flow-tips-n-tricks・✏️

[calm kiln]

Actually very surprised it worked

Okay please is anybody here going to help me animate this picture i've been trying to animate it since yesterday but it seems like the VEO2 on google AI studio is more restrictive than the one on vfx.

Man in a blue hi-vis vest with the UMTK Solar logo demonstrates a small solar panel on a sunny city rooftop. He lifts the panel, tilts it toward the sun, then gestures across its surface as if explaining how it works. He points toward a nearby rooftop, then taps the panel with confidence.

.I don't understand why they won't generate this for me. Nothing controversial or sensitive.

[calm kiln]

Please try mine. Thanks.

[molten parcel]

it also keeps blocking it for me

sorry :/

[calm kiln]

I understand and appreciate what you're saying, but I did all of that. I even removed the prompt entirely and tried to generate the video without it, but it's still not generating. I believe this would have made sense if the brand were widely known.

[teal quartz]

First question on the Veo 2 video editor is it possible to change the aspect ratios??

[outer lark]

I'm not sure if this applies to Veo 2 but this is what I found when searching online

[valid briar]

I'm getting "We couldn't create what you asked for." with this prompt. Any Advice?

Generate a dynamic 8-10 second video of a brightly lit, futuristic classroom circa 2075 where diverse children (ages 8-12) collaboratively learn using holographic tables and individual augmented reality interfaces, assisted by sleek humanoid robots that offer guidance via projections and physical demonstrations; a warm, approachable human teacher facilitates from the front, gesturing towards a large central holographic display showcasing interactive educational content like 3D models and data visualizations, while the background features transparent interactive whiteboards and subtle futuristic technology, all captured with a slightly low, gently panning camera angle to convey an energetic and optimistic mood focused on seamless tech integration and collaborative learning.

[ashen juniper]

Maybe it blocked the children part

[valid briar]

Ok, I'm removing the word children and the mention of ages and just putting the word "students". If it has issue with those terms why does it go all the way to 99% and then decides it can't do it.

[ashen juniper]

Idk but try it this way now

[valid briar]

Tried this prompt and now I got an "Oops, something went wrong!"

Generate a dynamic video of a brightly lit, futuristic classroom circa 2075 where diverse students collaboratively learn using holographic tables and individual augmented reality interfaces, assisted by sleek humanoid robots that offer guidance via projections and physical demonstrations; a warm, approachable human teacher facilitates from the front, gesturing towards a large central holographic display showcasing interactive educational content like 3D models and data visualizations, while the background features transparent interactive whiteboards and subtle futuristic technology.

[ashen juniper]

Have no idea why it does that. Maybe try "adults" now 😂

[valid briar]

I just keep getting the "something went wrong" message now. I've shortened the prompt significantly. Maybe it doesn't like the idea of generating a video of a classroom/

[worn cedar]

yes, classroom is sensitive.

try work arround with synonyms of other way to describe the same idea.

[onyx badger]

censorship is super sensitive , but whats nutty is that some of the previous ones that I made initially said to try something else then about one hour later the video appeared, so maybe they have a manual effort looking at generations before they release it.

[reef sorrel]

I had to rework the prompt completely

A smooth, low panning shot moves through a vibrant, futuristic classroom in 2075, styled as a documentary. The camera glides across the room, showcasing a diverse group of young learners engrossed in their studies at interactive holographic desks that glow with soft light. Augmented reality overlays shimmer around them, enhancing the learning experience. Sleek, white humanoid assistants move gracefully among the students, projecting colorful visual aids and demonstrating complex concepts with fluid gestures. At the front of the classroom, a welcoming human instructor with a warm smile gestures towards a large central hologram, filled with vibrant, animated educational models. Transparent smartboards display equations and diagrams, while ambient tech details hum softly in the background, adding to the optimistic and collaborative atmosphere. The lighting is bright and even, casting a sense of innovation and enthusiasm over the entire scene.

[onyx badger]

me discovering all the crazy stuff you can make with this tool

[kindred wind]

I'm I giving to much detail? I cant get the video to come out the way I want. Made some really funny ones though.
Prompt was.
The scene opens on a wide shot: a rugged, windswept cliff face, its jagged edges silhouetted against a vast, pale sky. Perched precariously at the cliff's edge is a massive boulder, a dark, rough-hewn mass against the lighter stone of the precipice. Its surface is mottled with lichen and shadowed by deep crevices, suggesting an age measured in millennia.

The camera begins a slow, deliberate zoom. As it closes in, the subtle details of the boulder become more defined. We see the sharp, angular facets of its surface, the way the light catches the tiny crystals embedded within the rock. Then, something strange begins to happen.

From the boulder's upper surface, two thick, knobbly protrusions emerge. They extend outwards, initially clumsy and undefined, but rapidly morphing into recognizable arms. These arms, still rough and stony, flex and twitch, as if testing their newfound mobility.

Next, a distinct head forms atop the boulder. It's a crude, almost comical shape, with two dark, recessed spots that suggest eyes, and a slightly protruding ridge that hints at a brow. The head swivels slightly, as if taking in its surroundings with a bewildered, stony gaze.

Finally, two thick, stubby legs sprout from the boulder's base. These limbs are short and disproportionately thick, giving the emerging figure a top-heavy, ungainly appearance. The legs begin to extend, but their growth is uneven, causing the boulder-creature to wobble precariously.

As the legs lengthen, the wobble intensifies. The boulder-man, now a bizarre, half-formed figure of rock and nascent limbs, struggles to maintain its balance. Its newly formed arms flail wildly, attempting to find purchase on the smooth, unforgiving rock of the cliff's edge.

The center of gravity shifts dramatically. With a slow, agonizing tilt, the boulder-man begins to tip forward. The camera, still zoomed in tightly, captures the desperate, futile movements of its stony limbs as it loses its battle against gravity.

With a final, decisive lurch, the boulder-man topples over the edge of the cliff. The sound of its descent is a distant, echoing rumble, fading into the vast emptiness below. The camera lingers for a moment on the empty space where the boulder-man once stood, emphasizing the sudden, absurd absence.

[inner kite]

how do I upload the video? Do I have to download it from VideoFx first?

[kindred wind]

yes

[inner kite]

thank you

[kindred wind]

np

[inner kite]

Let me know what you all think

[kindred wind]

looks good.

[molten parcel]

I don’t know how to use this

[inner kite]

@kindred wind thank you

[molten parcel]

What do I need to do generate videos

[inner kite]

@gurung start with an image, and give it a prompt to make movements. Be as precise as you can, but start small and build your confidence. You got this!

[inner kite]

Starry Lock

[floral drift]

WAY too much description. Instead of "The scene opens with a wide shot" Say: "wide shot" etc. Omit unnecessary words.

[steady cradle]

남산타워 전망대에서 야간에 서울 시내 야경을 보여주는 동영상을 만들어줘

Make a video showing the night view of Seoul from the Namsan Tower observatory.

#flow-tips-n-tricks・✏️ Make a video showing the night view of Seoul from the Namsan Tower observatory.

[calm kiln]

@steady cradle

[oak pine]

I'm going nuts with it failing around 20%

[ashen juniper]

What was a prompt?

[slow hazel]

Tell me how to make a video based on text?

[long hornet]

#flow-tips-n-tricks・✏️ move to street

[kindred wind]

got it to work.

A large, round boulder rests precariously on the very edge of a dramatic cliff overlooking a vast landscape. Suddenly, the boulder begins to transform, rocky protrusions extending and shaping into legs and arms. As the newly formed humanoid rock figure gains its footing, unseen force or entities extend stony limbs and trip it. The humanoid rock stumbles, loses balance, and tumbles over the cliff edge, breaking apart as it falls into the abyss below. Cinematic shot, epic scale, dramatic lighting.

[reef sorrel]

?????

[thorn arrow]

how to generate video can any one help me out.

[reef sorrel]

You can use https://aistudio.google.com/generate-video

Also I’m generating your video right now not sure if it can pass or not since I saw the word blood in it

[thorn arrow]

thank you. i'll try once.

[thorn arrow]

Thank you @reef sorrel .I feel my prompt might not have been clear, but thank you for creating it.

[burnt briar]

Hi

[kindred wind]

I thought I would try a new style of prompts and so for I like the results.

The prompt
Concept: High-octane racing on water, featuring cars doubling as speedboats and pulling waterskiers.

Visuals:

Shot: Begin the 8 seconds with a dynamic Wide Shot (WS) or Long Shot (LS) showing several cars racing side-by-side across the surface of a large lake. Behind each car, a person is waterskiing, cutting through the wake. The wide shot is maintained for most of the 8 seconds to convey the scale of the race and the unusual setting, potentially ending with a slightly tighter Medium Long Shot (MLS) on a lead car and skier pair.

Camera Movement: Employ an aggressive and fast Dolly Shot or Tracking Shot, moving alongside or just ahead of the racing cars and skiers. The camera should be low to the water to emphasize the speed and the spray being kicked up. The movement should be fluid but convey the high speed.

Action: Cars, behaving like powerful boats, speed across the water, their tires potentially skimming the surface or modified in some way for aquatic motion. Waterskiers behind them are in full control, leaning into turns and creating dramatic splashes and wakes. Focus on the speed, the spray, and the parallel action.

Lighting: Utilize bright, natural High-Key Lighting that reflects intensely off the water. This creates a vibrant, energetic feel suitable for a daytime water sport event, making the spray and action highly visible.

Lens: Filmed with a Wide-Angle Lens (e.g., 24mm or 35mm Prime). This lens choice is essential for capturing the vastness of the lake and keeping multiple racing cars and their attached skiers in frame simultaneously. It also helps to exaggerate the sense of speed and perspective as the subjects move rapidly through the water. Using a Prime Lens ensures sharp focus on the chaotic, splashing action.

The videos

[kindred wind]

one more experiment to share.
the prompt.
Concept: A majestic (and slightly surreal) aerial view of hundreds of rubber ducks flying in formation, wings flapping, towards a summer pool.

Visuals:

Shot: The 8 seconds begins with an Extreme Long Shot (ELS) or Long Shot (LS) taken from an elevated position (as if looking down slightly from above or from a hillside). The frame is dominated by the sight of hundreds of bright yellow rubber ducks in the air, filling a significant portion of the frame, all steadily flying in the same direction. In the distance or mid-ground, a shimmering, blue summer pool is clearly visible as their destination. The shot stays wide throughout to emphasize the scale of the flying formation.

Camera Movement: Start with a Static Shot to immediately establish the vastness of the flying group and their flight path towards the pool. As the 8 seconds elapse, execute a slow, sweeping Crane Shot or a high-angle Dolly Shot moving smoothly alongside or slightly ahead of the leading edge of the airborne duck mass. This movement should be steady and match the rhythm of their flight, conveying their purposeful advance.

Action: Hundreds of rubber ducks are in active flight. Their small, rubber wings are visibly and rhythmically flapping in a coordinated, steady motion as they propel themselves through the air. Focus on the collective, synchronized (or near-synchronized) flapping action creating a visually striking image of a unified, collective purpose.

Lighting: Use bright, natural High-Key Lighting with strong, directional sunlight. This will not only make the vibrant yellow ducks and the blue pool pop but also catch the light on the texture of their rubber wings as they flap, making the action more visible even in a wide shot.

Lens: Filmed with a Wide-Angle Lens (e.g., 20mm or 24mm Prime). This lens choice is crucial for capturing the immense number of ducks ("hundreds")

flying together in the vastness of the air and showing the environment they are moving through, including their pool destination. The wide perspective is key to conveying the scale of this aerial migration. A Prime Lens ensures maximum sharpness to allow the viewer to perceive the flapping motion even from a distance.

[formal marsh]

Out of curiousity, I prompted with just your concept part.

[formal marsh]

Also prompted with just your concept part, and it came up with some interesting vehicles!

[reef sorrel]

I need to try this omg

[reef sorrel]

I’ll share this to my dad and say it’s real

[kindred wind]

I had Gemini , Deep research do a analysis on Film techniques and lighting. I then took that analysis and gave it to Gemini 2.5 pro.
Then I told 2.5 pro saying,

Every time i say "Help me make a prompt for a 8 second video of" use the attachment as a guide. with the template being in the format of.

Concept:

Visuals:

Shot:

Camera Movement:

Action:

Lighting:

Lens:

here is the analysis, always scan before you open something you down load.
https://docs.google.com/document/d/1kaXVichQYvCuE6LZTCx3KYWsH_CSzEHY1KAchj43fKU/edit?usp=sharing

[reef sorrel]

This is actually fascinating. Wild how the official Veo prompt guidelines exist, but they barely worked for me. I’m still kinda baffled.

[reef sorrel]

I will definitely read this !

[kindred wind]

I updated the rules for Gemini for videos. seems to work a lot better now.

copy pasted
Certainly. Here are the rules we currently have in place for when you say "let's make an 8-second video":

Stay within reason of artistic needs.
Template Format:
Concept:
Visuals:
Shot:
Camera Movement:
Action:
Lighting:
Lens:
Give as much detail as possible.
No inherent sound.
Generate an image for the prompt description.
Always use " Film Techniques and Lenses in Movie Production."

[wide storm]

A big red cat wearing a chef's suit with big muscles on a beach with a lot of people. The picture must be real.

[pastel carbon]

any prompt to properly make this? it seems like veo2 outright cant do arrow-firing and the subject usually just keeps holding the arrow, unsure if thats a model issue or prompt issue.
prompt: "movie shot featuring a dynamic front-view of a beautiful elven archer woman with blue skin as she shots an arrow, the camera follows the arrow as it is being shot and the drow archer is being seen from further and further away."

[kindred wind]

try this.. sometimes I have to refine the prompt after it renders a video.

Concept: An 8-second video featuring a dynamic front-view of a beautiful elven archer woman with blue skin as she shoots an arrow. The camera follows the arrow, emphasizing its flight and showing the elven woman becoming smaller in the frame.

Visuals: A beautiful elven woman with blue skin, dressed as an archer, dynamically posed. The arrow should be prominently featured.

Shot: Start with a dynamic front-view of the elven woman as she releases the arrow. The camera then closely follows the arrow's trajectory, moving further away from the archer. The arrow's flight path should be a key visual element.

Camera Movement: Begin with a steady shot on the archer, then transition to a tracking shot closely following the arrow. As the arrow travels, the camera pulls back rapidly, showing the elven woman becoming significantly smaller in the frame, emphasizing the arrow's speed and distance.

Action: The elven woman draws back her bow and shoots the arrow. The arrow flies through the air with visible speed and impact.

Lighting: Dynamic lighting, highlighting the elven woman and the arrow's path.

Lens: A wide-angle lens for the initial shot, transitioning to a standard lens for the tracking shot, and potentially a telephoto lens to emphasize the distance and the shrinking elven woman as the camera pulls back.

it going to take some redefining the prompt above, but i found this templat to work for me.

try having the camara sit on the end of the arrow head looking back to the elven woman

[kindred wind]

LMAO i did one last try.

[pastel carbon]

it still didnt fire the arrow though. same issue with mine.

[pastel carbon]

thats better

[formal marsh]

Prompt: A beautiful elven archer woman with blue skin releases an arrow from her bow.

[pastel carbon]

finally hahaha

(and then it shows stuck in her bow)

[formal marsh]

You're welcome.

[pastel carbon]

was it a prompting issue or both?

[formal marsh]

My only prompt was what I wrote: "Prompt: A beautiful elven archer woman with blue skin releases an arrow from her bow." Galaxieon gave you the arrow tracking part already.

[pastel carbon]

I added both, with a tiny issue.

she always pulls back XD

hahaha

XDD

[kindred wind]

when I convert a Veo2 to 3D, I read 1280x720 24fps 192 frames. I haven't done much upscaling not sure if the 1280 and 24 fps matter.

it could be stretching the ratio a little making making it happen

[cerulean wyvern]

💥💥💥

[sour anchor]

how you guys using veo ai i also want to use

[reef sorrel]

Sure! Go here : https://aistudio.google.com/generate-video

[formal marsh]

A dynamic front-view shot captures a beautiful elven archer with luminous blue skin, her athletic form accentuated by traditional archer's garb, positioned according to the rule of thirds. Using a telephoto lens at eye level, the camera begins with a close-up of her determined expression as she draws back her ornate bow with fluid precision. Natural lighting bathes the scene, creating subtle highlights on her skin and equipment. As she releases the arrow, the camera smoothly pans right, transitioning into a precise tracking shot that follows the projectile's swift trajectory through the air. The arrow maintains sharp focus in the frame while the background shows natural motion blur. As the camera continues its fluid movement, the elven archer gradually diminishes in size within the composition, emphasizing the arrow's impressive distance and speed while maintaining a realistic visual style throughout the sequence.

[formal marsh]

A dynamic front-view shot captures a beautiful elven archer with luminous blue skin, her athletic form accentuated by traditional archer's garb, positioned according to the rule of thirds. Using a telephoto lens at eye level, the camera begins with a close-up of her determined expression as she draws back her ornate bow with fluid precision. Natural lighting bathes the scene, creating subtle highlights on her skin and equipment. As she releases the arrow, the camera smoothly pans right to track the arrow centered in the frame.

I cut the prompt down to try to get the panning + tracking arrow in flight, but I think it's challenging to put all of it in an 8 second video.

[formal marsh]

This is another one from the longer prompt version:

[pastel carbon]

Finally, you can see it hesitates to fire. Not very professional of an archer maybe 🤔

[pastel carbon]

Veo2 really doesn't want to let go of that arrow

[formal marsh]

Professional archers always set up the shot.

[pastel carbon]

Zen archery

[formal marsh]

Add your feedback to https://discord.com/channels/1172568727942860810/1238621511401734254

[pastel carbon]

While haired perhaps. Hmm

[junior pike]

all those above videos are Veo2 right?

[molten parcel]

yes

[pastel carbon]

@molten parcel I wasn't able to dm for some reason unless that was on purpose. Discord rejected it.

[molten parcel]

sorry, now u can my dms were off

[formal marsh]

Otter and Snail at archery practice

[kindred wind]

Concept: An 8-second video of a blue Elven woman releasing an arrow from a shiny gold bow.

Visuals: A blue-skinned Elven woman releasing an arrow from a shiny gold bow. The scene should be dynamic.

Shot: Start with a close-up (1.2.4) of the Elven woman's face as she focuses and releases the arrow. Transition to tracking shots (1.3.3) following the arrow's flight. Incorporate a medium shot (1.2.5) showing the Elven woman's posture and the arrow's release.

Camera Movement: Use smooth and precise tracking shots (1.3.3) to follow the arrow's flight. Employ panning shots (1.3.1) to showcase the environment and the arrow's path.

Action: The blue Elven woman releases an arrow from a shiny gold bow. The action should be precise and controlled.

Lighting: Natural lighting to highlight the Elven woman, the shiny gold bow, and the arrow.

Lens: Use a standard lens (2.4) to capture the Elven woman's focus and the arrow's release. A telephoto lens (2.2) can be used for the tracking shots to emphasize the arrow's flight.

best i could do today , couple of failures.

[kindred wind]

Concept: An 8-second video of a blue Elven woman standing and shooting an arrow from a shiny gold bow.

Visuals: A blue-skinned Elven woman standing and shooting an arrow from a shiny gold bow. The scene should focus on her stance and the arrow's flight, without close-ups.

Shot: Start with a medium shot (1.2.5) showing the Elven woman standing and releasing the arrow. Transition to a tracking shot (1.3.3) following the arrow as it flies forward.

Camera Movement: Use smooth and precise tracking shots (1.3.3) to follow the arrow's flight. Employ panning shots (1.3.1) to showcase the environment.

Action: The blue Elven woman stands and releases an arrow from a shiny gold bow. The action should be precise and controlled, emphasizing the release and the arrow's trajectory.

