flow-tips-n-tricks・✏️

[surreal torrent]

If you're just doing this to create yeti videos to make your friends laugh, make sure they venmo you money

[carmine basin]

how much you will charge if I give you text to make me this short vlog?

[surreal torrent]

how long is "short"?

[carmine basin]

script is for 48 sec

@surreal torrent prompt is with 12-13 shots total video is 40-48 seconds!

[surreal torrent]

I am heading to bed. Been up all night. Signup for a trial and give it a shot.

[proper pulsar]

anyone got good prompts for commercial ad style

[dry hornet]

Is anyone else not satisfied with the veo3 fast?

I was generating my perfect videos, and suddenly the quality got terrible.

[carmine basin]

what happened?

[dry hornet]

I made this video here yesterday, and look at the quality, it's perfect:

Today he's only making videos like this:

[dry hornet]

You can notice a difference, right?

[carmine basin]

yeah! it hase big difference!

is there any prompt you changed new

[dry hornet]

No, I just changed the character and the location of the video. But I always make it very clear that I want realistic hair, pores on the skin, etc.

I think it's a temporary bug. I hope it is

[quiet harness]

Here's another one generated by that "VEO3 Prompt Details v2.0.txt"

Extreme close-up. A pair of luminous, gracefully moving celestial hands are in a dark, cosmic void. Floating, abstract ingredients surround the hands. The scene is dramatically lit by the ingredients themselves. The right hand elegantly takes a slice of shimmering, woven moonlight. The left hand carefully places it down onto a floating slab of polished obsidian. The right hand then adds a thin, translucent sliver of solidified gravity, which visibly warps the light around it. Next, the hands delicately arrange a cluster of whispering nebulae that drift like lettuce. Finally, the hands place another slice of woven moonlight on top. The celestial hands pick up the completed creation. The camera pulls back RAPIDLY. The scene is revealed to be a simple, suburban kitchen at night. The "celestial being" is a man in a bathrobe wearing glowing novelty gloves. He is holding a perfectly normal peanut butter and jelly sandwich. He takes a HUGE, messy bite out of the sandwich. Jelly squishes out the sides. The style starts serious and epic, then ends comedic and mundane. Hyper-realistic, cinematic, 8K, 16:9 aspect ratio.
Audio: Starts with a low, majestic orchestral score and soft, magical chimes. The music abruptly cuts off when the camera pulls back. It is replaced by the sound of crickets chirping outside a window. The shot ends with a loud, satisfying, and slightly obnoxious bite sound, followed by a happy, muffled "Mmmph!".

[autumn oxide]

#flow-tips-n-tricks・✏️ Ad Prompt: Airtex Washateria & Daiquiris To-Go 4th of July Special
Visual: A vibrant, celebratory scene with a strong 4th of July theme.

Main Subject: A cheerful, diverse woman in her late 20s to early 30s, smiling brightly and holding up a half-gallon jug of colorful daiquiri. The daiquiri should look icy and refreshing, with a bright, enticing color (e.g., red, blue, or green to fit the 4th of July theme).

Secondary Subjects: Clearly visible on a table or surface next to her are:

Two small, clear plastic cups of jello shots (one red, one blue, or patriotic colors) that look festive and fun.

In the background, or slightly to the side, have a full gallon jug of daiquiri also visible, with four jello shots arranged neatly next to it.

4th of July Theme: The background should feature classic 4th of July elements. Think:

Subtle fireworks effects (blurred or distant)

American flags (small, decorative ones or a larger flag draped in the background)

Red, white, and blue decorations (balloons, streamers, tablecloths)

A backyard BBQ or picnic setting (e.g., grill, picnic blanket, other partygoers blurred in the background enjoying themselves)

Warm, festive lighting, suggesting a summer evening.

Brand Integration:

The "Airtex Washateria & Daiquiris To-Go" brand name should be prominently displayed, perhaps on a sign, a cooler, or a subtle overlay in the ad. The font should be clear and legible, fitting a local business vibe.

The daiquiri jugs should have a simple, clean label that either includes the brand name or is clearly associated with it.

Text/Voiceover (if applicable for video ad):

Voice: A friendly, enthusiastic female voice (the lady in the visual)

Dialogue: "Hey everyone! Get ready to celebrate the 4th of July with Airtex Washateria & Daiquiris To-Go! Right now, buy a Half Gallon of Daiquiris, get TWO FREE Jello Shots! And for even more fun, buy a Full Gallon, get FOUR FREE Jello Shots! Perfect for your Independence Day party!

[drowsy canyon]

hey Ryan, i would like to say thank you for that prompt guide. It worked well with my series.

[surreal torrent]

@drowsy canyon I updated my VEO 3 Prompting Guide v5.0. I'm still testing it, I added a lot more information to it covering a lot more edge cases. So keep the other one around incase you get mixed results from this one. But this one should be able to do a lot more difficult things.

Here is an example of me using that guide.

Prompt:
Establishing wide shot. An ancient, gnarled apple tree stands on a floating island of iridescent, tessellated glass in a void of swirling cosmic dust at the golden hour. The camera performs a slow, low-angle dolly arc to the right, revealing a gravitational anomaly at the base of the tree. As the camera moves, the apples on the tree begin to slowly peel open, transforming into intricate, glowing clockwork mechanisms, each gear ticking in perfect, synchronized harmony. A single, massive, translucent eyeball with shimmering, binary-code irises then slowly emerges from the core of the largest clockwork apple, its gaze fixed on the viewer. The eye then pupils dilate and contract rapidly, reflecting fractal patterns of the cosmic void.

Surrealist hyperrealism, with a deep focus effect and a subtle, ethereal light bloom around the clockwork apples. The island's glass shimmers with an impossible, internal luminescence. Full-frame 16:9.

Audio: The gentle, rhythmic ticking and whirring of thousands of tiny, synchronized clockwork gears. A deep, resonant hum from the cosmic void. A soft, otherworldly "pop" as each apple begins to peel. A subtle, high-pitched electronic whir when the large eyeball emerges, followed by a low, hypnotic "thrum" as its pupil contracts.

[drowsy canyon]

i'm testing it right now and do a comparision between the two versions.

[dim gull]

What did you put in to get this prompt output?

[grim rune]

I'm trying to create a universal prompt for ASMR videos, can anyone help me?

[surreal torrent]

@dim gull Create a fascinating, surreal, and impossible scene for Veo3, featuring an ancient apple tree on a floating island, where apples transform into clockwork and an eye emerges from one. Include all required elements and advanced techniques from the guide.

[drowsy canyon]

tested in my series. It seems to work even better.

[surreal torrent]

Sick. This 5.0 guide seems pretty robust.

[drowsy canyon]

it seems to do exactly what i wanted too. I'm doing a horror style videos.

[exotic yew]

Give moment to this image

[surreal torrent]

Using my 5.0 guide this would be the prompt in Gemini or ChatGPT:

Create a beautiful scene of a dew-kissed forget-me-not petal transforming into a galaxy with a human face, with rapid camera movement and detailed synchronized audio, following the guide's advanced techniques.

[surreal torrent]

@drowsy canyon I ran into a bunch of things, if you look at the 5.0 prompt guide it fixes things on how to move the camera in complex scenerios for dealing with flat plane objects line paintings, transitioning between things, how to properly turn anything into something else instead of it fading ...

[drowsy canyon]

it works really well. I'm testing it in night time right now to see if it will be able to generate a night time horror feel to it.

[surreal torrent]

This was my prompt I gave to my guide:

make me a scene of leonardo devinci painting mona lisa and then have her jump out of the frame and give him a big hug and kiss and then he falls down because he's shocked and he says "mama mia"

LMAO

It's not exact, but pretty dang good

[lament juniper]

THIS ONE PROMPT

RYAN

[surreal torrent]

Yah. My guide if you load that into Gemini and give it those words it outputted this for me:

Medium shot. Leonardo da Vinci, in his detailed 15th-century studio, is meticulously adding a final brushstroke to the Mona Lisa portrait on a canvas at midday. Soft, natural light streams through a tall arched window, illuminating the scene with a painterly glow. The camera holds static, focused on the interaction. As Leonardo steps back to admire his work, the Mona Lisa in the painting smiles broadly, then, with a subtle shimmer, she smoothly steps out of the canvas and onto the studio floor. She immediately strides forward and wraps Leonardo in a surprisingly strong hug, followed by a chaste but firm kiss on his cheek. Leonardo, utterly flabbergasted, eyes wide, stumbles backward and dramatically collapses onto the floor, his painting palette clattering beside him.

Renaissance art style mixed with photorealism, emphasizing vibrant colors and deep textures, reminiscent of classical paintings come to life. Full-frame 16:9.

Audio: The delicate scrape of a brush on canvas. A soft, almost magical shimmer sound as Mona Lisa steps out. The rustle of her dress as she moves. A warm, embracing (Mona Lisa, clear, joyful voice): "My dear Leonardo!" A brief, surprised gasp from Leonardo. The sharp clatter of a wooden palette on stone. Finally, a stunned, drawn-out (Leonardo, shocked, Italian accent): "Mama mia!"

[fossil stream]

@surreal torrent

did a bit of expanding

[fossil stream]

https://github.com/snubroot/Veo-3-Prompting-Guide/blob/main/README.md

[surreal torrent]

@fossil stream what did you enhance beyond the 5.0 one?

[fossil stream]

yessir

added you some music video information that could help as well

[surreal torrent]

oh nice.

[dim gull]

https://promptalchemist.com/share/VIfc73LDDDB5Yr1eIRGr

[dim gull]

@surreal torrent Threw your original prompt into my prompting tool and came up with this. Snub and your prompt guide seems to trend toward that cosmic galaxy style more.

[fossil stream]

yeah that is what my content is focused on , I am about to create serveral guides for different types of video types

Stay tuned

[surreal torrent]

Mine is more generic

Nice... I've found the guide solves really complex problems

[dim gull]

Was this output not with your 5.0 adapted guide? It's pretty spacey without anything space related in your og prompt.

[surreal torrent]

I used the 5.0 guide

VEO 3 is very contextual ... if you put someone in brazil they start speaking portugues even if you don't tell them to

so if you're turning apples into eyes it thinks you're doing something freeky

[dim gull]

I haven't really had any issues with that, here's one from a while back. I do have prompting methods built in for getting them to speak different languages if needed though.

[surreal torrent]

It's not easy to get things to turn into other things.. i've been playing with that and I think I got it figured out and added into my prompt guide now..

[drowsy canyon]

how would you guys prompt in Veo 3 9.16? or do you guys just do it 16.9 and then crop the video to make it 9.16?

[arctic umbra]

16:9 then crop when edit... I don't think veo3 does any other aspect ratio

[dim gull]

Just use 16:9 until they natively add 9:16 support. Trying to force it to do 9:16 with prompting isn't consistent enough to be worth trying.

[drowsy canyon]

yeah i tried to edit in 9:16 and losing a lot of shots in the video.

[arctic umbra]

I saw an app that does tracking + cropping, but I forget which one it is..

[drowsy canyon]

Using the prompt guide 5.0 to make this scene.

[fossil stream]

https://chatgpt.com/g/g-6862c990b0608191bb3f024750204dd0-veo3-prompt-director

I made this also , hope it helps

outputs prompts like this

Wide establishing shot. A silver-fur snow leopard perches on a jagged granite outcrop high in the eastern Himalayas at blue-hour dawn; chilled mist coils through the valley below as the first indigo light brushes the peaks.

The camera begins a slow, six-second crane rise, starting waist-level behind the leopard and drifting upward to reveal the full sweep of snow-capped ridges and a distant, ribbon-thin waterfall catching pale sunlight.

As the camera clears the leopard’s head, icy wind gusts across the ledge; billowing mist swirls, refracts the newborn sunlight, and fluidly coalesces into a faint prismatic rainbow arcing behind the cat before dispersing with natural atmospheric physics—no cuts, totally continuous.

Photoreal, National-Geographic documentary style, crisp 4 K detail, full-frame 16:9.

Audio: Sub-bass Himalayan wind, distant waterfall roar, granular stone crumbles under paw, soft feline exhale; the wind intensifies precisely as the rainbow forms, then fades to gentle valley ambience.

[weary pewter]

@fossil stream Not sure if you created this guide or you are just sharing it but, first of all thank you so much for sharing this, second whoever created this did AMAZING job of summarizing key aspect of VEO 3 promting.... oh now i realize that this was actually created by @RyanAZ , Dude, you did a FANTASTIC JOB of creating this EXCELLENT guide.... Thanks a bunch. In the example section can you please create some examples for "Character consistency" Also for proper text display, becuase in my personal experience, I have face problems in display text properly, VEO3 just display grabage wordings several times.

[fossil stream]

I started it , now me and @surreal torrent are both building and collabing on it together !

Glad it could help

[fossil stream]

lol i puked a lil but that is a great output , was it close to what you were trying for lol

[surreal torrent]

My son made me do it, it was exact lol

Anyother one my son wanted to do. This was his prompt to the Prompt Guide 5.0:

"make me a mountain that packs up his bags then leaves"

[weary pewter]

@drowsy canyon I have been using a trick to convert my VEO 3 16:9 videos in 9:16 videos. The trick is to use Runway ML for that. Here are the steps and yes i am using paid subscription for Runway ML as well. 1. Create your VEO 3 video
2. Save it on your computer.
3. Open Runway ML https://app.runwayml.com/ and go to generate video option
4. Select model Gen-3 Alpha Turbo model - Its older model
5. Click on expand video, it will show either on left side menu or somewhere right on top where the modle name is displayed "Gen-3 Trubo"
6. Click on that button and select the video from your computer that you want to convert to 9:16
7. make sure that 768x1280 option is selected.
8. Click on generate.
Pro-tip: Sometimes if your actual 16:9 video has top and bottom black section / black bars, the generated 9:16 video shows black bar in the middle of the video, in that case, you need to use chat mode of Runway ML, ask it that you need a customize prompt to convert this VEO 3 16:9 video into 9:16 video and give it your original VEO 3 prompt that you use for VEO 3, Runway ML will convert your VEO 3 prompt to generate correct 9:16 video. Now go back to that expand video screen, specify your 16:9 video, specify your customized prompt for conversion, and hit generate, this method worked for me almost 100% times.

[sinful spoke]

Subject:
An ultra-realistic fly-through of Molten Core, a vast volcanic cavern deep beneath Blackrock Mountain. The environment is filled with rivers of flowing lava, scorched black stone, massive rock pillars, and glowing magma vents.
Context:
Interior of a legendary raid zone, lit only by lava flows, molten cracks, and occasional burning torches fixed to the rock walls.
Action:
The camera flies forward slowly through winding molten corridors, passing towering obsidian formations, chains hanging from the ceiling, and glowing red embers drifting through the air. Occasional remnants of past battles — charred weapons and ancient bones — lie near the path.
Style:
Photorealistic cinematic. Heat distortion from lava, detailed rock and fire textures, subtle ambient particle effects.
Camera Motion:
Smooth forward fly-through with gentle turns, shifting depth and scale as the camera moves deeper into the cavern.
Ambience:
Low rumbling, flowing magma, occasional bursts of flame and echoing drips of molten rock.

😦 tried 5 or 6 different thingy by now, it just doesn't want to give me a "real" version of this it always looks like this

tried "ultra realistic", "Photorealistic" , "real" ..all lead to pretty much the same result

[dim gull]

Try removing the videogame references. It thinks you want a video-gamey style cause that's what you're asking it to replicate. It does poorly with conflicting prompting.

[muted prism]

Hi subjects are broad

[drowsy canyon]

how would you guys prompt about if a person is sleeping on the ground and then a shadow figured gliding by outside the windows? I kept trying and it is either giving me another person walking by the window or an image of person gliding by the window.

[north jacinth]

The following prompt gave these awesome output:

Scene Description:
The battered Stormtrooper barrels into a vast Hoth hangar bay, turret fire churning smoke overhead. He ducks behind stacked cargo crates, then pops up—blaster drawn—to thumb a fallen trooper’s E-11. With frigid breath fogging his visor, he sprints across the ice-coated floor toward an open starship ramp as Rebel commandos pour in through shattered blast doors.
Visual Style:
Photorealistic grit: frost-coated metal girders, frost-encrusted cargo containers, and oil-slick puddles reflecting red warning lights. Sparks rain from scorched ceiling vents, and every dent and scuff on his armor shows in sharp detail.
Camera Movement:
Start with a low-angle whip-pan as he vaults over a crate, then a shaky handheld track-in over his shoulder as he fires back. Finish with a rapid push-forward racing down the ramp.
Main Subject:
The solo Stormtrooper in ice-caked armor, visor streaked with soot, gloved hands gripping a newly acquired blaster as he weaves between cover and charging rebels.
Background Setting:
A cavernous hangar lit by flickering red emergency strobes, crates stamped with Imperial insignia, a TIE fighter frame under repair at one side, and the hulking ramp of a Lambda-class shuttle looming ahead.
Lighting Mood:
Harsh emergency reds mixed with cold white floodlights that glint off icy surfaces, casting long, jittery shadows as the camera races forward.
Audio Cues:
– Rapid blaster fire echoing in metal corridors
– Rebel shouts (“Take him down!”) punctuated by clashing footsteps
– Groaning hydraulics of the open shuttle ramp
Dialogue (muffled, urgent):
“Almost there—hold it together!”
Music & Atmosphere:
Driving percussive electronic pulse underlies the chaos, rising in tempo

[woven condor]

This was the very next prompt using the exact same master template, gear specs (Sony A7S III, 24mm f/1.4 GM), and strict “no animation / no CGI” instructions.

And Veo3 gave me this.

It looks like PS2 graphics, nothing like what I’ve been building. I’ve lost MANY hard-earned credits on this — can anyone explain what happened or how to fix it?

(Uploading this hoping someone from the Flow team or community has answers 🙏)

[north jacinth]

Amazing 🤩

How many words does your prompt contain to be able to keep everything working?

[woven condor]

Appreciate it! 🙏 I usually stay around 600–750 words. Key is being ultra specific with camera gear, lighting, character blocks, tone, and realism. Every word has a job.

[north jacinth]

ok so camera gear inputs does play an effect. i was just using generic words like Sony FX3 camera, but didnt see too much of a difference in output

[woven condor]

Yeah, gear inputs absolutely matter, but not just dropping “Sony FX3” and calling it a day. It’s more about layering realistic context: lens choice, focal length, exposure details, visual tone, motion behavior, and how all of that ties into the scene’s intention.

A lot of people throw in camera names expecting magic, but unless it’s paired with human-like direction, the AI doesn’t know how to “use” the camera. You kind of have to direct it like a DP, not just list specs.

[north jacinth]

thanks
i definitely need to step my game here lol

[sinful folio]

Bro,what’s your prompts?

[fossil stream]

looks like no mans sky lol sick

[fathom onyx]

how are people doing to petri dish water drops then things grow or form quickly

[sinful spoke]

Subject:
An ultra-realistic cinematic fly-through of a vast underground volcanic chamber. The environment features glowing rivers of lava cutting through jagged obsidian rock, towering basalt pillars, and ancient stone pathways suspended over molten depths.
Context:
A colossal subterranean cavern lit only by the ambient glow of magma and scattered burning braziers. The air is thick with heat and drifting ash. Massive chains hang from the ceiling, swaying slightly above the lava flows.
Action:
The camera glides slowly forward through winding tunnels and open chasms, passing beneath stone arches, alongside glowing fissures, and over trembling bridges of scorched rock. Rusted armor fragments and burned banners lie half-buried in the ash, hinting at long-forgotten battles.
Style:
Ultra-realistic cinematic. High detail on lava shimmer, blackened stone textures, and rising heat distortion. Flickering firelight casts dynamic shadows across the cavern walls.
Camera Motion:
Smooth, continuous forward fly-through with subtle turns and elevation changes, emphasizing scale and depth as the camera moves deeper.
Composition:
Vast lava fields and narrow corridors alternate in rhythm. Light and shadow pulse with each lava surge. Perspective shifts to highlight vertical drops and ancient structures carved into the stone.
Ambience:
Low rumbles, the hiss of magma, crackling fire, and deep distant echoes from within the cave system.

Still think it could look more "real" 🤔 rock formation etc is ok but the lave isn't, is it?

[uneven rover]

Hi everyone,

I'm having trouble with the image-to-video feature. When I upload a starting image and add a text prompt, I get the notification: Switching you to a compatible model for this feature. Submit again to confirm (or check settings for details).

The problem is, even when I submit it again, it doesn't work. How can I get past this step and actually generate the video from my image?"

Using veo3

[dry hornet]

Can I use veo3 on my cell phone?

[zinc spear]

yes it works on mobile.

[unique cedar]

what´s the prompt?

[fossil stream]

No it doesn’t. If you’re using it on mobile may ask how? As the website states otherwise when visited from a mobile device.

[dim gull]

?

[fossil stream]

Huh guess I’m out with the times and need to download chrome on my phone 🤣

[indigo brook]

It's not working on my phone despite having a subscription when it reaches 50 to 70% it says something went wrong does anyone face this issue please help me with that

[odd birch]

Log in or sign up to view

[fossil stream]

oops one min and ill repost full file

theres 39 lol , that should suffice

@fathom onyx

[fathom onyx]

boss

[fossil stream]

theres the whole 50. Anyone who needs help , just ask , I can make guides of any kind.

[fathom onyx]

bro do you know how people prompt for the water drops into a petri dish and then a reaction happens and with crystals or structures

[fossil stream]

No but I can figure it out , send a video of one if you can and i will learn how to best do it.

https://github.com/snubroot/Veo-3-Prompting-Guide/blob/main/CartoonSand.md

[surreal torrent]

@fathom onyx the prompt guide that snubroot posted should be able to handle it. There are specific things that the prompt has to follow to turn things "into other things".. becuase we understand what that means in english, but the guide will break it into seperate transformations to make sure things happen correctly

A continuous one-shot sequence, starting with a bird's-eye view. The camera is positioned directly above a falling NOVA Sentinel unit named Zappie, looking straight down at the ground from a great height.

Zappie is a tall, humanoid android with matte-white armor plating over a dark charcoal mechanical endoskeleton. His head is smooth and featureless, distinguished by a single glossy black horizontal visor.

He is plummeting towards a brightly lit New York City street grid at night. The camera falls at the same speed he does, maintaining the top-down perspective. As the street rushes up, Zappie falls perfectly into an open manhole. The camera follows him without cutting, plunging into the darkness.

