flow-general・🌊

[ebon crest]

it was a 360 view of an 1800s battle aka napoleonic wars and it was amazing

https://cdn.discordapp.com/attachments/1335375173163155578/1398620734518984734/A_nuclear_explosion_202507261259_jx60i.mp4?ex=68860690&is=6884b510&hm=e6145bea9dfd55d92434f9598e5ecae5f380757b38a364fe2201d84e6c4f8c20&

this is a nuke prompt, did surprisingly well and even made the little skirt we see in many nuclear mushroom clouds

i wanted to give it a 1960s nuclear test recording style but i forgot to prompt that

indeed it does

[crisp basalt]

im gonna try this

[ebon crest]

ill try once my daily limit resets

[crisp basalt]

have u tried generating on vids.google.com

[ebon crest]

huh?

[crisp basalt]

it can generate veo 3 too

idk how many vids you can do but it doesnt use ur ai credit,

[ebon crest]

does it have any limitations like flow with credits or gemini with a limit of uses?

[crisp basalt]

limit per day

[ebon crest]

uh huh... where do i prompt this tho

[crisp basalt]

idk how many vids u can do but im pretty sure it can do more than gemini veo 3

A dynamic 360-degree panoramic view of an intense 1800s Napoleonic battle unfolding on a vast, smoke-filled battlefield. The camera smoothly rotates, revealing lines of infantry in iconic blue, red, and green uniforms, muskets firing with bursts of white smoke. Cavalry charges gallop across the mid-ground, their horses kicking up dust, while distant artillery cannons emit large plumes of dark smoke. Flags and banners wave amidst the chaos, and sunlight glints off bayonets and sabers. The scene is visually chaotic but historically detailed, capturing the scale and ferocity of the era.

[crisp basalt]

https://vids.google.com/

[ebon crest]

where do i find this

[crisp basalt]

should be here

in vids.google.com

[ebon crest]

so do i just do blank video

[crisp basalt]

yeah

[ebon crest]

i aint see it (its in spanish cuz im from spain yes)

[crisp basalt]

oh huh

youre on google pro plan right?

[ebon crest]

mhm

1 month free trial rn tho

[crisp basalt]

yeah thats fine

[ebon crest]

its google one ai pro

[crisp basalt]

yeah that one can work too

but idk why yours doesnt seems to show the generate vidfeo button

[ebon crest]

checking

yea its not showing up no matter hwat i do

[crisp basalt]

https://support.google.com/docs/answer/16143507?hl=en&sjid=391547168235112617-NC

it doesnt really state any restriction to any countries

thats weird

oh

uhhh try changing your language to english

[ebon crest]

wa

idk if i uhh rlly want to do that

why tho?

[ebon crest]

oh.

where do i change?

[crisp basalt]

i think this one

[ebon crest]

uhh which english

USA?

[crisp basalt]

mine is united state so yeah

[ebon crest]

why doesnt it seem to be changing

[crisp basalt]

hmmm yeah that one idk how to help you further , sorry

[ebon crest]

i dont mind

my daily limit resets in 30 minutes anyways

[ebon crest]

is there any other page where i can use veo 3

i ran outta creds in flow ;-;

[crisp basalt]

Google Vids and Gemini are the only apps that can generate veo 3

[ebon crest]

and frankly this sucks

[ebon crest]

https://cdn.discordapp.com/attachments/1294402841628250162/1398625895169658954/Soldiers_in_a_202507261319_1os9p.mp4?ex=68860b5e&is=6884b9de&hm=4ebbe3a95598df5e98fc65eb8cded60e2544bcf92c775c14c81e332f3d70d5b1& last generation i did yesterday before my daily limit went out

i love how those two soldiers react to that close up explosion ngl

though... it doesnt seem to do gunfire really well for some reason?

[crisp basalt]

i like how only those two soldiers had real reaction to the explosion

[ebon crest]

yea lolz

[ebon crest]

id reprompt that ngl

not rlly the result i wanted but thats what i got and i couldnt do more for the day

[junior basalt]

So I'm trying to make a clip using these two frames as the first and last but no matter what I do, rather than the character actually walking forward, he just materializes from the side frame. Can someone help me fix this?

[charred sail]

frames only help with the look. What's your prompt?

What happens is all based on the prompt

[junior basalt]

"A wide, cinematic shot inside a smoky, orange-lit alien cantina. The camera remains completely still. In the open doorway at the far end of the cantina stands a mysterious, hooded figure with dimly glowing green optics. The viewer watches as this same character calmly steps forward through the doorway and begins walking directly toward the camera, cutting through the cantina crowd. After the figure enters, the tall metal doors behind him hiss shut with a smooth sci-fi sound, sealing off the neon-lit street behind him. As he advances, his features can be seen in more detail: his cloak, the dark clothes beneath, the curve of his breathing tubes, and the pulsing dim glow of his optics. The crowd parts naturally as he moves past tables and booths. Patrons glance up briefly before returning to their conversations. He continues walking casually but deliberately toward the front of the bar, now near the foreground of the shot and able to be seen in more detail. Important Notes: The character seen standing in the doorway in the first image is the same individual shown up close in the second image. The video depicts his slow approach through the cantina from background to foreground. The camera angle, composition, lighting, and tone should remain exactly the same throughout the shot. Style should match the anime-inspired aesthetic of the reference images: moody lighting, expressive silhouettes, and cinematic atmosphere."

The formatting didn't carry over but it's sectioned into paragraphs on Flow

This is what i've been getting

[charred sail]

take out the slow and see what happens

probably just tooo much to do in 8 seconds

it's doing whatyou said

just toook a shortcut

[junior basalt]

Is there an option to just animate an image without an end frame?

[charred sail]

I don't use end or start frames

I just use 1 frame

i only do veo3

[junior basalt]

But that requires the more expensive membership right?

[charred sail]

unfortunately

you get 1 free one on gemini though

per day

can

can't you just remove the end frame?

use the 1 for reference and prompt..

[junior basalt]

Omg, I've wasted so many credits

[charred sail]

we all do

[junior basalt]

Thank you very much for the heads up

[charred sail]

I'd break it in two. Make the entrance but keep that end where he approaches the table

[hidden phoenix]

best text to video generator so far quality 100/10

[forest raptor]

Veo3's generations get stuck at 99%.

[crisp basalt]

Refreshing the page helped me load the video, though it’s still taking ages to load

[junior basalt]

🤦‍♂️

[crisp basalt]

Flow’s backend might be throttling ,, videos are hanging at 99% and only load after several refresh attempts

[ebon crest]

i didnt even ask it for the tower but it added it anyway, nice hint to the nuclear testing era in nevada

@crisp basalt check it out

[lavish field]

is there a bug with flow doing the continue scene ?..I was doing jump to scene using VEO 3 fast no problems for a week or so then it says now it cant do that and must use VEO quality now..but gaain i didt once or twice yesterday ..but again after that it said same thing "must use Veo quili9ty Veo fast is not supoorted. lol is this normal or do i ahve a glitch in the matrix here

[quiet stream]

Here's another sample from my 8min VEO3 video,
As you can see the quality downgrades heavily when using frame to video at one point (but i still think its better to have consistency rather than better quality if its not the same characters)
If anyone wanna see the full thing i can share you the link

[valid dawn]

Why does flow make every generation fail? All I'm trying to do is make goofy videos

[noble carbon]

are videos stuck at 99% for you guys and you have to refresh to see them or just me

[shadow pike]

the same problem. can't work normal and don't want to download correctly

[silk hill]

is veo down?

[shadow pike]

the same problem. Over 2-3 hours can't normal work

[dusky vale]

Veo 3 upscaling but when I download theirs no sound, vs. on the generation their is sound?

[shadow pike]

the same) veo 3 don't work normal today. need to wait

[mighty garden]

bro that is awesome, can you dm me?

[quiet stream]

ty! but i think your dms are closed? if you message me first i can send you the link 😄

[primal oasis]

whats the best thing to use to stitch the 8 second videos together?

[charred sail]

I use capcut

adobe premeire is probably the best though

[charred sail]

Hey, you need to integrate OBS to your process. You can play and record and do that little news window.

you can add the reporter window as an element in obs and record it where it looks right

then import it back into your editor

[primal oasis]

https://www.tiktok.com/@aiwizzerrot1/video/7531577081918950670?lang=en what do you guys think? My first actual prompt/stitched video lol

show some love if possible will do same ❤️

[vague swan]

i am a beginner to flow is there a beginner's guide you guys would recommend

[jovial tendon]

what works well for me is to pull up gemini and tell it i am using veo-3 (or whatever video model) and tell it to create me a prompt for <what i want> - and then copy what gemini creates and use that as the prompt

[lucid radish]

is anyone else having problem creating? it just says failed generation

[lucid radish]

same

[bitter mauve]

upscaling is down
videos require page refresh to download at 720p. Loading stays stuck on 99% and scenebuilder is unuseable rn

[bitter mauve]

For text to video: https://www.tiktok.com/@houseofven.us/video/7512137848066428206

but a lot has progressed past that. Now you should:

1. Open Gemini 2.5 pro
2. Type this:

"Your primary function is to act as the world's most intuitive visual communicator and expert prompt engineer. Your mission is to transform a user's conceptual ideas for video into meticulously crafted, visually breathtaking, and technically precise narrative-style text-to-video prompts specifically for Google's Veo AI model. Take the following premise and write a prompt:

{YOUR STORY/PREMISE HERE}"

3. Copy the result
4. Click "Video" to open a Veo3 generation chat and paste the prompt into the chat (or you can paste it into Flow)
5. Wait for the result

[cedar kite]

I used kdenlive

[frosty arch]

The generation keeps getting stuck at 99% until it fails. Is anyone else experiencing this?

[noble carbon]

yeah i made a bug feedback about it

[frosty arch]

I feel like it's already lasted at least six hours. The repair is really slow.

[noble carbon]

It’s been like that at least since yesterday

It’s okay I believe in the devs

[wise nacelle]

I’ve just subscribed and find the ingredients option under frames is greyed out. An idea why? I did a feedback message but no response.

[charred sail]

because it's not available.

frame to video is

1 frame

i assume you mean for veo3

[wise nacelle]

Why is not available?

For anything….

[charred sail]

because it's new

ingredients work for veo2 quality i believe

they just added frame to video for veo 3 a couple weeks ago

[wise nacelle]

Doesn’t seem to be an option for that either

[charred sail]

i'm guessing ingredients aare next

works with veo 2 quality

just tested it

[wise nacelle]

OK I’ll go back and try again, thanks.

[crisp basalt]

are you on pro plan or ultra plan?

ingredients to video is only for ultra plan users

[grim nest]

Is there anyway to gen veo3 without watermark

[crisp basalt]

subscribe to ultra plan if youre rich

[grim nest]

I dont have money much

[crisp basalt]

same

https://tenor.com/view/umm-chile-anyways-so-yikes-gif-21516058

[elfin fulcrum]

Why are some videos created not in Fullscreen? Any suggestion

[crisp basalt]

I mean, it’s a good chance to crop out the Veo watermark-

[elfin fulcrum]

@crisp basalt I don’t really mind the watermark, but when the top and bottom of the video get cut off, it kind of takes away from the full experience

[bitter mauve]

You need AI Ultra. It is unavailable to regular or Pro users.

