I'll begin this page with general advice, after which there is a section on the art of writing good prompts and a section on maintaining consistent characters. The first video is by Tao Prompts - he is excellent:
Google has given some advice on writing good prompts, including these tips:
Basics for writing prompts
Good prompts are descriptive and clear. To get your generated video closer to what you want, start with identifying your core idea and then refine your idea by adding keywords and modifiers.
The following elements should be included in your prompt:
Subject: The object, person, animal, or scenery that you want in your video.
Context: The background or context in which the subject is placed.
Action: What the subject is doing (for example, walking, running, or turning their head).
Style: This can be general or very specific. Consider using specific film style keywords, such as horror film, film noir, or animated styles like cartoon style render.
Camera motion: Optional: What the camera is doing, such as aerial view, eye-level, top-down shot, or low-angle shot.
Composition: Optional: How the shot is framed, such as wide shot, close-up, or extreme close-up.
Ambiance: Optional: How the color and light contribute to the scene, such as blue tones, night, or warm tones.
Here are other tips that I have picked up:
Veo 3 often displays subtitles that were not asked for and which are incorrect. To reduce this problem one can include sentences like this in the prompt: "No subtitles, no captions, no written elements or text should appear on the screen at any point. The screen must remain clean — only visuals and spoken dialogue." It is apparently also helpful to avid putting quotes around any text that is to be spoken within the video. Use a colon instead.
Many users say that it is useful to get ChatGPT to optimise your prompt. Here (Google doc) or here (corresponding YouTube video) is a possible instruction that you can give to ChatGPT so that it can guide you towards a good prompt.
In July 2025 it became possible to use a starting image in Veo 3 (in addition to a prompt) video generation, as opposed to just a prompt. That gives lots of control over how the video will look and therefore puts a lot of emphasis on the image, reducing the emphasis on the prompt.
Here are some more resources. The first video, by Kevin Stratvert (recommended) is good.
Many examples of interesting videos and their matching prompts can be viewed on Google Flow TV. Don't forget that the same prompt will never give the same video due to the random nature of AI. Actually these videos were made with Veo 2 rather than Veo 3. Veo 2 is cheaper to use but cannot generate audio.
Here are some very specific examples of certain types of video (selfie-style vlogger, street interview, cinematic, ASMR), by Cyber Jungle:
This format is designed to create videos that mimic the popular vlogging style, with the character holding the camera at arm's length.
Prompt Structure for Selfie Style Vlogger
A selfie video of a [character] in a [location]. [He/She] holds the camera at arm's length, and [his/her] arm is clearly visible in the frame. The style is YouTube Vlog like. The character [describe the action, e.g., walking, pointing, running]. He occasionally glances at the camera before doing something else in the scene. He says in a [describe tone or accent e.g., excited, nervous] tone: Lorem Ipsum. Sounds: [List specific ambient sounds and sound effects]. No subtitles.
Example Selfie Style Vlogger Prompts
Stormtrooper in Imperial Transport: A selfie video of a Stormtrooper in an Imperial Transport, with his squadmate Greg visible beside him. He holds the camera at arm’s length, and his arm is clearly visible in the frame. The style is YouTube Vlog like. The character gestures with his head towards a distracted Greg before looking back at the camera. He says in a casual, confident tone: "Alright boys, orders just came in from Vader's bitch ass and me and Greg are headed to Endor. Forest mission, easy terrain, little recon work. Should be chill." Sounds: The low hum of the ship's engines, metallic clanking of armor, muffled PA announcements.
Stormtrooper in Battlefield Trenches: A selfie video of a Stormtrooper in battlefield trenches, hunkered down next to his squadmate Greg. He holds the camera at arm’s length, and his arm is clearly visible in the frame. The style is YouTube Vlog like. The character is staring into the camera, looking exhausted, then glances at his squadmate. He says in a weary, fed-up tone: "Okay. We've been getting shot at for six hours. I haven't eaten." His squadmate Greg leans into frame and adds: "I don't want to die out here before Battlefront 3." Sounds: Constant blaster fire, distant explosions, dirt trickling down, the scrape of armor against the trench wall. No subtitles.
