. Walking plus wind usually breaks the uploaded face.
Landscape Weather
5 promptsClouds Cross a Ridge
11/30✨ What it does
A 12 second locked ridge where named clouds drift and the sun dips a half-stop. Keep the pass where no new building or hiker appears.
Turn [YOUR PHOTO] of a landscape into a 12 second Grok Imagine clip. 24mm, 16:9, locked. Clouds of [CLOUD TYPE] drift left to right across [RIDGE NAME LOOK] at real wind speed; foreground grass or rock stays almost still. Lighting shifts a half-stop as a cloud covers the sun, then returns. Audio: wind in [VEGETATION], a distant bird, no narration. Do not add hikers, cars, or a cabin that was not in the still. Photoreal. Duration 12 seconds. Imagine, not grok.com chat.
Pro tip: Name vegetation. Empty 'nature sounds' often adds a creek that was not there.
First Rain on Dry Asphalt
12/30✨ What it does
An 8 second locked street where first drops hit dry pavement and existing reflections grow. Keep the take where signs do not rewrite themselves.
Animate [YOUR PHOTO] of a street or plaza. 8 second Grok Imagine clip, 35mm, 16:9, locked. Dry [PAVEMENT TYPE] receives the first sparse drops, then a light sheet; reflections of existing signs grow, no new neon. Camera does not pan. Audio: drops, then a soft hiss, a drain gurgle if a grate is visible. No pedestrians invented. If the still has one adult, keep that identity; otherwise empty. Photoreal overcast. Not NSFW. Skip spoken weather reports.
Pro tip: Sparse first, then a sheet. Starting with a downpour hides the still.
Snow Dust on a Peak
13/30✨ What it does
An 11 second slow push as fine snow dusts the upper third of the uploaded peak. Keep the pass where the lower slope stays untouched.
From [YOUR PHOTO] of mountains, make an 11 second Grok Imagine take. 50mm, 16:9, very slow 1% push. Fine [SNOW TYPE] snow dusts the upper third; lower forest or rock stays as shot. Motion is fall plus a slight drift from camera left. Audio: high thin wind, no avalanche, no ski chatter. Do not add a lodge, a lift, or people. Photoreal alpine. Duration 11 seconds. Grok Imagine video from the still, not a chat description.
Pro tip: Keep snow in the upper third. Full-frame flakes often hide the original geology.
Fog Fills a Valley Floor
14/30✨ What it does
A 14 second high locked shot of fog pooling on the valley floor and stopping at a named landmark. Keep the take where ridge houses are not invented.
Animate [YOUR PHOTO] for 14 seconds. 20mm, 16:9, locked high angle. [FOG COLOR] fog pools on the valley floor and creeps up to [LANDMARK LOOK] without covering it. Trees in the still keep their species and spacing. Audio: muffled highway or river only if one is already visible, plus hush. No extra houses on the ridge. Photoreal dawn or the still's time of day. Grok Imagine, not grok chat. Safe landscape, no people invented.
Pro tip: Tell it where fog must stop. Unbounded fog erases the still.
Wind Runs Through a Field
15/30✨ What it does
A 9 second locked field where a wind bar bends stalks then lets them recover. Keep the take where the sky matches the still and no farmer appears.
Use [YOUR PHOTO] of [CROP OR GRASS] as the field. 9 second Grok Imagine clip, 35mm, 16:9, locked. A wind bar travels from left to right, bending stalks in a real wave, then they recover. Sky stays the photographed sky. Audio: dry hiss of the crop, one crow if rural, no tractor unless one is in the still. Do not add a farmer. Photoreal. Duration 9 seconds. Imagine video, not a chat caption. Not NSFW.
Pro tip: One wind bar. Looping gusts look like a gif, not a clip.
Like these prompts? There are full tutorials behind them.
Learn the workflows, not just the prompts. Easy-to-follow tutorials inside AI Academy, and the library keeps growing.
Steam, Flame, Water
5 promptsMug Steam in a Window
16/30✨ What it does
A 6 second locked mug with thin steam that breaks and vanishes. Keep the take where print on the mug holds and no extra arm appears.
Animate [YOUR PHOTO] of a mug or cup. 6 second Grok Imagine clip, 85mm, 4:5, locked. Thin [STEAM LOOK] steam rises, breaks, and vanishes; the mug print and [TABLE SURFACE] stay still. If a hand is in the still, lock that adult hand; do not grow a second arm. Audio: kettle tick from offscreen, quiet room, no sip voice. Photoreal. Duration 6 seconds. Grok Imagine, not grok.com chat. Safe, not NSFW.
Pro tip: Thin steam. Fog-machine clouds hide the cup and the still.
Soup Surface Simmer
17/30✨ What it does
A 7 second overhead simmer with a few bubbles and a drifting leaf. Keep the pass where the boil stays low and the bowl does not warp.
