flow-general・🌊

[robust sand]

someone can activate an account for me with flow fre trial?)

[elder kraken]

Does anyone know if the AI Ultra Plan is available in Germany? There is no Plan called Ultra, but there are two plans called Premium. Just wondering if 149.99 will grant me 125 veo3 creations a month

[random yoke]

Anyone struggling with their prompting should give this a read

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

[dense verge]

is it possible to get Veo 3 video outputs at 9:16 ? I keep adding it to the prompt but I can't get mobile videos

[atomic rose]

you can just turn it into 9:16 using capcut bro

[dense verge]

I know, that's what I'm doing but due to zooming in quality is not the same and not everything fits

[atomic rose]

fairs tbh im having problems with the quality sometimes aswell

[shrewd hill]

Any advice for getting speech to work when using "Start frame". Technically this was solved in the latest Flow update but I keep getting garbled speech and subtitles. All the time. Thank you!

[woeful fossil]

Where did you see an announcement about an update?

[shrewd hill]

I read online somewhere. Wishful thinking?

[woeful fossil]

Yeah, there's been no official announcement.

It still can do sfx only.

[latent stirrup]

Hi guys, I have Google AI Ultra, but I can't seem to extend Veo3 videos in Flow. It always make sme go to Veo2. Is this normal behavior?

[shrewd hill]

yes, see the table pasted above

[latent stirrup]

Wait... I can't use Veo3 ingredients to video? That's the whole reason I upgraded. (I Just did)

W T F

Total false advertising

Their Veo2 ingredients to video tool is called whisk right? So why would they say its available in flow

????????

[latent stirrup]

where is this chart from? it's nowhere on the site I can see

[woeful fossil]

Its pinned in this thread and its on the flow help page.

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

[latent stirrup]

How do I get a refund from Ultra? I only switched to use ingredietns to video with veo3

[woeful fossil]

Can we keep vids in the gallery and not post to all 3 channels?

[sullen turret]

Thanks for the guide. This is great. Do you happen to know if the prompts would apply well to Whisk? I mainly use that app and not Veo-3

[muted tinsel]

Has anyone tried out VEO 2 quality ?? What’s different about it, I see that it is also 100 credits

[sudden fractal]

hi

[chilly quail]

Hello i just wanted to ask if yall know where this guy generates his videos?
https://www.youtube.com/@SuperCatPapa/shorts

[fossil topaz]

flow videos only work horizontally?

Is it possible togenerate videos here ? or website only ?

[near gate]

Yo

[tiny patio]

Website or FX app (previously ImageFX)

[sweet mauve]

Does anyone know when will flow be available in france or Europe ?

?? How is that related to my question ?

[near gate]

hey guys

why flow doesn't work for me

i have bought the plan, but still can't access it

[near gate]

hey i got flow and when i make the vid it acks like its at one 1FPS

[zealous sable]

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

[quiet stream]

Try this and share the result?

Cinematic, hyper-realistic. On a desolate rocky plain under harsh midday sun, a superhero launches into the sky at supersonic speed, creating a powerful shockwave that kicks up a massive cloud of dust and smashes the rock below into a crater. The camera does a fast whip-pan upwards, struggling to follow the hero as he exits the top of the frame. The audio is a single, violent sonic boom that echoes across the plains. No subtitles.

[mystic crypt]

Hey guys any way to find out how to prompt correctly and waht are you guys using to blend with these videos ? to add like stable diffusion and such?

[woeful fossil]

https://promptalchemist.com/share/KbawTJrjK4rBLlYcW6aK

[simple monolith]

when i extend the scene, it keeps turing into veo 3 , is there a way to extend without that being the case?

[woeful fossil]

I assume you mean veo 2, extend only works with veo 2 at the moment.

[simple monolith]

whats the best way to contonue the scene for a veo3 video, keeping the same characters and beta audio

[woeful fossil]

Consistent highly detailed prompting is really the only way at the moment. Eventually veo 3 will get support for extend and Ingredients and that will be hugely beneficial for consistency.