Lighting: Natural lighting to highlight the Elven woman, the shiny gold bow, and the arrow.

Lens: Use a standard lens (2.4) to capture the Elven woman's stance and the arrow's release. A telephoto lens (2.2) can be used for the tracking shots to emphasize the arrow's flight.

[formal marsh]

I reran the same long prompt I posted earlier. 1/2 gens worked for the release. But, it still doesn't actually pan to track the arrow in flight.

[kindred wind]

I'll try again later but these 2 had me laughing

looks like shes trying to fly the arrow like a kit.

[stark sphinx]

Cool

[silver lagoon]

Lmaoooo

[kindred wind]

wow , i was having to much fun, I hit my limit of videos for videoFX this month. I thought it was never going to end. in less then a week my whisk resets so thats cool.

Food for thought add a subscription for poeple that want to pay for unlimited videos. I would love to make a movie with Google A.I. but the current pricing structure from what i understand is limited on the amount of videos being able to be made. Also when I work on a video i sometime go through 3-10 videos of fixing prompts or being to picky before i find the one video i like. That can also be limiting to a paying customer on veo 2 with a chance of more failed videos then successful is even more limiting. Don't get me wrong just seeing a possible issue/fix.

Thank you for the opportunity to be a trusted tester. it is a lot of fun making video's and the tech is amazing and will change the indie film industry for the better if done right with green screens.

[kindred wind]

getting better but still not getting it

[floral mica]

Un león animado jugando en la selva

[deft hound]

Does anyone know or have any experience with "negative prompts" for VideoFX?

Haven't tried yet, curious if you have and if so, the result.

[reef sorrel]

VideoFx doesn’t have negative prompts input (yet)

Only AI Studio/Vertex AI Veo and some APIs (like Freepik) support negative prompts input.

[simple remnant]

Section 4: The Strike – Execution (2:40–4:48, 16 segments, 128 seconds)
Purpose: Deliver an epic, action-packed strike sequence.

Visuals: Dawn sky, Indian jet takes off, camera tracks upward.

Audio: VO: “May 7, 2025…” Sound: Jet roar. Music: Heroic brass.

Purpose: Launch the strike.

Concept: An 8-second video of a blue Elven woman standing and shooting an arrow from a shiny gold bow.

Visuals: A blue-skinned Elven woman standing and shooting an arrow from a shiny gold bow. The scene should focus on her stance and the arrow's flight, without close-ups.

Shot: Start with a medium shot (1.2.5) showing the Elven woman standing and releasing the arrow. Transition to a tracking shot (1.3.3) following the arrow as it flies forward.

Camera Movement: Use smooth and precise tracking shots (1.3.3) to follow the arrow's flight. Employ panning shots (1.3.1) to showcase the environment.

Action: The blue Elven woman stands and releases an arrow from a shiny gold bow. The action should be precise and controlled, emphasizing the release and the arrow's trajectory.

Lighting: Natural lighting to highlight the Elven woman, the shiny gold bow, and the arrow.

Lens: Use a standard lens (2.4) to capture the Elven woman's stance and the arrow's release. A telephoto lens (2.2) can be used for the tracking shots to emphasize the arrow's flight.

[pallid delta]

New music Friday, which I like to use an unpredictable prompt. I had fun squeezing 7 costume changes into this YouTube short:

https://youtube.com/shorts/L-7uEdD7hJ0?feature=share

[clear valley]

people are using ai to make the prompts

[pastel crag]

Anima la imagen

[vagrant dune]

#imagefx・📷 creat girl

#start-here・📍 creat girl in new york city

[stuck epoch]

Hi

[chilly fox]

you can do negative prompts with veo on freepik?

[molten parcel]

Can someone try “An actual report from a Phillippine news channel, a woman news journalists interviewing a man who have witnessed a car crashed and describes what happend in pure modern tagalog language”

Just an experiment if the speech also does in different languages

[nocturne chasm]

Can someone please do a rick and morty inter dimensional cable type skit (not actual rick and morty style, just normal video) of like crazy weird show clips

[viral seal]

Why do all my prompts fail? Even something like "A flower blooming" is not generating, says its not possible try again later

[grand spade]

Yeah just redo. It'll restore credits if failed or refresh the page first

[gritty obsidian]

The evolution of condoms over the past century

[reef sorrel]

https://tenor.com/view/wtf-gif-18370552922777914358

[gritty obsidian]

hahaha

[lethal gorge]

#flow-tips-n-tricks・✏️ baby boy laughing

[tired coyote]

Yeah, similar issues with all AI video gen. A monster walking around is easy. Actually shooting an arrow (I see you got that to work actually!), or serving a tennis ball.... practically impossible. Though Veo3 is slightly better serving tennis balls. Sora can't even get close.

[kindred wind]

This is only the beginning, I can only imagine the possibilities a year from now.

[balmy moon]

#flow-tips-n-tricks・✏️ A megalodon breaching out of the water into the air and then landing on top of a ship, breaking the ship in half, debris and water everywhere, destruction and chaos

[thick haven]

"Obey to Optimize" A close-up of a man's face, lit by the pale blue glow of a smart fridge screen. His expression is blank, almost reverent. As the fridge softly hums, the camera slowly zooms in—handheld, slightly shaky, capturing a gritty realism like it’s shot on a late-90s Sony DVCAM, with visible grain and muted contrast. Behind him, the kitchen is sterile, almost surgical—white LED strips, smooth metal, no personal touches. A robotic voice whispers from the fridge: “Calories exceeded. Would you like to repent?” The man nods solemnly. Visual Style: Low-resolution texture with film grain overlay. Washed-out color grade, blues and greys dominate. Framing is claustrophobic, emphasizing control. Inspired by Tool and NIN aesthetics—industrial, uncomfortable, sterile dystopia. Mood: Haunting, absurd, prophetic. Feels like Orwell through a VHS lens.

[wild halo]

Hi all! qq - none of the videos I generated have sound, even after I purchased the membership plan, or use exactly the same prompt as ones on the official website. Anyone can help please? 🌹

[untold vale]

Are you using the VEO 3 toggle option and not VEO2. I noticed it sometimes defaults me back to VEO2.

[reef sorrel]

it'd work better with VEO 3 tbh 😅

[reef sorrel]

I refined your prompt just for this video A slow, slightly shaky handheld zoom-in, capturing a gritty realism with visible grain and muted contrast like a late-90s Sony DVCAM. The framing is claustrophobic, a close-up on a man's face illuminated by the pale blue glow of a smart fridge screen in a sterile, surgical kitchen with white LED strips and smooth metal. The visual style is low-resolution with film grain, washed-out blues and greys, evoking industrial dystopia. The man's expression is blank, almost reverent, as the fridge softly hums. A robotic voice whispers from the fridge, “Calories exceeded. Would you like to repent?” The man nods solemnly.

[reef coral]

#flow-general・🌊 humen looking at the phone

[stark sphinx]

veo 2 ingredients are working?

[calm zodiac]

Ho

#flow-tips-n-tricks・✏️ animate these image

[rustic smelt]

#Futuristic looped animation (3-5 seconds) illustrating thermoelectricity. A sleek, high-tech environment shows energy being generated from a temperature difference — glowing heat on one side, cold on the other, with animated arrows or lines showing electrical current flow. Stylized, sci-fi graphics, glowing circuits, blue and orange tones, smooth motion, cinematic lighting.

[drowsy pendant]

#flow-tips-n-tricks・✏️ slow drone flying over Roman ruins

[wispy harbor]

Create a cinematic documentary about Abdulrahman Hamdany — an Iraqi architect living in Dubai. Begin with nostalgic visuals of Baghdad: childhood streets, traditional architecture, the call to prayer echoing in the distance. Transition into the fast-paced, high-rise world of Dubai, capturing his evolution from dreamer to designer.

Include scenes of him sketching blueprints, walking construction sites, and discussing ideas with builders. Cut to quiet moments: journaling at night, phone calls with family, deep reflections while looking out over city skylines. Show the contrast between cultural roots and modern ambition.

Use natural lighting, slow-motion shots, emotional piano or oud-infused soundtrack, and voiceover narration telling his story — his values, challenges, and dreams. Overlay text with quotes like: “I design not just buildings, but possibilities,” and “My journey is a bridge between where I’m from and what I’m building.”

End with a sunrise over Dubai and a close-up of Abdulrahman smiling — hopeful, visionary, grounded.

[stable kiln]

"Animate a Studio Ghibli-inspired scene where a seven-year-old girl named Lia with dark hair, wearing a colorful dress, is running joyfully through a vibrant grass field. Beside her is a fluffy orange cat named Mochi, bouncing energetically in the wind, and a small yellow chick named Pip, flapping its wings excitedly. The grass sways in the breeze, and flowers sway as they run, with soft clouds drifting overhead. The wind moves Lia's hair and dress, and the cat and chick leap in sync with her movements. The sun glows warmly, casting soft shadows as they race through the meadow.

[graceful quartz]

Create a random minecraft video

[wispy harbor]

Change characters

[daring trail]

genrate video of miltirry video game

[reef sorrel]

Sorry can’t do that. You can do it yourself by purchasing Google ai pro plan

[fathom pagoda]

Create a Eurovision song movie done by a singer with red dress followed by clowned as dancers. Song is a classic Eurovision
Style. Extra real

[frosty oracle]

create a amazing space war video

[magic briar]

A close-up cinematic shot of a hyena walking cautiously through a sunlit savannah. The dry grass sways in the wind, and dust lifts as the hyena steps forward. Its eyes are sharp, ears twitching, and it briefly snarls before continuing. Shot in ultra-realistic 4K with dramatic lighting, shallow depth of field, and slow motion as the camera follows its movements

#flow-tips-n-tricks・✏️ A close-up cinematic shot of a hyena walking cautiously through a sunlit savannah. The dry grass sways in the wind, and dust lifts as the hyena steps forward. Its eyes are sharp, ears twitching, and it briefly snarls before continuing. Shot in ultra-realistic 4K with dramatic lighting, shallow depth of field, and slow motion as the camera follows its movements

[past quiver]

anyone not getting audio with their prompts?

[celest bear]

create a amazing space war video

[fossil cedar]

Is there a way for people in different geographies to try Veo 3 out? Also, when are they expected to launch in different geographies

[magic briar]

A cinematic shot in a peaceful savannah setting, showing a playful 3-year-old boy with black hair wearing dark blue clothes and his father in a black hoodie with a short beard, playing together and laughing in the grass. A curious hyena watches them from a short distance, then blinks one eye slowly while standing still. The camera focuses briefly on the hyena’s expressive face before panning to the smiling father and son. Shot in soft golden-hour lighting, with natural colors, shallow depth of field, and a warm, lighthearted atmosphere.

#flow-tips-n-tricks・✏️ A cinematic shot in a peaceful savannah setting, showing a playful 3-year-old boy with black hair wearing dark blue clothes and his father in a black hoodie with a short beard, playing together and laughing in the grass. A curious hyena watches them from a short distance, then blinks one eye slowly while standing still. The camera focuses briefly on the hyena’s expressive face before panning to the smiling father and son. Shot in soft golden-hour lighting, with natural colors, shallow depth of field, and a warm, lighthearted atmosphere.

[late ore]

#flow-tips-n-tricks・✏️ Scene 1: Han Li, dressed in green robes, holds a sect token at the mountain gate. Mist swirls around peaks as disciples fly on swords, with the sect’s protective formation faintly visible.Scene 2: Han Li follows elders into a secret garden glowing with spiritual herbs. Energy condenses into mist, and ancient broken formations float mid-air, creating an otherworldly aura.Scene 3: Han Li carves intricate runes into his chamber floor. A cauldron for refining Nine-Curve Spirit Ginseng sits centerstage, while spirit stones embedded in walls emit dim, oppressive light.

[gray tapir]

slow drone flying over the Ancient Mohenjo-Daro back in their time

[wild lynx]

Bebé presidente saludando ala multitud

[young yoke]

#flow-tips-n-tricks・✏️ High school students, a man looks at his loved one and another man happy at a prom; he is reminiscing about their moments. Suddenly a tear came from his eyes, and left without saying something, like saying "I wish it was me".

[cedar locust]

Person on the Street Interview talking about AI Video. The person is (excited, nervous, opposed) to the technology

Person on the Street Interview talking about AI Video. The person is (excited, nervous, opposed) to the technology

[reef umbra]

Crockett and Tubbs driving in Miami

[wispy jackal]

#flow-tips-n-tricks・✏️ Create a FB ad format of video with female in her 40s trying to do a workout, but feeling tired and saying "My energy recently is very low"

[jaunty pulsar]

create a scene with a kraken playing poker

[stark sphinx]

I mean the new reference feature, it's working? You have to create the images or you can upload?

[novel karma]

#flow-tips-n-tricks・✏️ A medium shot of a focused programmer in a hoodie, mid-20s to early 30s, seated in a dimly lit, clandestine tech workspace. His fingers rapidly type on a mechanical keyboard as complex Python, C++, or Unix programming code scrolls across a large, glowing monitor. Suddenly, a stern, authoritative voice from off-screen says, "Don't ask questions, just fulfill your end of the agreement." The programmer pauses typing for a split second, a flicker of resignation on his face, then calmly resumes typing as he responds, "Okay." The scene has a neo-noir, high contrast, cyberpunk aesthetic, dominated by blues, greens, and dark shadows from the monitor's light. Subtle lens flares from the screen. The camera is static, focused on the programmer, with the monitor's glow as the primary light source. The mood is tense, mysterious, and highly focused. Sound includes subtle keyboard clicks, faint server hum, and clear, distinct voices.

[reef umbra]

⁠flow-prompts┇✏ medium tracking shot of Crockett and Tubbs from Miami vice driving at night in a white ferrari listening to music

how does this work

#flow-tips-n-tricks・✏️ medium tracking shot of Crockett and Tubbs from Miami Vice driving at night in a white ferrari talking to each other

[molten parcel]

#flow-tips-n-tricks・✏️ Hey everyone! 🌊
What’s one thing you’re working on right now to stay in your flow zone?
Whether it’s studying, drawing, coding, journaling, or just vibing with music — drop it below and let’s keep each other motivated.

[primal ravine]

#flow-tips-n-tricks・✏️ datura trip mind blowing

[uneven mist]

#flow-tips-n-tricks・✏️ A unemployed graphic disigner living en belgrano drinking coffee and thinking about life. He is small and with a bear

[signal roost]

#flow-tips-n-tricks・✏️ one young girl go to shopping dress with boy friend and come with car sit night shot

[dire stone]

#flow-tips-n-tricks・✏️ A mysterious captain stands still in a pitch-dark room, framed in a dramatic vertical 9:16 composition. His silhouette is sharply defined — dressed entirely in black with a black captain’s hat featuring a subtle anchor emblem. The camera slowly zooms in vertically, from chest level up toward his head. His face is hidden in deep shadow, but his eyes glow faintly, casting an eerie light. A thin trail of smoke rises from a pipe in his mouth, barely catching the light. The room is minimal, with near-total darkness except for soft rim lighting that outlines the edges of his body. The mood is dark, powerful, and cinematic. No music, just low ambient rumble or silence.

[wild lynx]

#flow-tips-n-tricks・✏️ corredor llegando ala meta con 5 atrás

[urban tapir]

#flow-tips-n-tricks・✏️ The child's tearful scream of "Aama!" as his house burns around him paints a vivid picture of a devastating loss. It's a powerful and heartbreaking opening.
To build on this, consider the following elements for your script:
Expanding Scene
Here are a few options to continue from this impactful moment:
Option 1: Focus on immediate aftermath and helplessness.
[SCENE 1 - EXT. VILLAGE HOUSE - SUNSET]
(Soft music playing. A small house is burning. A child (Boy, 8) stands crying outside. Slow-motion effect.)
Child (tearfully screaming):
"आमा!" (“Aama!”)
(Fire crackles. Music fades to silence.)
BOY
(Eyes wide with terror, he stumbles back a few steps, his small hands clenching into fists. The heat of the fire is intense, licking at his face. He looks around frantically, as if searching for someone, anyone, to help. His cries become choked sobs.)
SOUND
Distant shouts, faint at first, slowly growing louder.
Option 2: Introduce a new character offering comfort.
[SCENE 1 - EXT. VILLAGE HOUSE - SUNSET]
(Soft music playing. A small house is burning. A child (Boy, 8) stands crying outside. Slow-motion effect.)
Child (tearfully screaming):
"आमा!" (“Aama!”)
(Fire crackles. Music fades to silence.)
OLD WOMAN (O.S.)
(A weathered hand gently touches the boy's shoulder. An OLD WOMAN, her face lined with worry and compassion, kneels beside him. Her voice is soft but steady.)
OLD WOMAN
"नानी, यहाँ आऊ." ("My child, come here.")
(The boy flinches, then slowly turns, his tear-streaked face looking up at her with a mix of fear and confusion. She pulls him gently into an embrace, shielding his eyes from the flames.)
Option 3: Hint at the cause or wider impact of the fire.
[SCENE 1 - EXT. VILLAGE HOUSE - SUNSET]
(Soft music playing. A small house is burning. A child (Boy, 8) stands crying outside. Slow-motion effect.)
Child (tearfully screaming):
"आमा!" (“Aama!”)(Fire crackles. Music fades to silence.)

[stark sphinx]

oh i get it, thanks

[dire stone]

I would be so happy if anyone could try my prompt, I got pro version but im in Sweden so I can’t use in unfortunately :(

[reef sorrel]

on it

[dire stone]

Legend 🫶🏼

[reef sorrel]

do u actually want this in 9:16 ?

[dire stone]

Yes it would be sick but if you can’t it’s fine with 16:9

I wanna use it on TikTok 😂

[reef sorrel]

ah

i think u can also do it as for 9:16 ratio, in https://aistudio.google.com/generate-video

somehow i cant use aistudio veo atm

[dire stone]

is this veo 2?

i can use veo 2 myself but not veo3 :( thats what im looking for unfortunately

[reef sorrel]

ah oops

[abstract panther]

hi

[dire stone]

np dude, thank you for taking the time

[dire stone]

yo!

[celest moss]

How much have u billed on veo 3?

[reef sorrel]

0

lol

[celest moss]

How?

[reef sorrel]

wdym how

[dusty hound]

A portrait of a super futuristic car, positioned slightly off-center to the left of the image. The car's body and structure are composed of numerous small, multicolored squares, each containing a different digit. The colors are vibrant and varied, including blues, purples, pinks, and oranges. The car's design is sleek and otherworldly, with smooth curves and unconventional angles. The front of the car, which appears to be the focal point, gives an impression of contemplation, with its headlights slightly dimmed. The vehicle seems to hover above the ground, defying traditional automotive design. The background is a mix of muted colors that blend into a soft, almost painterly effect. The composition is abstract, with the focus on the car and its fragmented numerical representation. The lighting is diffused and warm, creating a sense of depth and atmosphere. The image has a mixed media style, combining elements of digital painting, mosaic art, and futuristic concept art. The use of numbers within the composition creates an intriguing contrast between the technological subject and the abstract data. The perspective is directly facing the subject, captured in an ultra high-detailed, close-up camera shot. The overall mood is contemplative and thought-provoking, inviting the viewer to ponder the future of transportation and technology.

[celest moss]

Have u not used veo 3?