From [PRODUCT PHOTO] or a food still, make a 7 second Grok Imagine take. 50mm overhead, 1:1. [SOUP TYPE] shows a low simmer: two or three bubbles, a bay leaf drift, no violent boil. Bowl and [PLACEMAT] stay locked. Audio: tiny blip, spoon rest unused, kitchen fridge hum. No diner face unless it was in the still. Photoreal. Duration 7 seconds. Imagine video from the photo, not a recipe chat.
Pro tip: Cap bubbles at three. A rolling boil changes the food.
Candle Wick and Wax Pool
18/30✨ What it does
An 8 second locked candle whose wick flickers and wax pool shimmers while the label stays sharp. Keep the take where no extra candles appear.
Animate [YOUR PHOTO] of a candle. 8 second Grok Imagine clip, 100mm, 4:5, locked. The wick flame of [CANDLE SCENT LOOK] flickers in a small draft; the wax pool shimmers; the jar label stays readable. Audio: quiet burn, a distant wood tick, no birthday song. Do not add extra candles or a face behind. Photoreal warm interior. Duration 8 seconds. Grok Imagine, not chat. Not NSFW.
Pro tip: A small draft. Large flicker often melts the jar shape.
Ice Ring on a Tall Glass
19/30✨ What it does
A 10 second locked drink where ice rings once and a condensation track runs down. Keep the take where the liquid level does not drop.
Turn [PRODUCT PHOTO] of a drink into a 10 second Grok Imagine clip. 85mm, 16:9, locked. Ice in a [GLASS TYPE] rings once as a bubble escapes; condensation tracks down 2cm. Liquid level does not drop. Audio: one ice knock, bar hush, no bartender patter. Keep garnish from the still. No extra glass, no celebrity guest. Photoreal. Duration 10 seconds. Imagine, not grok chat.
Pro tip: One ice knock. Constant clinking reads as a looped stock clip.
Rain Beads on a Window Portrait
20/30✨ What it does
A 9 second locked window where rain beads travel and the uploaded adult may blink once. Keep the take where the silhouette count stays the same.
Start from [YOUR PHOTO] shot through or beside glass. 9 second Grok Imagine clip, 50mm, 4:5, locked. Rain beads on [GLASS TYPE] travel down, slightly distorting the adult behind; identity still locked. Audio: beads, gutter, indoor hush. The person may blink once. Do not add a second silhouette. Photoreal. Not NSFW. Duration 9 seconds. Grok Imagine video from the still, not a chat edit.
Pro tip: Say the person is behind the glass. Beads on a dry portrait look painted on.
Camera Moves on Stills
5 promptsSlow Push Into the Eyes
21/30✨ What it does
A 6 second smooth push from head-and-shoulders into the eyes, subject still. Keep the take where skin texture holds and the background does not crawl.
From [YOUR PHOTO], build a 6 second Grok Imagine push. Start 50mm on head-and-shoulders, end 85mm on the eyes, 4:5, gimbal-smooth. The adult holds still; only the camera moves. [OUTFIT] and [BACKGROUND HINT] stay as shot. Audio: a faint room tone match, no heartbeat SFX gimmick, mouths closed. Identity locked, no beauty morph, no extra person. Photoreal. Duration 6 seconds. Imagine, not grok.com chat. Not NSFW.
Pro tip: Stop at the eyes. Pushing into the pores often smears identity.
Three-Quarter Product Orbit
22/30✨ What it does
An 8 second 40 degree camera orbit around a planted pack, highlights moving on the finish. Keep the pass where type stays readable and the product does not spin.
Animate [PRODUCT PHOTO] with an 8 second Grok Imagine camera orbit. 35mm, 16:9. Camera travels 40 degrees around the pack on [SURFACE MATERIAL]; the product does not spin. Highlights on [MATERIAL FINISH] move as they would in a real studio. Audio: dolly whisper, no music sting. Keep type readable the whole way. No hands. Photoreal. Duration 8 seconds. Imagine video, not a chat description of the SKU.
Pro tip: Orbit the camera, freeze the pack. Dual motion doubles smear.
Tilt From Shoes to Face
23/30✨ What it does
A 10 second vertical tilt from shoes to face with one weight shift. Keep the take where the face at the end still matches the upload.
Use [YOUR PHOTO] as a full-length still. 10 second Grok Imagine tilt, 35mm, 9:16. Start on [SHOE OR FLOOR], tilt to the face, end in a medium. The adult may shift weight once; no walk cycle. Audio: a sole creak, indoor air of [SETTING], no greeting. Identity locked the whole tilt; do not invent a second pair of feet. Photoreal. Duration 10 seconds. Grok Imagine, not chat. Safe, not NSFW.
Pro tip: One weight shift. A walk cycle on a still usually breaks hips.
Crane Over a Still Landscape
24/30✨ What it does
A 12 second vertical crane from a named foreground object to the horizon. Keep the pass where new roads are not painted in.