The view instantly transforms into a surreal tunnel through the Earth. The tunnel's transparent walls show a rapidly passing cross-section of the planet: soil, bedrock, glowing veins of magma, a shimmering molten core, and then the layers in reverse.

Zappie bursts out of another open manhole directly into bright daylight. The camera follows as his momentum carries him twenty feet up into the air above a bustling Chinese street market. He then falls downwards, landing with a loud crash onto a wooden cart full of colorful fruit, which explodes in a shower of produce.

The shot ends looking down at Zappie tangled in the wreckage, as an elderly Chinese vendor stares at him in bewildered surprise.

Cinematic realism with a surreal, comedic, and action-packed one-shot sequence. Full-frame 16:9.

Audio: The sequence starts with the loud, roaring rush of wind. The wind sound is then mixed with a deep, geothermal rumbling as he travels through the Earth. This is abruptly replaced by the bustling sounds of a dense Chinese street market (chattering in Mandarin, sizzling woks, bicycle bells), ending with a final loud CRUNCH of wood and smashing fruit, followed by a surprised gasp.

[fossil stream]

copy the guide , create txt file named guide , boom you are locked in, just gotta mess around to get what you want

if gpt gets lost , ask it to refer to the guide

[surreal torrent]

@fossil stream what do you think of that one I just posted? lol

[fossil stream]

that was pretty dope bro !

[bright locust]

Thanks Snub, do you know where is the prompt guide 5.0? i see a lot of mention of that guide

its yours?

[fossil stream]

It is mine and @surreal torrent , we are working on it together but the 5.0 is the guide above , I just did not lable it as 5..0 haha

https://github.com/snubroot/Veo-3-Prompting-Guide/blob/main

this one yes ,

[bright locust]

thank you Snub ❤️

[fossil stream]

Anytime , let me know if you get stuck or need some help ! 🙂

[fossil stream]

Enjoy :)

@fathom onyx

[tropic jay]

After almost finishing my project with veo around 15,000 credits used but almost 25% of it was video prompts not rendering correctly anyone know if i can get some credits reimbursements for those ?

[frank robin]

https://tenor.com/view/hold-bold-gif-8800793

[tropic jay]

Lmaooo nah but fr each video is 100 credits super sad when they don’t come out correctly 💔

[frank robin]

I feel that fr

[dim gull]

Use Fast instead of Quality until you nail your prompting down. Many fast credit generations are perfectly usable as is.

[sinful spoke]

Just so you guys know the impact of your prompts 🤣 🤣 🤣 🤣 https://youtube.com/shorts/4IoCC7tFl-Y?feature=share

All fast btw, not a single quality vid in that

[wintry blaze]

Hi guys, I am new to veo 3 and AI editing tools. i was playing with veo 3 today. Just wonder if i can use myself on the video. Whilsy I am uploding picture with myself it cannot be uploaded cause I am living in the UK. Is there any way get around it ?

[stark torrent]

Has anyone bought a tik tok fully monotized account online?

[tiny sleet]

we are venommm

[tropic jay]

Fair but when you need the highest quality possible you deff see difference & all is good tho found out i can buy 20K extra credits for just $200 life is good

[vague axle]

Have you generated any fast videos within the last hour? they are coming out very stuttery for me and with awful audio, wanted to see if anyone else is having that

[tropic jay]

Tbh bro i tried fast for like maybe the first 10-12 prompts and couple of more there but never ever did i use them.. either wrong render or lack of quality but i am being pretty picky about what i choose so theirs that as well

[sand sparrow]

can you add a screen shot for each characters in this github pls so ppl know how they look without running the script

[heavy totem]

Do it in json

[fossil stream]

I will do this soon ! but prob not till tomorrow sometime , sorry !

[sand sparrow]

no rush, thank you, this is awsome by the way. so great job

[heavy totem]

When is veo3 getting image to video

[dim gull]

Yeah the top ups are nice. For credit pitchers and pro users, Fast is a fantastic option. Quality is quality though. 👌

[north jacinth]

made with veo 3 fast or veo 3 quality?

[woven condor]

Accidentally made a video that summoned a cult on tiktok. Now we're legally obligated to keep posting or the comments will riot.

https://www.tiktok.com/@hairysightings/video/7518982195353029919

[prisma schooner]

That’s funny, lol

[fading geyser]

does anyone have a prompt for the bread spreading videos.

[winter egret]

idk why I'm getting a cartoonish character....from 2-3 days. sim using the same prompt as before but now it looks all fake

[heavy totem]

Could a done a much better job with character consistency in my prompt but was lazy cool motion though!

[surreal torrent]

Are you using VEO 2 or 3? Did you switch from 2 to 3 or vice versa?

[surreal torrent]

what's a butter spreading video?

[oblique quarry]

I have found when trying to generate music visuals that I get a lot more luck when I prompt a musical genre and BPM, often the song is the same song everytime but that might just be because its in the same project on flow. But more importantly the action in the video is matching the regular beat of the soundtrack. Technically 120BPM would be perfect for the 8 second lengths, but I seem to get better results out of asking for a 128BPM house track. I am guessing cause thats more common of a BPM for a house track. I also have had good luck with prompting for a 174bpm Drum and Bass track. It might be worth adding that to the prompting guide. (Again this is for the goal of getting regular beat matched animation out of it, I throw the audio away when I am using it to mix visuals for events)

[fossil stream]

Noted , will look into it

thanks 🙂 !

[winter egret]

3 only

[tiny sleet]

veo3 fast audio

[remote snow]

what prompt did you use

[frozen escarp]

I have a story that I would like to use Google Flow to create a new video of the story. I have already used Gemini to make it into 4 different scenes. How do I get Flow to make each scene? And, how do I get Flow to remember the previous scene?

[north jacinth]

It’s crazy how the ai generates accurate dialogue for the character and scene. I’m not great at writing dialogue so I let the ai fill that part in and it usually turns out better than expected.

Generate an 8-second cinematic video that looks like it was recorded from the front-facing camera of a phone held by a Stormtrooper. The Stormtrooper is walking down a fluorescent-lit grocery store aisle, vlogging with deadpan humor. The camera angle must replicate true selfie POV — exactly what a front-facing phone camera would capture. His helmet should stay on, but allow slight bobbing and shoulder movement to convey emotion. Lip-sync accuracy must match this exact voiceover: You will create the dialogue for character based on the scene and personality. The shot should include slight handheld movement and standard indoor supermarket lighting. Do not include a phone or selfie stick. The video must look like the Stormtrooper is filming it himself as part of a hilarious relatable vlog.

[barren eagle]

How to make longer videos?

[fossil stream]

Scene continuation.

[barren eagle]

O

[fossil stream]

Meaning you continue the second in the next prompt by fully setting the scene again. Here a guide hope it helps

https://github.com/snubroot/Veo-3-Prompting-Guide

[fossil stream]

If you paste that guide into GPT you should get some good results. Just paste the guide and then ask it for what you want and to refer to the guide.

[barren eagle]

Check channels some guy promoting a scam

[fossil stream]

What do you mean

[fossil stream]

I’m not staff haha or I’d take care of it

[fossil stream]

I’m cackling 🤣

The geek bars 😭

[barren eagle]

U like it lol

[fossil stream]

Yeah good stuff bro. Did you use the guide to help with that?

[barren eagle]

Ye but didn't look like the prompt lol

[fossil stream]

Ye you gotta tweak it to your needs

Which seems like you’ve done a great job at

[barren eagle]

It is funny how he drops them innit

[north jacinth]

Did you write the dialogue?

[barren eagle]

Ye

[still burrow]

This is great! i have found to wasted so many credits but this will help for sure.

[fossil stream]

always use veo 3 fast , it literally will get just as quality outputs if you just input cohesive quality prompts. Only use Veo quality if you have tested and gotten succesful outputs from veo fast

[still burrow]

oh really? i have just inboxed you! i have several questions if you can help?

[nova siren]

The point of this thread is to post not only the result, but also the prompt used to generate the video.

?

[sage cloak]

No captions must be generated. Create a hyper-realistic shot of the first mate who has taken over the camera. She is standing at the back of the ship. She is a middle-aged pirate, wearing a cloth bandage around her head, revealing tangled, rain-slicked black hair. Her face glistens with fresh blood from debris falling all around her. Her wild eyes mixed with fear and determination. Her royal blue jacket hangs open, flapping violently in the wind, exposing a soaked, mud-streaked white linen shirt beneath her torn black leather vest. Her powerful hands grip the vlogging camera tightly at arm’s length, the lens catching every tremor of the storm as she shouts through the tempest, “She’s breaking up — we’re going under if we don’t reach the cove now!” The first-person lens captures the chaos surrounding her: jagged splinters of broken rail drift by in rising seawater on the deck, a snapped mast crashes behind with an earth-shaking crunch, and crew members scramble in the blurred background, silhouetted by flickering lightning. Rain lashes the camera lens directly, smearing vision momentarily as thunder explodes overhead. A monstrous wave builds in the distance, rising like a gray wall to swallow the deck. The entire footage must be seen exactly as recorded through the pirate's handheld selfie vlogging camera, with minor natural shakes, slight off-centering, and frantic repositioning as he struggles to steady the camera while bracing against the lurching deck. No external cameras, no third-person views — only the imperfect, handheld view from the pirate’s own perspective. No captions must be generated. No Cinematic Black Bars must be generated. Aspect ratio must be in 16:9. No watermark must be generated.

[eternal crown]

great! what prompt did you use?

[north jacinth]

That is amazing.

Did you use Veo 3 fast or Veo 3 Quality?

[weary pewter]

My ASMR experiments, besides just typical fruit cutting, i tried something different... https://www.youtube.com/watch?v=hLuoGNa8Lz8

[maiden marsh]

hi, please keep prompts entirely in English, thank you! As stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.

[sage cloak]

Thanks for the comment, I used Veo 3 fast then the final pass was to upscale it ( not this version, this is a clip).

[north jacinth]

Better to upscale using other applications than the output google veo gives?

[sage cloak]

I used Flow, it seems fine for this.

[woven condor]

Not sure if this is the right place for this so let me know if not, but I’ve been building something wild entirely with Flow.

It’s not some AI vlog gimmick. This is structured, cinematic storytelling with TV series, and movie-level quality, full realism, explosive pacing, and scenes that feel like they were shot on location.

The series (Jax & Finn) is about to hit a million views, with hundreds of thousands of likes and comments from countless people asking me for a full movie.

With background in filmmaking, I’ve reverse-engineered every shot, every camera angle, and every line of dialogue using Flow and turned that into a complete system for advanced cinematic prompting. It’s designed for both beginners and advanced creators looking to push Flow to its full storytelling potential.

I wasn’t planning to share this here but since I’ve seen people struggling to reach this level of realism, figured I’d mention it.

If it’s cool with y’all I’ll drop the link. If not, I’ll chill.

[barren eagle]

Huh?

[barren eagle]

Can I see an example of your videos

[pastel granite]

So cool!!! I can't seem to generate dialogue, even after trying a few different prompts. I'm using the Model 3 Beta Audio version. Any tips?

[sage cloak]

Try to cut and paste my prompt, see if that works maybe? I used Flow and set the video generator to Veo 3 fast.

[nocturne pivot]

any suggestion/idea how this user can maintain character consistency so well other that a "detailed" description every time?

[barren eagle]

Yes I know

DM me

[pastel granite]

Oh I see this issue - I was trying to use Veo 3 with a frame to video prompt, but it only seems to work with text to video.

[north jacinth]

Would love to see your work

[amber swift]

Does anyone have suggestions for flipping this badger upright?

[barren eagle]

Rotate video

[amber swift]

lol

Thanks trying it now

[north jacinth]

You unlocked 9:16 format 😭😭😭

[barren eagle]

?

So?

[amber swift]

I'm a beginner Veo3-er, I have the prompt I want to use but have been struggling with replicating the same video lol

[woven condor]

Here’s a one-minute sample built entirely in Flow using my own unique structure and advanced prompting formats for cinematic realism. Even when using ingredients-to-video, you won’t get perfect continuity every time...Flow just isn’t fully there yet. V3 is close, but we’re all hoping that level of consistency lands soon. Until then, this shows what’s possible when you push the tools right.

[amber swift]

I'd like to keep the same style for different scenes, what is the proper video prompt to keep it consistent

[woven condor]

There's a lot more than just a prompt for keeping continuity in V3

[amber swift]

What else do I need to keep it consistent like this?

[woven condor]

Ingredients to video helps

[amber swift]

What do you mean?

[barren eagle]

Dm me

[woven condor]

Just to clarify, Flow isn’t about writing one perfect prompt and getting continuity handed to you. When you use ingredients to video, you're giving the model specific visuals like subjects, locations, camera motion, and mood. But to get strong results, those ingredients need to be clean. That means using subjects or product references on plain or segmented backgrounds, and making sure your style and location images don’t include extra elements unless that’s part of your goal.

Your text prompt has to complement the visuals. If it says one thing and your frames show another, Flow gets confused. Everything should feel like it belongs in the same world. Consistency in lighting, style, and composition across your images helps the model blend everything smoothly and hold the look together. That’s the real foundation behind anything that even comes close to cinematic continuity right now.

[barren eagle]

Please can u use my prompt and show me result?

[midnight rampart]

https://youtu.be/tmonMZvoNTs

[next compass]

hell ya!!!!!!!!!

[earnest cedar]

I'd really wanted to say that Google and VEO 3 are SCAMMERS OVER 30 GENERATIONS WITH NO SOUND...

Google VEO 3 👉🏾big scammers

[dim gull]

What prompt are you using

[earnest cedar]

Cinematic ultra realistic scene in a Dakar street  the camera focuses on a 30 year old beautiful Senegalese man with long dreadlocks wearing a long sleeves blue dress that fits his strong athletic body and a big leather belt constantly holding a big megaphone speaker on his mouth while walking and saying with enthusiasm and happiness gestures : "AaythieuFaal thioonoylaxassaaydièkhneu'abadan BaayFaalyèkyaayfaalyii akwaakhidmeulakhassaay tol'lousabiirkeur wooté gnoudaadilakoylivré".

Tried It in Freepik and It worked but SO EXPENSIIIIVE

[dim gull]

Veo only officially supports prompting in english at the moment, but you can say (speaking in senegalese) before the dialogue and it might translate.

[earnest cedar]

It's not True I was the first in Senegal to Try It and It worked

Using WOLOF and Pulaar local languages

[dim gull]

I guess you're an expert then, and Google must be scamming you.

[earnest cedar]

Google is wrong but scams me all day long

Made a 100% using VEO 3 loosing lots of credits Audio missing

I'm from Fatick, Senegal West Africa. VEO 3 is something amazing but hurts a lot financially speaking

[dim gull]

You're intentionally using the model in a way they warn you that it is not designed for and then call them scammers. I just think that's funny tbh.

[earnest cedar]

You're scammers my testings are important and is helping out a lot of people in my country using AI in their cultural background. Google just has to award me for that

[tiny sleet]

What prompt did you use for this Bigfoot look... this is exactly the kind of Bigfoot character I wanted to create.

[woven condor]

Took a whole lot of trial, error, and structure to get him dialed in the way you are seeing him. Shoot me a DM and I’ll explain a bit more.

[tiny sleet]

sure buddy

[nocturne pivot]

can anyone ban this guy? he's contacting everyone promising to resolve whatever problem trying to push his prompt cause he has no credits.

[uncut robin]

How much of a difference are you noticing between Veo3 Fast vs. Veo3 Quality? Like, do you see much of a difference for the extra credits it cost? (20 credits vs. 100)

[dim gull]

I would say that Veo 3 Quality is not 5x as good as Fast. So, if you're worried about credit cost, stick with Fast generations.

[tepid mortar]

a video of a guy in a pickle costume that says super on it. he is in a hideout overlooking "taco timmys paradise of omega" he says "lets do it" and runs towards the building. a guy in a taco costume that says timmy on it is outside the building

[uncut robin]

I'm pretty much only using the fast, and getting decent results, When I got up to Quality, it doesn't seem to give THAT much better of results, but I wanted to know what others are experiencing

[dim gull]

Yeah, I'm not noticing nearly enough difference to justify losing 4 extra videos per generation. When each generation is essentially a gamble even with the best prompting. I'm down to play the numbers game and get more shots at getting everything right.

[zinc spear]

This is one of the multiple attempts I did. I'm having trouble on trying to make the cat talk along with the female character. This should have been a POV of a female vet. The issues I'm having are:

1. The female character's face keeps on appearing on the video. I instructed in the script that this is a first person perspective POV and that only the hands of the female character should be shown.
2. My script won't let the cat talk. 😦 The female character says all the dialogue including the one for the cat. 😦
   Any tips you could provide? I tried to change the prompt so many times. 😭

[dim gull]

Whats the prompt

[woven condor]

Totally feel you on that. POVs in Flow can be tricky because the model still struggles with fully committing to first-person framing, especially when it comes to faces. For the hands-only issue, try using a character reference that only shows arms/hands and make sure you're not including any images with faces. Also keep the phrase “first-person POV” early in the prompt, not buried later.

For the talking cat, Flow sometimes assigns all dialogue to the human by default. Try rewording it like a script, with lines labeled clearly like:

CAT (voiceover): “Hey doc, that’s a little cold.”
VET (speaking): “Sorry, just doing a quick check.”

Also throw in something like: “The cat speaks with a voiceover. It should not be the human speaking these lines.”

Sometimes just being extra blunt helps Flow separate characters better. Let me know if you want help refining the wording.

Also add that the cats mouth moves perfectly in synce with the voiceover

sync*

[zinc spear]

this is from the PA site prompt output:

Subject: Mr. Choco: A dapper tuxedo cat with piercing green eyes and a mischievous glint in his expression. Wears a single bell blue collar.
Beatrice: A kind, female vet with a warm smile and gentle hands.

Action: The tuxedo cat, Mr. Choco, sits on the examination table, allowing the female vet to inspect the inside of his mouth. As the camera slowly zooms out, Mr. Choco closes his mouth, his whiskers twitching mischievously, Beatrice pets his left cheek. He then leaps off the table and hops into a miniature toy car, speeding off with a confident air, donning black shades as he goes.

Scene: A brightly lit, sterile vet's office, complete with stainless steel examination table and various medical instruments. The walls are a calming pale green. The scene transitions from a close focus on Mr. Choco's mouth to a wider shot of the entire office as he makes his dramatic exit. The scene ends with a flourish as he puts on black shades, enhancing the comical departure.

Style: The camera provides a close up shot of Mr. Choco's mouth then zooms out to a medium shot of the cat, filmed at eye-level, all rendered in a hyper realistic style with soft, natural daytime lighting creating a joyful and celebratory atmosphere, 16:9 aspect ratio.

Dialogue: (Soft, concerned) Female Vet: You're not really sick are you Mr. Choco?
(Smooth, confident, lips synced)
Mr. Choco: I'm love sick, Beatrice! I just wanted to see you. Shall I pick you up tonight?
(Amused, gentle) Female Vet: Alright, but Ill let your mom know okay?
(Cocky, pleased, lips synced) Mr. Choco: Oh yeah!

Sounds: The gentle clinking of medical instruments, a soft purr from Mr. Choco during the exam, and the whirring of the mini car as he makes his escape.

Technical (Negative Prompt): subtitles, captions, watermark, text, words on screen, logo, branding

[woven condor]

Try adding this at the very beginning of your prompt:

“This scene is filmed entirely from the first-person POV of Beatrice, the female vet. Only her hands should be visible.”

Also, in the Style section, you’ll want to rewrite the camera details to say:

“Camera is fixed in first-person perspective from Beatrice’s eyes. Viewer only sees what she sees, including her own hands.”

Flow sometimes ignores POV when the camera motion or framing is written more traditionally, so that rewrite should help.

2. For the cat’s dialogue getting stolen by the human:
   This is a common issue. Flow tends to default all dialogue to one character unless you clearly assign lines like a script and specify the delivery method. Try reformatting your dialogue section like this:

VOICES:
Female Vet (spoken): “You’re not really sick, are you Mr. Choco?”
Mr. Choco (voiceover): “I’m love sick, Beatrice! I just wanted to see you. Shall I pick you up tonight?”
Female Vet (spoken): “Alright, but I’ll let your mom know, okay?”
Mr. Choco (voiceover): “Oh yeah!”

Also throw in this sentence at the top or bottom of the Dialogue section:

“Mr. Choco’s voice is heard through voiceover. His mouth should move when speaking, but the lines are not spoken by the human vet.”

[zinc spear]

I did have made changes but this is the first prompt I used to create the video.

[zinc spear]

makes sense. the videos I did previously for this cat uses the keyword "cat 1" "cat 2" and they did talk like a human.

[woven condor]

Worth a shot

[dim gull]

Taking a look now and seeing what adjustments can be made!

[zinc spear]

thanks guys. I'm taking notes of your advice and try it once more.

[woven condor]

YW!

Try this prompt. it has both, yours and the suggestion integrated.

Suggest maybe you review it first so you're confident to feed it to Veo

[zinc spear]

thanks!

[woven condor]

np

[zinc spear]

of course I will. I'm almost out of credit I need to review it first. haha

[woven condor]

Lol get it 100%

[unique maple]

I did a character and dialogue test to create a series of scenes. Since you can't use ingredients-to-video and extend for Veo 3, I was forced to write out detailed descriptions of each clip for 2 characters and struggled with consistency. So many trade offs - either the character doesnt look 100% the same, their voice comes out different, or the elements/shot/lighting in the scene aren't the same. Is there no other way to get all the characteristics to align without just tons of trial and error? Heres the project so far - https://youtu.be/gSVfHdNpWX4?si=6BUDPCtOBCo95ORR

[tiny sleet]

I sent you a DM twice but you haven't looked yet.

[dim gull]

I tried several times and after simplifying the scene it still couldn't get a well synced male voice from the cat. I would probably break this into multiple shots. Veo still has such a tough time with multiple subjects even with good prompting.

[woven condor]

Not seeing it anywhere

[zinc spear]

I tried this on flow. I don't know why it still won't let the cat talk. 🤦‍♀️

[zinc spear]

thanks for trying. You're right maybe I should break it to multiple shots separating scenes for each dialogues.