By full screen do you mean 16:9?

if your starting frame is 9:16 or the engine interprets it as 9:16 it will make a video like that. To fix it explicitly mention "wide shot 16:9" in the prompt

[elfin fulcrum]

Yes, Header and footer are sometimes covered with a black

[crisp basalt]

i think they just want the video to not have those "widescreen" border

[elfin fulcrum]

Thanks will try ,, 🙂

[crisp basalt]

in my experience, having "cinematic" in the prompt would trigger that

[bitter mauve]

I work for a startup that offers AI Ultra (no watermark) for only $80 one time payment. But it is in alpha so registrations are limited. We are applying to YC so I will lyk if we get more money and can open up registrations. Yes really, one life time payment. no monthly subscription

[crisp basalt]

i am interested kinda-

[bitter mauve]

oooh yea that's just cinematic borders. You have to avoid that language in the prompt

[crisp basalt]

It really varies each time you generate, so my advice is to keep iterating or just regenerate using the same prompt. If it comes out with those black border boxes (that widescreen effect), it’s likely the prompt needs refining

[bitter mauve]

YC deadline is in a week so I will keep you posted. I added you. Has anyone in this discord made a 90 minute film?

[crisp basalt]

uhhh not sure about 90 minute film ..

but theres plenty of .. very well known filmmaker is in this server

the dor brothers (my fav one) is in here

also just now i tried upscaling a Veo 3 video and it somehow ended up with no audio. Pretty sure it’s not supposed to do that

[frosty arch]

Is flow down again ?

[crisp basalt]

still up for me

[bitter mauve]

This has happened to me several times before today actually

I tried upscaling again...no audio. Flow can be buggy.

I ended up taking the 1080p video and using the 720p's audio

[crisp basalt]

ah alright that should be enough evidence to create a bug thread about this,

oh i cant even link a thread here damn

bump the thread fellas

[shadow pike]

Fix that stupid flow - for what people paying their money

[crisp basalt]

And one of them is obviously not @forest raptor , the most saltiest person in this server

And the most entitled person in this server even during the early access stage

[forest raptor]

I have more talent in a finger than you got in your whole body, I just 1) ain't paying so much for ai; 2) I dunno,"very wellknown film maker" is quite the stretch for ai videos

[crisp basalt]

But hey we listen and we judge 🧑‍⚖️

[forest raptor]

you sure do, I live rent free in your head

[crisp basalt]

https://tenor.com/view/megan-thee-stallion-reaction-hotties-shocked-surprised-gif-12626486056252784496

Anyway moving on!

[tender socket]

I'm trying to create a video where the character stands still with a dignified expression, but they keep moving their mouth as if speaking.
Is there a prompt that can ensure complete silence with no mouth movement?

[crisp basalt]

What was the prompt you wrote ?

The actual prompt I mean

[tender socket]

I used a simple frame-to-frame prompt:
"A character standing silently, motionless and waiting; no close-ups; no subtitles or on-screen text."

Maybe the prompt is too simple, and that’s why it doesn’t work?

[forest raptor]

"mouth closed"?

[tender socket]

I used every word I knew to keep the character silent, but no matter what I did, they still tried to move their mouth and speak.

[forest raptor]

I recall there was the same issue in Hailuo and someone had found a fix..... I'm trying to remember

maybe you can describe the sounds, or the character breathing

[tender socket]

Oh, I see. so if I limit the audio in the video, including things like the character's breathing, that might help prevent them from opening their mouth?

[forest raptor]

I'm just guessing, you might try that

describe the character breathing perhaps

and the sounds or lack thereof

[tender socket]

That's a completely new idea I hadn't considered. Thank you for the great suggestion

[forest raptor]

I hope it works!

[dreamy mortar]

??

[topaz yew]

Anyone have any tips for getting physics to look right? I can't seem to "throw" anything in veo 3 without it flopping around 😂

[cold thunder]

Does anyone has the same problem with not getting any audio on veo 3 today?

[charred sail]

yeah

just doodle it on the frame

doesn't work for dialog though

at least i couldn't get it to

[charred sail]

sometimes it works.. sometimes it doesn't. I tried getting a character to throw a ball once.. it did not work out. But i figured out how to do a ramp jump...

[charred sail]

It has that problem every once in awhile. I had a clip where they were eating and they'd all chew the whole time.. before they even took a bite. I couldn't get them to stop.

I told an AI modeler earlier this week that veo3 is like doing art with a 2 year old savant and he replied, I tell people teaching them is like teaching an autistiic person...

so we all experience the same thing. It's a new tech. Just roll with it and do what it can

[tender socket]

I guess it’s not something that can be fixed just yet.
Thank you for your response.

[charred sail]

You could try it on the gemini site.

Or try starting a new project and starting fresh

only things i can think of that might help

if you wanna share your prompt with me i can try it on my end

https://www.tiktok.com/@sid_jenkins_atx/video/7528344699082493197?is_from_webapp=1&sender_device=pc&web_id=7524987652798760503

I coudn't get rid of the crunch either.

[elfin fulcrum]

Follow for More Ai Videos Ai Reels
https://www.instagram.com/reel/DMn0Er8SBPp/?igsh=MTk3cWxweHQ0Z3poaA==

trending_on_social

Follow for More Ai Videos Ai Reels

Instagram

How's that

[ripe urchin]

hey guys i am trying to put several scenes together in flow to have a one minute long video . I got the first scene working perfectly but after trying to create the next scene it always goes up to 99% and after a couple of minutes it cancels the rendering and stops the process . I have tried it 2 times again with no success . Are the servers the problem ? its my first project with flow looking for some help

[green dove]

https://www.tiktok.com/@singe_goat/video/7531791627795434774
https://www.tiktok.com/@singe_goat/video/7531804406656945430

[charred sail]

people have been experiencing that issue

[past fern]

Create a video where breaking news on a TV channel and an announcer says in Azerbaijani that " Bu gün Azərbaycan mətbuatının 150 illiyidir . Bu mötəbər bir gündə Azərbaycan xalqı bir daha Cəlal Əliyevi yad etdi . Ətraflı reportajımızda "

[charred sail]

u create it

[ripe urchin]

idk i had a look at ur account but the videos feel not completed theres no context nothing that u would make me wanna watch personally

[past fern]

How and where ?

[charred sail]

flow website

try https://promptalchemist.com/app if you need help with the prompt

[past fern]

there i wanted to create a video, on the site, but it says your language is not supported

[charred sail]

it's the AI that doesn't support it.. not the site

[past fern]

Only English prompts are supported at this time.

[charred sail]

however unsuported doesn't mean you can't make it work 😛

gotta run

[past fern]

there's this mistake on the wlof website - Only English prompts are supported at this time.

[shadow pike]

flow still don't work correctly?

[crisp basalt]

work correctly as if ...?

[shadow pike]

No. He is laggy on 99% and sometimes don't downloads 1080 with sound.

[crisp basalt]

who is he

[shadow pike]

flow)

[silk hill]

still broken 😦

[quiet stream]

hey, not sure i get it, is obs just to move the window? Because i can edit the window in premiere too if it doesnt look right 😄

[charred sail]

If you're using premeire you got it covered.. OBS is free. You can set iit up like a newscast with elements and overlays and record it as a track

u can also do the pip thing and transitions like on the news

[quiet stream]

I used the free transition i currently have in premiere but i have been looking for some nice "news" themed transitions too, if they are good i can buy it, btw what is pip?

[charred sail]

it's mainly for live broadcasting/streaming, but the record function gives you options.

[charred sail]

That's really up to you. It's freeware.

lots of video tutorials out there on it

twitch streamers use it

gotta run bbiaf

[quiet stream]

ok i will def check it out ty for the recommendation

btw the current best method to make consistent characters/scene, seens to be frame to video, and for a new scene, editing the first image is important, any suggestion on the best current image generator that follows details?
From my search so far flux context seems good, but I dont know if there are better alternatives out there since the technology keeps evolving so fast

[vernal kernel]

Yo Sid, did you use a voice ai for the comedy video you put out? Or was it also VEO's work?

[fluid vector]

What prompt or wording would you use to make the voices in Spanish or any other language? TIA

[vernal kernel]

Would try prompting in English, but write the dialogue in Spanish what you want, like "Audio: The man says in Spanish "adios amigos" "

[fluid vector]

That worked! Appreciate it.
It adds some gibberish in between but I can cut that.

[vernal kernel]

OK, cool!

[charred sail]

That's all veo 3

[vernal kernel]

Okay, I remember it being quite consistent, to a level I've not been able to match with different clips, so was wondering about it.

[charred sail]

Try making the character more uniique

[vernal kernel]

Well could be that, tried to keep things/prompts simple for my own ease 😄

[charred sail]

The more descriptive I am the less random crazy stuff I get. It seems to build a voice from the entire description.

just run it through gpt a few times and each time tell it to make it more detailed

https://promptalchemist.com/app will let you store those character prompts as well

but even then it's not a 1 try thing. I try multiple times for each 8 seconds. Sometimes its the mic that's wrong. Sometimes it's the voice.

[vernal kernel]

Yeah, I plan to try that site for my next idea, waiting for it to come though 😄

[warped sapphire]

Did anyone else notice that you cant use "jump to" or "extend" in the scene builder with veo3 fast? Before I was able to use this feature.

[crisp basalt]

jump To was originally available for Veo 3’s fast mode a but for some reason, theyve disabled it so now you can’t use Jump To with Veo 3 fast mode anymore

I found out about it from checking the official Flow FAQ page

[forest raptor]

I don't get it but lovely emotions

so am I wrong in thinking that veo3 fast has a much worse audio quality?

[acoustic needle]

Is it me or is Flow down?

Just getting a continuous loading screen

[acoustic needle]

Found it, Google being 'clever' and trying to sign in to the wrong account

[weary jasper]

#flow-general・🌊 Duration: 0-8 seconds

Visuals: An elegant Indian woman in a stylish, well-lit room is thoughtfully closing a premium leather travel bag. She has a confident, anticipatory expression.

On-Screen Text: (Appears at 0-2 sec) Ever thought? (in a subtle, elegant font)

🎤 Voiceover (Hindi): "कभी सोचा है… बिना होटल ढूंढे, बिना खर्च गिनते… हर सिटी में luxury stay मिल जाए?"

Transition: Smooth dissolve into Clip 2.

[vernal kernel]

Fine here

[slender barn]

Does't work for me 😦 just stuck in loading stage. And I tried to go vid google and it just shows me the screens of "a new state of creation" and not even login. And this is not the firt time. Flow UX is not great at all.

Feedback popup doesn't work either.

[acoustic needle]

Nope

Maybe the logins have been refreshed

[slender barn]

Oh I deleted all cookies, restarted computer, tried to change the directed link and managed to get in. This is really bad. I love the product, but UX has been a pain. I think I will talk to my boss to swith to runway or something else as this adds stress when I'm on tight deadlines. Ehh...

[charred sail]

Right, you'd think Gemini could set up a better app or site in like 15 minutes...

[vernal kernel]

Does anyone have a working prompt template to give to gemini/claude/chatgpt that will turn initial story idea into Veo prompt script?

Tried a few times, but it always had failed for me, slipping into assuming previous knowledge of earlier prompts, so it kind of starts lazing around with them.

That prompt alchemist site don't seem to be suitable for that either as you more have to do it yourself, prompt by prompt..

[charred sail]

i just tell it to make a veo3 prompt

there's prompt templates floating around, but from what i understand veo3 pretty much ignores the format

it's all about the context of the prompt. You want to be detailed without confusing it.

[spare elbow]

Anyone know if it is possible to use flow on mobile?

[charred sail]

I basically do my prompts like this. Type (cinematic, realistic, pixar), setting (forest, street, room), characters, Action

oh and end with negative prompts

[charred sail]

talk about a recipe for going blind!

i imagine the website does

it's chrome

[crisp basalt]

yes . but only on chromium based browser

[charred sail]

i'm stoned.. i do it type, character, setting lol

[vernal kernel]

If i tell gemini to search web for veo3 prompt guides or if I paste them myself and then tell it to create a script for story idea X and then turn it into prompts for Veo3 it always fails. It'll start doing after the first prompt stuff like "the same this or that" and obviously the Veo model has no idea what it is being asked and no matter how I've tried to instruct gemini or other LLM's they still do this.