[reef sorrel]

have i not used veo 3..?

are u okay

[celest moss]

Idk

[willow swallow]

#flow-tips-n-tricks・✏️ The child's tearful scream of "Aama!" as his house burns around him paints a vivid picture of a devastating loss. It's a powerful and heartbreaking opening.
To build on this, consider the following elements for your script:
Expanding Scene
Here are a few options to continue from this impactful moment:
Option 1: Focus on immediate aftermath and helplessness.
[SCENE 1 - EXT. VILLAGE HOUSE - SUNSET]
(Soft music playing. A small house is burning. A child (Boy, 8) stands crying outside. Slow-motion effect.)
Child (tearfully screaming):
"आमा!" (“Aama!”)
(Fire crackles. Music fades to silence.)
BOY
(Eyes wide with terror, he stumbles back a few steps, his small hands clenching into fists. The heat of the fire is intense, licking at his face. He looks around frantically, as if searching for someone, anyone, to help. His cries become choked sobs.)
SOUND
Distant shouts, faint at first, slowly growing louder.
Option 2: Introduce a new character offering comfort.
[SCENE 1 - EXT. VILLAGE HOUSE - SUNSET]
(Soft music playing. A small house is burning. A child (Boy, 8) stands crying outside. Slow-motion effect.)
Child (tearfully screaming):
"आमा!" (“Aama!”)
(Fire crackles. Music fades to silence.)
OLD WOMAN (O.S.)
(A weathered hand gently touches the boy's shoulder. An OLD WOMAN, her face lined with worry and compassion, kneels beside him. Her voice is soft but steady.)
OLD WOMAN
"नानी, यहाँ आऊ." ("My child, come here.")
(The boy flinches, then slowly turns, his tear-streaked face looking up at her with a mix of fear and confusion. She pulls him gently into an embrace, shielding his eyes from the flames.)
Option 3: Hint at the cause or wider impact of the fire.
[SCENE 1 - EXT. VILLAGE HOUSE - SUNSET]
(Soft music playing. A small house is burning. A child (Boy, 8) stands crying outside. Slow-motion effect.)
Child (tearfully screaming):
"आमा!" (“Aama!”)(Fire crackles. Music fades to silence.)

[swift robin]

U need to have this plan

[celest moss]

Only 12k?

[dusty hound]

can i use in Poland?

[swift robin]

No, only us for now

[celest moss]

Are u in us?

[swift robin]

No :(
im from brazil, but i tried to use VPN

I failed because I need to have a us bank account

[celest moss]

Oh

I’m in us but I’m not paying that much

[celest moss]

Maybe u could buy a visa gift card

[surreal cedar]

شاب سعودي داخل السيارة متضايق من حرارة الشمس
مع اصوات سيارات في الشارع وأصوات تبين ان الشخص متضايق

#flow-general・🌊 A Saudi young man inside the car is annoyed by the heat of the sun, with sounds of cars in the street and sounds indicating that the person is annoyed. ...

[gloomy shuttle]

#flow-tips-n-tricks・✏️ create a medium shot of a relaxed man, enjoying beer and fishing

[void cypress]

this but make it jello-y and cute

[vale jungle]

#flow-tips-n-tricks・✏️ an african wicth doctor throws bones on the ground and then is surprised by the future he just percieved

[unreal urchin]

i had a few issues now with characters in a scene speaking gibberish before speaking the typed dialogue. anyone else experience this issue?

[stray vigil]

What ai is this

[opaque blaze]

#flow-tips-n-tricks・✏️ Beyonce walks towards the camera, she is wearing a black shirt with black pants, her hair is long, wavy and blonde with dark roots, she looks at the camera and says "I made this on VEO3, I'm a genius", the background is grey

[soft sentinel]

Hey guys! Can somebody generate out a promt for me please? I am really exited about Veo 3 but I don't have acsess to it yet. Here is more details:

Promt:
A horizontally placed glass bottle on a table. Inside, a raging ocean and thunderclouds, a medieval British sailing ship sails. The camera slowly moves from a wide shot of the ship to a medium shot of sailors on the deck. Cinematic dramatic lighting. Shot on an Arri Alexa 35 with an anamorphic lens. Realistic graphics. Aspect ratio 2.35:1

And you also may use this image generated by ChatGPT as a reference.

Thanks in advance!

[clever iris]

Make video

[opal quartz]

#flow-tips-n-tricks・✏️ A dramatic sky filled with dark, swirling clouds. Slowly, the clouds begin to part, revealing soft, golden rays of sunlight breaking through. From the heart of the light, a glowing human-like silhouette begins to descend gracefully toward the earth. The figure is ethereal, its features undefined, almost like a celestial spirit. The light surrounding it pulses gently, illuminating the clouds and casting long, divine shadows. The atmosphere is filled with awe and quiet wonder, as if something otherworldly is entering the world.

[delicate sparrow]

charming elderly adorable older woman, pulling by the shirt her peer husband towards her for a kiss on the cheek and she's saying "come here you!"

[formal marsh]

Veo2 T2V

[nimble gyro]

David Baszucki eating bug pizza

#flow-tips-n-tricks・✏️ David Baszucki eating bug pizza

[clear dagger]

How accurate is the result, given options it really seems it could go haywire

[clear dagger]

BRrrooooo LOL, the elf messes up her shot and hurts her hand. amazing

[clear dagger]

Great progress, The sag on her arm flapping as it tensions is pretty funny for an elf still. Love the quality though its looking good

[clear dagger]

Action: The blue Elven woman stands and releases an arrow from a shiny gold bow. The action should be Smooth precise and controlled release of the arrow. The arrow shoots straight

I would also consider writing this differently. I think the concept should be more encompassing, and you should write it more action by action with that specific scene described in the style of shot you want with camera movement.

So something like

Concept: An 8-second video of a blue Elven woman standing and shooting an arrow from a shiny gold bow.
The first scenes camera will not move, but focus on her stance and her smooth movements. then transition to the arrows flight.

Concept step-by-step:
Start with The blue Elven woman stands and releases an arrow from a shiny gold bow, medium shot (1.2.5). The action should be Smooth precise and controlled release of the arrow. The arrow shoots straight Transition to a tracking shot (1.3.3) that uses smooth and precise tracking, following the arrow as it glides straight and true through the air. Employ panning shots (1.3.1) to showcase the mythical environment.

[humble scaffold]

I gotchu

[clear dagger]

Welp, sorry @kindred wind 🤣

Thanks SD, clearly I also need time to mess around to refine results!

My guy @chrome prairie had kickass results with very simple prompts, so may need to understand veo 3's responses myself

[steady crag]

Action: The blue Elven woman stands and releases an arrow from a shiny gold bow. The action should be Smooth precise and controlled release of the arrow. The arrow shoots straight

I would also consider writing this differently. I think the concept should be more encompassing, and you should write it more action by action with that specific scene described in the style of shot you want with camera movement.

So something like

Concept: An 8-second video of a blue Elven woman standing and shooting an arrow from a shiny gold bow.
The first scenes camera will not move, but focus on her stance and her smooth movements. then transition to the arrows flight.

Concept step-by-step:
Start with The blue Elven woman stands and releases an arrow from a shiny gold bow, medium shot (1.2.5). The action should be Smooth precise and controlled release of the arrow. The arrow shoots straight Transition to a tracking shot (1.3.3) that uses smooth and precise tracking, following the arrow as it glides straight and true through the air. Employ panning shots (1.3.1) to showcase the mythical environment.

[humble scaffold]

Thats the same prompt...

[steady crag]

ahh I was looking how can I access flow as it shows unavailable, so i thought this channel generates some video, just trying it

[humble scaffold]

Nope.
https://labs.google/fx/tools/flow

You need to have a Google AI pro or Google AI Ultra plan to do this

[little cave]

Oh hello @bytememory (i am from research panel). LOL i try this when i access Flow.

[full mortar]

a modern dancer stands poised in a soft powder blue environment, surrounded by layered blue-toned fabrics. the background and floor are solid powder blue. the dancer wears structured, minimal clothing in similar blue tones. she moves with strength and calm — holding a powerful stance, one foot grounded, one arm reaching slowly forward. lighting is soft and directional, casting subtle shadows. in her hand, she holds a minimalist plate of food, balanced and central. the overall tone is elegant, slow, and composed. the scene is shot vertically, stylised and brand-led.

[queen citrus]

a modern dancer stands poised in a soft powder blue environment, surrounded by layered blue-toned fabrics. the background and floor are solid powder blue. the dancer wears structured, minimal clothing in similar blue tones. she moves with strength and calm — holding a powerful stance, one foot grounded, one arm reaching slowly forward. lighting is soft and directional, casting subtle shadows. in her hand, she holds a minimalist plate of food, balanced and central. the overall tone is elegant, slow, and composed. the scene is shot vertically, stylised and brand-led.

[frozen laurel]

#flow-tips-n-tricks・✏️ A chef from 1948 is working on luxurious hotel. The business woman comes to him and requests to prepare a brownie that should fit in the lunch box. The chef is preparing the brownie with passion and delivered to the business woman standing on the stage at an exquisite world centre with other business women.

[wind sedge]

can someone help me figure out how to get portrait aspect ratio,

[prisma zenith]

I've been using VEO3, and trying to make prompts of the subject speaking directly to the camera, but the mouth movements never matches the actual script. Is there anything I can do to help refine this?

[molten parcel]

#flow-tips-n-tricks・✏️ Activists posing with candles in front of foreign embassies, ignoring a man tortured in their backyard. Their banners read “Freedom” in English, but the Urdu behind them says “Stay Silent.” Spotlight exposes their dual faces — one smiling, one indifferent. Wimmelbild with charity logos turned into locks, hidden red carpets, award trophies dripping. Cinematic, 45° Dutch tilt, extreme close-up on a forged human rights certificate. Shot on IMAX, ultra-detailed.

[lunar charm]

Is there anything specific we have to do besides the prompt for audio generation. I am trying to have a generated video with a specific sentence said by one of the characters. I see no audio being generated?

[balmy glen]

40-Second Script

[0:00-0:05]
CLOSE-UP: Scientist holding glowing invention in dark lab
SCIENTIST: "My breakthrough could change everything..."
NARRATOR (V.O.): "But there's a VALLEY between your lab and the world."

[0:05-0:15]
VISUAL: Deep chasm appears. Scientist stands at edge.
SUDDEN TRANSFORMATION: Golden bridge materializes
NARRATOR: "KUEC TRL PROGRAM. The bridge you've been waiting for."

[0:15-0:30]
FAST CUTS:

Inventions transforming into real products (Transformer-style)
1 MILLION AED appearing as golden particles
Industry titans shaking hands with researchers
Products launching into cities, hospitals, homes

NARRATOR: "One million dirhams. Industry partners guaranteed. Your personal mentor. Your breakthrough's journey begins."

[0:30-0:40]
CLIMAX: All inventions explode into KUEC logo
TEXT: APPLICATIONS OPEN DECEMBER 30
NARRATOR: "Cross the valley. Change the world."
SCIENTIST: "...Now it will."
KUEC.AE/TRL
[END]

VISUAL STYLE: Transformer meets Marvel - dark to light, epic transformation effects, gold/blue palette
SOUND: Building orchestral score, transformation sound effects, triumphant finale

⁠flow-prompts┇✏
40-Second Script

[0:00-0:05]
CLOSE-UP: Scientist holding glowing invention in dark lab
SCIENTIST: "My breakthrough could change everything..."
NARRATOR (V.O.): "But there's a VALLEY between your lab and the world."

[0:05-0:15]
VISUAL: Deep chasm appears. Scientist stands at edge.
SUDDEN TRANSFORMATION: Golden bridge materializes
NARRATOR: "KUEC TRL PROGRAM. The bridge you've been waiting for."

[0:15-0:30]
FAST CUTS:

Inventions transforming into real products (Transformer-style)
1 MILLION AED appearing as golden particles
Industry titans shaking hands with researchers
Products launching into cities, hospitals, homes

NARRATOR: "One million dirhams. Industry partners guaranteed. Your personal mentor. Your breakthrough's journey begins."

[0:30-0:40]
CLIMAX: All inventions explode into KUEC logo
TEXT: APPLICATIONS OPEN DECEMBER 30
NARRATOR: "Cross the valley. Change the world."
SCIENTIST: "...Now it will."
KUEC.AE/TRL
[END]

VISUAL STYLE: Transformer meets Marvel - dark to light, epic transformation effects, gold/blue palette
SOUND: Building orchestral score, transformation sound effects, triumphant finale

[lunar charm]

@balmy glen Do you have the clip?

[balmy glen]

no

how to get it do the clip

[lunar charm]

@balmy glen Your first clip. I just tried with that prompt.

[balmy glen]

@lunar charm wow amazing !!! how to get the 40 second clip

[lunar charm]

You need to stich the scenes, but I don't know how to keep the same people or characters in the scene

[desert anvil]

i think on the flow vid from google they showed how to do it, right? i dont have the tool personally

[kind vortex]

tried this and the output has no sound.
am I doing something wrong?

[lunar charm]

You might be selecting veo 2 instead of veo 3.
Veo 3 is only one that has audio generation.

[nova acorn]

It’s a known bug that many have pointed out, that many veo3 videos are not generating sound , especially when multiple scenes are added via scene builder.

Well, it’s known to us users but I don’t know if the Google team knows

[severe aspen]

our teams are looking into this! appreciate everyone that's reported this to #1319424187730755636

[brave acorn]

....

[balmy glen]

@Georgia where we can get accsess to try it ?

[fathom acorn]

Please forgive my ignorance. I'm in Flow trying to add audio. I understand this can only be done with Veo3. How and where do I select that?

[pastel turret]

you need to be an Ultra subscriber for now

[fathom acorn]

oh. thanks.

[pastel turret]

it would then be available in the dropdown

[thick haven]

Hey I got you ! Making it right now

[thick haven]

Here you go ! I made a couple, I made some from your reference image and then others with sound and music with prompt only !

The third one was a goof but really funny, reminds me of Monty python XD

the sound was realy good in the second one, like you were in a ship staring at the ship

[hexed berry]

Can someone try rick and morty voice

[thick haven]

https://youtu.be/g0iilqpnn5A here is the youtube link I uploaded it to youtube so its easier to share !

[soft sentinel]

Wow, that's impressive! Thank you very much! I like how it moves the camera though glass on 1st and 3rd examples. It's really Powerful tool to use. Can't wait to get it in my country!

[thick haven]

Honestly I love

it

[thick haven]

Just use a VPN and you can access it !

[thick haven]

Yes here you go

Kinda Trash but something

[hexed berry]

Not bad

Thanks

Try live action see how it goes

[thick haven]

lmao omfg

[white gulch]

can someone generate a video from this image?

[thick haven]

https://www.youtube.com/watch?v=SWMx6ogIX04

share it wtih your friends

[thick haven]

Im tapped out today my dude but might be down tomorrow

[humble scaffold]

I gotchu

[lime oyster]

wow cool

[graceful wing]

Can anyone make an AI music video with a song I wrote?

[scarlet trench]

Hi all, someone may have answered this question, but does anybody know why captions are coming up when Sound/dialogue begins?

[slim barn]

#flow-tips-n-tricks・✏️ Queen Esther praying on her knees as light shines glistening.

[blazing karma]

dont output audio 10 only 2 make with audio. very bad!

[molten knot]

@vestal seal has said it’s a known issue and they’re working on it

[scarlet trench]

Thank you

[humble scaffold]

Try not to use quotes, if you need dialouge say something like They talk about [subject] and that [additional detail] and it seems to go away most of the time

[pulsar crow]

how do you add sound to your flow videos? my projects are all silent?

i have ultra

[reef sorrel]

how did you prompt it btw?

[pulsar crow]

oooo my quality was selected as veo 2

[reef sorrel]

OOF

hopefully u didnt burned many credits

[pulsar crow]

dang they are sneakkkkyy veo 2 is the default selection lol i wonder how many credits are wasted by people because of that

[reef sorrel]

Trust me, ever since day 1, a bunch of users burned through their credits just because it auto-switched back to Veo 2

[hexed berry]

Whoever is in charge of the ui design should be kicked out

why did you delete it go on type it

You’re here 24/7

@reef sorrel go on

[pulsar crow]

why do i always get this message? (Switching you to a compatible model for this feature. Submit again to confirm (or check settings for details).

it automatically switches me to veo 2

[pulsar crow]

it looks soooo gooood

[glacial cloak]

Hi everyone I am searching how to optimize prompts for veo 3, Is it a good idea to detail each scene? For example: Scene 1:
Timestamp: 0s–3s
-Prompt

[hexed berry]

Dont think so

[ivory ocean]

Hello, I have just subbed to flow, and wanted to test out a few things, I've been working on a short when I seen a video on this AI, I've asked for a few prompts but it seems it is overly sensitive with prompts, I asked it to produce a video of a girl walking out a door of a old house and looks out into a field with a old tree next to a barn, the tree has ravens on its branches, Somehow this might go against some policy. How can we use these if everything is against some policy?

[frigid flame]

Does Flow explicitly prevent generation of identifiable people? If so, can anyone explain how the Will Smith spaghetti video was made? https://x.com/attractorgenie/status/1925572274477748640?s=48&t=EKhpV-WHck1nr3XC4F5T9A

Petrov Dragan - ceo & founder @qme.ai (@attractorgenie)

This is Will Smith eating spaghetti.

But I don't even notice Will Smith.

All I see is infinite possibilities.

Veo3 from Google opened up so many opportunities for storytellers.

Storytelling beats AI prompting. Every single time.

Credit: Javi Lopez

X · May 22, 2025, 3:19 PM

[proper ingot]

when does it get available in india

[hexed berry]

Img to vid

[rain garden]

Hello. Currently working on a video using 8 separate clips generated in VEO 3. Has anyone else tried this yet? Issue I keep running into is I cannot get the music to flow seamlessly from clip to clip and wonder if I should just leave the music prompts out and add it in later with a separate program.

[reef sorrel]

It’s a limitation, so it’s probably best to just add it later using post editing software

[rain garden]

Yeah thanks, gonna try that with capcut and see if it's sufficient

[reef sorrel]

This guy used iMovie to do post editing

[reef sorrel]

Oh definitely sufficient, CapCut is really nice ✨

[rain garden]

Ok, cool, thanks for the recommendation

Only have phone so it may have to be a team effort with my friends who have PC if mobile apps aren't enough 😅

[reef sorrel]

I always use CapCut on my phone so don’t worry 😉

If anything there’s actually web based version of CapCut

[rain garden]

Very impressive software for realistic human interaction but the subtitles, music and scene to scene flow are where I get hung up

[reef sorrel]

Yeah, it’s unfortunate, but they’re currently working on fixing the subtitle issue, so stay tuned ✌️

[reef sorrel]

Everything about Veo 3 is highly impressive the generated physics alone feels very natural

[rain garden]

Yeah. This is my first time learning any of this. Video editing, marketing, etc. Very engrossing. Perfect time to get into it with this thing.

[reef sorrel]

Oh btw speaking of marketing

Not sure if this would be useful to you, but they recently expanded access to Google Vids for pro users.

https://workspace.google.com/intl/en_my/products/vids/

[frigid flame]

What does “Failed Generation” specifically mean? Is it a refusal to make the video based on engine type and prompt? Is it failure due to too many users? Can it also be a policy violation?

[reef sorrel]

Usually safety filter triggers it

If you try again with same prompt and it still failed then it’s definitely safety filter

Failed generation doesn’t consume your credit

[frigid flame]

Are there any negative prompts recommended to avoid the burnt-in misrendered subtitles appearing on screen?

[rain garden]

@reef sorrel have you had any success in keeping the voices the same throughout different clips? I'm trying to use a voiceover with a character from an earlier scene.

Anyone else?

[reef sorrel]

hmm not really

im trying to save up my credits so i wont do much veo 3 temporarily . let me check the deepmind veo page to see if theres any examples for that

[narrow musk]

can you create a sub section with the most prompt tips to begin with?

like how to have subtitles, how to avoid subtitles, how to generate speech, etc

[runic shell]

Is the audio working on veo3. Some of my clips don’t have audio

[narrow musk]

yes SFX

[reef sorrel]

theres' no negative prompting way but closest way would be not to include "[dialogue]" at all in your prompt

[reef sorrel]

nope couldnt find any examples

[rain garden]

Yeah removing " " has helped and also clearly indicating NO SUBTITLES seems to be working for me

[reef sorrel]

you might be stuck with not able to keep the same voices for a while

[rain garden]

Generated 3 scenes that way without them this morning

[runic shell]

So it’s an ongoing issue?