[woven condor]

Yeah Flow still fumbles multi-character voice assignments sometimes, especially when it assumes the human is the default speaker. Even if the lines are clearly labeled, it’ll still stack everything on one character. I’ve had this happen countless times, especially when the dialogue is a little long for 8 secs. VEO tends to buckle under too much back and forth. You’re definitely on the right track though. We all are hoping V3 locks that in tighter.

Sometimes feed it the identical prompt works after a few tries. its WILD, but total credit muncher 😭

[zinc spear]

I had done previous videos for this cat character but with another cat and they perfectly sync with the lip action. This is the first time I did it with a human character and it gave me a headache! hahaha

[north jacinth]

dialouge between characters is either a hit or miss even with the right prompting.
problems i would get is that other characters speak the lines only or both characters speak together at the same time

[lost minnow]

/A slow, realistic cinematic 4K zoom-in shot, focusing on the consistent old man character standing quietly by the side of a bustling New York street. This is his first clear, full-frame appearance. Emphasize his striking white hair and the deep wrinkles on his face, visibly telling the story of a long life. He stands calmly, a stark contrast to the moving crowd. Soft, diffused daylight illuminates him.

[zinc spear]

That's my problem, only one character speaks the line. 🤦‍♀️

[onyx bone]

Pls share the prompt of this videos

[odd birch]

As stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now. Please keep video content to English. Thank you!

[broken sail]

cutting board placed on a kitchen countertop. The strawberry is deep translucent red with embedded seed-like bubbles and a green glass stem. A regular stainless steel kitchen knife - slightly worn and malistic-anters quickly from the right and slions through the strawberry in one clean, sharp motion. Each cut is fast and precise, producing a crisp candy-ilke crack, followed immediately by a soft ciink as the glass silce lands on the wooden board. The texture of the wood and metal, the natural lighting, and the kitchen ambiance feel authentic. No transitions, no movement-just five quick, satisfying cuts in a row. ASMR-style audio with enhanced slicing sounds, natural knife handling, and realistic impact on wood, 4K cinematic macro detall, neutral warm lighting, fixed camera

CB Tempizz

Create a video clip 16:3 8 seconds

[nova heath]

VEO3 cant understand how an archer can shoot an arrow. How can I train your Server to "see " and understand how to shoot an arow?

[drowsy canyon]

How do you guys suggest Veo 3 to understand night time? "The forest and sky are in absolute darkness with no visible stars, moon, or ambient light. Scene is exposed for firelight only — all trees and background remain completely black." thats the line i have in the prompt. However, when i generated in Veo 3. It always come out daylight no matter what i try to do.

[worldly gate]

Show a young male DJ with dark hair, sunglasses, and a colorful shirt performing at a packed beach day club in Miami. He’s holding a bottle of tequila while beautiful girls in bikinis dance behind him. A blonde woman in a shiny silver dress and sunglasses playfully reaches for the tequila bottle. The DJ smirks, pulls it away, and hands it to his assistant—another beautiful girl in a black bikini. The assistant says “can we get this girl out of the dj booth please" and throws the bottle at the the blonde girl. The blonde girl falls to the ground dramatically while everyone laughs and keeps dancing.

Anyone know why this prompt violates TOS?

[dim gull]

Using the word "Young" alongside Erotic Content, Alcohol Use, and Violence. My man hit the trifecta.

[tame wasp]

omg how you do this, so cute, what is your prompts to make it upright?

[oak tusk]

Prompt - Gloss Apple
A hyper-realistio cinematic macro close-up of a full
tronsparent glass apple with a soft light red hue, plaoed
perfeotly oentered on a rustlo wooden outing board. The glass apple emite a subtle internal glow under warm, focused studio lighting, A realistio human hand grips a sharp stainless stedl knife just above the apple. In dramatie alow motion, the hnife makes the first alean slice -- the front portion of the glass apple breaks off amoothly, accompanied by delieate glass rcrack ASMR sounds and a light scatter of clear shards.The knife immediately mokes a second precise cut, alioing off another piece, No musio, no talking - only rich, detailed ASMR aliaing and crocking sounds, Only the hand, knife, and fruit are viatble Shot wsing an ultra-ehapp maero lens, ahallow depth of fiald sinematio lIghting with glowing edgea, 1280*720 resolution,30FPS, 5 seconds long

[rugged charm]

#flow-tips-n-tricks・✏️

#flow-tips-n-tricks・✏️ hi

Create a detailed and engaging script for an adult Hindi podcast episode. The content should be mature and tailored for adult listeners, incorporating culturally relevant themes, contemporary issues, or storytelling that resonates with a Hindi-speaking adult audience. Ensure the language is appropriate for mature audiences and the tone is conversational yet respectful.

Steps:

1. Identify the main theme or topic suitable for an adult Hindi audience (e.g., relationships, career challenges, societal norms, personal development).
2. Develop a clear structure: introduction, main content segments, and conclusion.
3. Include engaging dialogues, anecdotes, or interviews if relevant.
4. Use Hindi language with clear, expressive phrasing suitable for audio listening.
5. Avoid slang or expressions that might be offensive; keep content mature but tasteful.

Output Format:
Provide the podcast script as a formatted text in Hindi, divided into sections with timestamps or segment headings for easy reading and recording.

[amber swift]

A realistic brown-furred badger with expressive eyes and soft fur walks through a bustling New York City street during golden hour. He holds a selfie stick with a phone pointed at his face. The camera is vertical and framed like a handheld vlog. The badger adjusts the selfie stick while walking, looking directly into the camera, smiling confidently. Skyscrapers, pedestrians, and warm sunlight fill the background. The video has slight natural camera shake, and the badger begins to speak excitedly, introducing his stream.

[amber swift]

Since then I've used a different style that I've been able to replicate haha

[uncut robin]

It has trouble with unreal things especially if they don't have a mouth that looks like it can talk right now, I've noticed

[acoustic galleon]

why is it when i provide more information to my prompts they come out worse lol?

[dim gull]

It depends on what information youre adding. Highly detailed prompts are amazing. But if you're trying to add too much to a scene then it can fall apart, especially because its time constrained.

[acoustic galleon]

right i see that might be the issue but like sometimes i add information and it does the opposite lol

can i dm you if thats okay

[heavy totem]

Yeah I found the limit, it’s 3 characters in a single scent with distinct voices and VO lines but you need to sacrifice a lot of detail to get it to work , too much detail and the VO get mixed

Like the scenes look fine but you can’t get them to do much I had a shot where I had one character tending to a wound on another character both delivering VO while a 3rd character walks in and says something, each wearing specific wardrobe and a details character description the VO kept getting mixed up I removed a lot of detail to get it to work but then the scene didn’t work as well

I bet it would work if they upped the denoising steps

Runway does that during special events , allocates more compute per gen I’m 100 sure of it

[viral skiff]

Is there any way to convert a 16:9 video to 9:16 for free?

[dim gull]

Cropping

[viral skiff]

is that a page?

[mellow verge]

ingredients only work with v2 (no audio) i assume 95% of this is veo3 with just advanced prompting?

[honest hemlock]

DM me

[pearl solar]

I can’t download my videos pls help

[tiny sleet]

Check the top right corner...sometimes I can't see it either

[obsidian stag]

Why my flow is showing tv instead to start project

[naive halo]

hello, help please. What prompts do I have to put in the voice so that it does not sound like a robot or like the voice of the translators?

[vast tapir]

i had this issue and if you inspect the page there was a firebase issue with chrome... not sure if it needs a refresh install.

i was planning to wipe my machine and a pc wipe fixed it, i assume a fresh install of chrome may too.

[vast tapir]

setup the prompt to sound like someone.

[maiden marsh]

please keep self promotion to #off-topic・🐢 thank you!

[dusty bronze]

Is Flow/VEO3 just bad at spelling, or is there a way to prompt it to get the spelling right? I had a video where the planets popped up on the screen, but they where mostly spelled wrong despite being listed in the prompt.

[dim gull]

Text is still a major struggle for Veo, if you prompt for only a few words it can usually do well. But the longer the text the more likely it will fail.

[balmy wind]

A cinematic, slow-motion shot of a futuristic superhero landing on a rooftop at sunset, surrounded by glowing neon city lights. Electricity crackles from their hands as their cape flutters in the wind. They stare confidently into the camera, eyes glowing with energy. As the screen pulses with light, they say with a smirk: "Power? I am the surge." Epic music swells in the background.

A crystal-clear daytime skyline. On a mountaintop overlooking a city, a superhero in a flowing emerald cloak levitates, arms outstretched. Glowing green energy spirals around their body, lifting massive boulders into the air with telekinetic force. The camera sweeps around them in dramatic slow motion as their emerald eyes flash. With a calm, commanding voice, they declare: “I move the world—with my mind.” Epic wind and ethereal choral music play.

just fiddling around

[honest meteor]

Is there a way to input greek characters into veo3?

[tough parrot]

this is not the place I'm aware of it but did you try Sora? I did not, I just wonder how it is.

[worthy cobalt]

Visual Style:
• Futuristic dystopian London — neon lights, misty air, metallic towers, deep bass pulses in the atmosphere
• A massive gathering of humans and androids in a plaza-like setting, under a glowing sky
• The two mixed-race lead characters stand on a raised digital platform, hyping the crowd
• Crowd chants echo like a revolution is brewing
• Synth-heavy score with deep drops and glitchy FX

⸻

🎭 Characters:
• Commander Kai – Mixed-race male, cyberpunk look, glowing jacket, voice-modulated mic
• Oracle Nia – Mixed-race female, glowing eyes, radiates energy, part-machine
• The Crowd – Large futuristic crowd of humans, aliens, and androids shouting in unison

⸻

🗣️ SCRIPT – SPOKEN WITH INTENSITY

[CAMERA FLIES OVER CROWD, LANDING ON THE STAGE]

KAI (booming voice):
“London… are you receiving this frequency?”

NIA (echoed, powerful):
“The portals are opening… two dates. Two gateways.”

[CROWD RUMBLES, ENGINES HUM]

KAI (raises arm to sky):
“Say it with us—”

[CROWD CHANTS, SYNTHS RISE UNDERNEATH]:
“9TH OF JULY — DASH THE HENGE!”
“15TH OF JULY — CLUB SILLY, ARCH 555, BRIXTON!”

[FLASHES OF LIGHT, SKY PULSES, CAMERAS SWEEP THE CROWD]

NIA (grinning):
“Plug in. Power up. Perform.”

KAI:
“Bring your beats. Build the future.”

[BEAT DROPS – CROWD ERUPTS INTO DANCE / CHAOS / LIGHT]

[ON-SCREEN TEXT IN HOLOGRAPHIC STYLE]:
⚡ THE SONIC ALCHEMIST OF LONDON ⚡
🎛️ ELECTRONIC OPEN MICS
📍 JULY 9 — DASH THE HENGE, CAMBERWELL
📍 JULY 15 — CLUB SILLY, ARCH 555, BRIXTON
🔊 BE SEEN. BE HEARD. BE STRANGE.

[FINAL SHOT – CROWD CHANTS ONE LAST TIME TO THE SKY]:
“9TH OF JULY — DASH THE HENGE!”
“15TH OF JULY — CLUB

[static topaz]

<@&1173646484311384114> how to know when my credits reset?

[main rose]

same day of the nxt month you bought the plan

[static topaz]

idk when i bought

[main rose]

check in the subscription info

[acoustic galleon]

does anyone know how what prompt i need to give veo to make an object be sliced into multiple slices and make each slice fall over?

cause ive tried everything and it keep slicing the object in half 💀

[tough parrot]

use gemini for that kind of promt errors

[acoustic galleon]

i dont understand gemini ive tried it but idk how it works even though its similar

[pulsar musk]

https://app.parler.com/daveabao12?ref=RGHQIJyHD7

[warped compass]

Did anyone here create an “pov fast pace motorcycle racing” video prompt

[sinful spoke]

@dim gull Mr Guru 🙂 i do have 3 animated (pixar 3d) characters and in one scene i need to explain how they are all doing completely seperate things. Any Trick for Prompting 3 subjects+Action?

or..can i blend Videos somehow? 🤔

[dim gull]

3 subjects in one scene is going to be tough depending on what you want them doing. If you can direct it so there are multiple clips with different angles so you can focus on those characters specifically. Directing a bunch of actions for a single 8 sec clip might be better served over three clips if it can be helped. Otherwise make sure each subject is named, and then use the names to direct their actions.

EG: Action: @Character1 does this thing in the foreground, while @Character2 is doing this thing on the left side of the camera view, and @Character3 is here doing this thing.

Veo 3 is best when you have a single subject doing a single thing right now. The more layers of complexity you add the more you'll have aberrations like improper physics, or inconsistent characters/scenes which will ultimately burn creds trying to get it right.

[sinful spoke]

👍 ok let me try that, i have this chameleon and in one scene i want it to to kinda "break the 4th wall" crawl along the camera (the glas from the inside...does that make sense) and directly look at the viewer. Just my english skills are...well apparently not good enough to explain that to an ai 😂

[north jacinth]

Paste the character into Chatgpt and have it describe everything about how the character looks and should act
Then use that character information in the google veo 3 prompt

[sinful spoke]

god damn it i always forget i can do that

thanks

tooo many AI tools 😂

[sinful spoke]

maan....even with chatgpt this is tough, tried to get an image first, i can get it so far, its indeed in the foreground looking at the screen but i want it to be "stuck" to the glass from the inside, with his hands

like with two hands "stuck" to the glass

[sinful spoke]

goood damn finally @dim gull , not exactly the matching art style but finally getting somewhere

[glacial sail]

Can someone help me describe in prompt style a very fast super-hero launch that creates a crater underneath and has the character fly off the top side of the screen? My character just jumps forward

[quaint jungle]

@dim gull tried it out i think the output is cool

[mellow elk]

Create a vertical video styled like a viral Instagram Reel: Moses from the Bible holding a phone in selfie mode, recording himself while standing confidently in front of the parted Red Sea. He’s wearing flowing ancient robes, wind whipping through his beard and clothes. Behind him, towering walls of water churn violently with epic motion, and a stampede of Israelites in ancient desert clothing run frantically past him on the dry seabed, glancing back in fear. Moses speaks calmly to the camera as chaos erupts around him, keeping his focus as a leader. The sky is dramatic with storm clouds and golden light breaking through. The camera is in selfie perspective, slightly shaky like a phone vlog, with cinematic lighting and fast-paced energy. Add subtle lens flares, modern viral-style subtitles, and Instagram Reels-style caption overlay at the bottom: ‘Moses Vlogs the Red Sea Escape – #ExodusChronicles’. Sync movement to suspenseful, cinematic trap-style background music with bass drops timed to people running. High-resolution, fast-paced transitions, vertical format (9:16), made for Instagram Reels

/veo prompt: "Moses holding a phone in selfie mode, confidently vlogging in front of the parted Red Sea. Towering, vicious walls of water churn behind him while hundreds of Israelites run past frantically on dry ground, escaping Pharaoh's army. Dramatic golden light breaks through stormy clouds, sand and wind whip through the air. Moses stays calm as chaos unfolds. Cinematic lighting, viral selfie cam energy, intense biblical atmosphere." aspect_ratio: 9:16 length: 12s style: cinematic

[nocturne pivot]

any idea why this prompt is always flagged??

A young woman with a smooth, doll-like face. She has soft peach-toned skin with natural freckles across her nose and cheeks, large glossy hazel eyes with long curled lashes, and full bubblegum-pink lips coated in ultra-reflective gloss. Her platinum-blonde hair is styled in two high ponytails with pastel pink streaks. She has a delicate septum ring adorned with small diamonds centered below her nose. She wears dramatic, extra-long gel nails in translucent candy colors with rhinestone accents. Her expression is gentle and observant, often blinking slowly or reacting with subtle curiosity.

ASMR close-up. She sits at a wooden table under warm ambient lighting. In front of her is a vibrant red glass apple, shaped like real fruit with a glossy surface. She lifts it gently with both hands and brings it toward her mouth. As her teeth press in, a soft cracking sound is heard as the shell splits. A bit of thick syrup glistens at the edge.

She softly says:
“Wow… this is crunchy.”

She continues to chew slowly as light crackling sounds echo.

[worthy hatch]

why is this being flagged?

in vlog style: a group of unruly adults broke into a cloning facility to show evidence to celebrity cloning and explain it, behind them are pods with celebrities floating inside. they zoom in to one that looks like p diddy sleeping with a bottle of baby oil

[north jacinth]

bruh

[worthy hatch]

content creation at its finest

at least help me reword it lmao

[dim gull]

Try not putting young woman. Explicitly put woman in her early 20s.

[worthy hatch]

i did it!

[fossil stream]

Anyone struggling with any prompt rn

I’m bored and wanna help sum people out so just lemme know 1

[zinc spear]

here's an update on what I'm working. haha I tried to cut the scenes separately as advised here. The cat finally talked 🤣 but needs more improvement here.

[mortal reef]

Ok

[autumn patrol]

@fossil stream 👋

[fossil stream]

Sup

[sinful spoke]

Do we have anyone with a broom to remove that weird sca** up there? Wheres housekeeping? 😀🤣

[sinful spoke]

ahh thanks whoever it was 🤣

[tawdry sable]

this got me dying 💀

[valid oyster]

which prompt can i use to get that vlog type videos

[reef sorrel]

A photorealistic, selfie-style video from a wide-angle lens, held by a large gorilla wearing a bright orange prison jumpsuit inside a stark, cold, concrete solitary confinement cell under a harsh fluorescent light. He looks directly into the lens, his voice echoing slightly as he says, "So yeah, me and Bubbles got into a fight and now we are in solitary confinement." He turns the shaky, handheld camera to the side, revealing another gorilla in an identical jumpsuit gripping the cold steel bars of an adjacent cell. As the camera pans, we hear the faint clank of metal and the second gorilla's muffled voice, "I wish I had baby oil in that fight." The first gorilla turns back to the camera, his expression shifting to defiance as he declares, "Nah, I'm escaping this prison, I don't care what y'all say," with a low, tense hum of the overhead light filling the silence.

tried to replicate it, result may varies.

[valid oyster]

thanks

[tawdry sable]

nah people on tik tok going wild with veo 😭

[upbeat lichen]

Hello
how do you get a video that lasts more than 8 seconds please ?

[dim gull]

You need to generate additional clips and add them together. You can do this in flow with the Jump-To feature in the scene builder if you need to maintain audio, or you can use Extend if you do not.

[prisma zinc]

OMG omg

[timid schooner]

Hi

[coarse radish]

I need this

[north jacinth]

thats even more better

[iron axle]

why am I not able to generate videos of other people?

[wintry sinew]

Results may vary

[wintry sinew]

Just gotta describe the person a bit

[reef sorrel]

https://tenor.com/view/will-smith-close-up-huh-sad-gif-17800869

[iron axle]

I did

My prompt was P Diddy, Ava Kris Tyson, Dr Disrespect, Drake, and EDP445 chasing 100 kids like it's a MrBeast video

[dim gull]

So yeah using real people's names will likely flag the system and so will adding children, especially when you're implying something explicit and you've done both.

[frozen sonnet]

#flow-tips-n-tricks・✏️ A photorealistic, selfie-style video from a wide-angle lens, held by a large gorilla wearing a bright orange prison jumpsuit inside a stark, cold, concrete solitary confinement cell under a harsh fluorescent light. He looks directly into the lens, his voice echoing slightly as he says, "So yeah, me and Bubbles got into a fight and now we are in solitary confinement." He turns the shaky, handheld camera to the side, revealing another gorilla in an identical jumpsuit gripping the cold steel bars of an adjacent cell. As the camera pans, we hear the faint clank of metal and the second gorilla's muffled voice, "I wish I had baby oil in that fight." The first gorilla turns back to the camera, his expression shifting to defiance as he declares, "Nah, I'm escaping this prison, I don't care what y'all say," with a low, tense hum of the overhead light filling the silence.

[iron axle]

But it’s just chasing the children, I’m not asking it to show the whole kit and kaboodle of them… doing the deed

And I’ve seen tons of other AI diddy videos, how do people make those if using real names flags the system?

Like even doing the famous will smith eating spaghetti or eating lunch or whatever flags it, so how do others make those?

[reef sorrel]

by not using Veo 3

lol

If u r really that desperate to generate videos for your fantasy, there are plenty of video generation models out there that can handle it

[iron axle]

I’m just doing it because it’s the current trend, stuff about diddy, trump, labubu, squid game, etc are popular rn and the goal is just to maximize views so why not take advantage🤷‍♂️

Did you not see during the diddy trial the AI videos of the trial, the channel that was making them grew to 1M within a month

[reef sorrel]

like i said

theres other video generation models out there that can fulfill ur fantasy

im not judging

[reef sorrel]

Cool

[iron axle]

glad someone else had a similiar vision to me

@worthy hatch did you manage to find out how to do it with veo 3 or if not what other tool did you use?

[reef sorrel]

have u submit your concern in #1319424187730755636 yet

[iron axle]

I haven't, thank you for your help tho I think you answered it unless this other guy has figured it out

[reef sorrel]

if many ppl experiencing it it should be enough to get the staff attention /labs team

As for generating celebrities, they recently tightened the safety filters though that mainly applies to user-uploaded images

wait let me try on gemini

[iron axle]

@reef sorrel can you join a vc if possible rq?

[iron axle]

alr thanks, lmk when it is finished

[reef sorrel]

oh i actually cant do voice rn cuz im at cafe rn

[iron axle]

oh np then

[iron axle]

damn

gemini is what? X?

oh its also google

[reef sorrel]

yeah

[iron axle]

google veo is better right or gemini?

[reef sorrel]

u can generate veo 3 videos on various websites from google

Flow.google
vids.google.com
gemini.google.com

[iron axle]

yea but veo 3 costs a lot lmao

ohh I see

[reef sorrel]

i mean if u have pro plan, and ran out of credit in flow. u can always use vids/gemini

it doesnt use credit system , but it uses limit per day

[iron axle]

you can buy more credits right?

[reef sorrel]

yep

[iron axle]

do you have a channel?

[reef sorrel]

nope, im actually just doing this for funsies

[iron axle]

oh cool

Na I'm trying to print a little if you know what Im saying

[reef sorrel]

but i think theres actually many ppl here who got the same idea like u

[iron axle]

Until and if youtube, tiktok, etc demonetizes AI videos, I'm going to take advantage of it

I kind of highly doubt they will but I'm still gonna get in it early here if it gets demonetized later

@reef sorrel might be a little off topic but do you think I can get sued if I use real people in AI videos...?