From [YOUR PHOTO] of a wide view, make a 12 second Grok Imagine crane. Start low on [FOREGROUND OBJECT], rise to reveal [HORIZON LOOK], 16:9, 24mm. Motion is vertical only, no yaw. Audio: wind gain with height, a distant [AMBIENT HINT], no drone buzz unless you want that. Do not add roads or people missing from the still. Photoreal. Duration 12 seconds. Imagine, not grok chat.
Pro tip: Vertical only. Adding yaw on a still often tears tree lines.
Parallax Layers on a Street
25/30✨ What it does
A 7 second lateral parallax where a near prop moves more than the far building. Keep the take where empty sidewalks stay empty.
Animate [YOUR PHOTO] of a street or alley as a 7 second Grok Imagine parallax. 28mm, 16:9, 15cm lateral. Near [FOREGROUND PROP] moves more than the far [FAR BUILDING LOOK]. No new shop names. Audio: a passing bike or distant horn matching the still's city, no crowd walla with invented voices. If a figure is in the still, lock that adult; do not populate empty sidewalks. Photoreal. Duration 7 seconds. Grok Imagine, not chat.
Pro tip: Name the near prop. Without it, the whole frame slides like a sticker.
Go from copy-pasting to actually mastering AI.
AI Academy: hands-on tutorials on ChatGPT, Claude, Midjourney, and many other tools. New tutorials are added regularly.
Fabric, Flags, Water Edges
5 promptsCurtain Billow, Then Rest
26/30✨ What it does
An 8 second locked window whose curtain billows once and settles, floor light shifting. Keep the take where the room does not gain a new person.
From [YOUR PHOTO] of a window, generate an 8 second Grok Imagine clip. 50mm, 16:9, locked. [CURTAIN COLOR] fabric billows inward once, then settles. Light on the floor shifts with the fabric. Audio: cloth, a far street, no spoken line from inside. If an adult is on the sill, lock [YOUR PHOTO] identity; otherwise empty room. Photoreal. Duration 8 seconds. Imagine video, not grok chat. Not NSFW.
Pro tip: One billow. Constant waving looks like a fan pointed at the lens.
Scarf End in a Walking Still
27/30✨ What it does
A 6 second planted portrait where only a scarf end lifts and drops. Keep the pass where the body does not start a walk cycle.
Animate [YOUR PHOTO] for 6 seconds. 35mm, 4:5, locked. The adult stays planted; only a [SCARF COLOR] scarf end lifts and drops. Hair may move a little. Audio: wool rustle, outdoor air of [SETTING], mouths closed. Identity locked, adult, no extra walker. Photoreal. Duration 6 seconds. Grok Imagine, not a chat rewrite. Safe clothing, not NSFW.
Pro tip: Plant the feet in the prompt. Scarf motion plus a step usually warps legs.
Pool Surface From a Deck Still
28/30✨ What it does
A 9 second locked pool edge with a small ripple from camera right. Keep the take where swimmers are not added and tile stays sharp.
Use [YOUR PHOTO] of a pool or water edge. 9 second Grok Imagine clip, 24mm, 16:9, locked. [WATER COLOR] surface gets a small ripple from camera right; tile and [DECK MATERIAL] stay sharp. Audio: water tick, pump far off, no splash party. Do not add swimmers. Photoreal sun as in the still. Duration 9 seconds. Imagine, not grok.com chat.
Pro tip: A small ripple. Big waves hide the deck that sold the still.
Fountain Restart From a Pause
29/30✨ What it does
An 11 second locked fountain whose jets climb from a pause to the still's height. Keep the pass where the plaza stays empty of new tourists.
Animate [YOUR PHOTO] of a fountain. 11 second Grok Imagine take, 35mm, 16:9, locked. Jets of [FOUNTAIN LOOK] start from a near pause, reach the photographed height, and hold. Stone and [PLAZA LOOK] stay as shot. Audio: water climb, then a steady veil, plaza hush. No invented tourists. Photoreal. Duration 11 seconds. Grok Imagine video from the still, not a travel chat.
Pro tip: Start from a pause. Full blast from frame one hides the still's water shape.
Flag Luff on a Building Still
30/30✨ What it does
A 7 second locked building still where the flag luffs twice then hangs. Keep the take where the design matches the photo and the roof stays empty.
From [YOUR PHOTO] of a building or mast, make a 7 second Grok Imagine clip. 50mm, 16:9, locked. A [FLAG COLOR] flag luffs twice in real wind, then hangs. Brick, glass, and [SKY LOOK] stay. Audio: cloth snap, wind, a far truck, no anthem. Keep the flag design from the still; do not invent a national mark the photo did not show. No extra people on the roof. Photoreal. Duration 7 seconds. Imagine, not grok chat.
Pro tip: Two luffs. A constant snap looks like a fan on a stick.
Free tool
Image Prompt Builder
Build a rich copy-paste prompt for Midjourney, DALL-E, or Stable Diffusion.
Frequently Asked Questions
Prompts are the starting line. Tutorials are the finish.
A growing library of hands-on tutorials on ChatGPT, Claude, Midjourney, and many other AI tools. New tutorials are added regularly.
Related guides