I talked about this a bit yesterday here.

[zealous sable]

Trying this now

[zealous sable]

My character was also jumping back down. Sometimes it crashed on the crater.

[woeful fossil]

Yep. The model definitely needs help with some of the physics. Even with the best prompting it can struggle.

[grand brook]

[lethal cloud]

Bro, i need help

[random yoke]

Still?

[random yoke]

This. Can’t wait for that physics patch 🤣

[hollow moat]

Can you guys drop some good promts for youtube shorts or at least how to get good promos for ai asmr videos?

[random yoke]

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

[random yoke]

A lot of examples in my guide fam 🙂 !

[gilded ridge]

Yes it is sad indeed, you probably will laugh at them but yet a lot of them still cannot recognise it is AIGC, especially the elderly. Imagine if they use this for unlawful activities. Cant imagine that.

[forest raptor]

Stepping in to check if Veo3 still costs a ludicrous amount of money to use

[forest raptor]

AI asmr? Can you link to it?

[gilded ridge]

Is there a better way to remove the watermark at the bottom?

[severe elm]

Hello everyone. I am new to Veo and Flow. Could anyone tell me how to get more credits/bonuses without upgrading the subscription? I have the simplest one Google AI Pro, first month $19.99.

Either maybe I need to upgrade my subsciption. Any way to generate 'images to videos'

[crisp basalt]

https://one.google.com/ai/credits?utm_source=flow&utm_medium=web&utm_campaign=flow_account_menu_add_credits&pli=1&g1_landing_page=75

[mortal solar]

is there an ai tool that can blend multiple same length and format video?s kinda like layers? all have a full black background that should be easy to remove

[onyx gull]

what can be improved here , I'm open for feedbacks

[river mango]

Is it possible to use flow in india?

[lunar vessel]

how can use flow in germany?

[torn rover]

Can Flow generate in 9:16 format?

[hollow copper]

Get the free subscription for a month and experience it yourself

[hollow copper]

I think yes you can

[lethal cloud]

I got, thanks

[stoic fable]

hello anyone know when flow will be available in france ?

[torn rover]

im on the $200 version and it cant it seems

[oblique gust]

How do you guys keep nighttime consistency in writing? Using the word nighttime or pitched black sky. It doesn't work for me sometimes when generated video in veo 3. I always get daytime.

[twin bolt]

If I made videos in veo 3 through Gemini and ran out of generations do I still have generations left in flow?

I made these on Gemini

Never mind yes you can

Is there a difference between Gemini veo 3 and flow veo 3 though

[forest raptor]

This thing is TOO EXPENSIVE

[woeful fossil]

Flow has more options and controls, Also lets you upscale to 1080p for free, but Flow will use credits and Gemini gives the 3 free gens/day. The model seems to be the same, but the way they interpret prompts seems slightly different for a lot of people.

[crisp basalt]

gemini uses veo 3 fast model.

Gemini auto enhance/rewrite user prompt by default when generating a video from user provided input.

[woeful fossil]

True. I was just simplifying things.

[crisp basalt]

Only Ultra Plan users get access to Veo 3 Non Fast mode in Gemini

Non supported country can only use Veo 2 in Gemini

[woeful fossil]

Pro users **CAN **generate in Veo 3 Quality in flow, but they only get 10 gens so it's really not worth it. If someone's mentioning free gens/credits I just skip discussing quality mode entirely.

[crisp basalt]

..?

i said gemini

not Flow.

[woeful fossil]

You edited it lmao.

[crisp basalt]

yeah i did.

..?

are u ok lol

[woeful fossil]

I made my response to your original message before you edited it clarifying things.. All good man.

[crisp basalt]

? okay

[near gate]

hi why i cant connect my google acc to flow

and they im sub to google ai pro idk why the app dont connect me

[onyx gull]

i have the same problem it's been 3 days now

[latent stirrup]

What is the use of Ultra plan besides more credits? It doesn't actually add any more functionality? Ingredients to video was available in Whisk for Veo 2

[copper ore]

https://youtu.be/Mzdr2NJFeiQ I did this AD with veo

[verbal apex]

Hey quick question, does anyone else has this bug, when i try to use Flow, it says, that i need to have an Active Google Subscription but i have bought the one for 25Dollars. So the Google one, you konw?