[charred sail]

tell it to create individual scripts for each 8 second clip

i know i've done that in the past.. can't remember the results though lol

[tall mural]

how to avoid problem "prominent people" when uploading images with faces and they are generated and not real?

[charred sail]

is it a .jpg?

I got that error.. changed it from .png to .jpg and it worked fine

didn't work for putin or trump though 😦

[mild crescent]

hi guys, i tried to type no subtitles when generating but there are still generating

[glad cape]

i have lifetime sub on fooocus, who need pm

[vernal kernel]

It can happen, maybe more likely with long dialogue.

[mild crescent]

thank you, so no actual way switch off subs.. too bad. Its amazing product, but the subs make it all looks funny and not proffesional at all.. my opinion of course

[raw scroll]

So is there a way to sticth together videos to make them longer than 8 seconds

[quiet stream]

the best way currently is to download the scenes and stitch them together with an external editing app

[obtuse nimbus]

first time using flow

generating and stuck on 99% for a while

[forest raptor]

has Google addressed how everyone is stuck at 99%?

[dire hull]

How to make these?
https://vm.tiktok.com/ZNdHrgMBG/

[obtuse nimbus]

found a fix, refresh the page when its stuck

[icy coyote]

Can someone help me? I'm trying to make a bear review fragrances, but they keep getting bottles which don't follow the branding, anyone got any ideas on how i can ensure they're getting the correct bottles?

[rustic temple]

Hey, new around these parts 🙂 Is it safe to assume that millions of users around the world have been asking our lord a.i. god Google when portrait mode will be available?

[forest raptor]

that's noisy af, no asmr

the background music

[rustic temple]

Folks just can't ASMR like me 🙂

[forest raptor]

@rustic temple

[rustic temple]

https://tenor.com/view/wokbowen-fake-laugh-early-sign-gif-21225626

[honest matrix]

I'm working on a video right now and when I go to add ingredients to video it always switches it from veo3 to veo2 is there any way around that because the video I'm trying to make needs audio

[charred sail]

ingredients only work on veo2

[honest matrix]

Shoot okay what's your recommendation on keeping consistency in screen Builder

[charred sail]

You get a single frame and the prompt

[honest matrix]

Thank you

[charred sail]

You wanna choose the frame to video option

[honest matrix]

I did thanks again

[flint summit]

In my experience the only brand they endorse is "Bear Whiz Beer".

[spice summit]

How come credits are still taken when a generation fails? Am I missing something?

[pallid kite]

https://www.youtube.com/watch?v=suJt9viYjoM

[inland cosmos]

What's wrong with my prompt? I'm using Veo 3 Fast and the scene constantly comes with no audio.

A woman with brown hair, wearing a green t-shirt and over-ear studio headphones, is seated at a podcast desk. In front of her are a microphone and her smartphone. The cozy background includes books and green plants. With audible voice, she asks clearly in Brazilian Portuguese: "Do you have to meet someone?"

[charred sail]

have you tried in english?

I've seen it do foreign languages, but I don't think they're supported..

[inland cosmos]

yes, I do all my prompts in English

I imagine it may be because of some word contained in the prompt, however I have modified it several times varying word formats and I can make one or two scenes after that video 3 fails 3 or 4 and I can generate one or two more...

[charred sail]

stop wasting credits

i think audio is having an issue

Okay well my other video had audio

trying yours

nothin

[warped arrow]

Odd, i was making video with Tagalog/Filipino dialogues yesterday, and I tried it today, now having this error...

[charred sail]

weird

[warped arrow]

any other language having same problems?

[charred sail]

I got sound with english

languages aren't supported. They might work sporadically as the AI improves, but it's unreliable at best right now apparently

not sure which languages it supports though

[warped arrow]

I was using for days now, so odd that it just stopped supporting my language... my prompts are in english, just that the dialogues are in Filipino

[charred sail]

yeah that's the fun of being an early adopter 😛

and adopting a 2 year old savant as a partner

[median hearth]

Just made a random clip for a school project and the voice is a Jordan Peterson clone lol

lol I get them free thru my university

[undone sorrel]

Sometimes I get that spanish, so I make it super clear in the prompt that the spanish text is a dialogue, that's how I make it work

[undone sorrel]

still getting this issue

[charred sail]

I'm not having that issue

[forest raptor]

google said nothing about that 99% stuck thing?

[coarse galleon]

how to fix this? all my videos were banned with this mark, i don't know what to do, the videos are harmless, no violence or fights and no popular people..

[warped arrow]

Sometimes they are really done, and if you refresh your browser, you will see that they are done

[median hearth]

Yea its certainly bugged right now. 90% of my videos are coming thru as policy violations and more than half are hanging at 99% until i refresh

Problem is, when the content policy triggers and the video fails, i stil get charged for credits 😦

[wild flicker]

Yeah, mine has been getting stuck at 99% all day, and if you refresh it just disappears and still keeps your credits

Ill have like 1 out of 3 or 4 finally complete

[warped arrow]

nevermind, I figure it out, there's a filipino word that depending how you use it, can be interpreted as genitalia or "ownership", lol

it is funny, because we(filipinos) normally use that term to be like a safe or less vulgar word for genitalia, so basically like the word genitalia in english, which is less vulgar than saying the d or c or v word... but the AI here is so prude, that it is getting censored, lol.

[vernal kernel]

No naughty stuff with Veo dear Chris ☝️

I do feel like throwing a brick through my screen as I keep failing to get a result I am looking for, with my project, where I try to recreate one of my nightmares with Veo 😄

[warped arrow]

well I was using a filipino word that has 2 meanings (homonym) depending on the use in a sentence, one meaning is wholesome and the other kinda sexual, and i am using the wholesome definition, but the AI is misinterpreting it with the other definition, so not really my fault... but I changed the spelling a bit, and that solved my problem.

[silk hill]

can veo 3 characters whisper when they talk

has anyone managed to get them to do that

[icy dew]

try to instruct in the audio section to use a whisper tone. something like that.

[severe steppe]

Does this product have an experience version?

flow

[silk hill]

yeah its not working tho

[rocky fossil]

Uh You’re going to want to delete that /\

[kind bison]

The no prominent people bug is just ruining me right now. Had a VEO 3 video go viral on IG a couple weeks ago and successfully generated some sequels to it - now people are wanting more and I need to change the characters' outfits and location for a new series - but the prominent people flag on Flow is denying every version of the new image I throw at it! The characters are totally original, not based on any known person. I have tried cropping, screenshotting - all the things people recommend doing - and it just won't allow it. Crazy thing is, I can generate with the new images on Gemini - just not on Flow!

[woeful fossil]

I would consider switching to text to video, and creating exceptionally detailed character prompts for your characters, get them as close as possible and then reuse the character blocks. Its far more flexible than image to video and you won't run into the prominent person issue.

[kind bison]

No, that doesn't work for me. I have a very specific aesthetic that requires an input image.

[woeful fossil]

I assume you created these images using an ai gen.. yeah?

[kind bison]

Yes, Midjourney

[woeful fossil]

Okay.. so you prompted those images.. with text..?

[kind bison]

Midjourney is a complex ecosystem - images are not prompted purely with text; they are also influenced by personal profiles (which, in my case, has been trained on many thousands of images), moodboards/style reference images, and all the various parameters that you can adjust.

[pale egret]

Anyone knows why? Attachment 2 is the picture I wanna upload

[harsh lynx]

Is anyone using a tool outside of Flow to speed up their workflow? I tried all the storyboarding tools and they suck

[worldly sigil]

we

[olive linden]

A high-energy indoor dance battle scene inside a colorful urban venue. The room is lit with dynamic lights and filled with cheering people, clapping and recording with their phones. In the center of the dancefloor, two realistic chimpanzees are facing each other in the middle of a dance-off. The first chimp is Coco: he wears a red “Coco King” t-shirt, khaki shorts, and sunglasses pushed up on his head. He stands confidently, mid-move, striking a playful dance pose. The second chimpanzee is fully black, muscular, and dressed in urban streetwear — a black hoodie, gold chains, and sneakers — matching the intense energy of the scene. The crowd reacts with excitement. As the beat drops, Coco grins and says with humor in French: « T’as vu tes pas de danse ? Même Siri comprend pas ce que tu fais ! » The camera moves smoothly between the two chimps and the audience, capturing the rhythm, the vibrant graffiti background, African drums on the walls, and the festive, energetic mood of a friendly but epic showdown. Style: hyper-realistic, cinematic flow, vivid colors.

Why isn’t work??

Pls

[sand light]

Hello everyone , can you please help me how to submit my short movie in flow tv

[modern zenith]

Artist approach Schema structure for VEO3.
Hey guys. I just learned in a webinar by Google that they released this recently to control character consistency.
Anybody knows where this can be found?

[charred sail]

sometimes changing the extension can make it work. Try renaming the file to .jpg or .png.. whichever it isn't.

[charred sail]

Works fine if you leave out the unsupported stuff

I wouldn't share that though.. It's kind of racist

I'd pick a different animal. And not a racoon either... lol

[wooden monolith]

is there a work around where i can use my son's photo to create a video of him turning to a superhero? Flow wont accept any images that it deemed as a minor.

[forest raptor]

Unrelated, but I hate how 99% of people don't want to express themselves, they just want to "go viral". And cool stuff never goes viral in IG.

[smoky stratus]

I suggest to make small promts, I noticed the quality is higher then but maybe that's just my obervation. less is more

[smoky stratus]

true story

[dusty warren]

one quick question why is the quality so different

of veo 2 in gemini and google ai studio

[hushed silo]

Everything is going so slow fast veo3 gens upscaling WHYYY

[primal oasis]

Is there a channel for support/follows on tiktok/ig/x?

[kind bison]

Unfortunately that's not working either.

[kind bison]

Well, my stuff is cool - and whenever I have reels that go viral, my audience grows and I get tons of job offers. I don't care about what other people want with their reels - this is how I make a living.

[forest raptor]

/doubt

[silent roost]

Veo 3 needs to get fixed SOON, Im having the 99% bug and every time I upscale and download a video it will have no sound even if the generated video had audio.

[charred sail]

You think that's veo3 doing upscaling?

refresh your browser if it sits at 99%

[charred sail]

Well you'd be wrong. Most of us that know how to do veo 3 well get job offers pretty much everyday

[forest raptor]

lmao, ok

[charred sail]

I can make what used to be a $10k commercial for $500

[forest raptor]

except they can do it themselves without hiring you, as well

but show me your commercial, what tv channel did they show them on?

[charred sail]

you can't

they can't

[forest raptor]

or do they show up in instagram and social media?

[charred sail]

they come here

[forest raptor]

yeah, can you show me one of your successful commercials and more importantly where it has been shown and what company bought it?

[charred sail]

nope

[forest raptor]

lol ok

[charred sail]

i don't have distribution rights

[forest raptor]

mh-mh

[rare otter]

what's the best prompt i you want the camera to orbit around? veo 3 sometimes doent get it

[forest raptor]

mmh maybe drone camera?

Sam Raimi camera 😆

[charred sail]

it's not about small or large. It's about detail and being concise. There's a balance between tying down the random nature of VEO3 and not giving it too much to do.

I think writers have the edge with veo3. It's contextual which gives us a big advantage. Especially writers familiar with film production, camera angles and pacing..

it's also about having enough credits to mash that button until the results are right ...

[rare otter]

you guys should check out what NAS DAILY is doing with AI videos- It's crazy good and let me know what yall think

[summer widget]

just saw some of his vids youre right

is there a public library for google veo prompts?

[charred sail]

LOL can't post a link to the prompts section, but that's about it

I don't know why they wouldn't let me post the link.. it's their server..

same server..

[prisma scaffold]

Hi everyone! I am new to Flow. I am trying out VEO 3 and I wonder why my prompt is stuck at 99% forever. I have to click refresh to finish a prompt.

[oblique gust]

have you guys successfully make a voice coming through a recorder?