[reef sorrel]

Uhhhh

[iron axle]

Like even for something like the will smith vids

[reef sorrel]

I don’t think so

[iron axle]

oh ok thanks

[reef sorrel]

Maybe you will get something

Which is shadow ban.. copyright strike etc

Or maybe ur acc will get demonetized

I think hailuo is the only website that can do celebrity ?

[iron axle]

Thanks I'll check it out

[reef sorrel]

I’m not sure if their new model can do it

[iron axle]

maybe I should delete that tho lmao

[reef sorrel]

Nah I think you’re fine

We’re just discussing ✨

[iron axle]

It would be kinda wild tho if diddy of ALL people actually sued me cause I made AI videos of him lmao

[worthy hatch]

They'll demonetize you if they find out. And if you use cues that resembles celebrities then the repository will build a likeness

[iron axle]

WHAT'D you use? did you use veo 3?

[iron axle]

I was asking if they demonetize videos or whole channels for being AI generated, not the celebrity thing but thats good to know

[worthy hatch]

Flow

And not yet. I haven't found anything out about ai being demonetized besides the fact that if you don't use the ai tag then it might

[frigid kestrel]

So if i am posting videos using veo 3 but there is no water mark, then i still have to have a tag saying this is AI?

Cause if i need to mention this is AI, no problem

in ANY post I do using any type of AI generated content?

[reef sorrel]

yea

[worthy hatch]

Yeah just to be safe, otherwise they'll demonetize if the content is too much or too sensitive

[iron axle]

if it's too much AI?

[frigid kestrel]

Thank you for the tip, really helps for future posting

[jovial yew]

umm guys how to access the free student version of veo3 and just wanna ask if it's worth it

[strong epoch]

Hi guys just want to know how can make the cut or the separation of two halves is where the exact point cut the knife of the object.

I have problem with this always the cut is on the center not where the knife cuts it. 😔

[worthy hatch]

Too much as in abrasive or unhinged content. Mainly false propaganda or slandering

[worthy hatch]

What's your prompt

[lyric leaf]

Does it not work iOS?

[strong epoch]

Subject: An apple made entirely of interlocking Lego bricks, each meticulously crafted from shining porcelain in a range of vibrant colors, predominantly reds, greens, and blues. The porcelain gives the apple an intensely glossy, reflective surface that catches and scatters the studio lighting beautifully.

Action: A disembodied hand, entering from the right of the frame, wields a gleaming stainless steel knife to precisely slice the Lego-porcelain apple in half. The two halves are now separated cleanly along the cut line, showcasing the internal structure of perfectly divided Lego bricks and smooth porcelain glaze on each side. On the lower portion of the screen, the words "CUTTING CRYSTAL" dynamically twist and morph into the word "SUBSCRIBE", completing the transition with a satisfying click.

Scene: The setup is a brightly lit professional studio with clean, uncluttered surfaces. The two halves of the Lego-porcelain apple rest separated on a dark, solid wood countertop that is designed to beautifully reflect the studio lighting, clearly showing the clean division. The background remains a seamless white, ensuring all attention is focused on the object and action.

Style: A close up shot at eye-level keeps the focus entirely on the Lego apple and the cutting action, all enhanced with a cinematic, dramatic film grade and clean, professional studio lighting creating a joyful and celebratory atmosphere, 16:9 aspect ratio.

Dialogue: N/A

Sounds: The gentle clink of the knife against the Lego bricks, followed by a crisp, amplified slicing sound highlighting the clean separation, and a final distinct, playful click as the text transforms.

Technical (Negative Prompt): subtitles, captions, watermark, text, words on screen, logo, branding

[reef sorrel]

A cinematic, eye-level close-up shot focuses on a vibrant apple made of interlocking, glossy red, green, and blue porcelain Lego bricks, centered on a dark, reflective wood countertop against a seamless white studio background. A gleaming stainless steel knife descends vertically into the frame, making a gentle clink on contact before slicing cleanly from top to bottom with a crisp, amplified slicing sound. As the knife lifts away, the two halves slide perfectly horizontally apart, remaining upright on the countertop to reveal their smooth, cream-colored interior cross-sections in a physically logical and satisfying final composition.

i tried

[dim gull]

Sometimes it's the weirdest things that give Veo problems.

https://promptalchemist.com/share/pfzALih5tKd3wwuZ6iXD

[ripe parrot]

https://www.tiktok.com/@nonsensedaydreaming/video/7520536340463635742?is_from_webapp=1&sender_device=pc&web_id=7519210840357864966

Anyone hungry?

https://www.tiktok.com/@nonsensedaydreaming/video/7524011705379179806?is_from_webapp=1&sender_device=pc&web_id=7519210840357864966

[strong epoch]

Thank you guys for the help. ☺️🙏

[dusty schooner]

https://x.com/IamEmily2050/status/1941834470640021669?s=09

Emily (@IamEmily2050)

Veo3 fast on the Gemini app

{
"shot": {
"composition": "Alternating close-ups and medium shots, 50mm lens for close-ups and 35mm for medium shots, shot on ARRI Alexa Mini, shallow depth of field",
"camera_motion": "slow handheld with subtle push-ins during emotional

X · Jul 6, 2025, 12:19 PM

https://x.com/IamEmily2050/status/1941885053619068984?s=09

Emily (@IamEmily2050)

Veo3 fast on Google Gemini App, let's hope so many people go viral with these videos 🙏😀

{
"concept": "The 23-year-old Japanese woman and her 16-bit pixel avatar perform a TikTok trend dance in perfect sync, and at the end of the choreography, they both snap back into the

X · Jul 6, 2025, 3:40 PM

https://x.com/IamEmily2050/status/1941912850060165293?s=09

Emily (@IamEmily2050)

Veo3 fast on Google Gemini App

{
"concept": "A surreal, heartfelt wedding between a 23-year-old Korean woman and a pixelated 16-bit Yeti. They stand together at the altar, gazing into each other’s eyes with joy as the pastor asks if they agree to marry each other. A gentle,

X · Jul 6, 2025, 5:31 PM

https://x.com/IamEmily2050/status/1941896040485138602?s=09

Emily (@IamEmily2050)

Google Veo3 fast, on Google Gemini App 😀

{
"concept": "Inside a Korean supermarket at night, a 23-year-old Korean woman and a pixelated 16-bit Yeti perform a trending TikTok dance in the middle of the aisle. Shoppers stop and watch in awe, amusement, and disbelief as the real

X · Jul 6, 2025, 4:24 PM

https://x.com/IamEmily2050/status/1941573122836328636?s=09

Emily (@IamEmily2050)

Veo3 fast

{
"shot": {
"composition": "Extreme close-up, 50mm lens, shot on ARRI Alexa, shallow depth of field, filmic texture",
"camera_motion": "whip pan from tight on face to mid-shot circling around",
"frame_rate": "24fps",
"film_grain": "fine Kodak 5219

X · Jul 5, 2025, 7:01 PM

https://x.com/IamEmily2050/status/1941520109371363717?s=09

Emily (@IamEmily2050)

Veo3 fast

{
"shot": {
"composition": "Epic aerial wide shot, 80mm lens, shot from a helicopter using gyro-stabilized camera system, deep focus, high-resolution documentary style",
"camera_motion": "smooth tracking pan following the arc of the whale’s breach",

X · Jul 5, 2025, 3:30 PM

https://x.com/IamEmily2050/status/1941483840243105933?s=09

Emily (@IamEmily2050)

Veo3 fast

{
"shot": {
"composition": "Macro close-up, 85mm lens, ultra-shallow depth of field, locked-off static shot, clean tabletop composition",
"camera_motion": "none",
"frame_rate": "24fps",
"film_grain": "soft organic film texture overlay"
},

X · Jul 5, 2025, 1:06 PM

https://x.com/IamEmily2050/status/1941126453715948005?s=09

Emily (@IamEmily2050)

Veo3 fast

{
"shot": {
"composition": "High-angle tracking shot from a helicopter, 200mm telephoto lens on a stabilized gimbal system, shot on RED Helium 8K S35",
"camera_motion": "aerial tracking following the emus' path, with gradual zoom-in",
"frame_rate":

X · Jul 4, 2025, 1:26 PM

I will keep sharing more prompts, hopefully it will help someone 🙏

[north jacinth]

how does the json script output compare to a regular paragraph prompt?

[iron axle]

Hailuo doesn't seem to generate any audio for it tho 😭

[iron axle]

Which one do you guys think is the best out of these?

[finite briar]

that's what I was thinking, like does the json script make a better outcome then it would with a regular prompt?

[north jacinth]

Paragraph, no json script input to the prompt builder

Spider-Man sits comfortably at a low table in Tony Stark’s private office suite, the New York skyline glowing through the tall windows behind him as he records a relaxed sushi mukbang vlog. The table is laden with artfully arranged platters of nigiri, maki rolls, sashimi, and hand rolls, illuminated by soft studio lighting that casts even, diffused illumination and subtle backlighting for depth. Fully suited in his red‑blue‑gold MCU Nano Suit with a fully sealed mask displaying only expressive white eye lenses, he uses chopsticks to pick up intricate sushi pieces and gestures animatedly as he offers quips—randomly chosen about the spiciness of wasabi, fresh sashimi, Queens seafood stands, or his chopstick skills—delivered in a natural New York accent with lip‑sync inferred through subtle suit movements. A single, static wide‑angle camera set to 9:16 aspect ratio at 60fps sits fixed on a tripod, perfectly framing Spider-Man at the center of the shot and the sushi filling the lower third, with no tilts, zooms, or motion. The visual style is photorealistic and cinematic, with sharp detail on suit textures, glossy fish scales, and rice grains; minimal HUD overlays show sushi names and ratings, and there are explicitly no morphing, distortions, warping, disfigurements, or disconformity. Ambient city hum is faintly audible through the windows, but Spider-Man’s clear human voice remains the focus of the playful, nerdy, self‑aware downtime vibe.

Json scrip input version

[vale field]

anyone notice in the past few days any prompt that involves cutting, the knife incision and the actual cut are different directions? mostly perpendicualr to each other?

[robust vessel]

Imagine spenderman drink a milkshake

[north jacinth]

anything is possible lol

[robust vessel]

@north jacinth I'll try 😂😂

[balmy wind]

An animated white and grey striped munchkin cat wearing Gen Z streetwear — puffer vest, cargo pants, gold chain, and tiny sunglasses — struts down a palm-lined street in Los Angeles during golden hour. The cat holds a glowing, holographic bag of catnip and approaches a trendy Gen Z man with curly bleached tips, a mesh shirt over a vintage band tee, ripped jeans, and colorful Crocs, sipping an iced matcha. With a suave voice, the cat asks, “Would you rather take this catnip... or double it and give it to the next feline?” The guy smirks and replies coolly, “Double it.” The camera cuts to a dramatic slow zoom on a middle-aged woman standing on the sidewalk — wearing yoga pants, an oversized UCLA hoodie, and carrying a small dog in a sling — who instantly grins and blurts out, “I’ll take it!” Warm sunlight flares behind her as the cat tosses the doubled bag her way. No music — just ambient LA street sounds and comedic timing.

[rare rover]

can any one tell how to veo 3 video in portait mode in google flow

[barren eagle]

<@&1366831844502274180>

It is in every channel

[orchid jay]

I'm having trouble making a prompt with Miniature characters working on regular sized items

Some of the actors don't move or do random stuff.

[atomic sorrel]

hey! i'm new here, could i kindly ask you how you managed to get the 9by16 format?

[barren eagle]

Say it in the prompt and say make it sideways and slanted

[icy scaffold]

Hello, I would like to know if it is possible to make a video by using a prompt & and a picture. My objective is to create a video with a picture of my friend and have him talk about something with an unrealistic person.

[atomic sorrel]

goated man! thank u

[barren eagle]

Yw did it work?

[atomic sorrel]

i'll try now real quick!

nope, it gave me some cinematic bands for some reason, as if it was shot on an anamorphic lens

[hot lynx]

Gemini being dooki rn

[faint frigate]

#flow-tips-n-tricks・✏️ A stunning full-length portrait of a caramel-skinned African American woman with flawless purple makeup, glossy caramel nude lips, and dramatic full lashes. Her long black wavy hair features striking blonde ombré tips, cascading over her shoulders. She’s wearing a sleek purple halter top adorned with subtle black snake print patterns, paired with sparkling silver diamond dangling earrings. Her long purple nails are intricately designed, adding to her bold and glamorous vibe. She poses with one hand elegantly resting under her chin, exuding fierce confidence. The background is a dark urban alleyway, softly illuminated by glowing purple puddles and vibrant reflections from nearby neon signs, creating a moody yet mesmerizing atmosphere.

[rancid hound]

Dm'ed

[jaunty igloo]

Yes possible with any image-to-video gen model, also a lip-sync AI tool.

[weary flume]

Hello everyone, how can i generate videos on discord? Is it even possible or i may do it only through web?

[jaunty igloo]

Only on web.

[jaunty igloo]

Hi 🤗

[sinful spoke]

From a fun challenge i Posted in #flow-general・🌊 , does anyone know why this is Happening? The exact prompt in the screen. Result as seen. I had the same with "Man Puts in USB Stick first try" which always resulted in a man failing to....did the API train those "Myths" technically? Its fun to play with it after all.

[real mural]

Elven princesses, cloud spirits, and glowing fairies visiting the boutique, admiring shimmering magical gowns, gowns glowing with soft enchantment, fairy customers surrounded by sparkles"

[supple shard]

/imagine a dimly lit Italian catacomb with arched stone walls, viewed from the upper corner like a grainy security camera angle, a small rectangular glass coffin resting on a stone pedestal lying lengthwise in the center, inside is the vague outline of a preserved child in a lace shroud, eerie bluish-green tint, faint fisheye distortion, timestamp overlay that reads "CAM 4 – Capuchin Crypt", candleless, clinical lighting, low-res surveillance camera quality --v 6 --ar 16:9 --style raw

[sullen glen]

Effect Name: Fiery Explosion
Category: Thermal
Key Triggers: explosion, fiery, fireball
Visual Description: A classic fireball, thick black smoke, and scattering, burning debris.
Sound Profile: A loud BOOM, the crackle of fire, and a deep rumble.
Effect Name: EMP Burst
Category: Energetic
Key Triggers: EMP burst, electromagnetic pulse
Visual Description: A rapid flash of blue-white light, a visible ring of distortion in the air, and electrical arcs. No fire.
Sound Profile: A rising hum, a sharp VWOOMP, and the crackle of static.
Effect Name: Cryogenic Shatter
Category: Thermal / Kinetic
Key Triggers: cryogenic shatter, brittle fracture
Visual Description: A blast of cold white vapor. The object is covered in frost and shatters into brittle fragments.
Sound Profile: A sharp HISS, followed by a loud CRACK like breaking ice.
Effect Name: Gravity Implosion
Category: Kinetic / Gravitational
Key Triggers: implosion, gravity well, crushing force
Visual Description: The object doesn't explode but is crushed and collapses inwards to a single point. Light and shadows are distorted towards the center.
Sound Profile: A deep WHOOMPH, followed by the sounds of groaning and tearing metal.
Effect Name: Arcane Nova
Category: Energetic (Fantasy)
Key Triggers: arcane nova, magical blast
Visual Description: A sphere of multi-layered, iridescent energy (violet, green, gold) with fractal patterns.
Sound Profile: A rising choral hum, with the sound of crystalline chimes.
Effect Name: Shrapnel Burst
Category: Kinetic
Key Triggers: shrapnel fragmentation, shards
Visual Description: A brief flash, followed by hundreds of sharp metal shards projected outwards, creating sparks on impact.
Sound Profile: A sharp CRACK, followed by the chaotic whizzing and ricochet of shards.

part 2

Effect Name: Organic Growth
Category: Creation (Flora)
Key Triggers: bursts with life, instantaneous growth
Visual Description: Thick vines erupt from the object, rapidly growing and covering themselves in bright flowers.
Sound Profile: The crunch of the breaking object, followed by rapid rustling and "growing" sounds.
Effect Name: Bio-Transformation
Category: Creation (Fauna)
Key Triggers: biomechanical flesh, pulsating organic mass
Visual Description: The emergence of a pulsating, wet, organic mass mixed with chitinous elements.
Sound Profile: Wet, squelching sounds, with a rhythmic heartbeat.
Effect Name: Particle Beam Overload
Category: Energetic
Key Triggers: particle beam overload, containment failure
Visual Description: A beam of heat strikes the target, causing it to overheat before detonating in a blast of pure energy.
Sound Profile: A high-frequency hum, the crackle of an overload, and a sharp energy explosion.
Effect Name: Volumetric Smoke Puff
Category: Pyrotechnics (Non-fiery)
Key Triggers: volumetric smoke puff, movie pyro
Visual Description: A powerful burst of dense, opaque white smoke that rapidly fills the space.
Sound Profile: A deep, muffled "FOOMPH".
Effect Name: Underwater Cavitation Blast
Category: Hydrodynamics
Key Triggers: underwater cavitation, vapor bubble collapse
Visual Description: A sphere of vapor forms underwater, which then violently collapses, destroying the object.
Sound Profile: A deep, muffled, underwater "BLOOMP", followed by a sharp "CRACK".
Effect Name: Genesis Pulse
Category: Transmutation
Key Triggers: genesis pulse, coalesce and form
Visual Description: An object is converted into a sphere of energy, from which a living creature is "sculpted".
Sound Profile: Chimes/bowls, followed by wet, organic, formative sounds.

[muted prism]

How to make film w AI

[tawdry violet]

Hey guys, how’s it going? I’m from Brazil, and I create my content in Portuguese, but I’d love to chat about some prompts and the challenges I’m having while creating.

How are you managing to keep the characters you generate consistent?

Oh, and don’t be surprised—my character always wears a leopard-print speedo. lol.
That was the best way I found to make it eye-catching and have a signature style.

I play around a lot with the character, changing their look in different videos. This was the video where I managed to keep the character the most consistent so far. https://www.instagram.com/reel/DLGG07fPWao/?utm_source=ig_web_copy_link&igsh=MzRlODBiNWFlZA==

amaurydesunga

Desvende o segredo para nunca mais ter medo do Imposto de Renda! 🤫

A receita do sucesso (ou quase isso) segundo o nosso Contador Criativo:
Menos verdade, mais criatividade. 😉

No final das contas, o importante é... dar um jeito!

🗣️ O que você achou da filosofia do Amaury? Genial ou caso de Polícia Federal? Comenta aí! 👇

#veo...

Likes

1425

Instagram

[woven condor]

Hey, what’s up man — that leopard-print speedo cracked me up.

I’ve been working on a character-driven series too, and yeah, Veo3 struggles sometimes with continuity, especially across different scenes. What’s helped me the most is locking in the exact wording of the character description and reusing it without changing anything. Even small phrasing shifts can throw Veo off. If I change anything, I duplicate the old prompt and build from there so I don’t lose the structure.

Also, scene complexity isn’t really the issue. I’ve use 700+ word prompts with detailed motion and emotional beats that still came out clean. The key is being precise and intentional. If you overload randomness, it derails, but if everything in the prompt is supporting the same visual logic, it can absolutely handle it.

Veo3 still has its quirks, but when I anchor key traits and keep the character’s tone, emotion, and motion consistent across scenes, it tends to lock better, even without Ultra.

[olive temple]

immense skill

[hybrid owl]

The character looks in the mirror. He doesn't recognize himself in the mirror. He was once human, but now he's a mutant. Realistic and highly detailed lighting, rough texture, and a long shot that transitions to a close-up of his face, showing the distinctive texture of his fur and the intensity of his gaze.

[tepid garden]

#flow-tips-n-tricks・✏️ how to write prompt to generate video and where ? In flow prompts or where ? Plz help

[severe aspen]

hey there! welcome to the community. in order to make videos with Flow, you'll need to visit flow.google and have a Google AI Subscription: videos will not generate directly here in the server.

if you have a prompt you'd like to have generated, you can share it here and there's a chance that someone in the community can generate it for you. please note that this is up to members of the community and is not a requirement from Flow creators.

check out Flow here
here's more information on Google AI Subscriptions

[humble galleon]

A stylish and mysterious woman named Pérola, with indigenous features, warm brown skin, 1.58m tall. She has long, straight dark hair with a short fringe and a soft half-up hairstyle. She is walking confidently and professionally on an outdoor runway surrounded by greenery during golden hour. She wears a monochromatic off-white outfit: a ribbed off-the-shoulder top, high-waisted straight-leg pants with a subtle cut-out at the waistband, and nude platform sandals. Her posture is graceful, her expression elegant yet subtly smiling, exuding charm and power. Makeup is minimal and sophisticated, enhancing her natural beauty. The background is softly blurred with modern decor and plants. Cinematic lighting, hyperrealistic detail in skin, fabric, and movement. Shot in luxury fashion magazine style with a shallow depth of field, 85mm lens, f/1.8. Target audience: successful, elegant women who appreciate refined, minimalist beauty without exaggeration.

/generate image: [faça upload da imagem da modelo]
prompt: Animate this real photo of a stylish woman posing in front of a mirror... [prompt completo]

A beautiful Brazilian woman named Pérola, 1.58m tall, with morena skin leaning towards dark brown, subtle freckles on her cheeks, and a confident yet soft expression with a mysterious smile. Her hair is straight, shoulder-length, black, with short bangs above the eyebrows. Her body is sculpted and naturally feminine. She stands confidently, radiating elegance and natural beauty, targeting a high-end, successful female audience. She is wearing a stylish off-the-shoulder ribbed knit top in cream color and high-waisted off-white pants with a cut-out detail at the waist. She also wears nude platform heels. Her look is sophisticated and modern, without exaggeration. Minimal makeup, elegant earrings, and a subtle glow on her skin. The background is softly lit, studio-style or boutique-like, with neutral tones to highlight her outfit. Ultra-realistic, full body, fashion editorial style. --v 6 --ar 9:16 --style raw

[polar marten]

anyone know how to get a good sounding female voice.. not a robot. using veo 3 and keep getting a robot female voice.