Stormtrooper on Tatooine (Sarcastic): A selfie video of a Stormtrooper in an Imperial Transport, with his squadmate Greg in the background fiddling with a nav-computer. He holds the camera at arm's length, and his arm is clearly visible in the frame. The style is YouTube Vlog like. The character looks at the camera, then glances worriedly towards his squadmate. He says in a sarcastic, resigned tone: "Alright, this is it. We're shipping out for Tatooine. High heat, endless sand, probably locals that hate us. Sounds amazing." His squadmate Greg says proudly from the background: "I pre-loaded the nav-route myself." The vlogger quickly looks back at the camera and says in a dread-filled tone: "Oh no." Sounds: Ship engine humming, armor clanking, frantic beeping from the nav-computer. No subtitles.
Stormtrooper in Crashing Ship: A selfie video of a Stormtrooper in the co-pilot seat of a crashing ship, with his squadmate Greg visible in the pilot's seat. He holds the camera at arm's length, and his arm is clearly visible in the frame. The style is YouTube Vlog like. The character is being shaken while yelling at his squadmate, who is frantically pulling levers. He says in a furious, terrified tone: "I told you, Greg! I told you not to touch the nav system! You plotted it upside down, Greg! Greg! What the f*** are you doing?!" Greg shouts back: "Correcting for wind shear." Sounds: Loud ship alarms, groaning metal, rattling panels, the roar of atmospheric entry. No subtitles.
Stormtrooper on Tatooine (Post-Crash): A selfie video of a Stormtrooper in the Tatooine desert, with the crashed ship smoking behind him. He holds the camera at arm's length, and his arm is clearly visible in the frame. The style is YouTube Vlog like. The character shows the wreckage, then points the camera at his squadmate Greg who is climbing out of the cockpit looking dazed. He says in a disbelieving, deadpan tone: "We've… landed… if you can even call this landing." His squadmate Greg stumbles out and says: "That was the smoothest one yet." The vlogger stares at the camera and says: "There is no way this man passed flight school." Sounds: Hissing steam from the wreckage, strong desert wind, a distant Krayt Dragon roar. No subtitles.
Man in Tokyo Subway (Strained): A selfie video of a man in his late 20s crammed into a violently swaying Tokyo subway car during rush hour. He holds the camera close to his face, and his arm is barely visible amidst the crowd. The style is YouTube Vlog like. The character is sweating, grimacing at the crush of people before looking into the lens. He says in a strained, breathless tone: "Welcome to the Tokyo crush. You can't even breathe… and yup, someone definitely just farted. Sounds: Loud Fart, the screech of train wheels. No subtitles.
Man in Tokyo Subway (Wide-Eyed): A selfie video of a japanese man in his late 20s, his face pressed against the glass of a subway car so crowded that people are a solid mass. He holds the camera, only his face and the faces of those crushed around him are visible. The style is YouTube Vlog like. The character is wide-eyed and silently pans the camera focusing on another man who is helplessly trying to play his own acoustic guitar among crowd. Sounds: guitar music playing. No subtitles.
Bigfoot and Yeti in Nevada Desert: A realistic selfie video of a massive, hairy creature that looks like Bigfoot, standing in the dark Nevada desert with a companion that looks like a Yeti. He holds the camera at arm’s length, and his shaggy arm is clearly visible in the frame. The style is YouTube Vlog like. They sprints to the Area 51 fortified gate, presses a large red button, and then both creatures turn and run away into the darkness. He says something funny and starts laughing. No subtitles.
Bigfoot and Yeti in Area 51: A realistic selfie video of a massive, hairy creature that looks like Bigfoot, standing in the dark Nevada desert with a companion that looks like a Yeti. He holds the camera at arm's length, and his shaggy arm is clearly visible in the frame. The style is YouTube Vlog like. They are inside running in the Area 51 corridor, chased by aliens. He tells a joke. No subtitles.
Hannibal with Elephants: A selfie video of a rugged, bearded man in ornate Carthaginian armor, who looks like Hannibal, riding atop a massive war elephant. He grips the elephant’s harness, his armored arm clearly visible. The style is YouTube Vlog like. The character grips the elephant's harness, panning the camera to show his elephant smashing through a line of terrified Roman soldiers, then turns the camera back to his own determined face. He says in a grimly satisfied, commanding tone: "Surprised them with elephants. The legions are breaking. They have never faced such fear. Sounds: The terrifying, trumpeting roar of elephants, the panicked screams of men, the splintering of wooden shields, the rhythmic, ground-shaking thud of elephant feet. No subtitles.