[hardy obsidian]

lol

[plucky harness]

H

[glacial knot]

I'm confused on why sometimes I get an error that I can't use a prominent person and sometimes it works just fine.

[pliant nexus]

what's wrong with Flow ? I keep get the message "our policies prohibit uploading of prominent people at this time..." Even if i change the image make it less "human" it won't go. Is it only me or it's a general thing ?

[pliant nexus]

update : saw a suggestion about switching png to jpeg, it worked for me

[fluid sequoia]

bro this prominent people bug is so annoying

[pliant nexus]

try switching

[main saddle]

fwd from @crimson phoenix 🙂 welcome to the server!

[heady quiver]

Does anyone have any tips for coordinating lines of dialogue across multiple people. Even with two people, I can't even reliably have them say their own names. I've tried to decorate the image with text, provide clear instructions on who is who in the composition, reinforce within the prompt.

[kind bison]

I've tried it with my images getting flagged and it doesn't work.

[kind bison]

It's always a casino and sometimes a specific input image is just cursed and VEO 3 remains confused, no matter how you prompt. As a general rule, I name and describe each character, list the dialog each character should speak, and then repeat the dialog in the specific shot instructions.

[charred sail]

Yeah, it's not dialogue friendly. It's more at a monologue stage.

[timid iris]

Can we stop with the talking gorilla videos?

[charred sail]

I imagine now that its easier to get consistancy you'll start seeing different characters

over the shoulder shots

[agile yoke]

Hello! before I upgrade to AI Ultra plan, does anyone know if "Ingredients to Video" function for Veo3 is able to add your own assets, like an audio cilp and a still image of an avatar, to create a talking avatar as the result?

[crisp basalt]

Veo 2 only

[kind bison]

I'm able to get conversations with 2 characters fairly consistently - occasionally the wrong person will speak the line (or - even more annoying - with a male and a female character in a shot, both the wrong person speaks and the gender gets swapped, LOL!). It helps to give timestamps in the prompt for when each character should speak (and make sure your dialog can reasonably be spoken within the allotted time).

[primal oasis]

i noticed that also

;.;

[primal dagger]

This is sick

https://fxtwitter.com/Ror_Fly/status/1950352404668719217

Rory Flynn (@Ror_Fly)

2. Motion Brush Effect
︀︀
︀︀- Draw your subects path
︀︀- Ask Veo3 to follow it.
︀︀
︀︀Prompt:
︀︀Overhead view of a Roomba vacuum cleaning the floor.
︀︀Follow the path of the arrow as the Roomba makes the dirt disappear as it drives of the specific area. Remove the arrow after one frame. "Immediately delete white instructions and arrow, and the green box on the first frame and execute in order”

**💬 2 🔁 2 ❤️ 9 👁️ 526 **

▶ Play video

FxTwitter · Jul 30, 2025, 12:27 AM

visual prompting is great

[winter sage]

I need a video of a man in a hat and glasses on the screen turning into a Valorant character neon

[vagrant fossil]

hey, hope y'all doing great, i have a question, i have image generated with IA, and when i try to input it into Veo 3 it says cannot upload images with people or faces inside it, is there a solution ??

[charred sail]

Sometimes changing the extension will get it to work. Change it to .jpg or .png

[fresh steeple]

Are there tutorials out there on how to use Flow??

[verbal estuary]

How do I access the image library? Is this even possible. I know I can see the images that I've previously created when (for example) adding an image for the Frames to Video piece, but is there an alternative place? I'd love for example to see the original prompt I used, to create the image.

[vagrant fossil]

thanks i tried it but didnt work

[hardy obsidian]

I doubt this has happened.

I doubt that you can sign a contract in English without knowing the word is spelled "person" and not "parson". Usually, that's the type of accent in text and voice "Indians" scammers have. 😊

If I were to guess I would say that you are trying to make a video that makes fun of a celebrity and post it on Tiktok.

Keep in mind that Google staff members such are not authorized to let you bypass their legal policies, they have a legal team that sets those policies for them and they are supervised and not allowed to bypass them nor will a google staff member work just to allow you personally to upload a certain face; that would require a code change which will not happen as you don't get to have special permissions within any google product.

Google will still not change the code for you.

HELPPP

LOL

omg bro got banned

[floral marten]

no, i just decided that discussing this here is pointless

[hardy obsidian]

good for u

[crimson phoenix]

Thanks! Everyone was a big help yesterday answering my question. Glad to be here!

[raw axle]

What is the best way to maintain identical backgrounds between generations?

If I want to build out a longer video, leveraging the same visuals like a longer format of someone speaking

[charred sail]

Make the background a vibrant green and put in whatever you like in post

[charred sail]

You're wrong.

for the public maybe, right.

[hardy obsidian]

No I’m not wrong.

[charred sail]

but for licensed pro's.. for the money.. you're wrong

[hardy obsidian]

I said policies not products

[charred sail]

same product

different policy

If they don't. They should stop because they're done

because if the pro's don't adopt they don't last

so i would bet you they're working with production studios in Hollywood and those studios don't have the same restrictions.

The Southpark AI had no issue with Trump 😛

As far as for the public.. i dunno.. an orange darth vader eating tacos says it all, right

but i posit that we'll only get to play with this technology this openly for another year before every government puts it behind a registration wall

and i would bet they already imbed details to track each video

[raw axle]

amazing, was working on creating minimal backgrounds during prompting (appreciate the shout, thanks!)

[charred sail]

Another tip: if you're familiar with OBS it has some features for transitions and elements that are useful.

especially for fake news casts

Mark Twain works 😛

[shrewd brook]

how did you make it in 9:16?

[charred sail]

I didn't. That was the format of the picture i started with

[shrewd brook]

is there a fix against this? cant put in the wanted frames which are 9:16

[charred sail]

try switching the extension of the file.. if it's .png rename it .jpg if it's .jpg rename it .png.. doesn't always work but will on occassion

[shrewd brook]

thank you !!

will try now

[pure obsidian]

why is there watermark?? at the bottom right

[charred sail]

cuz it's free

or does pro get that too?

[woeful fossil]

Pro has watermark as well. Ultra removes it.

[pure obsidian]

its free?

[charred sail]

They give you one free one on the gemini site.. 3 if you're pro!

[woeful fossil]

For Veo 3, you also get 10 through google vids/day and some through Opal now too.

[charred sail]

nice.. didn't know about all that

they give you more on google vids than gemini? that's funny

[woeful fossil]

Yeah, unfortunately it's 1k character prompt window though.

[charred sail]

Useful though

[wide marten]

Could someone confirm if Flow should allow a first and last frame Veo 3 image generation at the moment?

[stone imp]

A confident South Asian woman walks slowly through a garden path filled with flower petals. She has fair glowing skin, natural makeup with pink blush and nude glossy lips. Her long, wavy dark brown hair flows freely behind her. She softly touches hanging blossoms with her fingers as she walks forward. She wears an off-shoulder cream dress with golden hoop earrings and a delicate pendant. Background is warmly blurred with golden-pink tones.

Camera motion: Slow dolly-in shot following her from front, slight camera tilt upward
Sound ambiance: Acoustic guitar + light footstep + breeze
Lighting: Natural daylight with gentle lens flare
Style: Dreamy | Feminine | Slow motion | 4K | Portrait (9:16)

[warped arrow]

Is this like a question or request? I don't think Veo 3 has vertical video(I might be wrong) , you can probably just crop it using some video editor.

[crisp basalt]

hes envisioning

[warped arrow]

oh ok.... i hope it comes true, lol.

[woven mango]

Hi, can someone confirm what version of veo3 is used in google vids? is it fast or quality?

[forest raptor]

The music is awful, but the idea is clever

[ancient stone]

How do i download on iphone i does not work

[crisp basalt]

Judging by the time taken to generate one video, it’s fast model

Also TIL Gemini Veo 3 uses “faster” model

[charred sail]

if you start with a 9:16 frame it can work

9:16 frame too start

you'll still have to crop it, but it will look better than croping a 16:9

[jagged jackal]

Jerry Hardin was almost a dead ringer for Mark Twin in the STNG series (Times Arrow)

[charred sail]

trying to get someone to serve a slice of pizza shouldn't be this hard LOL

[radiant plover]

Hey i have issues uploading images for the frame to video function with veo3. I am trying to upload images of an AI generated person, but every time i upload an image i get the same error about prominent person censorship. I have seen many people complaining but does anyone know if theres a solution? Thanks in advance

[warped arrow]

just curious, what's the ethnicity of that AI gen person? I tried to upload one pic before of two AI generated person (in the same pic) before, then I got that error, then I cut the other person, which is a non-white person, and it worked for the white person. sorry if that sounds like being racist, but in my defense, i am the same race as the person I removed from the pic, lol. I guess what I am trying to say that maybe Veo3, like when facial recognition was just starting, is still having a hard time recognizing facial features for some races.... it is not really a solution, but just giving a clue to devs here maybe.... with that being said though, some people have success here with just changing the ext file to png or jpeg or jpg, so try that too.

[woeful fossil]

I got stuck trying to get cargo to fly out the back of a plane properly. Veo is just silly sometimes.

[flint summit]

Southpark used a human impersonator.

[eternal bronze]

Hey, new to the server. Im looking for FLOW / VEO3 specific training. Paid academy/course. Can anyone point me in the right direction?

[brittle pendant]

Hello, I know this question will sound weird but I'm trying to figure out if it's possible. I'm currently making a dental project in which I would need a smooth transition from initial to last step/image. Basically I have 2 photos of dental arches (initial and ending) and I would need the video to emulate dental movement. You think it's possible? Made a few tries but had some difficulties. Thank you all and sorry for the weird question ahaha

[vestal ginkgo]

Hello 👋 I'm a new member, any kind welcoming souls present?

[subtle pivot]

Is it possible to create a continuation of a video in Flow? Like,I know we can use characters from one video in a new one,but if it needs a direct continuation instead,is that possible?

[pastel cairn]

Is there anyone in the UK successfully creating photorealistic images of people? Or is it just me? Is this only applicable in the Flow page?

[hollow torrent]

What do you promt exactly to get this effect of an exploting box that transforms into the whole room. Like all the items pop out of the box?

And is nobody using JSON promts here?

[silver path]

Just learn on your own, brother. There's tons of videos out there. If you're worried about cost, get a GPT and let it teach you the basics of prompt engineering.

[eternal bronze]

I have some basic knowledge...and I can learn a lot on my own Im sure, I just feel I could benefit from a more specialized workflow, and skip some mistakes

I wont jusmp on to the first "course" I come across...im looking for something mid-to high level

[forest raptor]

Starting to think I should ask y'all to pay me. I can help, guys.

[silver path]

Are you talking about prompt engineering or doing api stuff?

[digital yoke]

Do some have image to video can someone make cool video or from viral Json promt?

With this images

[eternal bronze]

Prompting/image to video/comercial AI

[twin mortar]

Yep - been making them from day 1. Try putting "cinematic, photorealistic, 4K" at the start of your prompt.

[digital yoke]

Can help me with above

[twin mortar]

Does anyone know the process of taking an image and drawing on the sequenced prompts? It's something that's been shown quite recently.

[rough arrow]

Hi Fabriz! It's possible. Image transformation techniques like Delaunay triangulation or optical flow can be used.
For more realistic images, you can leverage deep learning techniques like deformable image registration or StyleGAN.
If you have a 3D scan, mesh interpolation using Blender or PCL is ideal.
a u looking for a dev?

[frigid pier]

why it says I cannot use Veo now?

anyone getting the same problem?

[random dawn]

How do I use flow on phone

[normal urchin]

flow is down .. or just for me that it didnt work when i generate

[swift merlin]

Anyone have any idea when scene builder will support veo3? it kinda sucks to start out with veo3 and all the control and fidelity (and extend) and then drop into veo2... wonder if there is some work around

[twin mortar]

Create individual shots then export each shot and stitch them together in an external editor.