[visual basin]

Define an specific Accent, and also if the Dialogues are too long for 8 secs the voices speeds up to fit the 8 secs

[night cove]

wow this looks so real lol

[rugged yew]

magine A hyper-realistic crystal avocado being sliced open with a sharp knife on a wooden table. 4K, cinematic lighting, ASMR slicing, slow motion

/flow prompt: A hyper-realistic crystal avocado being sliced open with a sharp knife on a wooden table. 4K, cinematic lighting, ASMR slicing, slow motion

The character looks in the mirror. He doesn't recognize himself in the mirror. He was once human, but now he's a mutant. Realistic and highly detailed lighting, rough texture, and a long shot that transitions to a close-up of his face, showing the distinctive texture of his fur and the intensity of his gaze.

[rugged yew]

A beautiful indian woman named Pérola, 1.58m tall, with morena skin leaning towards dark brown, subtle freckles on her cheeks, and a confident yet soft expression with a mysterious smile. Her hair is straight, shoulder-length, black, with short bangs above the eyebrows. Her body is sculpted and naturally feminine. She stands confidently, radiating elegance and natural beauty, targeting a high-end, successful female audience. She is wearing a stylish off-the-shoulder ribbed knit top in cream color and high-waisted off-white pants with a cut-out detail at the waist. She also wears nude platform heels. Her look is sophisticated and modern, without exaggeration. Minimal makeup, elegant earrings, and a subtle glow on her skin. The background is softly lit, studio-style or boutique-like, with neutral tones to highlight her outfit. Ultra-realistic, full body, fashion editorial style. --v 6 --ar 9:16 --style raw

[fallen pier]

lol

[robust grotto]

What techniques are everyone using to ensure character continuity from prompt to prompt?

I've had some success with an initial ingredient-to-video and then using that to create frames for subsequent clips, but it's expensive to create the first shot

[woven condor]

Continuity in Veo3 can definitely get shaky across scenes. What’s helped most is locking in the exact character wording — even small changes throw it off. I usually duplicate older prompts and build from there so I don’t lose structure.

The issue usually isn’t complexity. I’ve used 700+ word prompts with detailed motion and emotion, and they still came out clean. As long as the visual logic is tight and everything supports the same idea, it works. But if there’s too much randomness, it breaks.

Veo still has quirks, but when I keep tone, traits, and motion consistent, it usually stays locked.

[dark nacelle]

Yeah remembering to keep things consistent across a(n eventual) bulky document can get daunting.

[brisk solstice]

https://www.tiktok.com/t/ZT6derEwK/

[stiff root]

i have this prompt
but red suggestion on how to add our bottle of hot sauce

Golden afternoon light, rustic Louisiana kitchen.
Slow-motion close-up: bubbling pepper sauce in a copper pot, steam curling like smoke from a fire.
Hands stir gently with a wooden spoon, then lift it — glowing red sauce drips slowly back in.
Camera pushes in on the Parish Heritage Pepper Co. bottle beside a handwritten family recipe card.
Warm fiddle music swells under ambient kitchen sounds.
Mood: cozy, nostalgic, homemade heat.

Final frame: Text overlay — “Tradition You Can Taste.”

⸻

Want a version with dialogue, or one focused more on the peppers themselves?

[empty rapids]

A hot girl dressed in formal attire - top and shorts with vibrant texture
Having long boots in her feet,shining formal sunglasses with a beret on his head and formal wrist watch narrow belt
She confidently standing posing in a way her left foot little far from other one as she is walking through bustling futuristic city
She hold a formal hand bag with her hand,
Create a hyper realistic cinematic image

[stoic bluff]

Hello, does anyone have the prompt character of the yeti and the bigfoot that I put in the photo? I've been trying for days to get closer to these characters but impossible, thank you

[livid canopy]

could u share 2 prompts sequence mate? ❤️

[unreal girder]

Hi all, new here and just signed up for the Ultra package. I have some ideas for videos, but cannot get Veo to follow the prompts. I gave it a very simple and structured skit, but it consistently gets the roles reversed and now Im out of credits for the day with nothing to show for it. Any tips on getting Veo to follow the prompts in the skits?

[woven condor]

I’ve spent a lot of time fine-tuning my builds, so I try not to just drop full prompts out of context, but I’m always down to share tips if you’re stuck on something specific. Let me know what you’re trying to create.

[woven condor]

That’s a common issue, even with structured prompts. Veo tends to prioritize visual logic over written roles, so unless the contrast between characters is extremely clear and visually grounded early, it’ll often flip them.

One thing that’s helped me is locking in distinct visual cues + motion for each character before the dialogue starts...like posture, accessories, or where they’re positioned in the frame.

Stick with it, even Ultra takes a few rounds to tune in.

[echo storm]

Anyone got any tips on how to generate a character speaking so they don't have massively substandard audio?

A few days ago audio was fine - nows it absolute balls.

[severe aspen]

hey all! if you're a Flow user, we'd love to hear from you~ (poll can be found in #labs-announcements・📣!)

[stray elk]

The Yeti King decides to have some fun by sneaking into a group of campers deep in the forest. Pretending to be a friendly vlogger, he films himself creeping closer to the tents, whispering funny commentary about the campers’ reactions. As he scares them with spooky noises and sudden appearances, he keeps the camera rolling, capturing every hilarious scream and surprise. Throughout the video, the Yeti King cheekily talks to the camera, acting like he’s on a wild adventure vlog, making jokes and reacting to the campers’ panic. At the end, he reveals himself with a big laugh and invites the campers to join him for a friendly campfire party.

https://tenor.com/bNhDq.gif

[shadow pendant]

Hello I’m new here
I want to make an ai podcast where 2 people are speaking about something. For example, I want one character to speak and the other to respond. But each time, the wrong person is speaking.
How can I fix this?

My characters is a bigfoot and one normal person. I want the gorilla to make the first sentence but the person is speaking first

This is my prompt

[drowsy canyon]

How do you guys introduce Bigfoot and Yetti with different outfit and have both to be able to speak to each other in prompt?

[woven condor]

A realistic Bigfoot sits at a basic wooden podcast table next to a human male. Both characters are seated side by side in front of a neutral white backdrop. Two podcast microphones are clearly visible and placed in front of each speaker. The shot is perfectly still, as if filmed from a locked-off camera on a wooden tripod. The video begins already recording.

The Bigfoot, sitting on the left side of the frame, clearly moves its mouth and speaks first. The human remains completely silent and still while Bigfoot speaks. Bigfoot finishes speaking, then pauses. After that pause, the human calmly responds — only after Bigfoot finishes speaking. There is no audio overlap. The camera remains focused on both characters at a medium distance, with both clearly in frame the entire time.

There is no laughter, no camera shake, no unexpected movement. Just a serious, awkward moment. Lighting is clean and even, with no visual distractions. Dialogue should be clear and spoken by the correct character, matching their position and lip movement. The scene should look natural and grounded — like a real podcast setup filmed in a minimalist studio.

[woven condor]

Veo 3 Fast can get confused with speaker order, so this structure helps keep things stable without overloading it. You don’t need to write dialogue word-for-word — just guide the behavior like this. Try this as well:

A realistic Bigfoot sits at a simple podcast table with a human man. They sit side by side in front of a plain white wall. Each has a microphone in front of them. The shot is steady and filmed from a fixed camera, with both characters clearly visible in frame the whole time.

Bigfoot is on the left side of the frame and speaks first. His mouth moves naturally while the human stays silent. When Bigfoot finishes speaking, he stops moving. Then the human, sitting on the right side, responds calmly. Their voices do not overlap. Each character only speaks when it's their turn.

The lighting is flat and even, like a normal studio. There are no camera movements, no background distractions, and no visible camera. The scene is quiet, serious, and feels like a real podcast recording.

[woven condor]

Simpler with dialogue:

A realistic bigfoot sits at a simple wooden table in front of a white backdrop. A human man sits next to him on the left. Both are seated facing forward. Two microphones are clearly visible on the table. The camera is locked in place, as if recording from a tripod.

The scene begins with no one speaking. The bigfoot turns his head slightly and says: “Do you think white people should be allowed to say the n-word?”

The human calmly replies: “Nah.”

The bigfoot then pauses briefly, looking straight ahead, and says in a serious tone: “Oh sh*t, the table is leaving.”

[drowsy canyon]

@woven condor how would you introduce outfit for both characters then? I had the first character have the exact outfit and stuff. However when i put in the second character details. Veo 3 doesn't understand i think. So it messed up both character details.

[woven condor]

That was a reply to Pablo.

[drowsy canyon]

i know but i was reading what you are saying. I just have a quick question too 😄

[woven condor]

If you give Veo a long run-on description with two characters in the same sentence, it blends their traits, especially clothes.

The fix is to describe each character in a separate paragraph, starting with their position in the frame and making sure each outfit is tied clearly to a unique subject.

[drowsy canyon]

sent you a pm.

[shadow pendant]

@woven condor thank you will try this out later

[woven condor]

NP!

[amber swift]

Can anyone help me with a prompt that would make this style of kangaroo get run over by a emu? The style is comedic dead pan and has a Larrikin casual aussie accent

I'm still learning prompt structure but I have three videos that work for this so far

[dim gull]

https://promptalchemist.com/share/ANnk3VwjP3BzuslyVte9

Give this a shot? Tweak to your satisfsction first.

If you want it to be exactly the same style, use the image you provided as the first frame in frame to image.

[wind mica]

GUYS HELP ME LOOK AT MY MESSGAE IN GNERAL PLSS

[dim gull]

Veo doesn't support vertical formats yet. You'll need to crop it in another video editor.

[sand sparrow]

@dim gull Going to try your site to make the prompt as what ever I do my main prompt keep changing my character

[dim gull]

The character manager will definitely help make sure your prompt uses the same characteristics. Make sure to save and tag the character!

[sand sparrow]

Will do thank you, I will do that and test I will update you once I create 2-3 scenes with your prompt or should I send you some of current videos I did and the prompt to see if I should add anything

[sand sparrow]

@dim gull see how the character changed

Main Prompt: Use consistent dragon design across episodes: emerald-sapphire scales, violet eyes, antler-like obsidian horns, golden claws, prismatic feathered wings, glowing runic necklace.

see how these two are diffrent look the head one with horns one without it

[dim gull]

I see, and these were made with this character prompt?

[sand sparrow]

yes

Description: Use consistent dragon design across episodes: emerald-sapphire scales, violet eyes, antler-like obsidian horns, golden claws, prismatic feathered wings, glowing runic necklace. Baby Aurex spots a slightly elevated, glowing rock. He jumps onto it with surprising agility, his golden claws easily finding purchase. As he lands, his prismatic feathered wings automatically flutter open and closed for a brief moment, showcasing their full, brilliant colors before settling. His shimmering emerald and sapphire baby scales are prominent, his oversized, glowing violet eyes survey his new perch, and his tiny obsidian nubs are clear. His glowing runic necklace is visible.

Camera: Medium shot, capturing Aurex's jump and wing flutter in detail. (8 seconds)

Description: Use consistent dragon design across episodes: emerald-sapphire scales, violet eyes, antler-like obsidian horns, golden claws, prismatic feathered wings, glowing runic necklace. As Baby Aurex lies at the bottom of the slope, his shimmering emerald and sapphire baby scales gleaming, a group of ethereal, glowing butterflies gently descends and begins to circle him. His oversized, glowing violet eyes follow their movements with fascination, and his glowing runic necklace pulses softly in sync with their luminescence. His prismatic feathered wings flutter with excitement, and his tiny obsidian horn nubs are clearly visible.

Camera: Overhead shot, looking down at Aurex surrounded by butterflies. (8 seconds)

[dim gull]

So here's what I do. I add the character features I want into the character manager and it creates a more detailed description. And then you can use that in your prompts.

https://promptalchemist.com/share/nv1FXuxiukxyNOp5TV9q

[sand sparrow]

can you make for both promots as they have different actions

so i can try to make it right now and i will share here how it comes out

or i can just change the action part right

and the rest will be same

[dim gull]

https://promptalchemist.com/share/9ZAA4WMkTMeRlbTuWP20
https://promptalchemist.com/share/tLb6P3e99W03AvYDagHw

[sand sparrow]

making now thank you

See same issue look the horns and the spikes

although you used young and i am using a baby version

@dim gull

i like the background much better in yours though

[wind mica]

How I do it help meeeee any TIPS?!

[sand sparrow]

use capcut

[raw oyster]

is there a way to make Veo not add subtitles or captions?

[dim gull]

If you need near perfect character consistency you'll need to use image to video, or you'll need to get even more precise with prompting. Obviously tweak everything to your needs, but I hope I provided a better jumping off point!

[sand sparrow]

yes that make sense i could use a screen of the one i like better to do frame to video i guess but yes thank you so much i like your version better and more details i iwll try to work on the character then will continue with your layout

see this one i love it its pefect

i may use this for frame to video

[visual veldt]

Hey guys I could really use a hand with this stuff

[fringe hare]

Can you share the prompt you're using that is helping with consistency? I'm having major output consistency issues even though I keep saying at the beginning of each prompt to use the exact same character & voice over.

[dim gull]

Stating to use the exact character and voice over will not help as it doesn't have a memory of what you've previously prompted. You need to make the descriptions of subjects and scenes more detailed themselves. Or use an alternative method like ingredients with Veo 2 or Frame to Video with Veo 3.

[woven condor]

Veo doesn’t “remember” anything it just rewards consistency and logic. That’s why I duplicate working prompts and evolve from there. Even tiny changes throw things off. Without knowing what you’re building, what version you’re on, or how your prompts are structured, it’s hard to give advice that would even apply. I totally get the struggle, but it’s not as simple as just “sharing a prompt.” What’s working for me now is the result of weeks dialing in structure, tone, motion, and ultra-specific continuity , character blocks, visual logic, even the way I frame voice cadence.

[fringe hare]

I can't even get it to output correctly from the script.

My script has 2 voices and it keeps giving output of bigfoot saying the other characters line thats off-screen. Idk how to fix that. I feel like im just waisting away credits each time I click generate.

[next sentinel]

looking for help; starting image is a woman with hands raised, glowing hot pink, shooting lightining bolts. I want the glow and lightning to intensify to take over the shot. can't seem to do it. is there a ban on certain words?

[pine locust]

#flow-tips-n-tricks・✏️ Scenario
You’re dispatched to a major fire at a commercial warehouse just after midnight. On arrival, you see flames breaching through the roof and thick black smoke pushing out from multiple loading bays. Visibility is low, and the structure appears unstable in some areas. As your team begins setting up lines, you hear shouting from inside — workers are still trapped. A site supervisor urgently warns you that a nearby section stores industrial chemical drums that could ignite if the fire spreads further. The layout is unfamiliar, and mutual aid is still 10–15 minutes out. You only have your crew of four on scene, and conditions are rapidly deteriorating. You need to make a quick decision on how to allocate your team.

[pastel pawn]

How to make consistence same character to videos, how use correct prompts to audios background and voice?

[crude juniper]

on google chrome

[amber swift]

If you use cap cut you can change the ratio for an easy fix but you're going to have to a blur canvas or something to fill the background

[molten parcel]

I’m working on a commercial using Google Flow Ultra and Veo 3, and I’m running into a few challenges. I’ve generated images of my characters and want to place them into scenes together. When I try to add multiple characters manually using the plus sign, I get an error saying that the Veo 3 quality model doesn’t support that, and it suggests using the Veo 2 Fast model instead, which ends up stripping out the audio. Additionally, I can’t find a way to label or reference these assets directly in the prompts, which makes it harder to keep things organized. If anyone has tips on how to manage multiple characters in one scene without losing audio, or how to label and reference assets more easily, I’d really appreciate it!

[static tusk]

Hola, como puedo acceder a veo 3 ?

[woven condor]

Para acceder a Veo 3, necesitas estar suscrito al plan Google One AI Premium (que incluye Gemini Advanced) o al plan AI Ultra. Con cualquiera de los dos planes, puedes usar Flow, la herramienta de creación de video basada en la tecnología de Veo.

Puedes acceder a Flow desde este enlace: https://labs.google/fx/tools/flow
También está disponible dentro de la app móvil de Gemini; solo debes buscar la opción de video o el ícono de estrella.

El plan AI Pro te permite generar una cantidad limitada de videos al mes. El plan Ultra incluye funciones adicionales, como generación de audio, en caso de que lo necesites.

[fringe hare]

@woven condor & @dim gull

This is what I'm using and it's still having bigfoot say the directors part. I don't know how else to explain the details so it outputs correctly

Scene 1: Bigfoot Refuses to Run (Split into 3 parts)

Part 1

Bigfoot stands stage-left with arms crossed, slouched posture, fur slightly ruffled. He glances at the crew with narrowed eyes, visibly annoyed.

The director is off-camera but gives bigfoot instructions on what to say next by saying the following line to bigfoot "Okay Bigfoot—UFO’s chasing you. Say 'They’re trying to steal my DNA!' and run."

Bigfoot sighs and shifts his weight dramatically.
Bigfoot: "Seriously?"

I had 145000 credits yesterday and now am down to 100000 and haven't even had a solid scene 1 created.

[sand sparrow]

They definitely need to add request credit refund as I am on same boat I lost over 1000 credit nothing to show yet

[dark nacelle]

You're learning though right? That goes for everyone that has tossed credits out from the start, myself included.

[fringe hare]

Definitely learning! Burning through 45k credits without getting a usable Scene 1 feels a bit rough, especially when the tool seems to misinterpret pretty clear direction.

[static tusk]

Muchas gracias, actualmente tengo un plan de Google one , puedo migrar a unos de los planes que mencionas sin perder los datos de mi Google drive?

[woven condor]

Sent you a pm

[dim gull]

45k credits?.. You spent nearly $500, on 450 Quality generations or on nearly 2500 Fast generations and didn't get a single usable clip?

[woven condor]

Lol

[dim gull]

Literally my second attempt with Fast, so 40 credits total.

[woven condor]

Sí, puedes cambiarte a uno de los planes que incluye Gemini Advanced (como el de Google One AI Premium) sin perder tus datos de Google Drive. Todo tu almacenamiento, archivos y configuraciones se mantienen igual, solo se actualiza el plan con acceso a herramientas como Flow y Gemini.

[fringe hare]

I upgraded to Ultra which was $125 (I believe) and no. Don't have a single usable clip. Each one keeps generating Bigfoot saying the directors line who is off camera and stated in the script that I gave above. That's just latest script that I've built up after each fail but still missing something because it's still generating bigfoot saying the directors line EVERY time.

[dim gull]

Okay so 4.5k credits then? Thats a pretty big difference but still far too many when you're not getting results. Check the clip I posted above, is that what you're going for?

[fringe hare]

K, but how long have you been doing this for?

You're using your experience and knowledge that you've built up with the tool over time to justify the situation and yours is showing the director in the scene. Did you add that to the prompt to help with the output?

[fringe hare]

No. 45K credits. This number in the screenshot was 145000 last night. This is it today without scene 1 generated.

[fringe hare]

Yeah, it's a lot better than what mine is spitting out but yours might be more trained than mine? I think mine still doesn't have a sense of direction and it's also not including the back of the directors head in the shot but wondering if I need to have that to help it with the output? Maybe having the director fully out of the shot is confusing it more?

[dim gull]

That's only 10.7k credits.

Also, we're both using the same model, mine isn't more trained than yours. I'm just using highly detailed prompts to get the output I'm looking for. It's true I've spent a lot of time learning what that looks like, but one of the biggest things you can learn quickly is if you're not getting the result you're looking for move on, you'll save credits and learn to be more creative in the process.

Here's the prompt I used to generate the scene above.
https://promptalchemist.com/share/zrJG8J22s5KCAsOSdBht

[woven condor]

Sharing this one because not every output goes to plan.

A little editing can go a long way. Both of these clips were originally fails. Characters were off, motion didn’t land, but there were a few frames that still had something. Ended up combining the best moments from each with some editing, and the final version took off.

Sometimes it’s not about waiting for the perfect output. It’s about spotting what can work and building from that.

[fringe hare]

Shoot. You're right. I've been at this too long and my brain is fried. Haha. I'm sorry!

I had to literally go look at it again because I was like "No it's not!". Ooi. Serious palm to the face moment. Smh.

I'll take a look at that prompt you sent me. Thank you btw for the help!

[dim gull]

It happens to the best of us! That's why I was initially shocked, I was like 45k hot damn!!

I built my prompting tool to help people save credits cause each failed generation is just time and money out of your pocket. Best of luck creating!

[gritty mesa]

Hello. I am trying to get Flow to render a scene with this prompt:

Photorealistic. The scene, typical in some action movies, is of a woman being rescued from a helicopter above, from which is descended a rescue ladder. However, only the woman and the rescue ladder is seen, but the helicopter or any part thereof, is not rendered in any part of the frames.

The first frame of the video shows a woman balancing herself on the tightrope above an abyss, all the muscles in her body are taut with desperate concentration, fear and adrenaline, her hair is moving wildly with the wind. She is grabbing the first rung of a rescue ladder made of two parallel white nylon chords, joined by thin brown cylindrical pedals about the same thickness as the chords and made of steel (the rungs) in between, that are 1 foot in length and evenly distributed 1.2 feet apart from each other, that starts to descend continuously and without pause, one by one rung, from above and in front of her as as she helps to control the descent of the rescue ladder until the bottom pedal is at about the height of the tightrope and her feet, and the fourth rung from the bottom is about the her eye level, and stays suspended in front of her. As the rescue ladder descends in front of her, the woman looks up, grasps the rescue ladder and when the lowest rung is in position, climbs on top, by putting one foot first, and then the other foot, on the lowest pedal (rung) of the rescue ladder. Then, after the woman has climbed onto the rescue ladder, the rescue ladder with the woman ascend towards the top of the frame and away from the tightrope. The helicopter, or any part of it such as the rotor blades, skids, etc. are not seen in this scene.

But then, it does it all wrong, materializing parts of the rescue ladder out of nowhere, dropping the rescue ladder behind her, materializing a second rescue ladder to her side, showing parts of the helicopter rotor blades above, etc.

[orchid jay]

My last day of 1 month free google one. Goodbye flow 😫 goodbye whisk 😫

[lean star]

Sounds like you aren't using gpt o3 to help craft and iterate off the jump. Plus @dim gull is correct as well

[crude reef]

@gritty mesa its not gonna anywhere near perfection i spent a day working on a batman type trailer and it turned into a comedy

[gritty mesa]

🤣

Im certainly no expert, but a beginner. What i am going to do is end up choosing start and end frames that are sort of usable from all the bloopers flow is making, and editing these with gimp, in the hope that with two frames, flow better understands what i want it to do

[gritty mesa]

If you can make a trailer in a day, youre pretty darn good i'd say

[crude reef]

no its jjst hot nonsense

https://www.youtube.com/watch?v=pb2YQq0k5rE

i feel like i deserve an oscar for it though

trailer 2 has fighting and explosions but its all so ridiculous its better not to think bout the blows landing and whats exploding

[maiden ice]

What kind of prompt structure would you suggest to unify the faces of people?