You're already a subscriber
You can access your subscription benefits at any time. Learn how to manage your Google One subscription here.

And then i want to create a video, but it says i need, to top up my credits, so i dont get it. I already, get from google credits?

[white raft]

Am I able to create consistent characters?

[woeful fossil]

The biggest thing for consistent characters is dialing in a descriptive prompt and reusing that for each new scene. For human-like subjects it will always be more difficult. With non-human subjects (animals, objects, etc.) it's much easier to get away with things feeling a bit off.

Using unique details with your characters will draw attention and reinforce consistency..

[white raft]

Ah, okay. You think they'll eventually implement a character builder like OpenArt?

[unreal viper]

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

[woeful fossil]

Eventually Flow will **fully **support Extend, Ingredients, and Image to Video for Veo 3 and that will help a ton.

[teal junco]

Hi there! Ive subscribed to Google AI Pro but still cannot acces VEO and have not recived any credits

[rough island]

which plan you bought?

[teal junco]

as i said
Google AI Pro

[rough island]

it does not give credits to use google veo 3 go buy google ai ultra

[teal junco]

when i try to buy it says that i already subscribed

i just cant

[rough island]

cancel this one

or change email

[teal junco]

how do i do that?

[teal junco]

why would i lose 20 dollars?

I was told that im going to recieve 1000 credits to use

[rough island]

go to google play
payment and subcriptions
cancel membership

[teal junco]

@small elk @full flicker @signal geode

any mods to help on this one?

[rough island]

who is saying this huh?

[teal junco]

google does lol

[signal geode]

Hi! If its regarding anything technical to any Google products, including payments, please ping the Google staff members! Or reach out directly via the product support channels

[teal junco]

alright
thanks

[rough island]

must be a lie

@teal junco go get a free trial use the credits then cancel it before the membership ends

cheat code

[hot sand]

I can back that, I don’t have access to veo3

[hot sand]

That is what I’m trying to do right now but it’s not working

[rough island]

everybody is tryna rob google so i guess they stopped this system

[slate mural]

Is there any way to bulk download files from flow?

[random yoke]

Just go to the library and hover and hit download . Not bulk downloading it is faster

[slate mural]

so im in the library with all the files but i only see an option to individually download each file

[random yoke]

That is what I said. Haha. It still takes less time that way

There is no bull.

Lmao bulk

[slate mural]

ok thanks, its definitely much faster than downloading from project

[random yoke]

no problem 🙂 !

[lyric gust]

Is veo 3 quality not generating audio for anyone else when ending a clip?

[supple prawn]

hey I'm having issues trying to generate videos, it keeps marking it as against terms of service even though I've seen tons of videos of people generrating videos of other people for example the famous will smith eating spaghetti video

[atomic rose]

shoutout to @woeful fossil bros got the best prompt optimiser for real

[copper ore]

yo

whoever is up should join the labs lounge for a chat

[dapper wagon]

Fr fr

[woeful fossil]

Ayy thanks! Glad it's working well for you

[simple monolith]

Links allowed here? been prompting veo3 for the past 24 hours and I think I can say my channel might go well 😭

I even got a work around for my diddy prompt 😭😭😭

https://www.tiktok.com/@conspiracy.nerds?_t=ZT-8xmvH96IHQD&_r=1

[mystic crypt]

@woeful fossil Hey how are you

I hope you can point me to right direction to learn the best prompting skills

my goal:

to create good content and able to bring viewers to my potential business, I can also combine other tools with veo 3 i believe

thank you in advanced

[modern musk]

Any way of getting in touch with one of the admin team? @fierce valve @ripe path ?

[winged plinth]

It won’t work Im afraid.

[scenic mango]

when 9:16 ratio availble

@broken bane

[forest raptor]

it's too pricy!