Moses Parting the Sea: A selfie video of an old man with a long white beard and robes, who looks like Moses, caught in a massive, surging crowd on a wet seabed. He holds the camera high, his arm clearly visible over the heads of others. The style is YouTube Vlog like. The character stumbles forward, panning the camera across the impossible scene: a dry path carved between two impossibly tall, curving walls of translucent water. Within the trembling liquid walls, the giant shadows of whales and sharks drift past. He then spins the camera to show the endless column of people pressing forward. He says in a breathless, excited tone: "Hey guys, quick update. We are literally walking on the bottom of the sea right now. This is insane. Sounds: A deep, powerful hum vibrating from the water walls, the echoing murmur and shouts of a massive crowd, the wet squelch of thousands of footsteps on sand and rock. No subtitles.
Spartan Warrior: A selfie video of a ripped, muscular man in a bronze Spartan helmet and red cape, standing in a narrow pass with hundreds of other Spartan warriors behind him. He holds the camera at chest level, his heavily muscled arm visible. The style is YouTube Fitness Vlog like. The character flexes his bicep, then dramatically points down the pass as if facing an unseen enemy. He then winks at the camera. He says in a booming, confident voice: "The Persians are outnumbering us, but my testosterone game is strong! My new Thermopylae Test Boost gives me peak male performance. Order now, link below! Stay strong! Sounds: The clang of swords and shields in a with a significant splash. The reporter and the vendor both turn with expressions of shock. No subtitles.
Moon Landing Conspiracy: A selfie video of a man with wild eyes and a tinfoil hat, standing on what looks like a dusty, dimly lit film studio designed to be like the moon's surface. He holds the camera at arm's length, his arm is clearly visible in the frame. The style is YouTube Vlog like. The character frantically gestures around the set, knocking on a lunar module prop which wobbles and makes a hollow sound. He peels back a piece of duct tape to show the cardboard underneath. He says in a triumphant, manic whisper: "You didn't believe me, but I have proof now! They're staging the whole thing right here in Hollywood. This lunar module? It's cardboard and duct tape! Sounds: The low hum of studio lights, the distant, muffled chatter of a film crew, the hollow thud of knuckles on cardboard, the ripping sound of duct tape. No subtitles.
Jesus at the Wedding in Cana (Conversation):
Prompt 1 (Mary): A selfie video of a man who looks like Jesus, standing with his mother Mary in a quieter corner of the wedding celebration in Cana. He holds the camera at arm's length, and his arm is clearly visible in the frame. The style is YouTube Vlog like. The character has a slightly exasperated but loving expression as his mother, Mary, leans into the frame, looking directly at the camera with a knowing smile. Mary says in a playful, matter-of-fact tone: "Hi guys, I'm J.C.'s mom, and I just told him the wine's gone." Jesus looks at the camera and adds in a flat, resigned tone: "Totally out." Mary finishes, gesturing towards Jesus with her thumb, saying in a conspiratorial tone: "So now, it's officially his problem. Sounds: The distant, muffled sound of wedding music and chatter, the clinking of empty clay wine jars being set down nearby. No subtitles.
Prompt 2 (Jesus): A selfie video of a man who looks like Jesus with beard and kind eyes, holding a clay cup filled with dark red wine. He holds the camera at arm's length, and his arm is clearly visible in the frame. The style is YouTube Vlog like. The character is beaming, taking a sip from the cup and giving an enthusiastic thumbs-up to the camera. In the background, wedding guests are cheering and servants are rushing to serve the new wine. Jesus says in a relieved, triumphant tone: "It worked! Crisis averted, vibes restored. And let's be honest, if we'd ordered wine for delivery, it would have taken years." Mary says: "yes absolutely."
Monkey and Lion in the Jungle: A selfie video of a small monkey with dreadlocks and a Jamaican-colored beanie in a lush jungle. He holds the camera at arm's length, and his long, hairy monkey arm is clearly visible in the frame. The style is YouTube Vlog like. The character is sitting next to a large, calm lion, cautiously reaching out with his free hand to gently stroke the lion's furry cheek. The lion seems completely unbothered, blinking slowly. He says in a shaky but excited Jamaican accent: "Look at dis now! De king of de jungle, big and mighty, but him just a big kitty cat, see? Pure vibes, no danger. We're bredda's now. Sounds: The low, gentle rumble of the lion purring, the chirping of tropical birds, the rustle of leaves in the breeze. No subtitles.