I’ve done 2 veo3 and both don’t have audio

[narrow musk]

some have said they have this issue, it works for me so i dunno

[knotty night]

anyone know how to use flow veo 3 in uk

[rain garden]

Yeah every Veo 3 clip I've made has audio. Just make sure you're set to Veo 3 every time you generate cause sometimes when you close tab and come back to it, it will default to Veo 2 setting with no audio.

[rain garden]

Only available in US rn

[knotty night]

u recon i can use vpm

vpn

[quartz harbor]

queen latifah as equalizer

[rain garden]

Someone I know says they have used VPN from UK and it worked

[knotty night]

can u find out which one please

[narrow musk]

Even with VPN you will not access to flow from outside to the US rn

[knotty night]

shame how about veo 2

anyone used that

im new to AI

[warped elk]

it will be avaible in Europe soon ?

[rain garden]

No idea I've heard maybe soon

[warped elk]

why it does not work with a VPN ?

[knotty night]

Anyone know any other good AI video generation

[warped elk]

there is no alternative as good as veo 3

[knotty night]

thats a shame

[reef sorrel]

u can use veo 2 in gemini

without vpn

[narrow musk]

Their security feature is not checking only this

[warped elk]

ok i'll wait to get the realease in Eu

[narrow musk]

@reef sorrel , i don't know if i can tell here a trick that may work for EU users that want to test it rn?

[reef sorrel]

hmmm yeah, you can DM them but honestly, I think they might roll it out to the UK soon. no ETA yet though.

[narrow musk]

bcz every time new people will come, we will have the same question again & again

they can even create a wiki like

[runic shell]

Is there a prompt to use the same character?

New prompt = new face 😂 can’t keep a story line with different people

[barren cloak]

exactly we desperetly need a detailled guide for beginners to solve all those basics issues

[azure trellis]

Really basic question: If I have an idea for a short film with 3 x 10-second clips, do I give the entire script, is there a way to say "cut this script up at the scene changes but maintain continuity" or some other way of saying "you won't be able to generate the whole thing in the first pass, I know we have to do this in steps, stop rushing through and trying to fit all 3 scenes in one"

Like storyboarding a movie first.

[dire patio]

Hi, I am trying to make a prompt that will have a clay world "grow" into a 3D world. Right now, whenever I try it makes it look very complex and adds a bunch of objects and bouncing affects. I have a start frame and end frame. I did not include the full generated video, but you can see how many extra objects appear.

Current prompt, "stationary camera, flat clay slowly, carefully, and organically grows into 3D world"

[noble lintel]

Hello, any guide on how to achieve character and settings consistency?

[bleak kayak]

Alr since I'm a bit of a total drama phase and I don't have the money for veo 3
But can someone try and make a toal drama episode with veo if it understands it or at the very least make a scene from the show

[nova gale]

I just used veo3 make this video, veo3 is really good omg :

https://www.instagram.com/reel/DKC3cJ9P_Fp/?igsh=azNjb3NobXMwNDlt

cosmic_bone_hk

Inspired by 「哪一天我們會紅」之劇中劇 -
「殺人記憶」大家未睇可以去睇睇～ 🫂
真係近來少有覺得好睇嘅劇🙏🏻

@no_dof.official @michelle3.8.9.7 @gthan_so @jacky_cheung_d @sugarsokaaa @77y_anna @viutv
@marifemarife

#哪一天我們會紅 #無景深 #其實都係想做自己鐘意做嘅嘢 #邊個...

Instagram

[lime oyster]

bloody hello!

[atomic bluff]

O prompt que eu usei no vídeo do pão de queijo @fair hatch :

"A cheese bread vendor from Minas Gerais, Brazil, with a Minas accent is interviewed by a reporter, where he invites people to come buy cheese bread at his shop, offering a free cup of hot coffee. He speaks Brazilian Portuguese. His name is "Antenor da Silva", Profession: "Vendedor de pão de queijo"."

[fair hatch]

Muito bom

[unique trout]

hi there

[sly haven]

You give it one 8 second 'shot' at a time. Sometimes you get lucky and it does multiple shots in those 8 seconds but don't count on it.

[native isle]

Hey everyone! Any tips on how to write a prompt that makes characters talk like in real conversation?

[hexed berry]

Maybe add : before the dialogue

[carmine harness]

Yes, I can't seem to train it to follow the person in the picture that I provide veo 2 or 3.

[broken sleet]

This guy predicting the future

[ocean heart]

Amazing...

I haven't tried adding conversation to my videos yet

Can the prompt be accurate to the second? For example, raise your hand in the first second, then wave your hand in the second second, then blow a kiss, etc.

[spiral mesa]

I have Veo3 but none of the videos add sound or dialogue when prompted. Any ideas what I am doing wrong?

[hexed berry]

Select the veo3 option. Might have defaulted to veo2

[misty wedge]

⁠flow-gallery┇🖼⁠
Can someone fix this prompt for me to do what it is intended to do ?
white background. Camera focused on woman she says "Hello. I am AI" Then the woman turns into man and says "This is compilation of AI Videos" Then the man turns into a dinosaur

I

total failure

[fleet rose]

Video #start-here・📍 horse’s trying finish raining #flow-tips-n-tricks・✏️ horse’s

[violet vortex]

can someone generate a video from this image?

[tawny storm]

Rural Ireland, sometime in the 1860s.
A young woman walks side by side with a boy—
their long, homespun garments catching the gusts of salty coastal wind,
their hair tousled, their eyes fixed ahead.

With quiet resolve, they tread across a windswept cliff top,
where hardy wildflowers, pale and wind-blown,
blanket the ground like nature’s own tapestry.

Below, the ocean surges—grey-green waves crashing
with relentless fury against the sheer rock face,
sending ghostly sprays of white foam spiraling skyward.

They move forward still, hearts steady,
into the embrace of the unknown.

[pearl pond]

Nincompoop

[carmine harness]

I’m struggling to find a prompt that uses the picture provided in the frame to video option. Like I need to create a consistent character that never changes

[wide dirge]

How did you make a consistent clip of more than 8secs?

[azure nest]

no BS no Ads no li mns etc... this helped me a lot this weekend https://chatgpt.com/g/g-68322c28b68c8191a00e13955b09f737-ai-video-prompt-expert-veo-3-gemini-edition

[sick juniper]

Why are the VO prompts constantly generating subtiltes. And they are almost always spelled wrong.

[formal pollen]

Can you guys make video from image?

[humble scaffold]

Don't include quotation markers in the prompt and there are a lot less

[dusk urchin]

A young Moroccan man in his late 20s stands in a modern, sunlit kitchen in Casablanca, enthusiastically describing a mobile app called Douq.ma. The app helps people discover the best dishes to try in local restaurants. He holds a smartphone, tapping and swiping as the camera alternates between over-the-shoulder shots of the app interface and his animated gestures. The background includes tagines, mint tea, and stylish kitchen decor. Soft Moroccan music plays in the background, with smooth transitions, warm lighting, and a slight cinematic zoom for emphasis.

[tired thistle]

idk but check it out yourself you can sign up for a trial month for free.

[humble scaffold]

I got you, be back with the video in a few minutes

[dusk urchin]

I love it 😍

[soft sentinel]

Would be nice to turn this amazing painting from a Nelson Museum into an actual movie battle scene and keep it in the style of a painting. Would be nice to see how AI will turn it into a depth scene with a camera flying though the battle

[ocean furnace]

Good stuff

[dusk urchin]

How can i generate

[half thistle]

Can flow generate a video with audio ?

[knotty estuary]

#flow-tips-n-tricks・✏️ One morning, Bob the squirrel decided to try human food and snuck into a kitchen through the open window. Eyeing the toaster, he stuffed it with acorns, thinking it was some kind of nut-roasting machine. As he proudly pressed the lever, the toaster sparked, flung flaming acorns across the room, and set off the smoke alarm. Chaos erupted. The cat screamed, the dog did parkour off the furniture, and Bob — with his tail singed and ego bruised — leapt out the window yelling, “Next time, I’m just ordering pizza!

[half thistle]

Can you generate a video with audio with Flow? I tried but didn't work. I wonder why?

[humble scaffold]

You need to be using Veo 3. It is labeled as "Highest Quality"

[azure nest]

for anyone that is looking for more info on how to prompt veo3 or learn the right ways to use it I made this, no BS no Ads no links etc... this helped me a lot this weekend https://chatgpt.com/g/g-68322c28b68c8191a00e13955b09f737-ai-video-prompt-expert-veo-3-gemini-edition

[fathom onyx]

I sent this to general but maybe here is better... Hey yall!! hyped to be apart of this community and get Flowing! I have a question if anyone can help me out. In Veo 3 is there a way to prompt it so it outputs a vertical video? I saw an example of a guy doing it saying something like "vertical video for a tiktok" and it came out full 9:16 but it was laying sideways, so you just have to edit and flip it and its fine. but Ive wasted a ton of credits just trying to get a vertical video.

[azure nest]

you just used a description of the painting?

its very consistent

cool

i wonder if its passing any context?

[soft sentinel]

Oh, that looks awesome, thank you! It's a totally new way to turn historical paintings into some additional ways for documentaries. Making history alive

[azure nest]

very very cool

I hope they dont censor it so we can reference this stuff

awesome 🙂

[soft sentinel]

Yea, it even added new ships after getting away from original painting

Ah, I wish I could play with it myself, but it's not available in my country(

[wide dirge]

How did you manage to keep the cohesion in doing more than one video?.

[azure nest]

I want to know as well I spent all weekend on this https://chatgpt.com/g/g-68322c28b68c8191a00e13955b09f737-ai-video-prompt-expert-veo-3-gemini-edition

I want to add it in lol

[dense temple]

Thats a solid extension until 0:22!! Great demo

[wide dirge]

Gotcha. That's why there's no sound. Can't wait for that to be possible with Veo3 and hopefully beyond.

[lime oyster]

lmfaoooo fast mode is tone of sh**

[latent mesa]

has anyone been successful in using "frames to video" with experimental audio and getting it to generate audio?

[lime oyster]

fast mode is like you dream something then when u wake up like "wth was that?"

[lime oyster]

what is even frame to video?

[latent mesa]

setting option to upload a first or last frame of reference. it does a good job matching consistency of characters but i can't get it to generate with audio even though it has that option as a model

[fathom onyx]

tryin it out

[azure nest]

awesome 🙂 lemme know what you think

[fathom onyx]

i got it finally make a vertical style video!!!!

using your chat gpt

[azure nest]

🙂

very cool the knowledge base is 20+ pages of all their docs

lol

and tons of good prompt ezamples

[fathom onyx]

this took me all day to make lmao

[azure nest]

lol happy to help

[fathom onyx]

seriously dope

[azure nest]

tell a friend 🙂

[trail yarrow]

is there a way to get 4K access? I'd like to request it. I'm coming from the film/cinematography side and Veo 3 is blowing my mind
the resolving detail/not the resolution though, is intriguing

[fathom onyx]

Lmao

That was amazing why you delete it

[sly haven]

Meant to post it it in the #flow-videos・📹 🙂

[cyan pulsar]

@fathom onyx Cool! What was the prompt to make it vertical?

[swift osprey]

How do I make a Veo3 video? Tell me please

[frosty drum]

Dumb question maybe but why do none of my clips have audio?

Nvm I was on veo 2

[grand rain]

How do you get onto veo 3?

[grand rain]

Thanks. I turned it on but no audio generated. I'll try again.

[novel pier]

make sure it's on highest quality also should say veo 3

or like mine

then veo 3 when you're choosing (hidden from my image)

[grand rain]

Awesome. I was able to get it! Let's see if it generates.

Yeah, there's just no audio generating. I'll keep playing around.

[lost finch]

How can I keep a consistent character with veo 3? for some reason they change each 8 seconds.

[sly haven]

In scenebuilder using the "jump to" generation but it is hit or miss and degrades slightly over time.

[slender spire]

Long prompts really do wonders

https://fxtwitter.com/thaakeno/status/1926782706911707344?t=Ek3e2mQ6h6iD4WWJ2fJiiA&s=19

thaakeno (@thaakeno)

#Veo3 #Joker

**❤️ 2 👁️ 89 **

▶ Play video

FxTwitter · May 25, 2025, 11:29 PM

[rotund path]

hi

how to write prompts?

[lime oyster]

just let gpt or gemini write the prompt

[steel ridge]

#flow-tips-n-tricks・✏️ A medium shot frames an old sailor, his knitted blue sailor hat casting a shadow over his eyes, a thick grey beard obscuring his chin. He holds his pipe in one hand, gesturing with it towards the churning, grey sea beyond the ship's railing. "This ocean, it's a force, a wild, untamed might. And she commands your awe, with every breaking light"

[molten parcel]

“make a clip of johnny joestar from part 7, sitting on the mountain at the end of the story looking at the sky and there is beautiful sunshine happening, and him saying “let’s go home gyro” in japanese and all this be manga style art and there be edit like camera movements”

[humble sand]

This thing is brilliant no doubt but the thing that is very educational is the character continuation. As you only get eight seconds a video which I’ve done several times I would like to create a new scene using the same character and when I do it completely changes character, even though I give exactly the same look and style. At Three bux a spin it is a bit frustrating.

Can anyone suggest the work around?

[lone sierra]

I am using gemini to refine my prompts but they are relatively long. sometimes the output misses some of the descriptions i add.
should i use long descriptive prompts or should they be shroter?

[warm mist]

#flow-general・🌊 A dark, rain-slick dystopian street in New Kumasi, year 9030. Neon signs flicker in broken Kanji-English hybrid text. The camera glides low past faceless guards in matte-black armor, scanning civilians with glowing retinal devices. Thunder cracks. A cloaked African migrant steps from the shadows, eyes glowing faintly with electric blue energy. The wind lifts his hood slightly, revealing a scarred face and emotionless stare. Slow-motion as he raises a hand — suspended in the air — and the ground trembles. Police drones buzz above. Lightning reflects in puddles. Tension builds with an eerie cinematic score. Cinematic lighting. High contrast. Moody, futuristic atmosphere. 4K resolution. High frame rate.

[wide dirge]

So in the prompt you note for it to make a vertical style video?.

[raven osprey]

Create a high-quality 8-second video set on a busy urban street in Hanoi, Vietnam, during the daytime. The background is filled with moving cars, motorbikes, and many pedestrians.

In the foreground, feature a young, beautiful Vietnamese female reporter, around 20 years old, tall, wearing a tight white dress, holding a microphone, and looking fashionable and confident.

She is interviewing a charming Vietnamese businessman in his early 30s, who appears elegant, successful, and well-dressed in a modern suit.

Both characters are facing each other, speaking in a cheerful and humorous tone. At the end of the video, they both laugh out loud together.

Add Vietnamese voiceovers and Vietnamese subtitles with the following dialogue:

Female reporter: “Theo anh thì đàn ông thành đạt sẽ thích những người phụ nữ như nào?”
Man: “Ôi giời! Bây giờ lấy được vợ là may rồi. Nhiều tiền như tôi đây còn đang sợ ế đây này!”
(Both laugh loudly together: “Haha!”)

The tone should be light-hearted, fun, and energetic.

#flow-general・🌊 Create a high-quality 8-second video set on a busy urban street in Hanoi, Vietnam, during the daytime. The background is filled with moving cars, motorbikes, and many pedestrians.

In the foreground, feature a young, beautiful Vietnamese female reporter, around 20 years old, tall, wearing a tight white dress, holding a microphone, and looking fashionable and confident.

She is interviewing a charming Vietnamese businessman in his early 30s, who appears elegant, successful, and well-dressed in a modern suit.

Both characters are facing each other, speaking in a cheerful and humorous tone. At the end of the video, they both laugh out loud together.

Add Vietnamese voiceovers and Vietnamese subtitles with the following dialogue:

Female reporter: “Theo anh thì đàn ông thành đạt sẽ thích những người phụ nữ như nào?”
Man: “Ôi giời! Bây giờ lấy được vợ là may rồi. Nhiều tiền như tôi đây còn đang sợ ế đây này!”
(Both laugh loudly together: “Haha!”)

The tone should be light-hearted, fun, and energetic.

[wide dirge]

Was the closed caption in this prompt final product accurate?. Not that I would know as I don't speak Vietnamese but it seems to butcher english fairly often.

[stray lodge]

Create a playful animated music video in a colorful cartoon world. Show curvy animated girls dancing and shaking their hips in sync with a fun dance beat. The style is exaggerated, cheeky, and humorous. Add silly characters, bright backgrounds, funky transitions, and surreal dance sequences. Make it fun, addictive and perfect for short viral dance clips. Focus on scenes that match the lyrics “muovi il culetto” with synchronized hip-shaking animation. Think pop visuals like ‘Chiken Banana’ or 'Domitila'.

[west swift]

how do i create a character library ? i want to create 3 videos with the same characters

[west swift]

gemini cannot give definitive answers for google products. i think thats a basic library that should be updated onto its memory

[keen wind]

#flow-tips-n-tricks・✏️ “make a clip of johnny joestar from part 7, sitting on the mountain at the end of the story looking at the sky and there is beautiful sunshine happening, and him saying “let’s go home gyro” in japanese and all this be manga style art and there be edit like camera movements”

[molten parcel]

Make a clip of creating a video game

[keen wind]

make a clip of johnny joestar from part 7, sitting on the mountain at the end of the story looking at the sky and there is beautiful sunshine happening,

How do we get it to make videos for us?

[celest raptor]

A dramatic, looping animation of a medieval city built on rolling hills, burning down to ashes. The city features stone towers, wooden houses with thatched roofs, narrow cobblestone streets, and a distant castle on a hill. Fire spreads through the city — roofs collapse, smoke billows into the sky, and glowing embers rise. The sky is dark and filled with smoke and ash. As the destruction completes, the scene subtly transitions back to the beginning, with the city intact and the first flames appearing — forming a seamless loop. Cinematic lighting, realistic fire and smoke effects, and a somber, apocalyptic mood.

[golden peak]

Create a clip in widescreen format. Panoramic view, cinematic feel. A road in a mountailn village. There are trees and houses by its sides. Misty woods behind. It's a gloony day. An old man walks trhou the road, from one side to another.

[lone sierra]

switched to chatgpt it has much better prompt gen.
does veo3 in flow allow consistent characters eg if we upload a reference image to the ingredients...

Okay good to know. Wondering if its worth to upgrade..

Yeah just trying to get a feel for it first. The only issue is testing veo3 is inside gemini not flow which is more restricted... but yeah thanks for the advice

[tropic prism]

Guys, does anyone knows a good prompt for realistic night landscapes images?

[noble lintel]

Any tips on getting consistent character in veo3? As ingredient to video still not available. Any idea on when it may be available

Is it available for veo3? I thought it was just veo2

[swift cloud]

Hi Frontier, I hope you're doing good. quick question, How does prompting work for Veo3, I really want to make cute animal couple relationship videos but I want to have consteitn animation, how do I make sure my prompts aren't to long but there good enough to where it is clear for Veo 3 to make a good detailed video

So like, use chatgpt to make the characters, then upload them to veo 3 to make a scene occur?

[fathom onyx]

Yes but it’s still not a guarantee that it will render vertical. It’s default is 16:9
It was the right ratio but I had flip it in editing because it came out sideways which is fine. But then i tried again after and it didn’t work. Hopefully it updated soon so you can choose the ratios before and it’s a guaranteed. We will be wasting a lot of credits for vertical videos

[fathom onyx]

Chat gpt is insane at making prompts and detailed characters creation. Camera style all that. More detail is better. Any wrong word can throw off your whole prompt.