[crude reef]

like get consistent character faces?

i know people have done it with faceswap software but i havent gone that deep yet

[opaque kraken]

Is there a protocol in place to replace an objects background. For example: if i had a still of a coke can sitting on a clean white table and background. but wanted the can to remain but the background to be replaced with the can sitting on table outside in a park?

[crude reef]

is the can prompted in ?

[gritty mesa]

I use pica faceswap (art guru) its pretty good but not so good with profile views of faces

[crude reef]

i just started looking into using comfyui to generate images and stuff locally im hoping some mac freindly video models become available for the mchip

[crude reef]

i need to figure out lip sink.

[prisma zinc]

Btw love Prompt Alchemist

[dim gull]

Thank you!!

[spice meadow]

anyone have any tips for accents? I put an australian man says this with an australian accent and still doesnt work. Sometimes yes and sometimes no.

jamaican one is pretty consistent

if i can figure out accents it would help

[dim gull]

Here's an example where I just mentioned the subjects nationality and the model used a stereotypical accent that worked well.

https://promptalchemist.com/share/HMVoYI1GPpdnTazGP1Mj

[dim gull]

Other times I actually put the accent im going for in the dialogue section with inflection/intonation notes.

https://promptalchemist.com/share/ANnk3VwjP3BzuslyVte9

[sudden crown]

Ha the guy with the gator made that so much better. Great job

[spice meadow]

Crazy! still could be better. Ive also noticed that when you do 2 maybe 3 prompts it helps. I used to only do 1 and then have to redo them

Im working on a commerical and instead of dialogue im doing a voice over to tell the story

[frigid kestrel]

Thank you for this btw, can i be a member?

[sage cipher]

Is terrible how difficult it is to make a simple interaction between 2 characters in order to make one of them say onoe phrase and the other one to say a different one. Have tried a lot of times with a very soecific prompt that says a man says "x" and a woman responds "y".

Being the result almost all the times that the man says the 2 phrases totally ignoring the prompt that is very very specific about who says what and when to say it.

All the rest of the instructions are brilliantly followed by Veo 3.

The who says what thing is failing constantly and consuming credits. One of the attempts even got me a no audio result.

[crude reef]

What’re @sage cipher what’s your prompt

[crude reef]

Nobody seems to get that joke

[crude reef]

So ive managed to get captioning on the last several videos by promoting it now I’ll try without it

{ "video_prompt": "A slow-motion close-up of a sloth with dilated pupils and twitchy eyes, holding a coffee cup halfway to his mouth. The sloth is frozen in place, moving extremely slowly. The camera slowly zooms in on his face. Despite barely moving, the sloth’s wide eyes and subtle facial tension suggest extreme internal panic. Background is a peaceful jungle or cozy café, with gentle lighting and visible coffee steam.", "length_seconds": 8, "visual_style": "comedic documentary, surreal tone, high-detail animation or CG, dramatic lighting", "camera_direction": "slow zoom-in from medium shot to extreme close-up on the sloth’s eyes over 8 seconds", "audio": { "background_music": "soft ambient café sounds or calm jungle ambience", "sound_effects": [ { "name": "internal_scream", "type": "voiceover", "start_time": 1.5, "end_time": 7.5, "volume_curve": "fade-in crescendo", "description": "high-pitched inner scream, growing louder and echoing as the camera zooms in" }, { "name": "glitch_pop", "type": "effect", "start_time": 7.8, "end_time": 8.0, "description": "brief record scratch or audio glitch at final zoom moment" } ] }, "text_overlay": [ { "

this got me this

[humble oxide]

Yo i want to make some too can you help me ?

What’s the steps

On making these videos

[crude reef]

you gotts be in flow on gemeni

[crude reef]

lol

[white needle]

#flow-tips-n-tricks・✏️ A realistic 3D animated scene of a sleek futuristic spacecraft called "Maeme's Roaster" descending and landing on a green screen background. The spacecraft has smooth metallic textures, glowing thrusters, and realistic shadowing. As it gently lands, dust subtly kicks up under it. After landing, the cockpit opens to reveal a charming, humanoid robot pilot. The robot smiles warmly, winks, and gives a confident thumbs-up to the camera. The background remains solid green for post-production keying. Cinematic lighting and smooth camera motion.

[solar vortex]

#flow-tips-n-tricks・✏️ Close-up of the credit card being tapped on a sleek payment terminal. Upon tapping, golden glowing particles rise into the air, transforming into a stylized Etihad Airways airplane taking off into the sky. Ends with a clean, cinematic fade-out. No text.

[gritty mesa]

Where can we access your prompting tool?

[sage cipher]

I checked the prompt you posted with the zombies and tried to use a similar structure you used and it worked.🙂 Thank you

[sage cipher]

I can´t imagine that mother clucker prompt 😂

[crude reef]

it was in json format

i used chatgpt

it suggested the json format

[sage cipher]

That would be my next question 😛 Thank you, it seems to work better that way

Lip Sync A reallistic gorilla walking in new york using a hand held selfie camera, describes the noisy city, on the background we can hear the new york noises, cars, sirens, people talking. The shot is taken with the phone camera on his extended arm while he walks and talks.

[analog inlet]

Created with Google veo Just tried first time

The Prompt:
Generate a cinematic video. Camera: Slow, gentle push-in and a subtle left-to-right pan, gracefully revealing three 'Soul Karma' organic product bottles (e.g., face oil, serum, moisturizer). Environment: Products are elegantly arranged on a large, smooth, light-grey river stone. The background is a deeply blurred, sun-dappled, serene forest clearing with tall, ancient trees and soft moss on the ground. A gentle, soft-focus mist subtly rises from the forest floor behind the products. Lighting: Warm, golden hour light filters through the leaves, creating soft bokeh. Text Overlay: First, 'Soul Karma®' appears with a gentle, pure white glow. Then, 'PURE HARMONY' fades in underneath it. Mood: Tranquil, pristine, inviting, pure.

[sage cipher]

how did you made 14 seconds?

[sage cipher]

Is a shame they don't do lip sync. I have tried a lot with dogs and it seems that the model only is able to make lipSync with humans or characters with similar mouth structure as humans (monkeys, big foot, etc). It seems that it works almost fine if is a short snout animal (pug dogs, racoons, etc.).

[crude reef]

Have you tried the anthropomorphic

[crude reef]

try something using this its the best i could get

@sage cipher

[dim gull]

This one was better, dialogue slightly scuffed, I just prompted for a non-anthropomorphic dog this time.

https://promptalchemist.com/share/POjxazWysP1O5ERj02Mu

[dim gull]

Yeah, unfortunately Ingredients only works with Veo 2 for now. Image to video, or strict prompting with Text to Video, are the only ways to work with any consistency in Veo 3 right now.

[warped compass]

Do you guys think veo would be able to generate a pov fast pace motorcylce Ride, Dodging Cars.

Or would the Cars not generate reasonable. I’m thinking of a prompt for this

[livid canopy]

what song is this? in the end 7'

[sage cipher]

I think is better that You create an image with both of them in they same image and then explain in the prompt wich once is wich. Actually in that way You can already put them in they background you want

[livid canopy]

u can screenshot last frame and generate an expansion with veo 2-3

[sage cipher]

I tried that but don't used to work in Veo3 with audio

[livid canopy]

better veo2 fn

U can do it w V3 only using 1 atm. 2 = veo2

[crude reef]

Allanis morriseyttw

[rocky needle]

Hi @dim gull , How do you keep your voice consistent? I often have a robotic voice

[crude reef]

Alanis morisette

Ironic

Or something like that,

Cause the cow is eating a burger

[crude reef]

someone asked bout cars

[short crown]

Do you have the promtp of this master piece?

[wicked mortar]

can you share prompt?

[crude reef]

yeh

give me a second

@9100365 that was several clips generate and combined

@wicked mortar a neon-drenched city at dusk, styled like a Miami-meets-Tokyo metropolis in a GTA VI setting for the playstation. in game footage of a guy and a pig in. high speed police chase

[tired moss]

Hi - created a person on Midjourney and Flow won't let me upload because of "prominent person" flag. Anyone get around this?

[woven condor]

You're not wrong, but as long as you understand it’s not just about the prompt itself, it’s the logic inside it. Character structure, consistent roles, camera rules, pacing between lines... it all has to make sense like a real scene would. Most people just stack info and hope for the best, but that’s why their outputs fall apart.

Got to a point where my characters actually talk to each other in different scenes and stay fairly consistent. personalities, same pacing, no weird swaps.

--- SCRIPT & DIALOGUE --
CHARACTER: _________
DIALOGUE:__________
[ACTION/EMOTION TAG: seconds 0-4, out of breath, walking, natural pace,, ect]

[ACTION/EMOTION TAG: seconds 4-8, deadpan delivery, chewing pineapple, ect]

[heavy totem]

Does the TAG help

I have my emotion inside a character performance tag

[heavy totem]

That’s very good I haven’t seen many where the VO and performance isn’t robotic or over acted

[woven condor]

I have tone, pacing and emotion in my character blocks as well.

[heavy totem]

I’ll try “tag” nomenclature

That’s pretty similar
To how I do it

Just one VO here but there actual nuance like yours

[zinc spear]

I'd like to thank this group. I finished my talking cat episode. I just couldn't show the output here. but big thanks to everyone who helped me ❤️

xoxo

[dull garden]

https://youtu.be/t7fthKOjaI4?si=FR523zy7It1uhBDi

Trying to complete a whole series but have trouble keeping the prompts continuity

Any suggestions would be helpful

[winged ice]

How to implement the prompt words for the surround shot?

[versed urchin]

شارعٌ حديثٌ مزدحمٌ في طرابلس ليبيا خلال النهار. في الزاوية، يقف مبنى زجاجيٌّ عملاق مستقبليٌّ بشاشة LED منحنية ضخمة تغطي طابقين. فجأةً، تنبض الشاشة بالحياة مع إعلانٍ واقعيٍّ للغاية بتقنية CGI. ينبثق حذاء ميريل الشهير agility J068697 ثلاثي الأبعاد ببطءٍ من الشاشة، كاسرًا الإطار كما لو كان شيئًا حقيقيًا، باستخدام تأثيراتٍ ثلاثية الأبعاد وغبار متطاير. يطفو الحذاء في الهواء مع أضواءٍ متوهجة، وتتلألأ انعكاساته على المباني المجاورة. يتوقف المارة في الشارع ويحدقون بدهشة. يتميز المشهد بإضاءةٍ سينمائية، وظلالٍ ناعمة، وحركةٍ بطيئةٍ مع ظهور علبة الحليب، وقوامٍ فائقة الواقعية. أسلوبٌ بصريٌّ مستقبليٌّ وغامر، مستوحى من إعلانات اللوحات الإعلانية ثلاثية الأبعاد الواقعية في طوكيو وشنغهاي.

[sage cipher]

Yes, is a little painful (and expensive) to learn how to make some things work, thank you for your advice 🙂 I appreciate it.

[worthy ether]

I have this simple script, but no matter what I add different people say the line that belongs to Tamara. And ghost runs off the screen instead of the man.

[crude reef]

@worthy ether {
"visual_style": {
"style": "Cinematic realism",
"camera": {
"description": "Single wide-angle shot",
"movement": "Mostly static or subtly handheld"
},
"framing": {
"position": "Low and slightly behind Tamara’s left shoulder",
"details": "Showing part of her side profile"
},
"shot_type": {
"description": "One continuous shot",
"cuts": "No cuts"
}
}
}

try this

and chang it around and see if tht doesnt do something

[echo storm]

What’s with the use of the {} and “”: etc.

Do these actually help Veo with the scene creation or are they for your own use in keeping your data readable?

[crude reef]

yeh thaht do

[worthy ether]

I think the problem is not in the prompt, cause my prompt has been growing with detailed directions with no good results. I think veo 3 gets confused who does or says what when you have more than 2 characters in frame

[crude reef]

they keeo scebne very close between each change

[echo storm]

Where did you learn how to do that? Is there documentation somewhere?

Is there an agreed structure to the {}:”” etc in how they should be nested?

[crude reef]

oh wow they blocked my post on it

i posted a zombie clip 4 times with diff gender or zombie and dif type of cg

all 4 exactly the same shooting text and eeven shirts

if you have chatgpt

or gemeni

[echo storm]

How does this differ from just having a prompt with bullet points etc?

[crude reef]

ask it to create a json script

im not too educated on this whole topic but i do know that everytime i use json scripting i get exactly the same thign i wanted with a slight adjustment

thihnk of json as a over head views of the a spot you can pick things out on

like on a map

like i said im no authority ive been very happy with the output though

[dark nacelle]

A script would be a better term. It gives direction.

[crude reef]

i had one video it qued a record scratch sound

[worthy ether]

still no good. even tried to only have 2 characters, but still misunderstands even in json format

{
"title": "Ghosted Listing",
"genre": "Real Estate Comedy / Supernatural",
"duration_seconds": 8,
"style": "Cinematic realism",
"camera": {
"type": "wide-angle",
"movement": "mostly static or subtly handheld",
"framing": "head-to-knees shot of man, positioned on right side of frame, facing toward the center of the room",
"shot_type": "continuous, no cuts"
},
"characters": [
{
"name": "Man",
"description": "Chubby, balding, in his 50s. Wears a black Ghostbusters T-shirt with the classic red circle-slash ghost graphic from the movie (graphic only, no text)."
},
{
"name": "Ghost",
"description": "Male, dressed in formal 1800s attire (long coat, cravat, high collar). Pale white, softly glowing, clearly human-shaped, and fully translucent."
}
],

"scene": {
"setting": "Dusty Victorian foyer with cobwebs, chandelier, and sunlight shafts. Still, eerie atmosphere.",
"initial_positioning": {
"Man": {
"position": "right side of frame",
"visibility": "head to knees",
"orientation": "facing toward center of room",
"action": "glancing around as he delivers line"
}
},
"dialogue": [
{
"character": "Man",
"line": "Hmmm, charming."
}
],
"actions": [
{
"character": "Ghost",
"entry": "enters from left side of frame",
"movement": "glides slowly across frame in front of man",
"gesture": "none",
"visibility": "fully translucent, softly glowing, human-shaped",
"line": "Get… out…"
},
{
"character": "Man",
"reaction": "gasps immediately when ghost appears, squeals in high-pitched frightened voice (not musical), throws arms up, and runs off-screen to the right in full panic"
}
]
},
"audio_notes": [
"Ambient wind",
"Faint creaking floorboards",
"Ghost line: 'Get… out…' in whispery voice",
"Sharp gasp from man as ghost appears",
"High-pitched terrified scream from man (not operatic)"
],
"visual_guidelines": [
"Man must not face the camera; he should face toward center of room",
"Ghost must not transform, wave, run, or disappear",
"Maintain cinematic realism — humor comes from contrast between ghost’s calm presence and man’s dramatic reaction"
]
}

[crude reef]

can you pinpoint the issue?

[worthy ether]

Ghost was supposed to say "Get... out...", then the man was not facing the center of the room as directed. And the ghost was not suppose to run off as specifically directed as well

[crude reef]

have you had succcess with sghosts

[worthy ether]

the ghost gasped instead of the man too

[crude reef]

thats odd it had you put wht the ghost didnt do

thsts seems like me saying "dont push the button, and you def dont tell him to push the button"

],
"visual_guidelines": [
"Man must not face the camera; he should face toward center of room",
"Ghost must not transform, wave, run, or disappear",
"Maintain cinematic realism — humor comes from contrast between ghost’s calm presence and man’s dramatic reaction"
]
}

[dark nacelle]

Remember that anything you don't specify is left up to the AI.

[crude reef]

that prompt though t the end , the ghost ran

[worthy ether]

finally, but still does not look towards the center of the room at the start. and the ghost screamed the line instead of whispering

{
"genre": "Real Estate Comedy / Supernatural",
"style": "Cinematic realism",
"camera": {
"type": "wide-angle",
"movement": "mostly static or subtly handheld",
"framing": "head-to-knees shot of man, positioned on right side of frame, facing toward the center of the room",
"shot_type": "continuous, no cuts"
},
"characters": [
{
"name": "Man",
"description": "Chubby, balding, in his 50s. Wears a black Ghostbusters T-shirt with the classic red circle-slash ghost graphic from the movie (graphic only, no text)."
},
{
"name": "Ghost",
"description": "Male, dressed in formal 1800s attire (long coat, cravat, high collar). Pale white, softly glowing, clearly human-shaped, and fully translucent."
}
],
"scene": {
"setting": "Dusty Victorian foyer with cobwebs, chandelier, and sunlight shafts. Still, eerie atmosphere.",
"initial_positioning": {
"Man": {
"position": "right side of frame",
"visibility": "head to knees",
"orientation": "facing toward center of room",
"action": "glancing around as he delivers line"
}
},
"audio": [
"Ambient wind",
"Faint creaking floorboards"
],
"visual": [
"Man must not face the camera; he should face toward center of room and look around casually before delivering his line"
]
},

"dialogue": [
{
"character": "Man",
"line": "Hmmm, charming."
}
],
"actions": [
{
"character": "Ghost",
"entry": "enters from left side of frame",
"movement": "glides slowly across frame in front of man",
"gesture": "none",
"visibility": "fully translucent, softly glowing, human-shaped",
"line": "Get… out…",
"exit": "remains onscreen and floats steadily across the frame",
"audio": [
"Ghost line: 'Get… out…' in low, whispery voice"
],
"visual": [
"Ghost must enter from left, remain translucent, glowing, and human-shaped throughout",
"Ghost must not run, transform, wave, or exit the frame; he stays onscreen gliding calmly"
]
},
{
"character": "Man",
"reaction": "gasps immediately when ghost appears, squeals in high-pitched frightened voice (not musical), throws arms up, and runs off-screen to the right in full panic",
"audio": [
"Sharp gasp from man as ghost appears",
"High-pitched terrified scream from man (not operatic)"
],
"visual": [
"The man must see the ghost immediately and react with fear—gasping, squealing, and running off-screen alone"
]
}
]
}

[dark nacelle]

Try center of frame. Keep the language consistent.

[worthy ether]

Ugh, still looking to the right in the beginning and the man now says Outie, and ghost still does not whisper wth

{
"genre": "Real Estate Comedy / Supernatural",
"style": "Cinematic realism",
"camera": {
"type": "wide-angle",
"movement": "mostly static or subtly handheld",
"framing": "head-to-knees shot of man, positioned on right side of frame, facing toward the center of the frame",
"shot_type": "continuous, no cuts"
},
"characters": [
{
"name": "Man",
"description": "Chubby, balding, in his 50s. Wears a black Ghostbusters T-shirt with the classic red circle-slash ghost graphic from the movie (graphic only, no text)."
},
{
"name": "Ghost",
"description": "Male, dressed in formal 1800s attire (long coat, cravat, high collar). Pale white, softly glowing, clearly human-shaped, and fully translucent."
}
],
"scene": {
"setting": "Dusty Victorian foyer with cobwebs, chandelier, and sunlight shafts. Still, eerie atmosphere.",
"initial_positioning": {
"Man": {
"position": "right side of frame",
"visibility": "head to knees",
"orientation": "facing toward center of frame",
"action": "glancing around as he delivers line"
}
},
"audio": [
"Ambient wind",
"Faint creaking floorboards"
],
"visual": [
"Man must face toward the center of the frame from the beginning of the shot, not toward the camera or right edge",
"Man should look around the space casually before delivering his line"
]
},

"dialogue": [
{
"character": "Man",
"line": "Hmmm, charming."
}
],
"actions": [
{
"character": "Ghost",
"entry": "enters from left side of frame",
"movement": "glides slowly across frame in front of man",
"gesture": "none",
"visibility": "fully translucent, softly glowing, human-shaped",
"line": "Get… OUT!",
"exit": "remains onscreen and floats steadily across the frame",
"audio": [
"Ghost line: 'Get… OUT!' — delivered in a low, slow, creepy whisper"
],
"visual": [
"Ghost must enter from left, remain translucent, glowing, and human-shaped throughout",
"Ghost must not run, transform, wave, or exit the frame; he stays onscreen gliding calmly"
]
},
{
"character": "Man",
"reaction": "gasps immediately when ghost appears, squeals in high-pitched, loud frightened voice (not musical), throws arms up, and runs off-screen to the right in full panic",
"audio": [
"Sharp gasp from man as ghost appears",
"Loud, high-pitched terrified scream from man (must register clearly as panic)"
],
"visual": [
"The man must see the ghost immediately and react with intense fear—gasping, loudly squealing, and fleeing right out of frame"
]
}
]
}

[dark nacelle]

Ugh. The fine tuning for stuff can really drain on you (and your credits stash naturally)

I read a quote that the Ai will get you there 90% of the way every time and the 10% can/will drive you crazy.

[crude reef]

it can def be maddening

[barren cloak]

are you guys experimenting the "no audio generated issue" ?

(veo3 fast - textovideo)

[barren cloak]

4 videos generated with this "no audio generated issue" message 🫠

[sharp wagon]

anyone else having the issue where veo 3 quality stops at 40% and then spits out some low quality video?

[unreal torrent]

This is the prompt format for fixed characters and dialogue in veo3?

[unreal torrent]

Yes, it worked for me, converting text to video. Today I made a video with my mobile phone, and it did the dialogue... of course, it did it frame by frame.

[wraith holly]

why is it that the audio only works sometimes haha

ive input the same script like 5 times and then it finally gave me the audio when i use the jump to function

i cant use extend unless i want no audio. i feel like i should just go no audio at all or find a way to take out the audio and make my own

[dim gull]

Yeah extend doesn't work with audio. It will switch you to Veo 2 instead.

[livid zodiac]

hey @dim gull im having an issue with your site / service i paid for

[zinc spear]

Is there a character limit for text overlay to be shown on the clip? It kept on misspelling the supposedly title of the video to show at the first seconds of the clip.