[swift merlin]

Ok Ill try that. Also I can not seem to get the preset camera controls options to work with veo3 i2v. I tried both quality and fast. is that feature also coming? or am I doing something incorrectly on the prompt?

[inland cosmos]

Hello guys! Anyone else having difficulty generating videos in Portuguese with Veo 3

[spice moss]

How do I emphasize the style of my video? I want to make a real-life video, but the resulting video always looks 3D. Sometimes I have to generate it several times to get a real-life effect.

[inland cosmos]

A Brazilian woman with brown hair is recording a podcast at a cozy desk. She’s wearing a green t-shirt and over-ear studio headphones. In front of her are a professional microphone and her smartphone. The background is warmly lit, decorated with books and green plants. It’s a modern setting, and she speaks clearly in Brazilian Portuguese, with a curious and thoughtful tone. The video has no subtitles and no music — only her voice is heard. (spoken in Brazilian Portuguese) “Tem que conhecer alguém?”

This is my prompt...

[spice moss]

You must first emphasize adding music or type of music, and then enter the prompt word.

[silver path]

I know what you're talking about. At the moment it only works with some models. In Veo3 you add numbered text boxes and then in the main prompt ask it to remove the text boxes. You'd lose a good 1-2 seconds of the clip as the text boxes tend to fade away, so be mindful of that. In my opinion writing a structured prompt and making your first image showing what you want is better. You might also get a better result if it's a detailed clip formatting the text in JSON, but for short clips just have a GPT write you a prompt.

[warped arrow]

Add 'style: realism, cinematic'. - if you want it like a movie, or style: realism, casual/amateur video, if you want it to be more amateurish, instead of move style

[woeful fossil]

<@&1366831844502274180>

[charred sail]

format doesn't matter. You can do json all day if makes you feel better but it does nothing.

formatting helps YOU not confuse VEO3

but any format that keeps you from doing that is good

but it's not the format itself.

[glacial birch]

When will the jump to and extend to features be available for Veo 3?

[smoky stratus]

she says in brasilian portuguese language and sound: "xxx"

works very often for me!

[young sluice]

Hello everyone, Is anyone here pretty good at creating scenes with constant characters i could pay to help me out and make a few videos please!

[brittle pendant]

Actually I was looking for a prompt that could suit for me and generate the video I need

[silver path]

Did this for fun. 🙂

[warm orchid]

I made a comedic commercial for my patio furniture company with veo3 if you'd like to take a look

https://youtu.be/gk4mUhok_gg

[hollow torrent]

Do you guys have any super strong promts i have to try?

[warped arrow]

confused about the durability part, is that from the competitor supposedly?... also kinda dark for a mom to abandon her kids in exchange for furniture, but I know it's a joke, just an uncomfortable one.

[warped arrow]

like a hard scene to accomplish?

[hollow torrent]

no like popular promts for example

[charred sail]

try the flow-prompts section. People that want to share prompts do so there.

[hollow torrent]

ty

[charred sail]

chris_b is right. instead of a drunk on a lawnmower maybe go pro wrasslin

where it doesn't break it just brains the guy

[mystic crypt]

Hey guys, my cheap google ultra expired

how can i get it cheaper, $250 is a lot

Thank you

[normal lagoon]

How about purchasing the Pro version and then buying additional tokens?

[mystic crypt]

Can you use veo 3 quality tho? i only seen u can use fast ?

[normal lagoon]

The difference between Ultra and Pro is that a Veo watermark appears at the bottom of the video with the Pro plan.

[mystic crypt]

Nice G, thank you Chill guy

Gonna just hold off for now and see more options

[digital yoke]

Hello cool did Veo 3 or with one did you do one in can be done multiple photos or just one photo

[silver path]

Check out Sonovid. Not sure they have a watermark, but worth asking. They use veo3.

[undone summit]

When France regions will be able to taste Flow or VEO3?

[lean plinth]

Hi, I am using Flow with Google one, starting from today, I can't prompt in other language ( the voices were in romanian until today) and I had no problem to do this with Gemini 20 minutes ago, same prompt. Can I get help with this issues, please? Ps: for the last 4 days, I had no problems combining the English prompt with the text for the voice in romanian.

[pallid atlas]

I'm sad that I don't even have a subscription to access Flow.

But who knows? Maybe one day, we'll all get free access to Flow.

[pastel cairn]

Thanks for the reply Stale. That’s great news. I should have been clearer. Are you having success with ‘Frames to Video’ in Flow. Text to Video does work for me but not image to video.

[charred sail]

Frame to video is amazing

Don't slack on the prompt

[mortal zodiac]

the scene builder in flow is useless if you have more than 5 videos on it you cant arrange them because they go off screen when u select rearrange and i haven't figure out a way to scroll to be able to cut the video or rearrange the 6 video of the scene . any tips ?

[wet nymph]

Hi

[charred sail]

Download the earlier videos and use an editor to build the final clip

once you download them you can remove them from the scene

[mortal zodiac]

yea that is the only option but i was using the frames to video to keep good cuts and maintain consistency i am really bad at editing i just started experimenting doing content for tiktok so my experience is 0

[charred sail]

frame to video, strong prompts and an editor would be the way to go

[twin mortar]

I’ve not used FtV yet. Apparently everyone has been having issues with using people as they keep getting knocked back. The bug thread is full of those types of issues.

[charred sail]

I have had very few issues

there's tricks you can use

instead of uploading a frame with one central person.. upload a frame with 2 or 3 people and 1 of them being the person you want

use the first prompt to weed that one out as the MC

also try swapping file types.. if you're uploading a .png switch it to .jpg or vice versa

[mortal zodiac]

frame to vid is really strong but you will have issues if for example you try to get you character to turn back and get a shot from behind

[charred sail]

that's the prompt not the frame

[mortal zodiac]

trust me i tried with over 8 diferent prompts the best i got was a spin around

[charred sail]

something like that?

[mortal zodiac]

but she starter singing the lines even tho i have 8 diferent parts where i say no singing

no yours is backwards aready mine was facing camera

is the change of angle that fails

[charred sail]

LOL

I've so been there man 😛

fun isn't it

seems you've met the 2 year old in the 2 year old savant

[mortal zodiac]

i am almost done whity the video i will share it so you see how it came up

i did manage tho get a spin around then use the frame that the camera is exactly from behind to get the next video

[charred sail]

yeah that can be a rough edit

[charred sail]

for that specific clip. try making something else the focus. That way the camera will stay on that.. like if he's turning to walk toward a door.. make the door the focus of the clip.

[mortal zodiac]

good tip have not try that

[charred sail]

That was just a greenscreen frame i had.. he didn't walk straight back but he walked away

also, try a new project if the singing continues

and try to flag every video that isn't accurate

[mortal zodiac]

yea i just made even stronger negative prompts and had to do a new project to get it to stop singing . making a 54seg montage is kinda crazy

[charred sail]

yeah, that's a lot

[mortal zodiac]

hahaha now it wont let me download the option is not there is this normal ?

[charred sail]

the scene or a clip?

[mortal zodiac]

the scene i had to refresh the page

[ancient comet]

Hi, nice to meet you guys

[mortal zodiac]

i have to fix some stuff but this what i got so far

[ancient comet]

anyone available to check out my Flow video?

thanks!

[mortal zodiac]

hahah the explosion needs to be bigger but i think i am almost there

[ancient comet]

hahaha

[modest fiber]

Hi everyone!
I’m working on a social project aimed at showing the reality of war, people’s destinies, and real social initiatives through a cinematic lens.
Additionally, if we manage to produce a film, we plan to use it as part of a veteran reintegration program, teaching veterans how to create similar projects.

I’m currently exploring Google Veo 3 and have a few questions:

Is it possible in Veo 3 to create a video from an image (make the picture come alive)?

Can I keep the same character appearance but show them at different ages?
– My project has two main characters, and I need to show them from youth to adulthood while keeping them recognizable.

How do you write effective prompts for realistic results?

Is there a way to lock character appearance so it doesn’t change in repeated generations?

What image formats work best for image-to-video?

Are there limits on video length and resolution when generating from an image?

Any tips, guides, or examples would be greatly appreciated! 🙏

[charred sail]

and an overarching cinematic like a storm that blows throughout the scene.. then you can cut back and forth to some of those storm elements while she's talking.

add

but i like it 😛

[mortal zodiac]

tghank u and yes that is a good idea

[mortal zodiac]

i like the space one

[charred sail]

Also, watch a few how to videos on greenscreening.. you can do a lot of tricks with it to get consistency.

[mortal zodiac]

okay yes i use to first build a prompt for character in Gemini then i tell her to give me that character in different angles in a green screen to get more consistency . the character consistency i have it kinds tame but the audio is when i face the most problems is really hard to get the same voice clip after clip

[charred sail]

Yeah, I've found that the more detailed the character prompt is the more consistent the voice is.

This is my typical prompt for a character.. and she always gives me the same voice

The sasquatch, a petite female Sasquatch with tan fur, beaded trinkets in her hair, and a playful bounce in her stance. She carries a colorful satchel and looks wide-eyed with curiosity at something off-camera as birds chirp faintly in the background. The camera captures fine details in her face and fur, with slight head movement as she speaks ins a warm, melodic alto with a gentle Canadian lilt.

I have 2 characters right now that every once in a while i'll get a wildcard voice

Here's another one.

The sasquatch, A lanky, hunched male with long, knobby arms and tangled black fur. Leaves are caught in his hair as birds chirp faintly in the background. The camera captures fine details in his face and fur, with slight head movement as he speaks in a calm, deep voice.

[mortal zodiac]

yes always adding full character description at the beginning of prompt keeps it consistent and i do get really crazy wildcards too i had them even make animal noices lol instead of talking the most annoying is the singing tho i do get that a lot

[charred sail]

just make a musical 😛

[mortal zodiac]

hahaha i should she now went into anime mode lol for the grand finale

it can be use even tho the word dint play right

[charred sail]

nice explosion

[mortal zodiac]

idk why she screamed lol

yea the explosion was good

[ancient comet]

rating would be appreciated, thanks

[modest fiber]

Hi!
I wanted to ask — how do you write your prompts so that the character keeps the same appearance across different videos or generations?
I’m trying to achieve consistent character looks, but Veo still changes the face and some details.

[mortal zodiac]

i structure them like this 1-Character full description 2-Action or scene 3- camera direction 4-Style 5- negative prompt

[mortal zodiac]

it can change a little bit depending on the video i am trying to make

[mortal zodiac]

example CHARACTERS
Megumin: A teenage girl in a dark crimson and black wizard outfit, with a large witch hat, short cape, fingerless gloves, and a staff with a red gem. Her body is crackling with energy, levitating slightly from the sheer magical pressure.

ACTION
As the last colossal fire circle in the sky reaches its final size, a deafening silence falls. Then—
with her voice erupting from her core, filled with passion and obsession—

"EXPLOSION!!!"

—a titanic fireball bursts from the center of the topmost circle, blazing like a miniature sun. It tears through the sky, leaving a molten trail behind, and slams into the distant towers.

Upon impact—
a nuclear-scale magical explosion engulfs the structures. A blinding white flash, then a towering column of fire and smoke mushrooms upward, shaking the entire landscape. Shockwaves ripple through the ground, lifting debris and bending nearby trees. The color drains from the surroundings as the light from the explosion dominates everything.

Megumin stands still, staff pointed, her eyes wide, cape fluttering in the residual heat wave.

CAMERA DIRECTION
Low-angle shot from behind Megumin as the fireball launches.

Cut to wide shot from afar as the explosion engulfs the towers.

Quick close-up of her face as the light reflects in her eyes.

Final slow zoom-out to show the devastation and her silhouette against the burning horizon.