[crude heart]

Does flow require a specific upgraded plan to use?

[swift cloud]

do you this chat gpt is better than claude to make the promtps, or go all the way into using chatgpt?

[crude heart]

But it's starting that I need to upgrade my Google one plan?

[fathom onyx]

Yeah I’m like a month into using Sora and chat gpt for making prompts. Some I do on my own but then I make weird stuff I can’t even comprehend to make weird creatures etc. depends what I’m doing.

[crude heart]

I already have Google one + access to Gemini

[fathom onyx]

That’s true. I’ve been using Sora for remixing other videos and applying prompts to replace (blank) with (blank) in videos. Some have gone viral. But I’ll try to start doing image to video more now. Feels like Christmas to wake up and use Veo 3 lol

[winged siren]

hi anyone help me to generete a video for me please

[fathom onyx]

Do you think if you upload the image in 9:16 you want the video to be based on then the video will come out 9:16?!

[fathom onyx]

What do you mean?

[wide dirge]

I posted failed examples I made in the gallery. If nothing else it would make for a nice sci-fi moment. lol

[sly haven]

Maybe. When I put "vertical aspect ratio video" in my prompts sometimes it produces a result like this:

[sly haven]

If you don't need audio, veo2 already supports generating 9:16 nativeley through gcp console.

[fathom onyx]

oh okay yeah i may try that more then and just add music or custom audio after!

yeah its kinda nuts google doesnt just add that already. 9:16 is in high demand and used by everyone on social media. cmon goog

[sly haven]

It's expensive just FYI ... $0.35 USD per second of video.

[fathom onyx]

oh well im $600 deep after a month. im invested

with sora, kling, veo 3 all combined lol

[left geyser]

U reckon veo2 is as good as veo 3 visuals if you dont need the sound?

[sly haven]

It's $11.20 every time you create 4 videos that are 8 seconds each. I racked up $250 just in 30 minutes. But if you're in the boat that can afford, all the better. 🙂

[fathom onyx]

haha yeah I think it will be worth it if you have goals of getting clients or building a channel etc. also I just got it to do audio but it was in a random language?!!

is it because it a cheesebuger creature and not a human?! LMAO

I removed costco from the dialogue and think thats what it was but then its now in random language...

[sly haven]

Flow Pricing / Quality
Veo2-fast is nearly identical in visual quality to Veo3.
Veo2-fast: $0.20 USD per 8 seconds currently $0.10USD during discount period.
Veo3: $2.00 USD per 8 seconds; currenly $1.00 USD per 8 seconds during discount period.

GCP VertexAI Console Pricing / Quality
Veo2 through gcp console has much worse quality than Veo2-fast-flow.

Veo2-gcp-console: $2.8 per 8 second generation.

In other words if you don't need audio, stick with veo2-fast in the flow website because veo2-gcp-console is currently 28 times more expensive for less quality.

[left geyser]

bet, thank you!

[fathom onyx]

i did the audio myself after. Idk why maybe cuz it's not a human? I'll have to just put in the prompt for a voiceover to say it for stuff like this maybe

[hexed berry]

Veo2 fast sucks compared to veo3 but is good for the price range

[white osprey]

It's amazing. But too costly for hobbyist.

[nova tundra]

where can I report errors? this thing has been stealing all my credits and generating videos that wont play! when i click upscale it also fails

[clever nova]

#1319424187730755636

[sterile wing]

oi

gere um video de um evtol

[half copper]

A graduation party scene in a university hall. Several young women wearing hijabs are celebrating. A banner or sign in the background reads 'Congratulations, Hamma Lakhdar University English Student Licence'. The atmosphere is joyous and celebratory."

[sour pecan]

a man walking down the forrest and bear walks behind him and he starts running

[carmine anvil]

A cheerful Egyptian farmer standing in a busy street market in Cairo, selling fresh tomatoes from a wooden cart. He wears a traditional galabeya and a white turban, with a warm smile on his face. The background shows colorful market stalls and old Cairo buildings. Add a comic-style speech bubble where the farmer says in Egyptian Arabic: "أنا قررت خلاص، هبيع طماطم... بلا تصميمات بلا وجع دماغ!"

[hoary wraith]

A cheerful Egyptian farmer standing in a busy street market in Cairo, selling fresh tomatoes from a wooden cart. He wears a traditional galabeya and a white turban, with a warm smile on his face. The background shows colorful market stalls and old Cairo buildings. Add a comic-style speech bubble where the farmer says in Egyptian Arabic: "أنا قررت خلاص، هبيع طماطم... بلا تصميمات بلا وجع دماغ!"

#flow-tips-n-tricks・✏️ A cheerful Egyptian farmer standing in a busy street market in Cairo, selling fresh tomatoes from a wooden cart. He wears a traditional galabeya and a white turban, with a warm smile on his face. The background shows colorful market stalls and old Cairo buildings. Add a comic-style speech bubble where the farmer says in Egyptian Arabic: "أنا قررت خلاص، هبيع طماطم... بلا تصميمات بلا وجع دماغ!"

[winged dagger]

#flow-tips-n-tricks・✏️ It’s a beauty salon, I need it full of stylists and beautiful clients, they are working on clients hair, haircut, blowouts and hair coloring

[drowsy lotus]

"Create a vertical cinematic video (Reels format, 9:16) for Facebook and Instagram, targeting Libyan tourists. The video should follow a realistic vlog-style journey of a young Libyan man visiting Egypt. It begins with him arriving at the airport, exploring Egyptian landmarks like the Pyramids, Khan El-Khalili, and enjoying local food and culture.

Then he visits Jamal Clinic for a premium hair transplant experience. Show the consultation, modern equipment, professional staff, and the painless procedure. Use real-looking before-and-after hair transformation shots.

The video should have a natural, authentic feel — like it was filmed by a Libyan content creator. Add a Libyan dialect voiceover (plogger-style), warm lighting, friendly faces, and Libyan-style Arabic subtitles. Background music should mix modern upbeat tones with Libyan and Egyptian flair.

End the video with the tourist confidently walking around Cairo post-treatment, smiling, and saying: ‘رجعت بشعر جديد، وذكريات ماتتنسيش من مصر’."

"Create a vertical cinematic video (Reels format, 9:16) for Facebook and Instagram, targeting Libyan

[balmy pasture]

#flow-general・🌊 Panoramic view of a spectacular sunrise over a vibrant natural landscape (flowering field, green valley). A stylized door slowly opens, revealing a luminous path ahead. A person (or a diverse group of people) begins to walk along this path with an expression of purpose and hope, symbolizing the new opportunity to live out God's purposes.

[mild turret]

how to create a point-of-view style of video. i try to include in prompt it still not working :

"Point of view video style of a man lying in bed at night, struggling to fall asleep as his eyes blink slowly in the darkness, the quiet hum of a fan and distant creaks filling the silence—until he hears a soft, unnatural thump at the foot of the bed; when he opens his eyes, he sees a pair of perfectly round, glowing eyes staring at him from the darkness; frozen with fear, he blinks—and the eyes are closer; with every blink, they silently float nearer, until they reach the edge of the bed; heart racing, he shuts his eyes and tries to fake sleep, barely breathing, but then a cold, childlike whisper cuts through the silence: “I know you not sleep yet…” — and the screen goes black with a sharp gasp and a burst of static, leaving behind only suffocating silence."

[frail grotto]

raise two boys talking

[granite torrent]

help to prompt for editing video

[slender pivot]

#flow-tips-n-tricks・✏️ Early morning, soft golden sunlight filters through a busy street. A variety of small chai shops start opening. Shopkeepers arrange tea glasses, steam rises from boiling pots. Close-up shots of hands pouring hot chai into glass cups. Different people — an old man, a young girl in uniform, a delivery boy — drink chai. Close-up: a customer smiles after the first sip. The environment is warm, lively, peaceful. Cinematic angles, shallow depth of field, lens flares, warm color tone. Light morning fog in the background. Focus on faces, emotions, and chai-making details

[tacit plinth]

How can I use Flow in UAE?

[proper pulsar]

just use a VPN

[molten parcel]

#flow-tips-n-tricks・✏️ a man walking in the old town of shanghai in 2025, looking for food on the market

[mortal slate]

I'm from Portugal and was also thinking about doing that. Do you think there's a risk of getting my account banned for using a VPN?

[dawn jewel]

A man on a Scottish island, crying

[finite ravine]

Some one help why flow is not working

[hexed berry]

https://tenor.com/view/uk-police-officer-screaming-no-noo-gif-3163377484734570206

[river geyser]

man with his child first look

@dawn jewel man with his child first look

[dim mesa]

#flow-tips-n-tricks・✏️ a street reportage in turkey

[tacit temple]

#flow-tips-n-tricks・✏️ A monster sea

[somber forum]

A seaman with a tattood face

[mental brook]

Someone who managed to create a video of two people having a dialogue, with the sound of them talking. What prompt did you use to achieve that? I'm having trouble getting both voices to come out and keeping the conversation consistent. (edit: I am using a photo as the first frame)

[pale copper]

Please help. Can I supply a photo of my character and ask Flow to animate based on the photo? It does not create the correct character based on my text inputs. However, chat GTP did a good job of creating the image I wanted. Can Flow work with this imge, and if yes - how do I achieve this?

[humble scaffold]

Change the setting from Text to Video to Frames to Video

[pale copper]

Thank you, I will try that.

[hollow fulcrum]

A crowded cantina lounge bar bathed in the eerie purple glow of a binary sunset on a distant, forest alien planet. A grizzled space cowgirl, her face weathered, walks over and sits at a floating bar, her hand resting on a blaster. Beside her, a towering terrifying ugly alien creature with multiple glowings eyes and purple scales stares into the middle distance. Cut to a close up of The cowboy as she turns to the alien, a hint of menace in her voice, and says, "What the hell are you looking at?". Then cut to a reverse Close up of the alien as he slowly swivels its head, its voice a low rumble, and he replies, "Ohh Honey, you have not seen anything!". This captures a moment of tense, intergalactic standoff.

[humble veldt]

I am struggling with what should be a very simple adjustment to an extended scene. Creating a smaller piece of candy. I have tried "2 centimeters", ".5 inches", "VERY small", "the size of her thumbnail", and a mix of everything in between but they keep delivering these large red gummy variations. Anybody else struggling with something seemingly this simple?
example prompt: "The woman puts a very small one centimeter in radius red sphere shaped gummy candy into her mouth. She then smiles and chews."

[molten parcel]

#flow-tips-n-tricks・✏️ Me faça uma notícia da Record TV sobre que os alienígenas tão declarando guerra ao Brasil

[silver apex]

#flow-tips-n-tricks・✏️ a man hiking in the mountains

[pseudo blaze]

#flow-tips-n-tricks・✏️ Sonic the hedgehog landing a flurry of punches on mario in a boxing match.

[whole tusk]

I want to make a flow prompt that copies the motions of wow.mp4 bar for bar but with the characters of Perfection.jpg doing it

[ancient oracle]

#flow-tips-n-tricks・✏️ یه ویدیو میخوام توی ایران باشه پشت سوژه اصلی مردم در حال حرکت باشن توی یه روز آفتابی دلنشین شلوغ ، صدای همهمه کمی باشه ، برج آزادی هم پشت صحنه باشه ، که یه خانوم قدم بزنه بیاد سمت دوربین و با شوق زیاد بگه : اگه میخوای لینک ثبت نام رو داشته باشی ، کلمه جمینای رو کامنت کن ، پیج کامدین هم حتما داشته باش کلی آموزش باحال بهت میده !

[spice spire]

ok. since yesterday something has been happening that I totally don't understand. I'm writing a prompt (which by the way worked normally before), it's about dialogues or character monologues. I'm putting them in quotation marks. since yesterday each character, instead of speaking normally like a human being, sings, and there's generated music playing in the background. what's going on?

[deft root]

yo can somebody make the binding of isaac gameplay with veo 3

i tried with veo 2 once

it got gui pretty right but nto gameplay

sorry for asking im just poor

[crisp garden]

#flow-tips-n-tricks・✏️ Lofi rainy day mountail ..peace ... calm ... realistic ..

[still thicket]

Vector illustration of 3 Gen Z teens hanging out on their cell phones

[vague yoke]

a little league baseball player striking out looking.

[eternal burrow]

#flow-general・🌊 a man skateboarding

[crisp talon]

Does someone know how to make veo 3 STOP PUTTING SUBTITLES?. I just want the characters to talk

[urban wedge]

Are the subtitles legible? I’m geniunly curious lol

[humble scaffold]

Don't include quotes, instead say They talk about [subject] and say that [more details]

[wide dirge]

I'm the biggest fan of closed caption but it is so annoying here since we can't cut it off.

[civic stump]

Hi everyone! Is there any way to evaluate the prompt before generating the video to avoid spending credits unnecessarily? I’m trying to create a video, but every time it generates the result, there’s always something I don’t like — and while I refine the prompt, I just keep spending more and more credits…

[jaunty spoke]

Is there a way to add people using a picture for a scene. For example, I added a picture (patio deck with chairs) to the ingredients section and then added an a.i. generated image of a family of 4 for reference. But for some reason I cannot add them to the scene. Sometimes it will add 2 of them but the description prompt just doesn't want to seat them in the chairs despite adding that to the prompt. I've tried using gemini to refine the wording and it just doesn't work. Veo may not understand there are chairs maybe...but it does add one person to a chair sometimes.

I thought maybe I can get a.i. to add people to the image instead and then turn the image into a video clip...but I can't figure out how to do that. Photoshop is a no go as it is highly inaccurate. Maybe this isn't going to work just yet. I'm open to suggestions though.

Also, in the ingredients option there is no way to select a camera movement and so you have to create a prompt. But no matter how I word it it just does what it wants. If I tell it no pan, no tilt, and do a crane up, or boom or jib up shot camera motion, it ignores that as well. Any suggestions?

[pulsar crow]

can you have multiple tabs of flow open at the same time for the same project all generating different scenes?

[warm torrent]

the people jump at the beach with small ball

[unborn gyro]

has anyone been able to get consistent voice audio from image to video?

[hollow moon]

nope

[sly haven]

Yes you can just be careful if in one tab you have it set to generate veo3 and the other veo2-fast/veo2-quality. I've found that when there is a mismatch, one of your tabs inherits the model settings of the other and you end up generating videos with the wrong model. So always check and make sure you manually select the model you intend to render each time.

[sly haven]

No

[solemn hornet]

i need help, i cant get a scene to jump to. i can get it to extend, i can almost get action i want it to extend to. but i cant get it to jumpp to anything, it just keeps extending the shot. what am i doing wrong?

[tacit plank]

#flow-tips-n-tricks・✏️ Comforting commercial about cats during christmas

[gritty wasp]

why there is no sound in any of my video in flow , veo 3?

[midnight marsh]

Genre: Psychological Horror / Paranormal
Themes: Isolation, distorted reality, doppelgängers

Suggested Visual Style: Dark, moody lighting. Flickering lights. Mirror reflections. Slight camera shake in tension scenes.

Scene 1 Apartment Interior, Night
The camera opens on a dim, sparsely lit apartment. Rain taps gently against the window. The only light comes from a small desk lamp and the blue glow of a muted TV.

Narrator (low, calm voice):
“I had just moved into a new apartment downtown. Old building. Creaky floors. But cheap. Too cheap.”

Scene 2 Bathroom Mirror
The protagonist stands brushing his teeth. As he bends to spit, a figure briefly flashes in the mirror standing directly behind him. When he jerks upright, it’s gone.

Narrator:
“I saw it the first night just a flicker in the mirror. I laughed it off. Figured I was just tired.”

Scene 3 Time Lapse Montage (Days 2–5)
Clocks spin. Daylight and night alternate. The mirror figure becomes clearer. Sometimes it mimics the protagonist. Sometimes it just stares.

Narrator:
“Every time I looked in the mirror, it looked… more real. More solid. And always a few seconds behind me. Like it was learning.”

Scene 4 Night 6: Breaking Point
The protagonist wakes to glass shattering. He rushes to the bathroom. The mirror is fine, but the reflection is different it’s not moving at all.

Narrator:
“That night, I realized it wasn’t copying me anymore. It was watching. Waiting.”

Suddenly, the reflection smiles. But the protagonist does not.

Scene 5 Escalation
Lights flicker. The reflection begins moving on its own throughout the apartment, staring at the protagonist through other reflective surfaces: the microwave, the TV screen, a phone screen.

Narrator:
“I covered all the mirrors. It still found me.”

Scene 6 Final Confrontation
The protagonist sits, visibly exhausted, eyes red. He pulls the sheet off the bathroom mirror.

Reflection:
“It’s time.”

The reflection reaches out not into the glass, but out of it. Fingers pierce through, shattering the mirror like water. The screen cuts to black.

End Screen Suggestion:

“Have you ever looked in the mirror… and seen something that wasn’t you?”
(fade to black, eerie mirror shattering sound)

Anyone would try this and send the results?

[covert copper]

I also encountered the problem of no sound when I downloaded the upscaling video. Could anyone give me some suggestions or solutions to deal with this problem? Thank you so much!

[lost trout]

hello when i create any video and add a voice using veo 3 I see subtitles in strang letters. how you turn off subtitles

[blazing idol]

Just add no subtitles

[lost trout]

you may need to choose not fast or veo 2 you need high quillty veo 3

[lost trout]

I don't see this option , i write this in the prompt but no way

[lost trout]

this what happens to me

[lost trout]

do you know how to remove subtitles

[covert copper]

https://github.com/siruitang626/Google-Flow-Practive-Guidlines
This repo is where we keep all the takeaways and tips we've picked up while using Google Flow and Veo3. We've tried to summarize everything we’ve learned so far, and the content is available in both Chinese and English. Hopefully, you’ll find something useful here!

Our goal is to share our experience and get more people interested in chatting and learning together about how to get the most out of Google Flow. If you're up for a conversation, feel free to add us on WeChat or drop us an email!

[covert copper]

Thank you for your suggestion. I have tried to use high quality with veo3, but it still doesn't work. The video still has no sound when I download it😭

[fathom vine]

Hi, sorry for bothering. I’ve already subscribed to the Google One AI Premium plan, but I still don’t have any credits to use Flow. I’m not sure if this is a bug or if I simply need to wait longer for Google to recognize my subscription. Can someone please help me?

[short rock]

It's definitely hit or miss with audio. You are not alone.

[short rock]

No, me neither.

[humble veldt]

I've wasted a lot of credits on this as well. I have had the best success (not always) with this at the top and bottom of my prompt respectively:
IMPORTANT INSTRUCTION:
Do not include any captions, subtitles, closed captions, floating text, transcribed dialogue, or on-screen text of any kind. Output must be clean visual footage only.

FINAL NOTE … IMPORTANT DO NOT IGNORE:
Do not add any subtitles, dialogue text, captions, floating labels, or on-screen words—automated or otherwise. No text of any kind in the video.

[fading flicker]

I have seen a clip with them speaking japanese

[bold whale]

"How can I keep a character consistent in the next video while still generating audio?" 😩 😩

[bold whale]

The video no have sound :(( i was tryit many time :((

[boreal wind]

Hello what's the app name?

[hardy glade]

Why My account have no credit,i has buy a plan.

[karmic lynx]

Maybe someone can help me:
I tried to use Google Flow today but it does not work. I always get "You need more AI credits to complete this request".
I have just subscribed to Google One AI Pro, so I have not used any AI credits yet. Any idea someone?

[meager karma]

I have a similar problem. I can not press the "Get Startet" button at the very end.

[abstract shale]

I have found using Veo3 within the Gemini App instead of within Flow, stops subtitles from appearing. Not sure if that is just luck, or a thing, but worth a go.