[dim gull]

Responding in DM's we'll get you taken care of :}

[zinc spear]

I tried to create a clip with a text overlay to show the title of an episode. It never spells correctly. I admit it's a bit long. I tried 6 times 😦 I wonder if there is a character limit for text overlay.

[echo storm]

I’d use an external editor for things like text overlays. Same with any in-camera effects. Just get your scene down then add extra elements externally.

[dark nacelle]

There's even google movies for post processing

[heavy totem]

regional accents are broken

burned ALOT of credits

it can

it cannot get any clearer than this : "Crisp, clear, and professional with a modern, educated British accent, her voice now strained with dread.", it worked on the last 20 gens

now nothing

[barren cloak]

hi @copper oar or @maiden ivy could you please give some hints to avoir the 'no audio generated" bug ? spent way too many credits to get no result.

[worthy ether]

it's not fixed, there are still side effect which I mentioned on top of the video

[sleek dagger]

https://www.youtube.com/watch?v=aUjsfhkjZ5s
would also love to learn how this person achieved character and scene consistency as I am facing issues with achieving the same. Maybe a prompting masterclass from them wouldn't hurt😂

[heavy totem]

hmmm, i think things ar ebeing shared somehow, one of my characters randomly mentioned project chimera

[spare zenith]

Is anyone facing this issue where even after giving detailed prompt the final video will add something on its own ? And it keeps repeating the same mistake.

[random nacelle]

Here is a prompt I typed myself. I would like it if I saw a Veo 3 video (with audio) from this prompt:
"A realistic video set in a medium-sized 2000s room with a green carpet, blue walls, a white ceiling and dim lighting. There was a vintage television to the left with commercials about promoting a fun play zone in the area. About 4 seconds in the video, the power goes out and then a second after a letter Z appears and scary music starts playing. At the end of the video the video glitches with artifacts all over the screen."

[stark sphinx]

It's possible to do good fight scenes with Veo 3?

[crude reef]

hmmmm

yes but notwell

how extensive i guess woult be the question

[random nacelle]

Welcome back! Can someone copy/paste this prompt and share a video in the gallery so I can see (if possible)? I have waited 4 hours.

[spice meadow]

High fashion editorial shot inside a grand museum hall, showcasing a statuesque androgynous Rolls Royce constructed from countless black and white mechanical tiny flip dot screen modules. The car dynamically displays abstract black and white patterns that shift and ripple, catching the light in mesmerizing ways. The Rolls Royce sits confidently amidst classic sculptures and ornate architecture. The soft, diffused lighting accentuates the rolls royce texture. The pose exudes an elegant vibe.

Camera Style: Selfie POV — slightly shaky, handheld GoPro with a wide-angle lens, duct-taped to a bent selfie stick. The panda’s right paw grips the stick tightly throughout, giving it a gritty, vlogger-on-the-streets vibe. Character: A scruffy urban panda with matted, grimy fur, wearing a ripped, dirty denim vest with a missing button. One ear’s got a jagged scar, and grease stains mark his snout and paws. He holds a half-eaten slice of greasy pizza in his left paw, occasionally gesturing with it while filming himself. Environment: A dimly lit alleyway at night. A flickering overhead lamp buzzes weakly. Neon signs from nearby businesses reflect in shallow puddles. The alley walls are tagged with graffiti, and beside him is an overflowing dumpster he clearly just climbed out of — the lid still creaking shut. A few rats scatter at his feet. Action & Dialogue: The panda turns the camera toward his face as he chews a massive bite of pizza, his greasy paw leaving a faint smear on the crust. He talks with a raspy British accent — somewhere between cheeky East London and worn-out rock star: (smirking) “You call it a trash can… (chews, points with the pizza) …I call it a mystery buffet.”* He throws a mock salute toward the dumpster behind him with exaggerated reverence, then struts forward, chin high, like he owns the alley. The pizza dangles lazily from his paw. Mood / Expression: Cocky, cheeky, and self-assured — like a streetwise panda with a chip on his shoulder and zero shame about it. Mischief in his eyes. Voice: Raspy British accent, casual and sarcastic, with a grin behind the words. Audio: Distant sirens, a metallic thud from inside the dumpster, faint city buzz in the distance. Duration: ~8 seconds

Setting: Times Square, nighttime, lights flickering through smoke Subject: A T-Rex and a raptor rampaging through a chaotic street Camera Style: One continuous handheld shot, street-level, tracking backward Scene Description (0:00–0:08): [0:00–0:02] The camera moves backward down a wet, neon-lit street in Times Square, trembling slightly. Sirens blare. A raptor sprints into frame, snarling as it leaps over a crashed police car. [0:02–0:05] As the camera keeps tracking back, a T-Rex stomps into view behind the raptor, jaws open, letting out a deafening roar. Glass rains down from a shattered billboard above. [0:05–0:08] A flaming taxi skids past the camera. The T-Rex crushes it underfoot as the camera finally swings up, catching the dinosaur's head silhouetted against glowing Times Square screens, just before everything cuts to black. Mood: Dark, terrifying, cinematic chaos Lighting: Neon reflections, fire glow, flickering signage through smoke and rain Sound Design: Distant screams, thunderous footsteps, raptor shriek, T-Rex roar, crumbling metal

[pliant valve]

@dim gull Subject: Numemon: Numemon is a squat, amorphous Digimon resembling a sickly green slug. Its skin is slimy and mottled with irregular, indigo-blue spots scattered across its back. Two thick eyestalks protrude from the top of its head, each supporting a large, blood-red eye. A gaping maw reveals rows of small, sharp teeth and a thick, pink tongue that occasionally lolls out. Its arms, usually retracted, are short and stubby, with small, claw-like hands. The overall impression is one of sluggishness and decay, with a perpetual expression of vacant disinterest on its face. Action: The squat, slimy Numemon is face-first in a overflowing dumpster, its stubby arms flailing comically as it rummages through discarded banana peels, crumpled newspapers, and half-eaten takeout containers. Its thick, pink tongue occasionally lolls out as it sniffs at a particularly enticing piece of refuse. The scene captures its absolute, if disgusting, bliss. Scene: A grimy, back alley of a bustling city. Overflowing dumpsters line the alleyway, casting long shadows in the harsh midday sun. Graffiti covers the brick walls, and the air is thick with the stench of rotting garbage. Flies buzz lazily around the refuse, and a stray cat watches with cautious curiosity from atop a nearby fire escape. The scene is lit by the soft, diffused light filtering through the tall buildings, highlighting the textures of the grime and decay. Style: A medium shot of Numemon filmed at eye-level with a shaky, handheld camera effect, rendered in a realistic documentary style with natural daytime lighting creating a joyful and celebratory atmosphere, 16:9 aspect ratio. Dialogue: (Excited, lips synced) Numemon: Oh, this garbage is the best! I love it! Sounds: The sounds of rummaging through garbage, wet squelching noises, and satisfied sighs from Numemon. In the background, the distant sounds of city traffic and the cries of seagulls add to the atmosphere of urban decay. Technical (Negative Prompt): subtitles, captions, watermark, text, words on screen, logo, branding, music, musical score, soundtrack

I like Digimon

[tulip monolith]

Interested in help with this prompt. I am just learning about prompting video in Flow, but have used other video generators.

Archery is a tough one it seems, and so far I have not found a video generator that makes it easy.

The idea is a "birthday card" animation for a friend. The archer is supposed to be his character in a D&D game.

My current prompt:

Archery. An Elven archer stands, ready to fire, with an arrow knocked already. He has long blond hair and is wearing leather armour with a green cloak. He is standing on the battlements of a fortified manor house. The elven archer pulls the bowstring, then the arrow shoots from his bow. The arrow flies to the right at high speed. The camera whip pans right to the target, which is a birthday cake. The arrow impales the birthday cake. The cake falls apart from the impact.

The start is excellent. The archer looks the part and draws their bow successfully. After that it goes comically wrong. The understanding of requirements (or my bad prompting for them) fails.

Any help or suggestions appreciated.

[lime oyster]

this is nonsense but kidna funny

[granite hinge]

To anyone whos done animation like cartoons or something what the best prompt that you us in order to have a great quality ?

[dull garden]

This is how gemini suggests the layouts, im sure you all discovered this.. but I found it half way through creation and I think if you begin prompting a scene in a certain way and then try to use a different prompting strategy then you might as well start over becaue you will waste credits. I will say Gemni suggestion did help when starting from it from your first prompt.

[worthy hatch]

im having a difficut time withb this prompt working properly, any help with revisions will be awesome

Realistic 8-second video showing an extreme close-up of a tan Shiba Inu dog’s front paw, resting calmly on a soft, light-pink cushion. Her paw is clearly canine, with visible short, dark claws extending from soft, cream-colored fur. The claws have a natural curve and are fully exposed, separate from the paw pad or fur.

Wearing her pink plaid dress, the Shiba uses her other front paw to hold a tiny nail polish brush delicately on her Nail making them completely pink. She applies a layer of glossy pink polish directly onto the curved claw surfaces only, {not touching the fur or paw pad}. The brush glides smoothly across the claws, leaving behind a visible shine.

The camera captures the shine building on each claw, and the action focuses solely on the claws being painted, not the fur or skin.

Camera Angle: Extreme Close-Up
Camera Technique: Static Shot
Style: Realistic
Sound: None

[dim gull]

Review this prompt and give it a try. I suspect you may want someone else applying the polish? Not sure what issues youre currently seeing.

https://promptalchemist.com/share/pZbpo3pM8Yle2hIS5syg

[worthy hatch]

a human hand keeps applyiong the p[olish and i want the dog to be able to do it

[worthy hatch]

very smart to make a site for this homie

im trying to make the shibv do the audry nethery meme too

[dim gull]

Ahh, this one focuses on the shibu applying it a little more. Veo has some trouble with meticulous movement. But this might be doable.

https://promptalchemist.com/share/ltjFU3v7rs0whEH06ObU

[deft raft]

https://youtu.be/UAAozGcWyOo

Hi everyone, today I made a Veo 3 vlog of my cat. What do you think about it? Any suggestions to improve it and make it even better?

[pliant valve]

Subject: Squirtle: Squirtle is a small, light-blue reptilian Pokémon with a distinct turtle-like appearance. Its most striking feature is its smooth, brown shell with a pale yellow underside, separated by a thick, prominent white ridge. He has large, expressive purplish-red eyes and a subtly hooked upper lip, giving him a slightly mischievous expression. His small, three-fingered hands and feet are light-blue, matching the rest of his body. The tail is long, squirrel-like, and curls sharply inward at the tip. When standing, Squirtle is typically upright on two legs, maintaining a balanced and alert posture. Squirtle's skin is smooth and reflects light in a way that makes it appear sleek and well-hydrated. Action: Squirtle, having just emerged from the pool, sits upright on a small rock near the water's edge. He looks directly at the camera, takes a small breath, and begins to speak about himself. As he talks, he gestures with one of his small, three-fingered hands for emphasis, and occasionally glances back at the shimmering pool behind him. Scene: The sun-drenched, tropical oasis has a heightened sense of serenity. The crystal-clear water of the pool continues to reflect the lush greenery and swaying palm trees, but now a small, smooth rock sits prominently near the edge, acting as Squirtle's stage. The exotic flowers seem to bloom even more vibrantly, and the sweet scent of tropical blossoms is almost palpable. Style: A wide shot filmed at eye-level, lit by warm, golden hour sunlight creating a joyful and celebratory atmosphere, 16:9 aspect ratio Dialogue: (Proud, slightly boastful, lips synced) Squirtle: Well, hello there. I am Squirtle. People know me for my cool shell and even cooler water moves. I always make a splash wherever I go. Sounds: The gentle lapping of water against the pool's edge blends with the clear, articulate sound of Squirtle's voice and the chirping of exotic birds. Technical (Negative Prompt): subtitles, captions, watermark, text, words on screen, logo, branding

@dim gull

[patent rain]

Surprisingly and funny

[flint oyster]

@sleek dagger Wow 😮, this is a very amazing work ..
Please how did you achieve this incredible video...
I'll pay just to learn more from you if necessary please.
Help!! 😭

[flint oyster]

@dim gull Please I don't understand how this app works..
Can you put me through please 🙏

[dim gull]

What do you need help with?

[flint oyster]

I'm just seeing this app for the first time..
And I don't know where to even start from..
Please I want to know what I can use the app to do , and how to do it without any obstacle please 🙏

@dim gull ☝️☝️ Please 🙏

[dim gull]

Its pretty straightforward as long as you have access.Go to google flow. Describe your video idea in the prompt window and hit generate.

[dim gull]

If you have any specific questions look in this discord search window and see if its been answered already.if it hasn't then ask away.

[tender badger]

I notice in your clip that the camera is pulling out rather than pushing in. Looking at the prompt it seems like you intended for the camera to dolly in while panning right, and for the products to be revealed in the pan to the right. The funny thing is I've been trying FOR DAYS to get the camera to actually pull back. I've tried every single way of describing it : dolly out, pull back, reverse dolly, move away from subject, it just keeps pushing in, no matter what. And VEO3 just did it for you, so I know it can be done!

[frosty geode]

hi

Creat a video for me

[fading rivet]

Has anyone tried to do a music track with lip sync that stretches beyond 8 seconds by using prompts in Veo-3 scene builder yet? Is it just me or do Veo 3 lip sync song gens just not fit together? Even tried a solo acoustic player with a slow song and still can't manage to make it work, so If anyone has achieved it then I'm all ears for advice and an example of how you did it

[frigid kestrel]

Hey guys, is anyone using other tools to like stable diffusion them or make them unique?

[pine harbor]

hi guys, is there a way to get around this message that i keep getting every time i try to upload an image created in midjourney that does not look like any celebrity, but i keep getting stopped because it says is "Our policies prohibit the uploading of prominent people."

[summer wagon]

I have spent an entire months worth of credits trying to get flow to do what I want to no avail. it has been close a few times but I want it to do very specific things for my advertising campaing for our clothing line YOGAZ www.yogaz.com I want the make our manniquin come to life and speak and say the words, "YOGAZ clothing makes me come alive." I finally got it to say the name correctly by tell it."Yo is pronounced like, "Yo" adrian from the rocky movie and Gazz is pronounced like "Jazz". but it won't stop singing. It won't just say it and half the times the mannequin will speak and the other half it will just be music. I want the manequinn to come alive and say the words. but after hundereds of tries and going on 2 months worth of credits and unknown number of prompts from both myself and chat gpt i am about to give up.

[faint anvil]

this is super confusing

[dim gull]

What's confusing about it?

[oak drum]

Hello everyone. I'm trying to make a video from a doctor moving his index finger and drawing in the air two letters, but without seeing the letters. All I get is a two finger taps and never get him to make the desired action. The prompt I am using is "A doctor smiling directly at the camera, performing a hand gesture as if drawing with his right hand's index finger. The finger moves following the "G" letter shape at his left sides, but without any letters or symbols actually visible. No signs or text remain. Ultra-realistic, professional lighting, sharp focus. Then he crosses his arms and smiles again in a confident look. No speech. No audio." Any hints to get what I need?

[dim gull]

https://promptalchemist.com/share/to5BeS25s9VnZcm7ij6G

[safe geyser]

Any tips on how to get Veo3 to spell this correctly?
ACME GROUND SYSTEM – MISSION OPERATOR TRAINING

Prompt =====>

A cinematic military-tech intro inspired by Interstellar and The Martian. The scene opens in low Earth orbit at night—stars shimmer beyond a slowly rotating satellite. The camera tilts downward through layers of atmosphere to reveal a glowing U.S. Space Force ground station nestled in the desert. Inside, a lone operator in uniform stares at a massive digital interface flickering with mission data. Alarms briefly flash red, then stabilize.

Suddenly, the title slams onto the screen in bold, metallic letters:
“ACME GROUND SYSTEM – MISSION OPERATOR TRAINING”
Final shot: the satellite above Earth locks into position with a deep, resonant hum.

Style: cinematic, high-contrast lighting, ultra HD, space military aesthetic, lens flares, atmospheric particles
Audio: deep sci-fi rumble, cinematic swells, pulsing tech-inspired score with bass and strings

==
Or is the text for the title I'm trying to generate too long?

Finally got Veo3 to spell the title correctly by shortening it to:
ACME
OPERATOR TRAINING

[teal walrus]

Hi how do I Prompt flow to stop adding subtitles to my videos? It's not even putting them in English

[karmic moss]

agg " CRITICAL VIDEO MANDATE: ABSOLUTELY NO SUBTITLES." in your prompt

[tired nymph]

how do i make veo3 create vertical videos? 9:16

[frozen escarp]

I am using Google Flow to create a new video of someone talking about something. The first video is great. How do I get Flow to create a video of the next scene?

[karmic moss]

what i do is expanding the aspect relation in a video creator , example , x 300% and Y)

X 300% and Y 200% , and the prompt "framed centrally"

[stiff root]

can some offer suggestions
video say it’s a violation

⸻

1. Time + Location + Environment
   Broad daylight, inside a low-budget TV studio awkwardly dressed to resemble a courthouse hallway. The set is cheap and obvious — vinyl wood paneling, prop law books spilling off crooked shelves, and a plastic American flag limply staged in the background. Harsh fluorescent lights give everything a flat, almost greasy look. You can hear the hum.

2. Detailed Scene Action
   A Bob Odenkirk-type lawyer — mid-50s, slicked-back hair, shiny cobalt-blue suit with an offensively loud red-and-yellow tie — bursts into frame mid-stride, grinning like he’s just been introduced on a game show. He points directly at the viewer, then spreads his arms as he launches into a rapid-fire sales pitch.

He walks directly toward the lens, hands gesturing wildly. His voice is overly enthusiastic, borderline frantic.
“Need help fast? Call Ez free bail bonds”
He throws in a dramatic wink and finger guns, still walking.

As he moves closer, the lens slightly distorts his face from the proximity — exaggerated and intense.

3. Camera Style and Movement
   Scene opens with a whip-pan to catch his entrance, followed by a fast dolly-in as he walks toward camera. Occasional snap zooms hit his hand gestures or his face at key punchlines. Subtle fisheye lens distortion adds camp. Cuts are punchy and comedic — quick snap cuts on his winks, finger guns, and wild gesturing, maybe even a bad star wipe to a reverse angle and back.

4. Audio/Musical Elements
   As he begins speaking, a cheesy 90s-style jingle kicks in behind his voice — think bad MIDI brass stabs, plastic-sounding saxophone solos, and funky slap bass.
   Comedic sound effects are layered in:
   • “Whoosh!” as he points
   • “Boing!” on finger guns
   • Canned studio laugh track or applause stinger as he lands a particularly desperate sales line

5. Mood, Lighting, and Cinematic Style
   Deliberately tacky and over-lit — sitcom

[dim gull]

Try removing Bob Odenkirk.

[zinc spear]

needs improvement on this one. I wanted to see more cherries 😂

[frigid kestrel]

So in the prompt add this for the frame style?

[dim gull]

No, hes saying that's how he crops it, he just uploads the 16:9 and blows it up to fill 9:16.

[dim gull]

<@&1366831844502274180>

[frigid kestrel]

spam???

I seen the same thing last week to get done,

thanks @dim gull

[prisma schooner]

;

[hushed bloom]

Need help with a prompt. I'm making a superhero movie for my daughter. The girl in the video is basically supergirl. I'm having a hard time getting the character to fly from the ground towards sky (duh lol) and I can't get the character to do a soft landing. She just jumps around lol.

[dim gull]

You would think that something like that would be easy, but Veo just really has a tough time with somewhat simple physical movements. Also, in case you're trying to get audio. Veo does not love having prompts with kids in it. Here was a prompt example I had from awhile back, trying to achieve a super hero takeoff.

https://promptalchemist.com/share/KbawTJrjK4rBLlYcW6aK

[rough lily]

Generate a video by revealing the given picture in very grand way with effects

[dusky oyster]

#flow-tips-n-tricks・✏️ create a timelapse stop motion video of red fort, india

[heavy wren]

Hello guys, i need a prompt for a marketing video of jewels, enybody else know a good one ?

[timid fable]

#flow-tips-n-tricks・✏️ create una mariposa volando

[untold hearth]

guys to make videos u need to subscribe?

[spice willow]

hello
first time using veo
Create a 20-second vertical video (Reel format) of a young woman complaining dramatically in a humorous and relatable way. She should be at home or outside a tech store, speaking directly to the camera with expressive body language.

Add Portuguese (Portugal) voiceover with the following script:

"Desde que o meu namorado foi à comprar o PC gaming… nunca mais me ligou nenhuma. Agora só fala com o teclado, os fones e os amigos do Discord. Eu? Fiquei em segundo plano. Pelo menos ele não me trai, está sempre em casa, colado ao ecrã."

Style: Light-hearted, TikTok-style, slightly exaggerated acting.

Add subtitles in Portuguese (Portugal), synced to the audio.

Background: Cozy or casual, urban vibe, visually appealing for social media.

Final scene suggestion: She shrugs and rolls her eyes while the boyfriend is seen in the background, gaming.
say everytime only in english

[dim gull]

Apparently auto-mod wont let me tag the other 3 of you that need to see this as well.. 🙄

[pearl cedar]

U can continue video by selecting option expand scene in scenebuilder

[frozen escarp]

@pearl cedar I appreciate it. But, can you please try to tell me how to do that?

[severe aspen]

that's weird re automod 🤔 looking into this btw

[dim gull]

Thanks Georgia, it wasn't a big deal, I @'d the four above that tried to use this channel to make a prompt as well, but Carl-bot gave me a warning. 🤣

[severe aspen]

robots r rude 🤖 and i prob messed up a perm somewhere lol! ty for the flag

[twilit dove]

Guys! please help! I want to know how to make the same character without losing and changing any details in VEO3

and how to add logo to the project I build?

[rugged bison]

Same issue here, first clip is great but when i use jump to it changes the character and the voice

[twilit dove]

Yeah. There is should be a solution bcz I saw some videos long avoiding it

[hushed bloom]

Yeah!!!! That's the issue I'm having with the take offs and landings. Not too concerned with the dialog part. Eventually I'm going to want to do a face swap for some scenes. We'll see how it turns out.

[sharp spoke]

as long as you are not doing it frame to video, you cant do it, making super descriptive prompt of a character will get you 90% there, but there will still be minor inconsistencies from video to video, same goes about the voice over

[dim gull]

I would say with adequate prompting, Veo 3 text to video can get consistency equivalent to popular image generators. Imagen, Midjourney, Sora. They all have trouble maintaining consistency as well.