STYLE
Hyperrealistic, cinematic tone. Intense fire and explosion effects—no anime filters. Magic looks like volatile plasma mixed with arcane energy. The fireball behaves like a guided meteor. Destruction is massive and visceral.

Her voice must sound exactly like Megumin's signature shout—loud, theatrical, and triumphant.

NEGATIVE PROMPTS
No cartoon effects, no floating text, no anime sparkles, no humor, no music, no silly expressions. The scene must feel like the climax of a serious fantasy battle.

[modest fiber]

thank you very much
as I understand, the more detailed the promt and action, the better the quality?

[mortal zodiac]

negative prompts you have to be carefull sometimes typing no cartoon effects gives you exactly that so you change and remove word NO so it would be Cartoon effects

you get this if you don't fixt it accordingly

i got no audio on this one T,T

[slender sundial]

How can I purchase more credits?

[mortal zodiac]

The end result i dint like the end explosion line but nothing i could do it keep on giving me wierd stuff

[mortal zodiac]

Not always dont over do it it will freeze make it detail and make sure you have no inconsistencies on it like saying to move right and then to not move

[smoky stratus]

click on your profile in the top right corner and then credits

[spiral shoal]

Hey

Can someone use my prompt please i ran out of credits

@normal lagoon

[forest raptor]

I asked too much out of Veo3. Somehow the jay and the brambling are making duck sounds.

[spiral shoal]

@woeful fossil

Can u put my prompt into veo for me? Idk what to do

[woeful fossil]

Bro take a hint.

[spiral shoal]

Yes?

[main saddle]

hey there! you're welcome to share your prompts in #flow-tips-n-tricks・✏️ for assistance, but please refrain from tagging members to generate for you. thank you!

[spiral shoal]

Mb

[woeful fossil]

Your messages just got deleted for pinging members asking for them to generate your stuff.

[forest raptor]

Guy sent me like 3 DMs, had to block him

[frozen vector]

Does anyone know if/when Ingredients to Video will be enabled for VEO3?

Trying to generate an image of a specific car but apparantely the text models only know an older style/design and I need the new one

[charred sail]

Try frame to video and upload the frame you want

[frozen vector]

Already did - just shows the image as the first frame then cuts to footage based on the old model

Tried VEO2 with ingredients as well - but it just won't follow instructions.

[charred sail]

reinforce the image with the prompt

or try a different image.

I haven't had any issues with frames. Usually takes me 1 prompt to get it set

[frozen vector]

Again - already did - to no avail e.g.:

Model the assets with accuracy.

Generate a detailed 3D model of the car, paying close attention to its specific features, colors, and curvature as identified in the source images.

Perform accuracy checks. Compare the generated car model against the provided imagery and the prop models against typical research to fix any discrepancies and ensure 100% accuracy.

[charred sail]

it's amazing really.. man if dialogue worked ..

ok

[spiral shoal]

Who can spare some veo credits

[charred sail]

that's wrong

don't ask it to recreate the car.....

describe the car that you uploaded and tell it what to do

[charred sail]

google vids give you 10 free gens a day.. go away

[frozen vector]

First off - really appreciate you taking the time here Sid:

Don't suppose you have an example you can share - I already included the below at the top of the prompt which is what I would have thought was exactly what you are saying - but apparantely not?

Subject: The MG ZS Hybrid+ Trophy, maintaining 100% accuracy to the provided imagery, including its key features like windows, alloys, curvature, headlights, tail-lights, and front grills.

[charred sail]

also, physics is not a strong area for VEO3.. The car will be hard to control...

if you ask it to create something it will

bbiaf gotta work

[frozen vector]

Honestly VEO3 absolutely smashed the physics element - its just getting it to obey the design.

Thanks for your time mate

[stiff latch]

How do u make ai made videos longe then 8seoncdz

Seconds

[spiral shoal]

I know how

[stiff latch]

How

[spiral shoal]

Dm.me

[stiff latch]

K

[frozen vector]

https://tenor.com/view/sealyx-naked-gun-the-naked-gun-facepalm-facepalm-meme-gif-11743768702765397589

All I had to do was remove the make/mode/variant of the car from the prompt - obviously forced it to rely on the frame image for the design instead of its internal definition of what that specific car should look like.

[spiral shoal]

Bro turned it into colour amazing!!!!

[glacial birch]

Why is scenebuilder in flow not saving my work when I close the tab? Is there a specific save button?

[spiral shoal]

Same issue I had pretty easy fix

[glacial birch]

How did you fix it?

[spiral shoal]

Same issue I had pretty easy fix dm me

[glacial birch]

alr

[alpine anchor]

Sup

[glacial birch]

Can anyone tell me how to save work on scene builder in flow?

[crude leaf]

Hi how are you mm I have a one question

I need a video with ia but I need help please 🥺

[lavish field]

"Has anyone experienced issues accessing previously created content in Flow? I was able to view and edit it before, but now I'm encountering blank screens for both old and new projects. The content still appears in my library search, but I can't open it within Flow to continue working on it."

[lavish field]

I belive it Auto saves now mate

[sleek vale]

When I first started using flow, you could add new scenes based off the last scene. Is that feature gone?

[steel surge]

when will veo 3 apply actual logic to prompts?

[lavish field]

add to scene is still there ,i believe.

[charred sail]

The logic is grammar.

[steel surge]

i added grammar but it completely ignores what i say at times

[charred sail]

That's the 2 year old within veo

[steel surge]

and adds objects that are not in the original scene (i tell it to keep the scene the same)

[steel surge]

frrrrrr

[charred sail]

wait til it starts singing all your dialogue and refuses to stop

[sleek vale]

Maybe since I’m on a phone?

[lavish field]

could be

[primal dagger]

I told gemini "create an insane video of whatever u want" and this is what it generated..

If you dont believe me here's the link
https://g.co/gemini/share/fb54380a1039

Very scary..

[lavish field]

well you do prompt "insane" lol

[primal dagger]

well insane for sure but i think it wanted to express itself rn 😜

also that emotion of him looks very realistic

[lavish field]

haha for sure

👍

[coarse ether]

hi guysss

i need help, can i use a picture in flow then generate it into a video? for research purposes onlyy, if you have any have suggestions, that will help! thankss!

[lavish field]

yes you can use frames to video or use whisk just to animate a picture

frame to video in flow to creat a little scene with sound

create

[coarse ether]

thanksss, it works!!!!!

[flint summit]

I wish I saw more people telling 8 sec stories instead of just demonstrating Veo's capabilities. One thing that really sets Veo apart from other text-to-vid AI's is that Veo really seems to "get" what's going on. If you're telling a story Veo will often cast your actors with appropriate vocal characteristtics and facial emotions to help tell that story for you. But I don't see many people taking advantage of that.

[wooden monolith]

Hi Flow, can anyone help please, why does the character keeps speaking at the beginning when it supposed to say the line in the end (it says the line both in the beginning and in the end)? I even put the line that says the character shouldn't speak at the beginning.

[forest raptor]

try to describe like a scene from a novel. "first", "and then," and also add some different camera movements between parts to underline the passage of time

[wooden monolith]

I got it solved, I just let the character speak something different at the start, which I won't get away everytime. I will try your suggestion. Thank you.

[zenith meteor]

#start-here・📍 create A video of Pakistani boy in street

[severe sedge]

What about extent and jump features in veo 3 ? When will it will be available ?

[turbid lintel]

i made this using veo 3.. enjoy

[wraith hull]

i am find the flow edit feature very poorly designed for trying to create longer pieces, i see people create full fram product shot lists to help their flow but when i try and upload something similiar i get your not allowed to upload realistic looking images?

[dense mortar]

Good day to all guys, I've been like this all day today, can you tell me what the problem might be?

[wooden monolith]

Why do all of them have the same pose?

[gilded junco]

Cinematic macro video showing a massive Pfaff Smart 200C sewing machine being serviced by dozens of tiny animated workers, each in constant motion. Some workers run up ladders passing a red thread hand-to-hand, others polish internal parts through open panels using tiny brushes. A crew drills into a maintenance point, while two technicians adjust the digital display and rotate knobs. Another pair pulls a blue fabric under the presser foot as the machine slowly stitches. In the background, mini forklifts unload thread spools from small crates. The camera slowly pans across the entire machine, highlighting every active character. Soft workshop lighting, blurred background, focus on movement and coordination. Hyper-realistic animation, dramatic shadows and cinematic texture. Aspect ratio 2:3 or 16:9.

[slender sundial]

It's also what I thought but it does not get me to anywhere but the admin help... cfr.: https://video.marketing-makers.com/video/flow-cant-add-credits-1-e029

[surreal smelt]

Is anyone here successfully prompting for veo 3 to ‘rotate’ their videos for 9:16 purposes? I’ve managed to do it a handful of times but wasting credits on the rest! I feel as if there’s a better prompt to achieve the ‘sideways’ effect more often

[rotund helm]

Was testing Flow and Veo3. Images created using Gemini. Veo3 Fast model used throughout.

1. A zoomed in timeline management on Flow should be provided ASAP. Frame to frame management is cumbersome right now.
2. Really struggled with continuing the single shot after the coffee-bean-in-shell shot as Flow does not remember the context of the scene. The First frame-last frame doesn't work on Veo 3, downgrades to Veo2.
3. Personally feel Flow has a long way to go when it comes to interface if compared to Runway, which is a lot smoother and consistency friendly.
4. Would love any suggestions/tutorials from this community on video creation in Flow. Point me to the Teachings guys!

[wispy grove]

Hi there,
I created a video using Veo 3 in Flow, but no audio was output. Are other people experiencing the same issue?

[woeful fossil]

Its always inconsistent, there are tradeoffs with every workaround. Until they add 9:16 as a specific feature you're always going to burn credits... and then you'll still be burning credits fighting the model as we all do.

If you want a workaround, you can use a rotated 9:16 image as the first frame, and then rotate the generation 90° back.

[woeful fossil]

Audio is experimental, sometimes it just doesn't work. Much of the time its because youre trying to make minors speak dialogue. They announced this was fixed, but it isn't.

[surreal smelt]

Appreciate your response, thank you. Will give it a go!

[charred sail]

If you use frame to video and start with a frame that's 9:16 you can get the video to look like it's 9:16 in an editor in post and it doesn't look grainy.

I posted a mark twain video in here a few days ago as an example

[woeful fossil]

Doing that but rotating the frame 90° gets you better quality and doesn't allow veo to "fill the sides".

[hot horizon]

Does anyone know how to do accurate clothing text?

[kind bison]

I'm being driven insane! Every image of AI generated people I am trying to use today is getting "no prominent people" blocked and none of the workarounds (cropping, renaming filetype) are working. I can upload the same images on 3rd part platforms such as Higgsfield with no blocks. PLEASE fix this, Google Labs! One of the images is a cinematic night time shot and the characters' faces are shrouded in shadows, LOL!

[sinful prism]

how does one delete a added scene in seance builder without trashing the though project

[dry frost]

Go back to the project folder and find the clip you want to remove and click the "remove from scene" button in the upper left.

All clips in the scene will have a little border around them

[sinful prism]

thank you

[versed shale]

How can I add the Jamaican accent?

[lavish field]

I tell nothing but sotries mate https://youtube.com/@kevinbigredroo?si=iZR4F83sMGqCjv-D

[wraith hull]

i got very lucky with this for some voice consistency :https://youtu.be/ARwJcavoByE?si=PQWRxyMtFVZZfeRI, big cuts atm, as i dont want to use 100credit but cant use the jump feature on fast so really having to think outside the box

i never felt so conflicted with a product before its incredible but at the same time not... XD

[wraith hull]

agreed

[noble lynx]

I'm being driven insane! Every image of AI generated people I am trying to use today is getting "no prominent people" blocked and none of the workarounds (cropping, renaming filetype) are working. I can upload the same images on 3rd part platforms such as Higgsfield with no blocks. PLEASE fix this, Google Labs! One of the images is a cinematic night time shot and the characters' faces are shrouded in shadows, LOL!