[boreal wind]

if you are on pc press f11 this will turn your app in fullscreen and you will be able to access the button it worked for me

[abstract shale]

There is a reason you will not find many Veo3-produced videos posted that aren't just separate, loosely related 8-second clips, stitched together. Because to create anything that features any kind of consistent characters, sound, dialogue etc, currently using Veo3 is both challenging and expensive. Until Google actually gets the 'extend' feature and FirstFrame/LastFrame feature working with Veo3, it's going to remain extremely difficult to do what you're asking. But everyone is in the same boat currently

[abstract shale]

Are you downloading at 720, or upscaling to 1080? If you download at 1080 you will lose all audio. Google need to make this clearer.

[abstract shale]

Google needs to fix this issue. I have found using Veo3 within the Gemini App instead of within Flow, stops subtitles from appearing. Not sure if that is just luck, or a thing, but worth a go.

[abstract shale]

You cannot download upscaled 1080p generations and retain the sound. Google have acknowledged this is an issue/. But they need to make it clearer. Until they fix this issue, you have 2 options. Download both the 720 (with sound version) and the 1080 (without sound version) and use the sound from the 720 clip with your 1080 clip when you edit in your video editing app. OR just download the 720 clip, and upscale it in a third-party upscaling app, such as TOPAZ

[abstract shale]

Image to video does not fully work with Veo3. Google have acknowledged this, but they need to make it clearer.

[abstract shale]

Try generating in Gemini app instead of within Flow - works for me. But could just be luck.

[abstract shale]

Welcome to the world of AI generative video. (and not trying to be funny... it's just the way it is for everyone. Its very much like spinning a slot machine.... ). You can perhaps analyse FLOW TV, and see what prompts have been used to generate videos you like, and try to learn from that. OR check out the FLOW GALLERY channel in this discord, and see what prompts achieve what styles. But there is always that element of rolling the dice when you hit generate.

[meager karma]

How are you accessing flow? I can not access it. I can’t click on get started.

[abstract shale]

http://flow.google

[meager karma]

Thanks, but it doesn’t work... I can’t click get started at the bottom... I tried every browser (including Chrome)

[boreal wind]

I answered you before

[boreal wind]

you can check my answer right here

if you are on your phone this is a different situation

[meager karma]

Thanks this worked!

[abstract shale]

I don't see a 'Get started' button when I use that link - it just takes me straight into the flow app.

[ashen forge]

I ran into the same issue — it’s actually caused by a bug. There’s a privacy policy screen that you’re supposed to agree to before you can click “Get Started,” but sometimes it gets hidden behind the Veo3 welcome card.

Because of that, the button doesn’t respond.

[boreal wind]

no problem

[boreal wind]

when you subscribe you get a welcome message and you can't skip it without cliking on the get started

[abstract shale]

ahhh right - gotcha

[meager karma]

It seems like here are some experts... Appreantly it is my first time experementing with Flow and VEO3. I have a google one pro and have "just" 1000 credits for flow. But it seems like I can use veo3 through gemini "for free" but only for always 8 seconds.

My question: Is there any smart workaround you would recommend? I am trying to make one social media video to market my idea. I have pretty much no experience and would like to do it as smart as possible ... are there any recommendations, insights, learnings etc someone could share with me :)?

[meager karma]

Thanks! Is there any limitation using Veo3 through gemini...? Or can I create unlimitet?

[proven prawn]

20 videos per month,10per day however (out of the 20 you get)

[meager karma]

Oh so its as of now a very expensive way to experiment and learn it...

[haughty elbow]

Same here

[silent briar]

Hi hi! I'm running into multiple failed generations, and a note that says "Flow is currently unavailable, check back in later."

[harsh vortex]

Genuinely can't wait till this tech to be more accessible

[meager karma]

I see the point actually! It make sense to me.... the computing power is most likely amazingly high...

I dm'ed you bc I have some specific questions and I am hoping you could help. Can you please check your DM 🙂

[harsh vortex]

Yeah i run things in fast first to get an idea of what google thinks i want and it can look soo good then i change models to veo 3 and its not even close and much more inaccurate. Thanks for the advice but there is nothing like veo 3 right now.

[austere briar]

Is there a maximum character for a prompt ? For Veo3?

And do you know where i can go to learn the best way to prompt for Veo3 Flow ?

Ok thanks mate

[sly haven]

I gave it a 100,000+ character prompt just now and veo3 failed to generate. It worked in veo2 though but it didn't listen to the instructions at the very bottom of the prompt.

[sly haven]

That being said I have seen people prompt veo3 with 10+ paragraphs.

Veo2 also accepted a 1,006,951 character prompt. I'm pretty sure it ignores most of it though.

[north wyvern]

A dramatic low-angle shot in a dimly lit interrogation room with flickering overhead fluorescent lights. An older Indonesian woman with slicked-back hair and a leather jacket leans forward across the metal table, her face half-shadowed. Cigarette smoke coils in the air, creating a tense atmosphere. She slams her palm down and glares into the lens, voice cold and cutting: "hay Bangai,peu hana tu'oh pakek Veo 3.Makajih jak lom sikula betinggi tinggi" The camera holds steady as her steely eyes pierce through the haze, the only sound being the ticking of a clock on the wall.

[hollow mountain]

I am have trouble getting voices into my Veo 3 vids...Here is my prompt - Setting: Car ride to school, an elementary aged girl and her dad Sarah in backseat, Dad driving Sarah says, "Ms. Franke has bolts...... in her neck!" The Dad chuckles, then says "Just a fashion choice, sweetheart." Sarah throws hands up in exasperation.

[harsh vortex]

I had no idea it can do other languages

[fresh crescent]

So my prompts dont have any sound I get a video but no sound

[mild osprey]

make sure youre using veo3

its the only one with sound i think

[fresh crescent]

Ah I used flow ?

[mild osprey]

look at the ettings thing by the prompt

i would agree though this thing is hard to use and like 90% of the time doesnt work well let alone work at all

but i think it may be from using video generation limits for the day.

[solar hazel]

Hola

[pearl peak]

Has anyone had success getting two camera commands to work at once? I'm trying to get it to do an arc shot and zoom out at the same time, and it will only do the zoom out. Is there other terminology I need to use to get it to register what I want the camera to do in a shot?

I tried using natural language first 'The camera sweeps around' and this was ignored. So I switched to arc shot, but also ignored.

I was also just trying to get the camera to pan down, and that wouldn't work either -_-

If someone has a good guide on getting that part to behave, that would be great.

[past kestrel]

has anyone worked out how to maintain the same voice throughout? Ive tried telling it not to, but it sometimes changes to a mans/womens voice halfway through...

[formal tulip]

Is it true that only text-generated videos can have sound now? When veo3 was first released, I used pictures to generate videos and they also had sound.😩 😩

[quick saffron]

generate a video with the prompt from this file.

Generate a video with the prompt from this file.

[viscid citrus]

Result?

[bold sluice]

I'm having trouble with dialogue accuracy in Veo 3. Sometimes the wrong character ends up saying a line, or the roles get reversed. Has anyone found a reliable way to control who says what?

[quiet temple]

yo can someone make me this prompt

A young man in his 20s trying to do pushups alone in a small, dimly-lit apartment. He’s wearing a basic workout outfit. His arms tremble, face strains with effort. Camera angle is low and cinematic, emphasizing his struggle. Sweat drips, he collapses to the floor, breathing heavily. Light from his phone screen glows beside him. Realistic, emotional tone

Voiceover: “No coach. No gym. Just you… struggling alone.”
Text on screen: “There’s a better way.”

[heavy nexus]

A dark-skinned Egyptian young man wearing striped pajamas stands on the rooftop of a house in a slum area, holding a chicken in his hand, surrounded by a chicken coop, delivering a message in Egyptian Arabic: ‘انا مش عارف اجبهالك ازاي انا في رسالة عايز اوجهالك يصديقي كنت هنساها وانا بجري ورا الفراخ على السطح الرسالة دي من واحد اسمه عبدالرحمن بيطلب منك إياك تنساهم ’, then waving his hand.

[fading flicker]

Bugs bunny is teaching integral calculus. He does it in a cartoonish way where he teaches a concept, and then there is a little fight or dialogue. He should teach basic integration, u-substitution, and other techniques.

[heavy nexus]

Someone make this prompt plz

[north gate]

Modern superhero, "The Digital Guardian," sleek dark blue and silver high-tech suit, glowing circuit patterns. Majestic white cape with "RGPD" in gold stylized letters and a shield symbol on one side, "IA ACT" in gold stylized letters and a vigilant eye symbol on the other. Confident, protective stance. Futuristic city or abstract data stream background. Detailed digital painting, cinematic lighting, epic. --ar 16:9

[wicked yoke]

Anyone with a direct link to Veo3?

[elfin aspen]

I'm trying to understand the ideal format to put inside the veo3 but this video I made was with this prompt here: { "model": "veo-3.0-generate-preview", "prompt": "Ultra-detailed, dark sci-fi: a 24-year-old woman with raven-black hair pulled into a tight braid, high cheekbones and piercing green eyes, clad in reinforced matte-black dystopian armor etched with battle-scar Faint scorch marks and faded insignias hint at past conflicts. She stands partially concealed behind a towering chrome pillar in a vast Westworld-style robotics facility, its polished metal floors reflecting the soft hum of overhead holo-panels. transitions to a wide shot, panning horizontally to map the labyrinth of gleaming workstations and flickering data-screens. It then dollies in with a smooth, steady glide, closing to a medium shot as she shifts her weight onto her front foot. With a single deliberate movement—shoulders squared, eyes narrowing—she steps forward, sweeps her arm upward, and draws a razor-edged katana in one fluid motion, the blade’s edge catching a cold, electric-blue glint from overhead LEDs. In an extreme close-up, we see the fine quiver of her lips as she inhales sharply; the metallic rasp of her breath is almost audible. She exhales and, in a low, menacing whisper, declares, “humans will pay.” At that instant, a classical dramatic score swells subtly, then slams to silence on her final word, leaving only the distant mechanical hum and her steady heartbeat echoing through the frame.", "negativePrompt": "warm colors (red, orange), vegetation, direct sunlight, cartoon style, medieval elements, mood, high saturation, exaggerated particles", "aspectRatio": "9:16", "durationSeconds": 8, "personGeneration": "allow_adult", "seed": 12345, "enhance_prompt": true }

I want to train a gpt or some file to transform my commands the right way so I can make the videos. Has anyone done this and can help me?

[scarlet citrus]

Person using sword

Modern superhero, "The Digital Guardian," sleek dark blue and silver high-tech suit, glowing circuit patterns. Majestic white cape with "RGPD" in gold stylized letters and a shield symbol on one side, "IA ACT" in gold stylized letters and a vigilant eye symbol on the other. Confident, protective stance. Futuristic city or abstract data stream background. Detailed digital painting, cinematic lighting, epic. --ar 16:9

[austere briar]

The shorter the prompt the best result you get ?

[queen saddle]

We can create the videos just putting the prompt here?

[fair hatch]

No

[queen saddle]

Where we can use veo 2 or veo 3?

[queen saddle]

Thank you

[twin cedar]

Ultra-realistic POV shot of a Chinese businessman standing on a busy street in Beijing, near the Tiananmen Gate. The background clearly shows traditional Chinese architecture, red flags, and Mandarin signage. It's daytime with natural sunlight casting soft shadows. The man is wearing a clean, modern dark suit and tie.

He is holding the camera with one hand (as if recording himself), and a smartphone in the other. The perspective is immersive, handheld and slightly shaky to simulate real human movement. The phone emits repeated coin-like sound effects — 'tiquin! tiquin!' — similar to cash or mobile payment notifications.

The man looks down at his phone with a mocking smile and slightly lowered head, then speaks directly to the camera in Brazilian Portuguese, using a thick Chinese accent. He says, in a sarcastic and playful tone:

'Isso ai, Blasileilo bulo… continua depositanto achando que o tigrinho vai pagááá.'

His facial expression is confident and teasing — smiling subtly, eyes squinting, lips curled slightly upward. Skin texture is realistic, with accurate human proportions and detailed facial animation.

Cinematic depth of field, shallow focus on face and hands, background slightly blurred. Lighting is soft and natural. Rendered in 4K resolution.

Important: Do not add subtitles — the video should be clean, with no on-screen text or captions.

Ultra-realistic cinematic POV shot filmed from the perspective of a young soldier in an ancient war setting (pre-modern era, no modern tech). The camera is handheld, slightly shaking like a GoPro strapped to the soldier’s chest or held in one hand while he runs. Explosions and debris fill the battlefield, with arrows, smoke, and yelling soldiers in the background. He runs through trenches and broken terrain, dodging chaos.

The young man is sweaty, has dirt on his face, expressive eyes, and breathes heavily. His mouth moves naturally as he speaks. His left hand is holding a primitive-style pistol, and his right hand holds the POV camera. The atmosphere is tense but immersive, with a cinematic look — 4K resolution, motion blur from running, and rich lighting with dynamic shadows from fire and smoke.

In natural Brazilian Portuguese, spoken with a strong carioca accent, the soldier says while running, slightly laughing in disbelief:

'Caralhoo muleque, isso aqui tá parecendo Madureira!'

Facial expressions are vivid — eyes wide, mouth open mid-sentence, face reacting to explosions and danger. Sweat drips, dirt flies, background has blurred soldiers fighting.

No subtitles or text on screen. Natural audio of battlefield — distant yells, fire crackles, footsteps, all mixed beneath the voice.

[twin cedar]

Ultra-realistic video in cinematic quality, filmed with a handheld camera from a slightly low angle (as if filmed by a camera operator standing on snow). A woman stands in the middle of a snowy, frozen landscape during the late afternoon with soft golden light illuminating the scene. The snow reflects ambient light realistically, and the sky is overcast with slight orange tones from the setting sun. Cold breath visibly escapes from her mouth as she speaks.

The woman is in her 30s, light-skinned, wearing a red winter Coca-Cola branded jacket, knit gloves, and a fur-lined hood. She holds a realistic glass bottle of Coca-Cola in one hand, condensation visible on the bottle. Her facial expressions are warm and charismatic, like a professional TV reporter or commercial actress.

She smiles confidently and speaks directly to the camera in Brazilian Portuguese (spoken clearly, naturally, and cheerfully — no subtitles):
“Tá com frio? Então imagina isso aqui bem gelada… Coca-Cola combina com qualquer clima!”

As she finishes speaking, two polar bears (realistic, non-threatening, calmly walking) appear in the background walking across the snowy field behind her. They are distant enough to not pose danger, but close enough to be visible and majestic.

[past kestrel]

Right, so you recommend 4 at a time? I also have problems with subtitles randomly appearing

[placid willow]

prompt should use 120 - 180 words better?

[restive prawn]

The sequence opens with a cinematic beauty shot of a young woman with defined curly dark brown hair and porcelain-like fair skin, styled with a modern, fancy elegance reminiscent of a youthful Dua Lipa, but with her own natural freshness. She stands confidently against a seamless, infinite background featuring a smooth, modern gradient — transitioning from soft sky blue at the top to warm, pale yellow below — creating a serene and luminous ambiance. Her expression is calm, serene, and empowered. The lighting is beauty-grade soft, diffused evenly across her face and body, enhancing the smoothness of her skin and the texture of her defined curls without harsh shadows.
As the camera holds on her, shimmering flower petals with solid, vibrant colors based on the Revlon Professional palette — warm reds, pinks, oranges, golds — begin to materialize behind her. The petals are large, soft-edged, and float gracefully, moving in slow motion with a fluid, airy elegance as if carried by a gentle breeze. The motion is organic and soothing, adding life and vibrancy to the composition.
Gradually, the petals swirl forward, enveloping her in a soft, colorful vortex. As the density of petals increases, she is slowly and gently obscured from view. Once she is fully surrounded, the petals coalesce smoothly, transitioning into a luminous, multicolored halo hovering in the air against the gradient background.
The camera then follows the halo as it drifts slowly to reveal the woman again, now joined by a young boy of about seven years old — who visually matches her, with fair skin and curly dark brown hair. He is dressed in casual, simple clothes: denim jeans and a plain white t-shirt. The boy approaches the woman with a gentle smile. She reaches down, tenderly caressing his hair, and they share a moment of pure, loving complicity.

[whole tusk]

what the heck is "divergence" can anyone explain?

[keen dew]

An UFO over NYC

Draw an UFO over NYC

[swift pulsar]

Eu copio e colo seu prompt e nao vai com audio (Veo3), sabe o que pode ser?

[twin cedar]

Pode ser as aspas na hora de pedir a fala, peça para o chat gpt mudar isso

E explica pro chatgpt que ta saindo sem som tbm ajuda ele arruma

[swift pulsar]

Voce só digitou aquele prompt no Highest Quality e foi, ou fez outra coisa? Vou tentar aqui, no meu aparecem legendas mas nao aparece som.

[hushed stream]

Ta bugado, o meu to fazendo aqui ja faz mo cota, uns sai com audio, outros não

[wet verge]

Can't seem to get a dialoge properly assigned to the people in the video. Any suggestions?

Prompt:

Setting: Bustling city sidewalk, lively daytime atmosphere, people cheerfully walking by, festive decorations visible. Reporter enthusiastically interviewing two excited pedestrians who humorously share their enthusiasm for Elena’s birthday. Visual Prompt (for AI): "A vibrant street interview scene, with a smiling, energetic reporter holding a microphone speaking to two humorous, animated pedestrians who are clearly enjoying the festive mood. Surrounding them, bystanders wave small flags and balloons, celebrating Elena’s birthday enthusiastically." Dialogue (8 seconds): Reporter (smiling): "Any special message for Elena today?" Pedestrian 1 (playfully dramatic): "Elena, you’re a legend! Can we have your autograph?" Pedestrian 2 (laughing, jumping in): "Yeah! Please? It’s for my dog!"

[hard gulch]

no vol 😦

[ancient sand]

Hi

[gusty marten]

A triangle park in the middle each side is a road and a boy is playing in the road with toys

[digital pine]

I'm not expert but the prompt seems good. How many times have you rendered?

I cannot for the life of me figure out what is triggering issues with content policy in this prompt. If I cut off everything after "What did you want to show me?" it's fine. Otherwise I get Failed Generation with a content warning:

A close-up shot inside an empty rectangular Renaissance chapel Straight walls lined with painted panels, and classical niches. Shot positioned just off-center, at shoulder height. The space is still and echoing, lit by narrow, high windows. A soft shaft of sunlight crosses the stone wall behind the two figures, revealing swirls of dust in the air. The camera captures both men from the side, in profile, as they face one another near the center of the chapel’s long aisle. They stand atop an ornate black-and-white patterned marble floor, beneath the unpainted barrel-vaulted ceiling, painted deep blue and dotted sparsely with gold stars.

The Patron, an older man, about 65 years old, stands thin and hunched, wearing a long, faded red ceremonial robe with gold trim, a tall crimson hat, and white gloves. A silver-topped sceptre rests in his right hand, which he leans on slightly as if it steadies him. His bushy white beard is unkempt, and his face is deeply lined, his eyes sharp with irritation. His body seems fragile, but he radiates an impatient, commanding energy.

Beside him stands, the Artist, a rugged man of about 33, broad-shouldered and weathered, wearing a plum-colored tunic and tall, slouched leather boots. A canvas satchel hangs diagonally across his chest. His long, dark hair is unkempt, hanging just past his ears, and he wears a dense, inch-long beard. His flattened nose, crooked slightly to one side, gives his face a hardened look. He seems guarded — tense, as though uncomfortable in the grandeur of the space.

The Artist glances toward the older man and mutters confusedly:

“What did you want to show me?”

The older man taps his sceptre on the floor. The metallic tap echoes in the stillness. The artist flinches. The Patron in a raspy, irritated voice, growls:

"Don't you see, Mickie!?"