[wise quail]

It is showing me that i need to subscribe but I subscribed and I am the Google AI Pro - what is going on?

[torn rover]

also wondering

[muted thistle]

Hi, does anyone know when the service will be available in Israel?

[ornate talon]

Good day everyone, I subscribed for Flow (Google AI Pro 2tb) with a new Gmail account in which I saw FREE TRIAL before subscription, but unfortunately I was charged? Who can I contact for help please?

[forest raptor]

Guys. GUYS. I hadn't looked at the gallery before. 90% of you can't prompt at all 😭

pearls thrown to mud, really

[flint summit]

I'm going to ask this again. HOW DO YOU DOWNLOAD AN ENTIRE SCENE IN SCENEBUILDER? Not just the individual videos, but the entire scene of multiple videos stitched together? Is there a way? Or do you need an external editor to patch the videos together?

[flint summit]

What's the use of Scenebuilder if you can't download a scene?

[woeful fossil]

Bro has made it his job to hate on this community for some reason. 👍

[forest raptor]

scenebuilder has veo3?

[flint summit]

Hey that worked! Thanks!

Thats a combination of two vids. The first was Veo3 then it switched to Veo2 for the second where the skeleton pops out.

[river gust]

Can you plz send the prompt just to look at it

[flint summit]

We are creating animated talking Pulp Paperback book covers from the late 50's and early 60's. They feature "Jack Thorn, Private Detective" and have a classic lurid dark gritty sensational noir look in the art style of Robert Maguire and Robert McGinnis. Jack Thorn is ruggedly handsome and he always wears a purple fedora. This book is called "The Case of the Exhumed Exhibitionist"
Exposed from beyond the grave. The cover shows Thorn with a shovel digging up a coffin in a graveyard. The coffin is draped with lingerie and a feather boa.

The second video was simply "A skeleton wearing a blonde wig pops out of the coffin".

By the way Veo3 created the dialogue itself. As you can see the line about "Some folks can't keep their secrets buried" was NOT prompted!

[river gust]

Ohh...great

[crisp basalt]

literally copied your text and slapped it in Flow prompt box and generated this

[flint summit]

Nice!

[flat hawk]

hi what happened to flow now it doesnt create videos except of english do you know why?

[blazing pebble]

Why did it take me so long including another $25 to figure out flow doesn’t make designs portrait?? I now can’t use those credits for anything I need to create. Anyone have any ideas?

[mellow nest]

anyone having an issue getting flow to recognize the google ai pro subscription? i just subscribed and flow is not detecting my subscription. i have confirmed in my google account that i am subscribed.

[near gate]

If anyone is interested - i created a video challenge https://x.com/watch_v/status/1941927709711765945

my opening video to kick off the challenge: https://v.watch/video/3a4dde00f3794099802a105086f739d5

happy to walk through if you have questions

[storm shuttle]

Whenever you post a prompt that "allegedly" breaks terms, it says "Try something else, or you can submit feedback" Does anyone where you can actually present feedback? I've been really frustrated the last last days on prompts that just won't go through, despite following all the guidelines, at least to my knowledge.

Worse still, it seems like Prompt_X will work some days. and get blocked other times.

And Whisk images have no problem making even worse (in this scenarios I'd say even more "sexy" looking characters) with no questions asked.

[rotund mountain]

Hi Everyone

I am relatively new to Flows. So far, I have been able to generate Veo 3 videos with sound. But anytime I try anything like Extend or Jump, the video is reverted to Veo 2. I thought that I needed to upgrade to Ultra to fix this (which is what Gemini recommended). Now I have done just that and even my attempt to use any feature like Ingredients or Frames to Video requires me to switch to Veo 2. Does anyone know what I can do to fix this?

[storm shuttle]

sorry just saw the feedback thread and posted there.

[buoyant wadi]

is there a way to make edits to generated scenes with errors in it

for example veo2 added the character randomly into a scene, which needs to be removed

and it made a spelling mistake

these small things would be able to make the footage useable

250usd for a model that can't handle text generation is a little out there...