Monkey on a Beach with Volleyball: A selfie video of a small monkey with dreadlocks and a Jamaican-colored beanie on a beautiful, deserted island beach. He holds the camera at arm's length, and his long, hairy monkey arm is clearly visible in the frame. The style is YouTube Vlog like. The character is sitting in the sand, having a very serious conversation with a volleyball that has a red handprint face painted on it. He pats the top of the volleyball gently with his free hand. He says in a thoughtful, Jamaican accent: "Wilson, me bredda, you always listen. Dat is deep, man. You always know de right ting to say. Sounds: The gentle lapping of waves on the shore, the rustle of palm trees in the wind, the distant cry of a seagull. No subtitles.
This format captures the classic on-the-street news report, often with a humorous or unexpected twist.
Prompt Structure for Street Interview
A street interview style video. Over-the-shoulder shot of a reporter holding a microphone with a news logo. [INTERVIEWED SUBJECT] giving an interview. Background: [BACKGROUND DESCRIPTION]. Reporter says: Lorem Ipsum. [INTERVIEWED SUBJECT] says: Lorem Ipsum. No subtitles.
Example Street Interview Prompts
French Crepe Maker in Food Fight: A street interview style video. Over-the-shoulder shot of a female reporter holding a microphone with a news logo. A French crepe maker is giving an interview. Background: A chaotic food fight in international food expo in Bryant Park. Reporter says: "Sir, did you saw how it began?" Pierre, the French Man with french accent says: "It began with sacrilege! He takes my delicate crepe, a masterpiece, and he puts… a DOLLOP of sour cream on it! Not crème fraîche! SOUR CREAM! It is an act of barbarism! So, I slap him. It was a slap for all of France! No subtitles.
Man in Black Shirt at NYC Line: A street interview style video. Over-the-shoulder shot of the reporter. A handsome French man in his 30s, wearing a stylish black shirt, speaks with a charming French accent. Background: The diversity of the New York City line is on full display around him, with people from all walks of life patiently waiting. Man in Black Shirt says with french accent: "I'm a French tourist and I don't know what is going on but it's in my blood to stand with others and encourage them. No subtitles.
Colombian Man and Pothole: A street interview style video. Over-the-shoulder shot of a female reporter holding a microphone with a new news logo. An enthusiastic street vendor is giving an interview. Background: A dusty, uneven street in a vibrant Colombian town. Dominating the scene beside them is a huge, muddy pothole filled with water. In the background, a figure pushing a wheelchair with an elderly woman is approaching. Reporter says: "The community is hopeful this hazard will finally be addressed. You agree?" The Street Vendor says with a cheerful Colombian accent: "¡Claro que si! This hole has been a nightmare for too long. Today is a good day! Action: Just as he finishes speaking, the wheelchair in the background hits the edge of the massive pothole and tumbles in with a significant splash. The reporter and the vendor both turn with expressions of shock. The Vendor exclaims, his cheerfulness gone: "¡Ay, Virgen! ¡No puede ser! no subtitles.
Irish Man on the Titanic: A street interview style video. Over-the-shoulder shot of a reporter interviewing a hopeful but scruffy man in a simple cap and worn clothes. Background: The crowded, noisy steerage-class deck of the Titanic. People are dancing to a fiddle. Reporter says: "You're heading to a new life in America! This must be exciting! The Irish Man says: "Oh, it's grand! And this boat is so terribly big, I'm sure we'll be perfectly safe from any and all floating ice. no subtitles.
These prompts are designed to create videos with a high-quality, film-like aesthetic, focusing on visual storytelling and atmosphere.
VEO 3 Cinematic Prompt Structure
Cinematic realistic [Camera Motion or Shot Type] [Consistent Character Description] [Background] [Action] [Style] [Ambiance] [Time of Day] [Year] [Dialogue] [Sound effects] [Voice-over] [Ambient noise] [Music] No subtitles.
Example Cinematic Prompts
Gorilla Driving a Car: A dynamic tracking shot captures a classic Porsche as it speeds along a winding, sun-drenched coastal highway. Camera flies inside, showing a gorilla, incongruously dressed in a dusty suede jacket layered over a simple gray peacoat, grips the steering wheel with intense focus. He glances in the rearview mirror, a flicker of urgency in his dark, intelligent eyes as he expertly navigates the sharp turns. The vast, churning ocean stretches out beside him, its distant roar a constant undercurrent to the powerful growl of the engine, creating a scene of determined escape against a backdrop of rugged, contemplative beauty, cinematic shot.
Olympics with Animals: televised footage of [Animal, subject or character] is doing an acrobatic dive into a swimming pool at the olympics, from a 10m high diving board, flips and spins.