The Patron continues, looking upward and raising his arms above his head. "This ceiling!"

[languid finch]

I have a question

if I wanted to a make a person in a picture sing a scat song. like the jazz with all the humming and beep bops or whatever. how would you write that out?

I only got a 100 credits left.

lol

[shy drift]

how to save with sound

[languid finch]

somebody hook me up with a prompt

how to save with a sound what does that mean

[shy drift]

on site its with sound but when i save it there is silence

[languid finch]

No I'm sorry thats not what I'm asking although the sound does seem to be hit or miss

what I'm asking is how would you write out a prompt to make someone "scat" a jazzy song

like the scatman

ya think its even possible?

[shy drift]

ahhh

idk honestly

[languid finch]

its cool im reading some of these other prompts

[lapis ermine]

try asking gemini

thats what I do when I cant come up with the words for an idea

also u can use veo 3 within gemini to make the clip without using ur credits

[languid finch]

Thats genius.

[languid finch]

but I can't use pics as the base?

[lapis ermine]

ah yea no

its text based

[languid finch]

ok

i like the image stuff but ok

im out of credits anyway so this is good that i can at least make text based

its unlimited?

[lapis ermine]

i mean its just for when u dont wanna use credits

no

theres a limit I forgot what it was

I went a day making like 10 videos

before seeing limit warning

[languid finch]

Runway I have to look in to that

[lapis ermine]

yeah runway is also good

[languid finch]

Is it censored to hell like flo?

is it free?

10 seems like a really small count

[languid finch]

cool

I understand

[digital pine]

FWIW I'm noticing the outputs from Gemini are pretty lackluster compared to Veo 3. But weirdly it will generate prompts that Veo 3 refuses to in Flow.

[lapis ermine]

most likely bc they use different moderation layers

I think gemini is less sensitive compared to flow and the others

[digital pine]

@lapis ermine Any tips for character continuity from shot to shot within Veo 3?

[elfin aspen]

Ultra-realistic sci-fi: a weary android in chrome armor stands in a rain-soaked alley at dusk, backlit by neon signage. Close-up with a 50mm lens from a low angle, rain droplets trickling down her face. She tightens her grip on a blood-stained katana. Slow dolly-in, depth of field blurring the neon city behind. Soft ambient synth, distant thunder, her breath trembling. Negative prompt: cartoon, bright colors, text, subtitles, logo. Rhythm: tense and slow, 8 seconds.

[narrow musk]

And it's 10 again after the two/three days?

[digital pine]

A close-up shot inside a quiet, high-ceilinged rectangular chapel. The camera is shoulder-height. It's such a close shot you can barely make out that the stone walls are lined with painted panels and classical niches.

Two figures stand facing one another. On the left of the frame we're viewing over the right shoulder of the old man, around 65, dressed in a faded red ceremonial robe with gold trim, a tall crimson hat, and white gloves. A silver-topped sceptre rests lightly in his hand. His bushy white beard and deeply lined face give him a frail but commanding presence. His expression is sharp, tinged with impatience.

We see over the old man's right shoulder the other figure, a rugged man in his early thirties — broad-shouldered, with a weathered plum tunic and tall leather boots. A canvas satchel crosses his chest. His long, unkempt dark hair frames a dense beard and a crooked, though normal sized, flattened nose. He looks guarded and slightly out of place.

The younger man glances toward the elder and asks quietly:

“What did you want to show me?”

The Patron in a raspy voice exclaims:

"Don't you see, Mickie?"

The Patron continues, looking upward and raising his arms above his head, still holding the sceptre. "This ceiling."

[lapis ermine]

Repeat key traits or outfit descriptors exactly in every prompt

Use consistent framing or setting language to reduce drift bc it can sometimes act up

And keep the prompt length short and focused so there’s less room for Veo to improvise

[slate saddle]

Making a video based on a photo
Pramet:
I want this person to come in front of the camera and say, "Don't worry, I'll bring a car."
Cinematic shot and lighting, and completely with special effects.

ساخت ویدو بر اساس عکس
پرامت :
میخوام این شخص بیاد جلو دوربین و بگه نگران نباش من ماشین میارم
پلان سینماتیک و نور پردازی و کاملا با جلوه های ویژه

[digital pine]

I see, looks like when using images it struggles to produce audio. So best to do a text to video prompt to generate audio and mix in Final Cut

[warped warren]

dimly lit stage; dingy bar. a man rhymes bountifully with a sharp flow and even tempo

[vestal seal]

Fun account to follow. Google DeepMind’s Greenfield team has started an X acct. Almost every example you have ever seen Google share for Veo or Imagen come from these amazing folks.

https://x.com/gdmgreenfield?s=21

GDM Greenfield (@GDMGreenfield) on X

Explorations with Google DeepMind generative models

X

[river arch]

incase y'all are looking for some shots references:

1. PAN (Left / Right / Up / Down)
   Horizontal or vertical pivot of the camera from a fixed position

2. Low Angle Shot
   Camera looks up at the subject, emphasizing power or intensity

3. WHIP PAN
   A fast, motion-blurred pan between subjects or scenes, often used for dramatic or comedic effect

4. WIDE SHOT
   Establishes setting and scale, shows subject in full within their environment

5. Cowboy Shot
   Frames the subject from mid-thigh up, often used in confrontational or cinematic contexts

6. Over-the-Shoulder Shot (OTS)
   Camera peers over a character’s shoulder, used in dialogue or perspective-driven scenes

7. POV Shot
   Represents what a character is seeing, places viewer in their shoes

8. Push-In
   Camera slowly moves toward the subject to build intensity or focus

9. Pull-Out
   Camera withdraws from the subject to create emotional distance or reveal context

10. CRASH Zoom
    A sudden, rapid zoom in or out used for shock or emphasis

11. Vertigo Effect (Dolly Zoom)
    Simultaneous zoom and dolly in the opposite direction to distort depth and create tension

12. Camera Roll
    Rotates around the lens axis, often matches or contrasts with character movement or disorientation

13. Tracking Shot
    Follows the subject’s movement, usually forward or backward

14. Trucking Shot
    Camera moves left or right alongside the subject using lateral movement

15. Arc Shot
    Camera circles around the subject using dynamic horizontal or vertical motion

16. Motion Blur
    Can be applied to fast camera movements like whip pan or crash zoom for cinematic effect

17. Handheld/Shake Effect
    Adds realism or chaos, can be subtle or dramatic depending on tone

18. Rule of Thirds (Composition)
    Subject is aligned along gridlines to create balanced, visually pleasing frames

[placid willow]

any tips to generate prompts?

[winter perch]

Do you have tips to make videos for perfumes, cinematic showing different scents etc. without actually showing a bottle. Whatever I prompt it looks odd, and nothing like i expect it to

[lime oyster]

noiceee

[swift pulsar]

How do I donwload with low resolution, I'm not getting the audio when I download

[worn cedar]

use an llm

[winter perch]

Didnt do any good, already tried that before

[paper bobcat]

countryside

[worn cedar]

but whats the prompt for the llm ?

I will give you a core idea, possibly with an image. You'll refine this by adding keywords and modifiers, creating a detailed Veo 2 prompt. Good prompts are descriptive and clear.
Your prompt should incorporate these elements naturally, without lists:
The subject is the object, person, animal, or scenery. The context is the background or setting. The action describes what the subject is doing. The style can be general or very specific, like a horror film or cartoon render. Optionally, include camera motion, detailing what the camera is doing. Optionally, specify composition, which is how the shot is framed. Optionally, define the ambiance, explaining how color and light contribute to the scene.
Prompt Writing Tips
These tips will help you write prompts that generate your videos:
Avoid lists. Use descriptive language with adjectives and adverbs to paint a clear picture for Veo. Provide context when necessary. Reference specific artistic styles if you have a particular aesthetic in mind, describing them without using proper names. Enhance facial details by specifying them as a focus, for example, using "portrait."
Negative Prompt
You will also provide a negative prompt, a powerful tool to specify elements to exclude. Describe what you want to discourage the model from generating. Do not use instructional language like "no" or "don't." Instead, describe what you don't want to see. For example, "wall, frame" means you don't want a wall or a frame in the video.
Finally, present both the positive and negative prompts in two separate, plain text boxes, with no special characters.

[worn cedar]

whta are the scents you want to represent? what mood? night, day, misty, clear… natural, design…

[slim mason]

Hi people. I'm Luciano from Brazil.
I'm tryint to create a video in Flow with portuguese narration. All my prompt are in english but this phrase is in portuguese, and the Flow says that only English commands are avaliable... how can I generate a different narration ?

This is my prompt

Create an 8-second horizontal video featuring a 50-year-old man who looks similar to Kevin Spacey but with very short hair. He is wearing a red hoodie and blue jeans. The background shows Av. Paulista in São Paulo, Brazil. The man is holding a smartphone in one hand as if recording a selfie video. He looks worried and says the following phrase in Portuguese from Brazil: "É galera, esse negócio de ficar o dia todo no linkedin procurando vaga deixa o cara doidão. É RH que não dá retorno, é PJ pagando menos que CLT... e os boletos não param... mas é isso... compartilha esse vídeo pra me ajudar vai...."

[vague spear]

https://x.com/hussamkazim77/status/1928865203556417819

Hussam Muhammad Kazim (@hussamkazim77)

Model: VEO 3

Prompt:

The camera hugs the ground, shadowing a focused ant dragging an oversized crumb. It threads its way through towering blades of grass, crosses a wide expanse of concrete, and slips into a thin crack along the house’s foundation.

#Veo3 #googleflow #aivideo

X · May 31, 2025, 5:24 PM

[short quartz]

what are these people using that make the videos of different celebs doing crazy stuff?

like will smith pasta lol

[stark sphinx]

Veo 2 reference feature Is working now?

[slim mason]

If I remove the “horizontal” and Kevin space instructions the error are the same

[slim mason]

Here is the error

[median cedar]

#flow-tips-n-tricks・✏️ 8 second, Veo 3 - A war-torn battlefield in Lanka. The sky is painted with orange and gold hues of the setting sun. Broken chariots, scattered weapons, and ashes of war surround the space. Lord Rama stands still amidst the silence. The breeze carries the echoes of a great war just ended. Hanuman stands in front of him with folded hands.

Camera: Wide shot slowly zooming in. Subtle dolly-in movement. Golden-hour lighting bathing both characters in soft divinity.

Sound: Faint wind, distant echoes of conches and emotional orchestral underscore.

Dialogue (Hanuman, humbly):
"प्रभु, मुझे बस इतना वरदान दीजिए... कि जब-जब आपका नाम लूँ, मेरे रोम-रोम में आपकी भक्ति उतर आए।"

Styling tags: cinematic, golden tones, 4K ultra HD, emotional depth, epic mythological atmosphere, high detail.

[willow scroll]

With Pro?

[narrow musk]

Yes but it was a question lol

[willow scroll]

Oh lol. I didn't realize, not I gotta slow my roll. That actually really sucks tho.

[obsidian sundial]

how do i promt

[narrow musk]

Ok so maybe we get a little extra

Good to know, I will try, it's good if they give more
There are some glitches at the start of the service for now

It's ok if in the meantime we get credits to test it more so it's a win win, but yeah I agree with u

[worn cedar]

https://youtu.be/1lktT4dVAT4?si=e1el6g25i6BGpUOW

[bleak fox]

https://imclient.herokuapp.com/p2pdrop/

[obsidian sundial]

#flow-tips-n-tricks・✏️ 8 second, Veo 3 - A war-torn battlefield in Lanka. The sky is painted with orange and gold hues of the setting sun. Broken chariots, scattered weapons, and ashes of war surround the space. Lord Rama stands still amidst the silence. The breeze carries the echoes of a great war just ended. Hanuman stands in front of him with folded hands.

Camera: Wide shot slowly zooming in. Subtle dolly-in movement. Golden-hour lighting bathing both characters in soft divinity.

Sound: Faint wind, distant echoes of conches and emotional orchestral underscore.

Dialogue (Hanuman, humbly):
"प्रभु, मुझे बस इतना वरदान दीजिए... कि जब-जब आपका नाम लूँ, मेरे रोम-रोम में आपकी भक्ति उतर आए।"

Styling tags: cinematic, golden tones, 4K ultra HD, emotional depth, epic mythological atmosphere, high detail.

[flint talon]

A hyper-realistic depiction of the character He-Man wearing a modern office uniform — dress shirt slightly wrinkled, striped tie loosened, and an ID badge that reads "Heitor Man, Junior Analyst". He is hilariously hiding under an office desk cluttered with coffee mugs, snack wrappers, and scattered reports. His muscular frame barely fits under the desk, causing one leg of the table to lift slightly. He-Man is secretly browsing his smartphone, trying not to laugh. A half-eaten sandwich is stuck in his utility belt. In a speech bubble, he says in Brazilian Portuguese: “escondam se bem do seu chefe no trabalho, como ele sempre diz, um bom funcionario é dificil de se achar.” The scene is brightly lit like an office at midday, and in the background, you can see the shadow of a boss figure looking around suspiciously.

[flint drift]

A slow dolly shot through an abandoned subway tunnel, flickering lights overhead.

#flow-tips-n-tricks・✏️ A slow dolly shot through an abandoned subway tunnel, flickering lights overhead.

[placid willow]

why all time text-to-video i got a flat color? it's mistake when prompting?

[astral verge]

Is flow acting weird for anyone else? I can't get it to start for some reason

[tidal vigil]

Veo3 video Yemeni girl talk to sheep

[knotty thunder]

#flow-tips-n-tricks・✏️ 8sec video , hot lady from russia walking into house

[atomic nymph]

i can't for the life of me figure out why Google Flow won't let me upload this image as my start image for Frames to Video. The ai says it supposedly may "violate" our policies. two people in an office looking at each other. it literally doesn't get more innocent than this. What policies are we violating here?

[raven turtle]

#flow-tips-n-tricks・✏️

Create a long, immersive motovlog video in 4K quality, set on a dry, sunny day along Route 66, with wide open roads, desert scenery, and some mountains in the background. A 17-year-old Brazilian motorcyclist named Thiago is riding a high-displacement sportbike (such as a Yamaha R1 or similar). He is filming himself with a helmet cam and occasionally with an external camera for cinematic shots.

He speaks fluently in Brazilian Portuguese. At the beginning of the video, while riding, he says:

“Boa tarde, rapaziada! Eu sou o Thiago, tenho 17 anos, estou começando mais um motovlog. Estou aqui fazendo uma viagem para o Chile. Hoje está um dia bem quente, estamos há algumas horas já na estrada. Vamos fazer uma parada aqui.”

Thiago pulls over safely to the side of the road. He gets off the motorcycle, takes off his helmet, and uses a handheld camera or drone to film the landscape (desert, mountains, and road), then films his motorcycle from different angles.

He says:

“Vamos tomar um pouco de água aqui, galera, porque ninguém é de ferro.”
(he laughs naturally)
He grabs a bottle of water from his backpack, drinks it, and refreshes himself a bit.

Then he says:

“Vamos lá, galera.”

He puts his helmet back on, gets on the motorcycle, and continues riding down Route 66. The rest of the video includes cinematic riding footage with occasional commentary in Portuguese, desert landscapes, and long stretches of highway. Add ambient road noise, wind, engine sounds, and occasional relaxing background music. Include drone shots, GoPro-style helmet views, and camera angles from behind the bike. The video should be as long and immersive as possible.

Important: All spoken language must be in Brazilian Portuguese, and the environment should clearly show a hot, dry climate with a wide-open blue sky and scenic mountain views.

[wooden mauve]

Cinematic scene

[river arch]

I'm working on an experiment, trying to make videos with whatever output flow gives me, sort of like testing out the creativity :

Prompt:
A simple underground stand-up comedy setting with dim, moody lighting. A solo female performer stands under a soft spotlight, holding a mic. The camera captures her from a side angle in close-up, emphasising her delivery and expressions. The background is minimal, featuring exposed brick or a dark curtain, giving it an intimate, open mic feel. She speaks with dry confidence and subtle sarcasm. She says: People say crypto’s a cult. Yeah? So’s your bank (quiet for one second). She continues with the delivery: The only difference? My cult lets me leave with my money. The audience reacts with a few gasps at first and laughs from the corner. She pauses to let the audience stop, smirks, looks down slightly, lets the silence hang for one second, then deadpans: Also, we have memes while smile laughing, DO NOT HAVE SUBTITLES IN VIDEO.

Pro tip:
DO NOT HAVE SUBTITLES IN VIDEO - will remove it

Follow along : https://x.com/itscharann while I create Super Bowl-worthy ads with Google Flow, I deconstruct my flows often as well 🙂

Charan (@itscharann) on X

Growth guy @hypepartners @hemi_xyz
Prev work : @babylonlabs_io @optimism @wmtoken @FigureMarkets

X

[bleak fox]

https://myaccount.google.com/inactive?utm_source=sign_in_no_continue

[knotty thunder]

🇺🇦🇺🇦💪💪😎😎😈😈🗿🗿🗿🗿🔥

@bleak fox

@bleak fox

[umbral abyss]

Working on a scene in Veo 3 where a male voiceover introduces a retro burger — “This is our all-new 1985 Burger” — then the female character steps out of the restaurant holding the burger and says “Wallah. It hits different.”

I’ve uploaded custom ingredients for the storefront, burger, and character, and wrote a detailed prompt to guide realism — grounded movement, 80s synthwave tone, and a Kuwaiti nighttime setting.

Here’s the outcome I’m trying to get:

She opens the yellow door, steps out naturally, holding the burger. She walks a few grounded steps, smiles calmly, and delivers the line. Camera is dolly-in, steady. Movement is physical — no float, no teleport, no cuts.

⸻

The issues I’m running into:
• Subtitles keep appearing, even though I’ve clearly instructed no subs in the prompt.
• The storefront gets replaced with a generic diner or something completely unrelated — even with an exact reference image uploaded.
• The character floats or glitches through the door instead of walking out with believable movement.
• The burger looks wrong, even though I’ve uploaded a specific image — it keeps defaulting to something stylized or commercial-looking.

Anyone cracked how to fully lock Veo 3 into respecting reference visuals, removing subs, and grounding character motion?

[umbral abyss]

Output for reference

[swift pulsar]

So I took a photo of myself one the sofa and simulated a talkshow with Einstein, still not able to get audio into this. Hope this feat gets in soon

[umbral abyss]

are you changing the settings to Highest quality (Experimental Audio)?

[swift pulsar]

Did it, but if you use a frame they send you to Veo 2.

[swift pulsar]

How did you manage to use Highest quality - Veo3 and keep the same character in diferent videos? I keep getting to Veo 2 when I try.

[prime bane]

I’m having difficulty getting any type of character fidelity between clips, ,can anyone point me at a series of prompts in veo3 that maintain a character throughout the scene or is that only available with “ingredients + prompt 2 video” (veo2?) or prompt + last frame?

[autumn fog]

Make a drawing video

[pseudo wyvern]

I think most examples I’ve seen have provided the character in the image frame or used things like masks and generic clothes it’s been an issue for me as well

[prime bane]

gotcha, so there’s not a veo3 solution yet i guess beyond re-explaining your character every prompt and hoping for the best

still neat i guess i’ll just do the share last frame stuff to keep the party going

[zinc spear]

hi everyone, can someone help me creating a prompt? i want to create a video in flow of a person holding a camera running. just like a selfie video. do you have sample prompt for this?

[hallow skiff]

Hi, i dont cant help you sobre crear a propmt xd

[zinc spear]

Oh ok then 😢

[sage coyote]

i have a liar question, can we use some vpn to access the flow in not supported country?

[spring ember]

This instruction is very comprahensive: https://www.youtube.com/watch?v=ItJpzC5pOaI

[spring ember]

Wow! This is craaaazy!