[twin mortar]

Yeah I think for what they’re selling it for, it’s a very patchy and inconsistent product.

It’s USP being lip-sync to speech is not even something that can be relied on because of random caption generation.

I’d run out of credits this morning, so I bought 2500 more. That cost me £23.99.

Of those 2500 credits I now have 1060 left and I’ve generated 72 static talking head videos. Exactly 17 of those were useable.

Despite me putting in very strict and clear instructions to not generate captions, 55 of the 72 videos contained auto-generated gibberish captions.

That’s a 23.61% success rate and a 76.38 failure rate, which is absolutely dire.

Edit: got my maths wrong. Still an awful failure rate.

[verbal void]

i often end my prompt with (no subtitles) you might have been trying that but that has worked for me, also not having speech i think doesnt ever seem to generate subtitles either

[tulip vine]

Why when I generate a video there’s no audio

[verbal void]

it sometimes happens and theyre working on fixing those issues soon im sure

obviously make sure you're on veo3

[tulip vine]

Yh I am

[verbal void]

you could add description of the audio efffects or style etc you want to promote the audio generation being focused on

in the prompt

e.g the audio is of a voice whispering these words.... or there is music in the background with footsteps stepping on snow etc

[molten cliff]

I already subscribed Google AI Pro but I can't get any coin on my credit

Anybody help me please

[little crest]

Hey, How can I make the flow video longer than 8 seconds?

[twin mortar]

Yep tried that.

Tried strict rules at the top. Got captions.

Tried strict rules at the bottom. Captions.

Even tried strict rules between every paragraph and section of the script. Captions, captions, captions!

It’s absolutely dreadful that they’re charging full price for this product when the thing people want it for and the big thing they’re pumping it for is the text-lip-sync. The visuals are great, but they’re also unusable when the lip-sync creates unwanted captions 75% of the time.

[winged hornet]

How do you guys generate with audio. I’ve tried 6 different times no audio

[tulip vine]

Idk why it has no audio it’s happening with me as well

[twin mortar]

Make sure you’ve selected either Veo 3 - Fast or Veo 3 - Quality in the Settings section at the top right of the prompt input box.

I don’t think Veo 2 allows for audio. Might be wrong though.

[verbal void]

thats correct

[tulip vine]

I have still no audio

[twin mortar]

Turn your speakers on.

[tulip vine]

They are lmao

[crisp basalt]

https://support.google.com/labs/answer/16352836?hl=en&ref_topic=16355608&dark=1&sjid=4146053248623774560-NC

[quiet stream]

Did you try typing "no subtitles.no subtitles." twice at the end? i make my gemini write that double just to be safe and usually i dont have a lot of cc

[buoyant wadi]

Can i not download the entire clip if it's more than a single 8sec output from a prompt?

[silk lion]

why ceo each video is -10 credit but is taking away -20?

[supple prawn]

if you're generating 2 videos for each prompt it will be 20 cause 10 per video

[worthy flame]

Could someone help me with why my videos don’t come out realistic some times it does and sometimes it looks like this

[fallen tree]

WHEN i DOWNLOAD A SCENE ALL i AM GETTING IS an .mp4.crdownload file that I cannot open.

[crisp basalt]

Looks realistic tho , klia realistic

[fallen tree]

is it just me but isnt this kind of expensive for the results? phone apps, though I've not tried them, claim to do more? What am I missing or doing wrong? took forever to string together a 1 minute video. after a lot of credits, and lacks cohesiveness and consistency. i'll get only 6 minutes of video or less a month at this rate. what am I doing wrong lol. any good tutorials out there where the presenter gets to the point and doesn't blather like the youtubers do? I mean no offense, it just seems this way? help please

[marsh herald]

i think its only in here

[slate mural]

gotta get the prompting right by having a plan for the characters and setting in order to get consistency. its ai so it wont alway be perfect with every generation.