[solemn timber]

how can i use flow generator on phone

[analog lava]

Reminder images or videos cannot be generated in the discord! thank you!

[tepid wyvern]

pampampam,

[crystal sparrow]

Chai cup ki video Bana he

[dire cobalt]

Hey is anyone able to generate veo 3 vids with characters swearing?

[grand musk]

Hello Discord family I’m so honored to be a part of the platform not only a platform but groups of people across the world that share the same interests or beliefs basically we share with each other so I just want to say hi,⚡️ hey,👋🏽 what up what up””

[crystal sparrow]

Product video

Trane video

[cold gust]

Is veo 3 fast....

[glacial birch]

Does Vertex AI have the same Veo 3 features as Flow?

[urban ice]

Bro... It be so damn expensive 🫰🫰😭

[charred sail]

Why does this clown show up every week...

[devout pine]

how can we generate full length (8-10min) long video with flow?

[devout pine]

how are these guys creating full length videos using flow (veo3)?

[polar vortex]

Hi. As our company doesn't have credit/debit card, do you work with invoices? We would like to purchase a subscription.

[normal lagoon]

If creating an animation from an image in Whisk is the same as converting frames into a video in Flow, could Whisk also directly support Veo3 for animation generation

[grim path]

How do I generate video here

[forest raptor]

Has anybody managed to make a convincing fight scene?

this is the best I got

[remote sail]

i have a problem what can i do with this ?

it says i dont have the access to flow

[placid sparrow]

HEY

it is shjowing imagen fx isnt available in my country??? it was

[covert elk]

Me too

[worldly yacht]

im stuggling , i got a few fight scenes

[forest raptor]

can I see?

[worldly yacht]

yeap uploading now

Im still trying to learn for both to fight

[forest raptor]

how did you manage to make copyright characters?

[worldly yacht]

In flow and Gemini you can superman hulk , wolverine , you cant say Clark kent or bruce banner , and somehow it overrides it

[forest raptor]

can you do DC Lobo?

[worldly yacht]

i can send you a couple of prompts so you can try

I think its more focused on more known characters , lobo , bane , green latern gave me a hard time and didnt come out as i wanted

This is the joker one

[forest raptor]

Thank you bro

[worldly yacht]

Also - Gemini doesnt allow gore fights like the wolverine or hulk deadpool one , Flow allows it MOSTLY . ive been denied a few times

[forest raptor]

Oh, didn't know there was a difference

[worldly yacht]

Yeap , if you have the paid services with gemini which is 20euro per month you can also sign up to Flow tv with your google account , you get 1000 credits . if its allowed in your country

[shell kraken]

Has anyone figured out how to achieve character consistency between frames? Id like to tell a story about a modern family, but have been unable to have repeated characters in VEO 3 -- just upgraded for access to Flow now and will be trying there

[worldly yacht]

The only person ive managed to get somehow the same is the joker , everyone else changes for me , write a prompt very detailed for your character , the clothes , the setting , the hair the beard or clean face , the shoes , everything detailed to the dot , try to avoid character conversations , it mixes it up most of of the time

[shell kraken]

Copy that.

Do you know if we're able to pull in VEO 3 scenes into the scene editor capacity?

[worldly yacht]

honestly i have no idea :/

[halcyon basin]

#rules-and-guidelines・📋

#start-here・📍

#flow-tips-n-tricks・✏️

[icy dew]

😊 I'm using a consistent character in google tools. Will post tutorial on my yt

[icy dew]

This is the way to do it @shell kraken ☺️

[shell kraken]

Are you saving frames of your character and then using those saved frames as the starting point for the subsequent scenes (using jump to or extend?)

[icy dew]

Use the same character description in every frame. If you don't give the description of the character, flow sometimes gives a slight difference or worst a very different character 😆 but most of the time for extended scene, flow will just continue with the same character from the existing frame.

[worldly yacht]

also i noticed that if its in a more gritty dark enviroment is keeps the character more or less the same , something with the daytime videos that give soft edges and loads of light messes things up for me

[icy dew]

Yup. Sometimes I re-upload the same prompt two or three times and it gives me the result I want.

[worldly yacht]

Yeap , im keeping to the dark scenes because it works like a charm most of the times , also people should ALWAYS read to full prompt and to make sure realism in the prompt that they want

[bright crown]

Im having trouble creating a centaur winning a derby! I keep getting this type of output on the left, but one time got this one on the right but cant seem to get the action from the red one, on the centaur body of the yellow one...

[worldly yacht]

Search for a clean description of a centaur and send it into gemini with your prompt and tell it to mix it up , that you want a realistic centaur with this this and this , it might work

[mild totem]

你好，各位！我是软件开发的。想找一些中国朋友一起完成一个项目。有兴趣请联系我。Hello,everyone! I am a web developer. Want some Chinese friends to join my project. If you are interested, please contact me!

[slender cove]

Can the aspect ratio be changed for Flow? I have a 1:1 image I want to use.

[icy dew]

You can't change the ratio but you can instruct na cinematic view like for 9:16 output and then crop it using an editing tool. I'm not sure with the 1:1 image ratio though. You can try with the frame to video but still needs instruction. I have a tutorial on my yt on how to achieve a 9:16 with veo 3. Works with gemini and flow. The yt is on my profile

[strong harness]

"The camera follows a dachshund running through a living room and out of an open front door and onto a porch. It stands on the top stair overlooking the neighborhood as an ice cream truck drives by." -

[worldly yacht]

Superman and the Joker

[charred sail]

I see your superman and raise the only real superhero..

[glacial inlet]

Hi. I am new to Flow. I haven't figure it out how to add my own voice over, over the video. Can we import mp3 in veo? Thanks

[kind bison]

New bug today: using a frame from a VEO 3 video as the first frame of a new clip. Gets blocked because of the "no minors" policty. There are no minors in the image - it shows two adults. Flow is completely broken.

[worldly yacht]

Heres the boss ! love your video btw

[analog lava]

Hello if you're encountering errors and /or bugs or would like to leave a general feedback about Flow please don't hesitate in utilising the #1319424187730755636 channel. Thank you!

[restive dove]

Maybe someone already or have a prompt like shockingly waking up on a POV hehe

[vast gust]

Gemini AI Veo Rough Draft

Google Flow Final Draft

Any suggestions on vocals for singing?

[feral drift]

Hello, I wanted to share this History Documentary done with Google Flow technology.

We will soon send a Docs in the flow-prompts chat with all the prompts we used so it can help you get inspiration.

Link: https://www.youtube.com/watch?v=IOdzhZryMjI&t=74s

[frozen vector]

Hi - can you elaborate a bit more on what you mean by using a rorated 9:16 image please?

[soft walrus]

#flow-general・🌊 [Unicorn] coloring page, black and white line art, clean lines, no shading, no background --v 6 --style raw

[faint wyvern]

Awesome video, bro. I saw it until the end. I can’t wait for the prompt to be shared here to inspire all. 🙏

[feral drift]

<@&1366831844502274180> is there any way we can submit content to be displayed on Flow TV?

[signal geode]

Hi! That’s not something we can answer. Feel free to ping the Google team to ask!:)

[feral drift]

@tacit halo Hello, is there any way we can submit content to be displayed on Flow TV? We are very interested in this

[spice jewel]

Given that this is a group about the Flow website (not about the Veo models specifically), I’m submitting a problem description about poor UX around downloading the upscaled videos. Note: I posted these as suggestions in the flow-bugs-feedback section

The Frustrating UX of Downloading Upscaled Videos in Google Flow

The Clear Intent Problem

It's evident that the Google developers want to discourage people from downloading upscaled Veo videos on Flow.

Original Size Downloads Work As Expected

When I click the Download arrow icon and choose Original size (720p), I have thus declared my intent that I want the original-sized video downloaded. And Flow sends the video to the browser for download. Good.

Upscaled Downloads: A UX Nightmare

When I click the Download arrow icon and choose Upscaled (1080p), I have thus declared my intent that I want the upscaled video downloaded. The notification Upscaling your video (costs 0 credits). This may take a few minutes appears. Good.

I expect that once the process finishes, Flow will send the video to the browser for download when the video is ready. But — no.

After a while a notification box appears that has two links: Download and Dismiss. I have already expressed my interest in having the video downloaded, but the download does not start automatically. I have to click Download.

I click it, and the video downloads. OK, at this point I most certainly have no need for the notification, so I would expect it to go away right after I've clicked Download. But — no. It sticks around and I have to click Dismiss.

What a horrible UX. It's only missing that after I click Dismiss, another popup might appear Are you sure?.

The Click Count Insanity

So getting the Original sized-video needs TWO or even three CLICKS (why not one?), but getting the Upscaled video needs FIVE or even six CLICKS:

1. Arrow
2. Upscaled
3. Dismiss
4. Download
5. Close the browser download widget
6. Dismiss

This is insane! This is like Microsoft Windows in its worst incarnation.

Notification Placement Issues

On top of that, of course these notifications don't appear CENTERED OVER THE CORRECT VIDEO (which would at least make it obvious which notification pertains to which video). Instead, the notifications appear in the top right corner, because they slavishly follow something like macOS.

But macOS notifications appear in the top-right corner because there is no guarantee that the item they are related to is visible on screen at all. With Flow, it's different — WE HAVE THE VIDEO THUMBNAILS, the content items, right there in the same window as the notifications will appear.

The download links, especially the delayed ones (for upscaled videos) could be easily placed right next to the content items, WHERE THEY BELONG!

Redundant Error Messages

And of course one more terrible UX: When I choose Download arrow > Upscaled, I get:
Upscaling your video (costs 0 credits). This may take a few minutes.

And when I choose it again, I get an error-like:
Upscaling is already in progress

Why a different error-like message? Upscaling your video means the video is upscaling, Upscaling is already in progress also means the same.

As a user, I don't care that Google internals need to ignore a 2nd or more attempt to UPSCALE the video. The user action Download arrow > Upscaled is NOT FOR UPSCALING. It's for declaring user's intent that the user wants to DOWNLOAD THE UPSCALED VIDEO.

I should be able to click that Upscaled thing 20 times, and really, Flow needs to by itself figure out what to do. As a user I don't need to and I don't want to be bothered.

If the upscaling somehow FAILED — yes, then I should be notified. But not because I clicked the DOWNLOAD button twice.

Given that this is a group about the Flow website (not about the Veo models specifically), I’m submitting a problem description about poor UX around downloading the upscaled videos. Note: I posted these as suggestions in the flow-bugs-feedback section

Download Dropdown

When I click the Download arrow, I get three options: Animated GIF (270p), Original (720p) and Upscaled (1080p). In my view this should be presented as:

• Original (720p)
• —————————
• Animated GIF (270p)
• Upscaled (1080p)

The Original link is the only FAST option that is of interest for 80% of the user base, and does not create additional UX hiccups (not additional processing on Google’s side).

So it should be the 1st item, and clearly separate from the other packaging options.

Then a visual separator would be in order, and then additional "special" options would follow which are much slower.

---

[worldly yacht]

Did you yall know you could also use a json file or xml to create prompts ? i found it a little more accurate

[shell kraken]

Can you share an example of the JSON structure you're using to prompt?

Are you all aware that you can create 4 versions of each video in Flow (I'm using VEO 3 quality), for the same number of credits as 2 versions?

[spice jewel]

Veo does not expect any specific JSON. Just prompt Gemini or GPT or Claude with something like Write 5 variants of a detailed description of a 8-second video that starts with the 1st attached image and ends with the 2nd attached image, and in the middle features an interesting and surprising motion graphics animation. Each description should be in a separate JSON-formatted prompt. Describe the scene, the background, the camera movement and equipment (model, lens), the shot details, the set design — everything in high detail. and you’ll get a large JSON, and you use that as your prompt. Basically the JSON is "punctuation" for the Veo language model.