Although there is a new I2V method im looking into that may be even better. Still need to test.

[drowsy canyon]

have you able to figure it out a way to get rid of subtitle once and for all?

[dim gull]

Not once and for all. But, I get subtitles less than 1% of the time now.

[drowsy canyon]

do you use "negative prompt": no subtitle?

[dim gull]

Part of it is negative prompting. The other part is dialogue structure.

I format dialogue like a Script and dont use quotations. Like this.

https://promptalchemist.com/share/6RybUtlcaRbQqGe28Ccg

[drowsy canyon]

I did use your methods before as well. No quotations. It seems like a miss and hit.

[dim gull]

Sometimes Veo just does what it wants. Gotta learn to live with that unfortunately. My method has been working well for me.

[severe aspen]

fwd

[willow thistle]

#flow-general・🌊 A cinematic military-tech intro inspired by Interstellar and The Martian. The scene opens in low Earth orbit at night—stars shimmer beyond a slowly rotating satellite. The camera tilts downward through layers of atmosphere to reveal a glowing U.S. Space Force ground station nestled in the desert. Inside, a lone operator in uniform stares at a massive digital interface flickering with mission data. Alarms briefly flash red, then stabilize.

Suddenly, the title slams onto the screen in bold, metallic letters:
“ACME GROUND SYSTEM – MISSION OPERATOR TRAINING”
Final shot: the satellite above Earth locks into position with a deep, resonant hum.

Style: cinematic, high-contrast lighting, ultra HD, space military aesthetic, lens flares, atmospheric particles
Audio: deep sci-fi rumble, cinematic swells, pulsing tech-inspired score with bass and strings

[severe aspen]

hey there! welcome to the community. in order to make videos with Flow, you'll need to visit flow.google and have a Google AI Subscription: videos will not generate directly here in the server.

if you have a prompt you'd like to have generated, you can share it here and there's a chance that someone in the community can generate it for you. please note that this is up to members of the community and is not a requirement from Flow creators.

check out Flow here
here's more information on Google AI Subscriptions

[agile bone]

Hey guys, so i have been trying to use a prompt in flow with no success. I am essentially trying to make hyper-realistic "Bratz doll" content but woth a modern Social Media influencer twist. No matter how much I tweek the propt, I get a failed to generate message. Paahhhleaseee someone help me out.

[molten parcel]

Flow

[dim gull]

If its not generating at all, it's probably because you're using the brand name in your description of it.

[stray thicket]

Anyone figured out how to generate a realistic proboscis monkey (Nasalis larvatus) in Veo3?

I’ve tried:
Nasalis larvatus, proboscis primate, and detailed prompts like:
“adult male, reddish-brown fur, grey limbs, large drooping nose, round belly, long arms”

Also added exclusions like:
“other primates, humanoid features, generic apes”

Still getting inaccurate results — mostly wrong species or cartoonish outcomes.
Any working prompt strategies or tips?

[eager kiln]

Hello, so I tried to use this prompt for Veo 3 Quality and the people keep singing with the song in the background for some reason:

Location: A stunning and remote beach in Negril, Jamaica, at sunset. Palm trees sway gently in the breeze. The water is crystal clear and calm.

Characters:

Character 1 (Kai): Male, late 30s, Jamaican, warm and wise expression, has dreadlocks, wearing a comfortable, unbuttoned linen shirt and shorts.

Character 2 (Leo): Male, late 30s, visiting Jamaica, looks like he's finally de-stressing from city life, wearing a simple t-shirt and shorts.

Action: The video opens with a slow-motion shot of their feet in the wet sand as waves wash over them. It then cuts to a wider shot of them walking along the beach. They approach a small bonfire they've made from driftwood and sit down on the sand next to it. Kai pokes the fire with a stick while Leo gazes at the horizon. They share a quiet, meaningful look.

Camera: Use a mix of epic wide-angle shots to capture the landscape and tight, intimate close-ups to capture emotion. Gentle dolly shots and slow pans. Anamorphic lens flare is highly desired when pointing towards the setting sun.

Audio (Music & Sound):

Music: An instrumental, soulful, and authentic reggae track. The music is purely background and should not have any vocals.

Sound Design: High-fidelity sounds of gentle waves, the crackling bonfire, and a distant sea bird.

Dialogue:

Instruction: The following dialogue must be spoken clearly and naturally in a normal speaking voice. The characters are NOT singing, chanting, or performing.

LEO: (Voice is soft, full of emotion) "I'd forgotten what this peace feels like. It's been too long."

KAI: (Voice is calm and reassuring) "The island never forgets you, my friend. She just waits for you to come home." ```

[frozen escarp]

I'm still wondering how do I create a series of videos that I could download and combine them all to make a longer video?

Like, how do I get it to use the same person, talking about the same thing as the first video?

[summer mantle]

Steven, i've had some success with the ultra plan and using "ingredients to video" You have to use a prompt that says something to the effect of "Same man as previous clip"

[frozen escarp]

@summer mantle la1 Here's an example of what I'm trying to continue to see the rest of the intended video.
But, (shhhhhh, keep this quiet, LOL. I understand the restrictions for any medical content, so, I prompted Flow to make a video about a mannequin, and it's doctor, LOL) How exactly do I continue this video?

https://youtu.be/vmRG8-25H5w?si=PkmqvIbbz0pgXqxz

[maiden marsh]

please keep all messages in English thank you!

[trim thistle]

Hi team! 👋

I’m really excited to try Flow for a project I’m developing at Smart Rooms — a Brazilian coworking company with multiple premium locations, automated booking, smart lock access, and a fully self-service platform.

I’d love to create realistic influencer-style videos showing a professional woman walking through our real office spaces, talking to the camera in Brazilian Portuguese and presenting our core features (like instant booking, prepaid model, and WhatsApp integration).

I already have photos of our real environments and would love to test Flow to generate immersive, localized video content.

Thanks in advance — looking forward to joining the early access group! 🚀

[frozen escarp]

She looks similar to the previous one. But, I believe that it is not exactly the same person. But, I also believe that this is possible.
This is what I prompted:

The same doctor is in the same scene and she continues to explain it to the same mannequin

👍https://youtu.be/DYy9MmmaDig?si=UFcOjJWiBYpje2EH

But, I might be hoping too much, hoping to make a video of her continuing the same conversation.

[tacit lark]

#flow-general・🌊

[next sentinel]

Veo2 Fast - starting & ending frames, prompt is "scene starts with an empty chair in a restaurant, then a series of pink lightning bolts strike near the chair, then the woman appears out of thin air in a flash" the result lackluster: a woman sitting up in a chair and some cheesy bolts of lightning. is the server overloaded or something? quality is bad this time of day?

[obsidian hatch]

I do yes...

[dim gull]

Your prompt is lacking a lot of description and detail, Of course you're going to get a lackluster result. You need to prompt for the results you want. Make sure to describe the woman, the restaurant, the chair, the level of realism you want... the better you prompt, the better your results will be.

https://promptalchemist.com/share/GtayzbC9hiSfcIf0WaIE

[next sentinel]

I'll try this, much thanks. Yesterday I got much better results so thats it seems like there's a degradation in quality for no apparent reason.

[dim gull]

I havent noticed a decrease in quality today. But im also using Veo 3 fast for the most part. Good luck getting some better results!

[heavy totem]

I’ve noticed that for me quality falls off around 11 pm and on weekends starting afternoon

[next sentinel]

seems like the issue is a Veo2 vs Veo3 thing. Using PromptAlchemist and Veo3 seems to get what I need, thanks again!

[dim gull]

Glad to hear man. Thats awesome 👏!

[bronze saddle]

Anyone have a or some good prompts for making “stream starting soon” intros for twitch?

[summer dagger]

hello looking to make a minute video using flow, finding it diffucult to get consintency throughout the different scenes from character to objects to the setting of the scene

[next socket]

anyone got the prompt with the middle finger with animals for veo3 ?

[prisma schooner]

⸻

Setting:
A dense pine forest clearing just before dawn. Mist curls low over the mossy ground, damp pine needles crunch underfoot. The air smells of fresh pine and earth. Nearby, a rusty, camo-painted military crate rests against a fallen log, half-covered in dew.

Filmed from Bigfoot’s chaotic selfie-stick POV — camera jitters with every breath and step, raw, unfiltered YouTube vlog energy. GoPro strapped on Bigfoot’s head.

Bigfoot — fully covered in thick dark brown fur, ape-like facial features, no visible human skin, tactical vest layered under a dark hoodie, brown backpack half-zipped — holds the selfie stick with a grin, eyes darting.

Fred the Snow Yeti — thick white fur, ape-like facial features, ragged camo vest, iced-out “FRED” chain — crouched silently by the crate, inspecting it with intense focus.

BIGFOOT (whispering, hyped):
“YO, BOYS, CHECK THIS OUT — FOUND A MILITARY CRATE DEEP IN THE WOODS. NO IDEA WHAT’S INSIDE, BUT THIS IS ABOUT TO GET WILD!”

SFX:
– Crunching pine needles
– Soft morning bird calls
– Distant rustle of branches
– Low hum of forest insec

[granite hinge]

@bronze saddle I used a random prompt like this Create a cool design for a twitch overlay for Webcam. Needs to look professionally done 8k it not perfect but its a start lol

But with some tweaking I am sure your going to one amazing.

[frozen patrol]

anybody here been able to get veo3 to say .ai domain in Spanish properly? For example, I asked to say mycompany.ai but the pronunciation is wrong. It does not spells them as 2 separate letters a i, but rather one word ai

[rocky citrus]

write those words in capital letters, "A.I." like this

[strange siren]

Go

[lean spear]

Does anyone have a good prompt here for cooking spaghetti noodles?

[next sentinel]

I have an issue where this prompt results in the character just getting up and leaving the table instead of disappearing in a flash of light per the prompt. ```Subject: Velvette.exe.v2.0: Velvette.exe.v2.0 is an exceptionally attractive, slender AI woman appearing in her early twenties. She possesses a symmetrical face with distinct Middle Eastern or Italian features, framed by long, flowing black hair dramatically highlighted with thick, vibrant hot pink streaks framing either side of her face. Her forehead is partially covered by sleek, styled bangs. Her expressive eyes are accentuated by heavy black eyeliner and bold, sharply defined black eyebrows. She is dressed in a daring and fashionable style, typically wearing a form-fitting, low-cut black blouse paired with high-waisted, hot pink leather pants. A matching hot pink leather jacket, often partially unzipped, completes her look. Her posture is confident and poised, often striking alluring poses that emphasize her figure.

Action: The POV is set from the perspective of someone sitting across the table from Velvette.exe.v2.0 as her eyes begin to glow a vibrant hot pink. The glow intensifies, drawing the viewer's attention as electrical charges, manifesting as crackling pink lightning, start to emanate from her slender frame. The electrical display intensifies rapidly until Velvette, dressed in her signature hot pink leather jacket and pants, vanishes in a burst of chaotic energy, all that's left is an empty seat and the lingering scent of ozone.

Scene: An opulent French restaurant bathed in dramatically moody, low-key light during the evening. The POV emphasizes the table setting, highlighting the delicate porcelain china and crystal wine glasses catching the limited light. High-arching windows are draped with heavy crimson velvet curtains, diffusing light from unseen sources. The walls are adorned with ornate, gilded frames surrounding impressionistic paintings. The table is draped with a light pink linen tablecloth, and the architecture boasts intricate moldings and cornices. The room is illuminated by strategically placed, low-wattage chandeliers and wall sconces, casting long, dancing shadows that accentuate the rich textures of the velvet and polished wood. The color palette is dominated by deep reds, soft pinks, and touches of gold, creating a warm, romantic, and slightly theatrical atmosphere.

Style: The camera, set at eye-level from a point of view shot, remains on a static tripod, rendered in a cinematic, dramatic film grade with soft, natural daytime lighting creating a mysterious and ethereal atmosphere, 16:9 aspect ratio.

Dialogue: None.

Sounds: The clinking of silverware and hushed ambient chatter abruptly cut out as a sharp electrical crackle fills the room, followed by a brief, intense hum.

Technical (Negative Prompt): subtitles, captions, watermark, text, words on screen, logo, branding, singing```

@dim gull lemme know if you have tips or suggestions.

getting a lot of this. Also, I'm doing image to video

so afer a few re-tries, I got what I needed, FYI

but it occured to me it could be funny if the character left, then walked back across the shot to the right & said something awkward

here's what I settled on

[dim gull]

Yeah, ive had issues with trying to get things to vanish as well, veo just doesn't know what to do with things sometimes.. Frame to video is also pretty finicky. This one looks pretty good though!

[gleaming lava]

https://www.youtube.com/watch?v=ZcD8dbdxlZ8

[pure crest]

I signed up for a subscription, but it said I would get a free month, but then out of nowhere my card was charged, so I would like my money back, because I had already canceled the subscription.

Please my money!

[dim gull]

Here's one of the ones where I managed to get him to vanish, but still not as seamlessly as I was hoping for.

https://promptalchemist.com/share/XW63As65qMY5Lc7dZxZD

[pliant valve]

Subject: Numemon: Numemon is a squat, amorphous Digimon resembling a sickly green slug. Its skin is slimy and mottled with irregular, indigo-blue spots scattered across its back. Two thick eyestalks protrude from the top of its head, each supporting a large, blood-red eye. A gaping maw reveals rows of small, sharp teeth and a thick, pink tongue that occasionally lolls out. Its arms, usually retracted, are short and stubby, with small, claw-like hands. The overall impression is one of sluggishness and decay, with a perpetual expression of vacant disinterest on its face. Veemon: Veemon is a small, bipedal dragon Digimon with bright blue scales covering his body. He has large, expressive golden eyes and a V-shaped marking on his forehead. Two small, white gloves cover his hands, and he sports a yellow underbelly that extends from his chest to his tail. Veemon has a short, stubby tail ending in a rounded tip. His overall appearance is soft and rounded, giving him a huggable quality. He typically stands with a wide, cheerful grin, often tilting his head slightly. His posture is energetic and bouncy, always ready for adventure. He is small and stout, appearing youthful and vibrant. Action: Veemon, standing on a small rock, wrinkles his snout in disgust and dramatically gestures with his white-gloved hand towards Numemon. Numemon, oblivious, continues to ooze slightly, his slimy green body glistening in the light. Veemon then delivers his line, his bright blue scales shimmering slightly. After Veemon speaks, Numemon remains indifferent, continuing his sluggish movements. Scene: A desolate, rocky landscape, typical of the Digital World, stretches out under a harsh, orange sky. Jagged cliffs and metallic structures dot the horizon. A small pool of murky water reflects the sky's colors. Patches of digital vegetation cling to the rocks, providing a sparse contrast to the otherwise barren environment. The air feels thick and stagnant, contributing to the unpleasant atmosphere. Style: A medium shot of the two Digimon, filmed at eye-level with a static tripod, lit by warm, golden hour sunlight creating a joyful and celebratory atmosphere, 16:9 aspect ratio. Dialogue: (Disgusted, annoyed) Veemon: Ugh, Numemon, you smell terrible. Sounds: The squelching sound of Numemon's slimy body as he moves, combined with the sharp, exasperated tone of Veemon's voice. A low, digital hum permeates the environment. Technical (Negative Prompt): subtitles, captions, watermark, text, words on screen, logo, branding

@dim gull

[reef sorrel]

I’ve compiled all the prompt examples from the Inspiration Gallery section on the Google Vids website.

A loving grandmother with soft, silvering hair, wearing lavender pajamas, gently tells a story to her bright-eyed granddaughter in pink pajamas. They are nestled in a cozy bedroom bed under a patchwork quilt. Soft bedside lamp light illuminates floral wallpaper and a bookshelf. The grandma's warm, soothing voice telling a story. Script: And as the little star twinkled bright, it whispered secrets on the breeze.

https://ssl.gstatic.com/docs/generativeai/videogen/inspirationgallery/veo3/inspirationgallery_veo3_video6.mp4

Sunbeams pierce the dense Amazon canopy as the camera slowly pans across the misty, vibrant rainforest floor, revealing a tiny, jewel-toned frog. The immersive soundscape features water dripping from broad leaves, a rich chorus of exotic birdsong, and the distant, echoing call of a howler monkey.

https://ssl.gstatic.com/docs/generativeai/videogen/inspirationgallery/veo3/inspirationgallery_veo3_video9.mp4

An energetic middle school science teacher in a tweed blazer stands in a vibrant, modern classroom. He gestures passionately at a massive, glowing image of the Andromeda galaxy on a state-of-the-art smartboard. Dynamic camera movement slowly zooms in on his face. Script: Look at this. Trillions of stars, millions of light-years away. Every single speck is a sun, maybe with its own worlds. What do you imagine?
https://ssl.gstatic.com/docs/generativeai/videogen/inspirationgallery/veo3/inspirationgallery_veo3_video8.mp4

A 90's hip hop female rapper raps about Google Vids. Urban setting on a street corner, lowriders in the background. Lyrics: I'm on the mic, 'bout to drop this rap. Google Vids in the house, AI-powered, straight up, Makin' video for work, ain't no need to fuss.
https://ssl.gstatic.com/docs/generativeai/videogen/inspirationgallery/veo3/inspirationgallery_veo3_video12.mp4

[velvet isle]

📜 Visual Prompt (Use for all scenes):

A realistic monkey with a consistent face, dressed in the 1965 Indian Army uniform: olive green battle dress with canvas belt, leather boots, steel helmet, and ammo pouches. He is holding an FN FAL rifle (used by Indian Army in 1965). The setting is a dusty battlefield near the Punjab border, with bunkers, barbed wire, army trucks, tanks in the background, and Indian soldiers around him. Gunfire, tank blasts, and smoke clouds fill the background.

🎞️ Scene 5: – Outro (Vlogger Style)

Visual:
Monkey turns toward the camera, covered in dust and scratches, slightly smiling.

SFX: Slow fade-out with soft military bugle tune.

||📜 Visual Prompt (Use for all scenes):

A realistic monkey with a consistent face, dressed in the 1965 Indian Army uniform: olive green battle dress with canvas belt, leather boots, steel helmet, and ammo pouches. He is holding an FN FAL rifle (used by Indian Army in 1965). The setting is a dusty battlefield near the Punjab border, with bunkers, barbed wire, army trucks, tanks in the background, and Indian soldiers around him. Gunfire, tank blasts, and smoke clouds fill the background.

🎞️ Scene 5: – Outro (Vlogger Style)

Visual:
Monkey turns toward the camera, covered in dust and scratches, slightly smiling.

SFX: Slow fade-out with soft military bugle tune.||

_📜 Visual Prompt (Use for all scenes):

A realistic monkey with a consistent face, dressed in the 1965 Indian Army uniform: olive green battle dress with canvas belt, leather boots, steel helmet, and ammo pouches. He is holding an FN FAL rifle (used by Indian Army in 1965). The setting is a dusty battlefield near the Punjab border, with bunkers, barbed wire, army trucks, tanks in the background, and Indian soldiers around him. Gunfire, tank blasts, and smoke clouds fill the background.

🎞️ Scene 5: – Outro (Vlogger Style)

Visual:
Monkey turns toward the camera, covered in dust and scratches, slightly smiling.

SFX: Slow fade-out with soft military bugle tune._

[drowsy furnace]

please help... i have tried this prompt 10 times and every time the product rotates like it is on a turntable but the camera does not move! Looks so odd. I want the camera to rotate around the product to a 90 degree view so the bottles all line up. PROMPT: a product shot of three makeup bottles sitting on grass floor surrounded by flowers. Camera slowly moves from left to right, roating 90 degrees around the bottles and the grass and the flowers, as the the flowers gently sway

[slender parcel]

Can't we get 9:16 reels ratio in flow ?

[frank robin]

not atm

[frank robin]

your problem is in the prompt - dont tell it to rotate, the camera movement you are looking for is called an orbit (where the camera orbits around the product) - try asking for a 180 degree orbit around the product and see if that helps

[drowsy furnace]

Thanks. Did not work. Same result:(.

[dim gull]

https://promptalchemist.com/share/DA7BE0hbwAAW09UIyzkI

[reef sorrel]

first attempt

[reef sorrel]

2nd attempt

nvm 3rd attempt got her implode

[reef sorrel]

As the action begins, the cinematic POV remains fixed on Character A as her expressive eyes, accentuated by heavy black eyeliner and sharply defined black eyebrows, begin to glow a vibrant hot pink. The dramatic, low-key lighting subtly shifts, highlighting the intensifying glow. Small electrical charges, manifesting as crackling pink lightning, begin to emanate from her slender frame, visually building the Hot Pink Electrical Discharge effect. The high-arching windows draped with heavy crimson velvet curtains are visible in the background, dimly lit by unseen sources, reinforcing the luxurious setting.

Focusing closer, the electrical display intensifies rapidly around Character A. The hot pink lightning crackles with more vigor, creating a chaotic energy field around her signature hot pink leather jacket and pants. The Moody Low-Key Lighting creates stark contrasts, making the Hot Pink Electrical Discharge more visually striking against the deep red and gold backdrop of the restaurant. The static, eye-level camera emphasizes the direct impact of this Cinematic POV Shot as the viewer is drawn into the escalating phenomenon.

The scene concludes as Character A, consumed by the Hot Pink Electrical Discharge, vanishes in a burst of chaotic, Vaporizing Teleportation energy, leaving an empty seat. A subtle, lingering visual distortion, like a heat haze or shimmer, briefly remains where she was, illuminated by the remaining moody light from the low-wattage chandeliers and wall sconces. The opulent table setting, with its china and crystal, remains untouched, emphasizing her abrupt departure from the Opulent French Setting.

The soundscape features the subtle clinking of silverware against porcelain and the low, hushed ambient chatter of a busy, upscale restaurant. This gentle background abruptly cuts out, replaced by a sharp, piercing electrical crackle that rapidly intensifies, followed by a brief, intense, high-pitched hum that fills the room, suggesting a powerful energy discharge and the lingering, sharp scent of ozone.```

[dim gull]

<@&1366831844502274180>

[dull garden]

https://youtu.be/fOshzv2cLG4?si=SNMR2G_cqZk2Hc3Y this one was particularly hard because Flow still has trouble with continuity in dialogue scenes. After a little editing I still think it came out okay. What do you think ?

[austere iris]

Hello, please help. There is no sound on the video. The dialogue in the prompt is written, but there is no sound