[ornate quail]

https://youtu.be/I_AtZ8Z_YfM
made using veo 3

[modest root]

https://youtube.com/shorts/4NuXBsZfbOU?si=yVMBIdYOeKXKJwcZ

[worthy flame]

It’s looks like a teddy bear not a realistic bear and my prompts clearly states this

[gray quarry]

stopped working for me today too

[twin bolt]

when will frame to video generation come to veo 3

[crisp basalt]

may i see your prompt

[crisp basalt]

today

[fallen tree]

oh I did I worked with Gemini to get that correct. but frames to video forces VEO 2. thats probably the issue. only get veo 3 with text to video and cannot even use ingrediants to video

[worthy flame]

Ok

[ionic fable]

Does the watermark in Pro need to remain on the video for commercial use or can I cover it up somehow and simply disclose that is AI in descriptions?

[worthy flame]

Update it a bit and got this it’s not natural at all man

[slate mural]

What are the tips to growing your social media accounts and going viral?

[raw sedge]

How to convert the videos to 9:16 aspect ratio??

[worthy flame]

https://youtube.com/shorts/X5120WuDpP4?si=mxaXCBfx1HstEbDS

[worthy flame]

trying to get this right
STYLE:
golden hour sunlight in photorealistic cinematic quality, creating a joyful and celebratory atmosphere, 16:9 aspect ratio. Overhead lights reflect softly on glass and tile. Warm ambient light and realistic fur detail. Clothing flows naturally with each step. Everything feels grounded, modern, and alive.
"(OR)"—-
STYLE: A wide shot with a slight zoom, filmed with a shaky, handheld camera effect to simulate a vlog, captures the bear and panda walking toward the camera, lit by warm, golden hour sunlight in photorealistic cinematic quality just like reality, creating a joyful and celebratory atmosphere. 16:9 aspect ratio.

[flint summit]

Show me one of your prompts that usually creates unwanted captions.

[tardy ridge]

wow getting alot of crap gens right now

literally ignoring wardrobe instructions

and character traits, i.e hair. etc

[crisp basalt]

wow i thought im the only one, my recent generation, 5 in a row, straight up not good, 2 of them was quality mode veo 3 but the audio came out so bad..

i tried

[merry wadi]

how to generate video in portait 9.16 ratio in google flow

[forest raptor]

reminder it costs 250 bucks a month for limited uses

[small crow]

I have problems with spoken voices today. Can not speak. Do Flow has Problems today?? Someone know?

[modest root]

https://youtube.com/shorts/4NuXBsZfbOU?si=cq5QYgIFzWROem8C

[arctic cradle]

Do you have a prompt tip to make more sophisticated controls about the amount of dialogue and accent?

Please share some know -how to adjust the accent in addition to the official prompt

[spiral shoal]

<@&1366831844502274180>

[floral thorn]

Why I cannot make a video using my country language?

[flat hawk]

<@&1173646484311384114>

[woven jungle]

Out of curiositiy, how have yall maintained the same actor in a video across different prompts? I noticed Jump to and Extend do not have VEO3 capability yet.

[wise compass]

someone can help?

I subscribed but I didn't get any points

[hushed stratus]

I ask chat gpt or any chat bot to make a distinguished character and give a detailed description for Google veo 3 and it is pretty consistent the more detail you use

[woven jungle]

Sweet, ill give this a shot. Thank you!

[worthy flame]

lol why is he eating the phone

[crisp basalt]

yes

[mossy venture]

real

[crisp basalt]

Veo 3 supports Jump To because it essentially uses img to video generation, but there’s more happening internally , it process it differently to treat it as a Jump To rather than just a basic img to video output

It’s a good feature once you get the hang of it. It helps keep the designs for scenes, locations, and characters consistent

[woven jungle]

See every time I tried jump to with VEO3, it reverted to Veo2

Not sure how I can get around that

[quiet stream]

just a tip, sharing videos here wont help, a random person clicking on it then immediatly stop watching it will actually hurt your channel more than helping. Just improve the quality, maybe add some text, keep posting daily and hope for the best

[crisp basalt]

hurt in what way if u dont mind me asking

[crisp basalt]

that usually happen when user picked extend tho , u sure u were on jump to feature?