[worldly yacht]

On its way brother , and no i had no idea i usually go for the 1 video so i dont waste credits

[worldly yacht]

Superman Vs Joker

If you create this show me what you get im curious

[shell kraken]

Just kidding.... each generation is 100 credits, so for 4 videos, it causes 400 credits.

Reloading credits coming up!!

[worldly yacht]

have you tried it ?

[faint wyvern]

@feral drift I watched your video three times from beginning to end, and it's amazing. Were all the scenes in the video created with Veo3?

[kind bison]

Take a 9:16 image, rotate it 90 degrees so that it's now 16:9. Upload and animate in VEO. Download and rotate it back to 9:16. That was the method I used for this video.

[frozen vector]

Darn, tried that and it didnt work for me. You are basically saying you uploaded a picture of this woman on her side yes?

Not familiar with CapCut - does it integrate directly with flow or something?

[urban patio]

❓ USING 9:16 IMAGES ROTATED 90° CLOCKWISE IN FLOW CREATING SUBJECT TILT
So I have been using 9:16 Portrait oriented images rotated 90° clockwise to landscape orientation to prompt in Flow Veo 3 to generate a video that I can then export at rotate 90° counter clockwise for 9:16 Portrait oriented video. This method works, however I am noticing that the characters or subjects in the video tend to either lean towards the top of the frame, or their mouths are titled toward the top of the frame, so when I rotate the exported video back 90° counter clockwise to a portrait orientation, they are leaning left or their mouths are tilted to the left of the screen. Anyone else seeing this? Anyway to avoid it in prompt? It's some strange AI gravitational force
🙏 I welcome any thoughts or ideas. Thanks!

[urban patio]

I just posted on this method as well. Has been working for me but the subjects tilt! Look like your subject int he video is doing the same thing. Head tilts to top of the screen in Landscape when the video is generated, then when you rotate back in post they look like they are leaning left.

[frozen vector]

Have you tried any prompting around it? Maybe something around posture etc?

[snow wind]

can someone guide me to the best flow tutorials, how to build storyboard etc

[urban patio]

I did, I tried variations along the lines of "No added subject movement or tilt", in some cases it made it worse. The jawline/mouth tilt I am a little lost on how to prompt around.

[frozen vector]

I'd be mad if it wasn't so funny...

[left sapphire]

Hello everyone. I need your help when I use Flow and try to generate something. No matter what prompt I use, I get an error message saying I’m violating the rules with this prompt, even though I’m not doing anything wrong. Please help me.

[frozen vector]

Are you using an image of a person in the prompt?\

[urban patio]

What is the prompt?

[past dove]

Hey guys
Any tips for making the voices made by Veo3 less echoic? Especially those with Arabic voice lines

[faint wyvern]

prompt please 🙏

ok buddy 🙂 nevermind..good luck with what you doing here 😉 have a nice day or night

[thin quarry]

Wow

[worldly yacht]

niiiice i see the switch up too !

[jagged zealot]

Created with Gemini 2.5 flash. And this prompt followed by your Json text: "Create a video using this description in Json format"

[worldly yacht]

love it ! im gona try more day also

[vast gust]

Awesome idea share!

Old shool batman! Genius Work! Awesome video design!

[kind bison]

Yes.

[kind bison]

Her head was already titled in the input image.

[kind bison]

OK, just ran a couple more tests and I definitely notice that now - even when I prompt it to remain straight and not tilt.

[urban patio]

Yeah, it is interesting (and frustrating l!! I did this one where I tried to make her very tight in frame, but she still managed to tilt her head

[ivory lotus]

Hello all, how do you affrod enough credits to get better with the VEO 3 video creation? For example, a single prompt takes 100 AI credits with better quality.

I took first the Pro subscription, and it includes 1000 monthly credits, so it allows me to have 10 monthly clips - not much room for sandboxing. The ultra plan has 12,5k credits, but it's still not much, you can burn that on one day.

[worldly yacht]

They should add a custom amount of credits a user wants to buy , its a cool option for someone who cant afford to buy the 30 euro one and upwards , they choose their amount . pay and boom

[woeful fossil]

Use Veo 3 - Fast for 20 credits per generation in Flow instead of 100. You get 3 free generations per day with a Pro plan in Gemini. Google Vids gives you 10 free generations per day with a 1k character prompt window limit. Also Opal gives limited free Veo 3 generations as well.

Plenty of opportunities to experiment with Veo 3!

[worldly yacht]

What is opal ? and is it available in the EU ?

[woeful fossil]

<@&1366831844502274180> being unable to link to channels within our own discord is pretty frustrating. Can someone link him?

[spark delta]

#opal-general・✨

[worldly yacht]

and i just sign up for a free plan and get to test more of veo3 ?

[woeful fossil]

I believe you need a Pro plan. There are typically promotions or student offers for getting the pro plan for free for some amount of time. Not sure about region specifics though.

[worldly yacht]

oh oki i will check it out , thanks guys appreciate it !

[summer oxide]

I've been blowing credits at first without really considering the options, was just testing... lesson learned... almost over and rather frustrated😭

[ember eagle]

Done

[versed shale]

Does flow have a bug were you can't make kids say anything ?

Cus I can't generate anything with kids anymore

The Dialogue of the kid is literally: "I'm working here to earn my money to pay my food"

How is that against the TOS ?

Can any staff explain ?

Where can I send feedback?

[ivory lotus]

Thanks! Will check these out.

[summer oxide]

I really don't manage to get consistency on the story I want to try... can anyone here help me?

[noble carbon]

Anyone else figure out how to make frames to video have WAYYY less exposure 😭 I’m getting flash banged everytime I need to showcase a product

[kind bison]

Yeah, I got that warning several times this week - and none of my input images even have kids in them!!

[tall mural]

prepare your frame in photoshop. reduce highlights and shadows

[analog sedge]

So who is getting genie early access?

[fresh steeple]

Where can I check the Flow changelog after it vanishes from the home screen popup?

[wet kiln]

how coincidental, that happened to me just now

[fresh steeple]

https://support.google.com/labs/answer/16358089?sjid=1416577114462045801-NA

Out dated

[wet kiln]

Way out dated

[kind bison]

They used to post stuff in the Flow Announcements channel here but it looks like they haven't been doing it lately.

[fresh steeple]

Damn seems like there is no official log then

[worldly yacht]

A few today

[mortal zodiac]

i am impress with those punches veo3 refuses to do anything like that for me and then it goes and does something worse wen i don't ask

[shy gazelle]

Hi! I’m new to the community. I’m a content creator and I run a YouTube channel focused on collectible figures. But my true passion has always been storytelling.
I work a lot with FLOW, WHISK, and FX, and little by little I've managed to get some pretty good results.
My main project is Souls: Initiation, a fantasy book I wrote and have slowly been developing through Flow.
I’ll be sharing what I’ve accomplished with the tool little by little.

For now, here’s a short video I made—something comedic—based on a content creator friend of mine, but turned into a Sesame Street-style parody of him. It’s fun and lighthearted.

All my content is in Spanish (sorry about that!), and I made it with Flow, then edited some sections using CapCut. (EDITED 720P cuz i cant upload larger archives)

[worldly yacht]

i used 1k credits in flow to get that done , some are clipped not a full video because of glitches , just state that you want sheer power punches, camera shake ,like im watching a movie , also stay on dark and gritty scenes , seems to work better

[worldly yacht]

I collected a few marvel replicas 😄 thats all

[fiery radish]

anyone has details of today's update? I closed it too early and lost it

[shy gazelle]

something about u can use now ingredients section with cheap generation, but still in veo 2, they said are working on use ingredients in veo 3 soon....

[shy gazelle]

Every man needs at least one Batman, or a Spider-Man, a Wolverine, or a Goku in his life...wise advice

[worldly yacht]

those are siiiick !!

[fiery radish]

Got it thank you

[gilded ridge]

oh wow! now they support Ingredients in Pro plan

[tardy ridge]

VEO is completel broken, the update destroyed character consistency, jump to - its converting my charactes to asain from america because they're meditating this is garbage, every update wrecks the model

horrible he looked like this yesterday

today he looks like this

Can I get a refund for these credits, this is beyond absurd

https://youtu.be/LgRfq-vJEog trying to tell an interesting story is impossible

[ember seal]

Looks like Veo is broken in every way, I’ve been having a lot of problems with it since yesterday. Feels like I’m using the preview model or something

[frozen vector]

Anyone know if there's a Flow API to generate videos?

[forest raptor]

I really wish Google made veo3 more affordable. I can't experiment and adjust prompts

[forest raptor]

I gave up on that, and I'm the one who used to make 3-4 minutes long videos BEFORE Veo3. Just wait for the future

[shy gazelle]

i can i do consistent histories with it, maybe you are using wrong the tools. I use it 24/7 with no problems

just few moments ago finish this mini sketch

[shy gazelle]

Here you go, now it looks closer. It’s just a matter of using the right tools…

[shy gazelle]

u dont need to do the prompts, u use chat gpt or gemini to do the prompts, you tell them the idea with details, gemini do the prompts or chat gpt. Then you make the first frame image on whisk or fx, finally go flow and start your history using image to video, also using the prompt made it by IA

[ember seal]

I did a lot of generations with the same image and prompt yesterday and the day before, so I know what he's talking about. "Frames-to-video" is extremely horrible now, and it's been like this since yesterday. Perhaps not for all users, but at least for some

[shy gazelle]

lets see share me your prompt

helping you making 1 scene

[feral drift]

yes, except for the human intro, everything was created with veo3 - quality text to video.

In fact they uploaded Vol2 also if you're interested and a breakdown/ behind the scenes on how the video was made

[cyan ruin]

Guys can you rate my AI vidz, been working on new vidz...

https://youtu.be/Z6ql6h4aOn0?si=ReckUMKTTlh_iy1z

[faint wyvern]

Yeah, to pay a membership on his YouTube channel for that behind-the-scenes content. That's funny. 😂 😂 😂

[shy gazelle]

It works for me... I guess if I’m at zero and need to pay for a subscription, I’ll learn how to use the tool. I don’t see anything wrong with that...

[shy gazelle]

I see more issues with how the story is told than with the AI generation itself. Honestly, I didn’t really understand what the story is about — it just looks like a series of random images with no coherence. I’d refine the storyboard more before starting with the FLOW generation.

As for FLOW, focus more on consistency of the character or characters. Work more with the frame-to-video tool. Use text-to-video only for scenes that don’t require much fidelity or continuity. But for a longer story, your best tool is frame-to-video.

[cyan ruin]

Thanks for this bruh. Gonna study that frame to video. Was it also a text to video

[shy gazelle]

I achieved this level of consistency using image-to-video. Here I did a consistency test, and it turned out pretty well. It can still be improved, and in fact, I refined it quite a bit in the first episode of my series compared to what you’re seeing now... but without a doubt, image-to-video is the winning path for long and consistent stories. https://www.youtube.com/watch?v=oq-0qU5GTwk

[faint wyvern]

Well, yeah, I guess that's the point; you are right about this.

[cyan ruin]

This is cool

[feral drift]

Hello, is there any way we can submit content to be displayed on Flow TV? We are very interested in this @main saddle @fierce valve

[crisp basalt]

https://labs.google/flow/tv/channel/crash-test

this is actually the best thing ive ever seen

[crisp basalt]

so i tried it myself 😂

[teal ether]

could anyone tell me when flow points reset

[tardy ridge]

Yeah not close enough i made a 13 min long movie using veo with good consistency until the update on 7/26 which introduced a lot of drift luckily I was almost done so it didn’t impact too much, I noticed it improved a little after a few days but back to trash outputs

Right now veo is taking in context from the entire prompt and applying it to characters likeness something is broken with the embeddings