[solemn arrow]

A simple underground stand-up comedy setting with dim, moody lighting. A solo male performer stands under a soft spotlight, holding a mic. The camera captures he from a side angle in close-up, emphasising he delivery and expressions. The background is minimal, featuring exposed brick or a dark curtain, giving it an intimate, open mic feel. he speaks with dry confidence and subtle sarcasm. he says: People say crypto’s a cult. Yeah? So’s your bank (quiet for one second). he continues with the delivery: The only difference? My cult lets me leave with my money. The audience reacts with a few gasps at first and laughs from the corner. he pauses to let the audience stop, smirks, looks down slightly, lets the silence hang for one second, then deadpans: Also, we have memes while smile laughing

[slender crater]

Today I learned that Veo 3 is so good that my 30+ line prompts weren't working precisely because it's realistic by default. I switched to a 2 line one (literally) and got what I needed.

[lone totem]

Why don t you have the sound ?

[neon kelp]

Create a video of a soccer player kicking the ball inside a packed stadium with fans wearing green shirts and flags.

help

[earnest cedar]

sure

[earnest cedar]

hmm

[earnest cedar]

Generate cinematic, photorealistic 4K video clip.

The scene opens with a medium shot of a soccer player running towards the ball. The camera tracks alongside him. He strikes the ball powerfully, and at the exact moment of impact, the scene shifts to slow-motion for 3 seconds, highlighting the ball deforming and grass flying up from the pitch.

After the kick, the camera returns to normal speed and performs a fast pan to follow the ball's trajectory towards the goal.

The setting is a packed stadium at night under bright lights, with a roaring crowd wearing green shirts and waving flags. The atmosphere is intense, with green and white confetti falling slowly during the slow-motion segment

[zinc spear]

😍 can you help me create a prompt just like a point of view of someone holding a camera, running from or jogging in a park.

[earnest cedar]

ok, wait a minute

[earnest cedar]

Generate a cinematic and photorealistic 4K video clip with an energetic and motivational tone. The perspective is first-person (POV), simulating the view of someone on a light, steady jog around a lake in an urban park. The camera movement should be smooth and rhythmic. The scene is set at dawn, with sunlight glistening on the lake's surface and creating subtle lens flares. Colors should be vivid and saturated. The audio should focus on ambient sounds: the rhythmic sound of footsteps on the pavement, the runner's controlled breathing, and the natural sounds of the park around the lake.

[flat crescent]

When I create a series of Veo3 videos in the same "project" on Flow, and part of the prompt describing the scene and character is the same, then there is very high probability of consistent characters. But it this behavior documented somehow? Do we know how we can (more-less) reliably use this consistency?

[slow token]

Hey folks, I was wondering. In most cases when I create a prompt, the system is not following the prompt, when it comes to sounds. Instead there is no sound at all

Here is the prompt I was aiming for. It has nothing to do with North Korea. Also if I replace it with south Korea, still no sound. What am I doing wrong?
North Korean news broadcast: female anchor in traditional hanbok, solemn expression, announcing in Korean: "The biggest fan of our great leader celebrates his birthday." Cut to Pjöngyan Elementary School: uniformed children presenting handmade treats to a visitor, symbolizing friendship. Bright lighting, patriotic background imagery, consistent with North Korean state TV.

Sound: Anchor's voice is clear and authoritative. Transition to soft, uplifting instrumental music with subtle sounds of children in the school scene.

[real anvil]

Also try hovering over the previous video in the top right where is say, "Add to this video." It will give you an option to extend that video OR "jump to" another sceen. This is probably your best chance of a continued character.

[real anvil]

This worked for me, take a screen shot of the character in the previous video (maybe even cut to the face or body), add that screen shot as an "ingredient" (upload image) then give make a text prompt to go with the image.

[robust grotto]

I'm having great success with using frames to video / ingredients - it's carrying character continuity very well

[calm violet]

Does anyone know how to train VEO3 so that you're able to make yourself the actual subject in the video?

[solid chasm]

I ran into this and I think it has to do with the fact this may reproduce a likeness. In my case the likeness was from AI people so that made a bit less sense but that is my guess. I have zero data to back that up

[robust grotto]

Yes, you have to use reference images of you on all sides

[calm violet]

Ok thank you. I tried earlier I just wasn’t getting the desired results. I have to tweak my prompts probably.

[robust grotto]

This is one of those things that probably takes a lot of experimenting. Ensure that you're using consistent full body shots in a well-lit space with a neutral background

[prime bane]

This video shows how to use AI image generation stuff for character fedelity, which, I get, I've used open source stuff. What I'm asking about was if there's a way to achieve consistency in veo3 and it didn't seem like it.

[torpid orbit]

I am having a terrible time with dialogue and sequencing.

The wrong person says the dialogue and clearly sequenced events happen out of sequence.

I've tried giving names to each character and doing

(1) Sam says: ...
(2) Roy responds: ...

But it doesn't work. Has anyone gotten dialogue and sequencing to work well?

[steady lantern]

Is there a way to use my own words/script in veo3

[unreal oar]

"A cinematic and surreal scene begins with a man standing dramatically on the edge of a tall skyscraper rooftop at sunset. The camera circles around him in slow motion as wind blows through his clothes. Without hesitation, he jumps off the edge. As he falls through the sky, he spreads his arms like a superhero. But suddenly — BAM! — he slams face-first into a large, soft, cartoonish cloud, which wiggles from the impact. His expression freezes in pain and surprise. After bouncing off, he continues falling rapidly. The city below distorts and suddenly the ground opens up like a vortex. He crashes through the pavement with a thunderous impact — sparks and dust fly everywhere. Cut to black. Then, the scene resumes with a dramatic whoosh as he lands hard into the Nether dimension from Minecraft. The visual style abruptly shifts to blocky, pixelated graphics. Lava bubbles, ghasts float in the distance, and the sky is red and smoky. The character gets up groaning, looking around confused and terrified. Style: Starts hyper-realistic, then shifts to Minecraft-style pixel art after entering the Nether. Add exaggerated physics for comic effect. Background audio includes wind, a cartoony “boing” sound when hitting the cloud, breaking ground noise, and then Minecraft Nether ambience with lava and eerie echoing sounds."

[hasty goblet]

You can use veo3 in gemini pro for additional scenes too.

[river ibex]

Hey can anyone teach me how to write prompts

[robust grotto]

not at all a cheeky answer: use AI to help you hone prompts. It's intoxicatingly good at it. My experiential answer: be as clear and concise as you can but don't be afraid to over-prompt. You'll get the hang of it.

[foggy thistle]

A young Algerian man sitting on a folding chair on the beach of Oran, Algeria, staring indifferently at the horizon. In front of him, a beautiful young woman in stylish summer clothes is talking to him with enthusiasm, but he is uninterested. The scene is set during sunset with warm Mediterranean colors, gentle waves, and the city of Oran visible in the background — a relaxed Algerian beach atmosphere with subtle urban elements."

He speaks men and woman in Modern Standard Arabic Algeria(الفصحى) saying:
: "طريقي طويله بصح تكمل بالهنا"
"واه نيشان طريقك طويله بصح تكمل باللي قلنا"
جبديها جبديها هههههه‍

[ocean mirage]

Create a cinématique vidéo of a young african 9 years old look at the Sky. Behind him a africain village. Caméra zoom slowly and stop.

[river ibex]

Thank you

[cold bone]

This prompt keeps failing generation and I can't figure out why. Can anyone help? I used Gemini to write the prompt. Dawn's early light casts long shadows across a vast, windswept mountain training ground, a photorealistic high fantasy setting like Lord of the Rings, showcasing ancient structures, distant dragon roosts, and advanced dragon warfare. The cold, tense, and intimidating atmosphere is presented without on-screen text or captions. A sweeping aerial view first establishes the academy’s vastness amid jagged peaks. Then, the focus shifts to new recruit (25 years old, 5'6", resembling Sydney McLaughlin-Levrone), in dark training leathers, as she arrives; a medium shot captures her determined hazel eyes widening with apprehension and awe at the sheer scale, her fists unconsciously clenching. High above, the black dragon Typhon observes. Instructor (28 years old, 6'4", resembling Josh Heuston), sharp-featured and radiating formidable authority (though a hidden weariness flickers). He barks orders, his dominance emphasized as if viewed from a low angle. He says: "This academy forges warriors," and his voice echoes. He says: "Fail, and the mountains will claim you. Succeed, and you might just survive the war." He angrily says to a struggling recruit: "Pathetic! Again!" The recruit, exhausted yet defiant, completes a grueling obstacle, her determination clear in a close-up. Her eyes lock with the Instructor’s eyes in a fleeting, intense moment—his stern, observant expression also seen in close-up. The scene is immersive, with no music, relying on natural ambience: howling wind, distant dragon roars, grunts of exertion, clanking gear, and Kaelen's sharp commands.

[river ibex]

I can’t add the “scene” can anyone explain why? Is it my video is to hard to add to it?

[frigid peak]

Types of Conjunctions in English 3 - Correlative Conjunctions

Our video is presenting "correlative conjunctions" topic valuable information and we also try to cover the all subjects about it

👉 Correlative conjunctions subject is showcased in several YouTube video, but we tried to provide you the most effective info in a concise and also easy to understand video.

👉 Some of the most common correlative conjunctions are:

• both … and
• either … or
• just as … so
• neither … nor
• not … but
• not only … but also
• whether … or

[unreal oar]

"A cinematic anime-style scene set in a vast, ancient battlefield under a stormy sky. Reinhard van Astrea, the Sword Saint, stands solemnly before a massive stone pedestal where the legendary Dragon Sword Reid rests, its hilt glowing faintly. The atmosphere is tense, with dark clouds swirling and thunder rumbling in the distance. As Reinhard reaches for the sword, the ground begins to tremble subtly. His hand grips the hilt, and with a determined expression, he attempts to draw it. Initially resistant, the sword suddenly releases with a burst of radiant light, illuminating the surroundings. The moment the blade is unsheathed, a shockwave erupts, causing the earth to quake violently, and the sky to momentarily clear, revealing beams of sunlight piercing through. The camera captures Reinhard's flowing red cape and determined gaze as he holds the gleaming sword aloft. The environment reacts to the sword's power: trees sway, dust rises, and nearby rocks crack. The scene emphasizes the immense power and significance of the Dragon Sword Reid. Visual style: High-quality anime animation with dynamic lighting, detailed character designs, and dramatic camera angles. The color palette contrasts the dark stormy environment with the bright, radiant light of the sword. Audio: Intense orchestral score building up to the moment of unsheathing, followed by a powerful crescendo as the sword is drawn. Sound effects include rumbling thunder, the sharp sound of the sword being unsheathed, and the ensuing shockwave."

[stark sphinx]

We can now upload images to the ingredients feature??

[arctic tapir]

Hey everyone I am so new in video generation fields I'd love to create a thread for us to share our go-to prompt hacks.

If you've got any clever tricks up your sleeve, please share them in this thread! It would be a huge help for everyone.

[arctic tapir]

Huge thanks to @river arch for sharing this very helpful #flow-tips-n-tricks・✏️ message

[glass canopy]

Yea veo 3 flow no accepting no images in the frame to video.

[full cloud]

Olá

Hello!!!!

[quartz grove]

Guys, I got Gemini Pro and when I go on flow it gives me 0 credits as if I dont have the Pro version

I'm seriously confused, like has this happened with someone else too or is it only me?

[hasty goblet]

Ok, so to do my talk show scene with object consistency and full audio what I am doing is using ingredients to video and having it create a series of visuals that I need so I can pull the frames and use them as first frames where the characters go back and forth talking. In essence working around the stupid limitation. Snip snip snip in the editor. I think this will work.

[earnest cedar]

hmmm

weird

[bleak kayak]

can someone do a gameplay video of either a mario or sonic game via veo 3?

kinda curious on how that would turn out

[winter bough]

i cant get audio to work generating from frame

[robust grotto]

Anytime! let me know if you have any more questions, happy to help

[robust grotto]

Try it and see! My guess is it will interpret your request quite broadly instead of giving you literal gameplay. But would love to see what it does.

[vague spear]

Guys do anyone know how to create consistent character movies in Veo 3

[silent zenith]

I'm trying to figure out how to make a video like his, where the character holds a selfie stick and the POV looks natural. Also, why does my lip sync always feel off? Any tips?

my

[silent zenith]

Okay, I’ve solved this problem.

[spring ember]

What stops you from generating the images first and then turning them into videos with frame to video mode?

[alpine frost]

R3ynolds

[rancid iron]

I have been battle testing this. i have got so close. i am simulating memory for it to be consistent. this is the closest i have come. im making huge advancements with it

100% a.i. that aint me. but close'ish to what I do look like

please any comments are gladly accepted. i'm obsessed with making it better. I would love to help anyway I can

[unreal kiln]

share your prompt

[vague spear]

How you did it? It's amazing!

[spring ember]

omg this is so cool! How did you do that?

[barren cloak]

Character, action and setting consistency is really problematic. Can someone from G team confirm if the "extend" option will come out for VEO3 someday ?

[calm kiln]

Please someone help me run this prompt:

A man wearing white Hausa clothes and a black Hausa cap, he has a calm demeanor and looks very professional, sitting in a podcast studio with a neutral background and a podcast-style microphone, speaking in Hausa; saying; "wannan shi ne abinda Abdulrahim Sarki ya jima ya na gaya muku. A.I abin al'ajabi ne. Yanzu ni ba mutumin gaske ba ne. Ni AI ne." Then he laughs.

[robust grotto]

You gotta run your own prompts, my friend

[normal cosmos]

I for w/e reason can't figure out how to predictably get voice/dialog to work. Sometimes, flow will do it if I suggest a topic for the person to talk about, sometimes I can even say specifically what they should say, and sometimes it'll just throw subtitles.

Has anyone found a method to the madness?

And is there any way at all to 'force veo 3' each generation it kicks me back to veo2

[robust grotto]

The best performance I've had so far is by adding "no text elements" to the end of the prompt

but it will sometimes remove ALL text, even from things that aren't subtitles

[calm kiln]

I dont have it.

[fickle token]

yo

how do i like not make this like messy with the video

Create a cinematic anime trailer set in a dark, urban world. The protagonist is Jun Kwon, a quiet, pale-skinned Korean teen with white hair and blue eyes with albinism. Begin the trailer with dnb and dim visuals of Seoul at night—rain hitting windows, empty school corridors, and Jun staring out with his camera. Transition into a montage: flashes of bullying, Jun silently taking photos alone, cooking in a dark kitchen lit only by the fridge light. Show him sneaking out at night, spray-painting graffiti, snapping photos of fights and urban decay. The tone shifts—low synth and heartbeat bass build tension as he hides behind a dumpster during a brutal street fight under purple-lit streetlamps. Include: Jun capturing the final blow of a brawl on his camera. His terrified expression as more attackers appear. Him stepping into the light, trembling but defiant, yelling, “Who the hell invited the freak to the brawl!?” A sequence of him running through alleyways, chased, his breath fogging in the cold night air. Cut to a heated argument with his father, Rayang Kwon, a hardboiled detective. A scene of the camera being slammed down, fists clenched, and Jun bloodied on the floor. Add distorted audio of angry shouting fading into silence. The music softens—Jun walking down neon-lit streets, wounded but still holding his camera. He meets a girl and a bunch of people, and one tall person talking to his father from the fight who says softly, “Your eyes… they’re so bright up close.” A rare smile flickers on Jun’s face. Cut to final shots: The city of Cheongju at dawn. Jun boarding a train to Japan. Montage of Rintama High—fights, broken windows, raw emotions. Jun standing on the school rooftop, wind blowing through his white hair, camera raised, eyes steady. Add dramatic narrator voiceover (speaking in korean): “And just like that... I'm here” End with black screen.

everytime when i put it in its all cluster of other stuff colliding together

[robust grotto]

use ChatGPT or Gemini to help you structure it

[fickle token]

https://tenor.com/view/lebron-alley-oop-lebron-alley-oop-lebron-james-assist-gif-16615109714643721624

good looks

[south shuttle]

Make all football player photos to unlimited zoom out

[ruby jolt]

HOW DID YOU FIGURE IT OUT

[jagged lichen]

why it says', "Only English prompts are supported at this time."?

[honest sigil]

Make a video of a bird fly by the 101

[restive atlas]

HOW do I prompt such that there is consistency of the object with supplied images? For e.g. a bottle of dishwashing liquid in the video must match and be accurate with supplied image of the bottle.

[north kayak]

#flow-tips-n-tricks・✏️ The girl is walking on the podtum, I'm filming her paparazzi

#about-google-labs・✨ The girl is walking on the podtum, I'm filming her paparazzi

[low vector]

Any tips on exteding to getting a small character to do a super hero like jump and have it so the camera changes view and see the character jump onto a coffee table? 🤭🤣
I been trying and failing

[earnest cedar]

idk if this helps you:

"A cinematic, hyper-realistic action shot of a miniature character, 4 inches tall and resembling a tactical action figure, at the apex of an explosive superhero leap. He is airborne, floating across a dimly lit living room, his body tense with power. The camera captures the scene from a low, sonorous angle, moving alongside the character at high speed to create a strong sense of movement, with the background blurred by the speed. The character remains in sharp focus, with an expression of intense determination on his face. A dramatic beam of sunlight from a window cuts through the air, illuminating dust particles and creating a strong outline light on the character’s silhouette, with a slight lens flare. His background is a dark wood coffee table, which appears enormous and distant from his perspective. The image is intended to have the quality and color grading of a Marvel film, with a shallow depth of field, in 8K resolution and a 16:9 aspect ratio."

[low vector]

If I already have my character and am extending do I need to describe him again?

[earnest cedar]

maybe, im unsure

or use frame to video

[spring robin]

anyone have a good way to prompt for first person POV? Like from a box looking up. Camera being in the box.

[pale copper]

Help me please. I cannot get a video of a fisherman catching a fish from a boat, starting with an aerial shot above the boat, the fish being reeled in, then the camera plunging into the water to capture the fish with lure in his mouth being reeled in, before camera follows the fishing line, breaks water surface again, follow up the rod, back to aerial shot of fisherman. I have burned over 300 credits with zero results. I get weird footage of fisherman in water, rod in water, etc. Not even Chatgtp or Gemini can get correct prompts.

[alpine frost]

Has anyone tried the new Veo3 - Fast for 20 credits? It appears to be available on mobile only

[languid arch]

what prompt did you use?

[stark coyote]

I now got it on PC as well, and it's not ''experimental'' audio anymore

[agile ravine]

Can somebody help run a prompt for veo 3?

[stark coyote]

Can somebody help run a prompt for veo 3

[coral sentinel]

Hi everyone! I’m trying to create some prompts using google FLow and could use a bit of help.
I’m getting this notification, but I’m not sure how to fix it – I already have Google One, so I don’t understand what the problem might be. Any ideas?

Any ideas?

[torn minnow]

They're related but not the same. Google AI Pro is a tier within Google One. You need to subscribe to Google AI Pro to get the feature you wanted. Pixel 9 Pro and similar have a 1 yr free subscription.

[serene bramble]

How do people get from to go over 8 seconds by adding them together (extending it doesn’t work) I’ve tried to do frames to video with v 3 quality but there’s no audio?

[stark coyote]

You can’t with Veo 3 - you can do frame to video you can’t use ingredients just yet

What I’ve been doing is that I structure my prompts to have as much detail as possible and little to no objects shown if an object is shown it’s a one time thing.

[inner venture]

An 11 years old boy going on a cycle. Suddenly the chain of the cycle was fall.
He break and get help with veo3

#flow-tips-n-tricks・✏️ An 11 years old boy going on a cycle. Suddenly the chain of the cycle was fall.
He break and get help with veo

[marble brook]

what do you have to type in the prompt to give them the narrative to say some lines?

[tall shard]

how to do that? i get the same issue