This section explores creating Autonomous Sensory Meridian Response (ASMR) videos, focusing on the satisfying sounds of cutting various objects.
Prompt Structure for ASMR
A hyper-realistic AI-generated ASMR video shot in extreme macro perspective. A razor-sharp knife slowly slices through [insert the object you want to cut, e.g., glass strawberry] — The object rests on a rustic wooden cutting board, bathed in warm, cinematic lighting. Each intentional cut delivers a deeply satisfying ASMR audio experience — crisp, juicy and clean - echoing in an otherwise silent, meditative space. Sliced piece drops gently to the cutting board. The overall pacing is slow and soothing, crafting a tranquil ASMR experience.
Example ASMR Prompts
Glass Strawberry: A hyper-realistic AI-generated ASMR video shot in extreme macro perspective. A razor-sharp knife slowly slices through an opaque glass strawberry in original fruit color — The object rests on a rustic wooden cutting board, bathed in warm, cinematic lighting. Each intentional cut delivers a deeply satisfying ASMR audio experience — crisp, juicy and clean - echoing in an otherwise silent, meditative space. Sliced piece drops gently to the cutting board. The overall pacing is slow and soothing, crafting a tranquil ASMR experience.
Opaque Glass Watermelon: A hyper-realistic AI-generated ASMR video shot in extreme macro perspective. A razor-sharp knife slowly slices through an opaque glass watermelon in original fruit color — The object rests on a rustic wooden cutting board, bathed in warm, cinematic lighting. Each intentional cut delivers a deeply satisfying ASMR audio experience — crisp, juicy and clean - echoing in an otherwise silent, meditative space. Sliced piece drops gently to the cutting board. The overall pacing is slow and soothing, crafting a tranquil ASMR experience.
Glass Banana: A hyper-realistic AI-generated ASMR video shot in extreme macro perspective. A razor-sharp knife slowly slices through a glass banana — subtle reflections and internal texture reminiscent of glowing amber. The object rests on a rustic wooden cutting board, bathed in warm, cinematic lighting. Each intentional cut delivers a deeply satisfying ASMR audio experience — crisp and clean - echoing in an otherwise silent, meditative space. The overall pacing is slow and soothing, crafting a tranquil ASMR experience.
Metal Lemon: A hyper-realistic AI-generated ASMR video shot in extreme macro perspective. A razor-sharp knife slowly slices through a metal lemon — subtle reflections and internal texture reminiscent of glowing amber. The object rests on a rustic wooden cutting board, bathed in warm, cinematic lighting. Each intentional cut delivers a deeply satisfying ASMR audio experience — crisp, wet, and clean - echoing in an otherwise silent, meditative space. The overall pacing is slow and soothing, crafting a tranquil ASMR experience.
Detergent Bottle Ice Cream: Hyper-realistic ASMR video. A big detergent-bottle-shaped block of ice-cream stands upright, its exterior mimicking glossy HDPE plastic—molded handle, injection seam, bright printed label, micro-condensation beads. A polished metal ice-cream scoop enters frame, slides and lifts one perfect ice cream ball from the bottle’s surface: creamy interior contrasts with the plastic-looking shell. The scene is bathed in warm, cinematic key light with soft bounce fill; background fully black for isolation. Each scrape and release produces soft, wet pops and muffled crunches that echo in an otherwise silent, meditative space, crafting a slow, soothing ASMR experience.
Opaque Gold Bar: A hyper-realistic AI-generated ASMR video shot in extreme macro perspective. A razor-sharp knife slowly slices through an opaque gold bar in original gold color — The object rests on a rustic wooden cutting board, bathed in warm, cinematic lighting. Each intentional cut delivers a deeply satisfying ASMR audio experience — crisp, metallic and clean - echoing in an otherwise silent, meditative space. Sliced gold drops gently to the cutting board. The overall pacing is slow and soothing, crafting a tranquil ASMR experience.
Podcast with Woman: A hyper-realistic AI-generated ASMR video. Static medium shot of a cozy podcast studio with warm lighting. A calm Asian woman sits behind a microphone, whispering softly and creating gentle ASMR sounds.
Podcast with Dog: A hyper-realistic AI-generated ASMR video. Static medium shot of a cozy podcast studio with warm lighting. A calm wiener dog sits behind a microphone, eating some delicious looking nicely prepared food, creating soft ASMR sounds.