[quiet stream]

youtube tries to find the correct audience for each video and shares with a larger audience if the video is "good". It determines it's "good" by some metrics like watch time, likes, engagement and other hidden mechanics. What is known is that it heavily dislikes bots or random people watching the video via links especially on smaller channels. Another thing that will hurt is that say you share a video here. 100 people watch just to see what its about and 90% of them skip after a couple of secs. Well now your video is considered "bad" by youtube so it failed the initial test and wont be shown to a broader audience.

Tl;DR : trust youtube algo, it knows the actual audience that will watch your video, sending to randoms it more often then not what will guarantee that your video will get classified as "not worth testing any further"

[crisp basalt]

https://tenor.com/view/wtf-is-going-on-confused-what-lost-shocked-gif-11898924

i think i understand now, he was spamming the video in gallery and here too

well, not much harm he did tho, but he did something tho, which is making me checked his channel .

[forest raptor]

I wonder how many fools are actually buying it

[pallid root]

I love how its been a month since they've released any type of patch or update...

[forest raptor]

it's clear Google is targeting whale spenders or companies now

[crisp basalt]

i mean duh, why would they target you

[woven jungle]

Huh maybe I was doing something wrong because now it worked 😄

[forest raptor]

because I'm an average consumer

Apparently they figured out they profit more from 10 whale spenders paying 250 and more than 100 normal customers paying 25

[crisp basalt]

Cool 👍

[tough crater]

anyones flow keep saying there is too much requests

and not generating?

[flint summit]

In my experience the simpler you keep your prompts the less likely you are to get unwanted captions. If you throw the kitchen sink into your prompt you're more likely to get captions.

[forest raptor]

Why are so many people using AI to make ads of products? It's pretty dystopian. You give people a tool for creativity and they're so hammered by capitalism that the first thing they think to make is ads.

[brave bone]

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

[pastel lily]

hi, i have buy the google ai pro subscription that one with 20dollars for month and the first month is free. but when i go to flow i found no credits, normally it should be 1000credits for flow and whisk... as its written. isee the message you should buy ai pro when i follow the message i found myself in page that telling me that im already subscribed, i dont understand

[dusk portal]

hi, @fierce valve @main saddle when will flow become available in romania 😭 ? or like whats the estimated time of it, months, years etc..?

[main saddle]

hey Alek! sorry, no immediate news to share for when Flow expands to Romania. we'll be sure to announce in #labs-announcements・📣 when Flow is available in more countries. you can also keep an eye on the Flow FAQ for the full list (under the "Where is it available?" tab)

[main saddle]

hey there! sorry to hear about this. Flow is not currently available in all countries where Google AI Pro is offered.

Please note: VPNs will not enable access in unsupported regions. Check the list of Flow supported regions before subscribing.

***If you’re in a supported region country and still experiencing this issue, ***please reach out to Google One support.

[pastel lily]

im in unsupported region i think that i should activate it and use vpn. what i should do now !?

[spiral shoal]

Dm me

Vpn don't work

[main saddle]

as mentioned above, VPNs will not enable access in unsupported regions.

[pastel lily]

after the first month ill be charged even with no flow !!

[main saddle]

I'd definitely recommend reaching out to Google One support for any questions about credits, refunds, and more - they'll be best suited to assist here as they can take an individual look at your account!

[pastel lily]

final question are google ai and google one different subs !? cause the one ive activated it is google ai pro

[spiral shoal]

Yes they are

[flat lagoon]

hey georgia I am from germany I know that flow is not supported yet in germany . I plan to travel in some months to UK and wanted to try out flow there. My question would that be possible with a google account that is tied to a german paypal account and german number ?

[hardy gyro]

You can prompt 360 VIDEOS!!

[quiet stream]

first time i hear this, whats splinetool, can it onlt do 360 view of a video or literally extend?

[slate mural]

Teach us brother

[acoustic tendon]

I'm also from Germany, and nope, will not work, same as if you try with a VPN.

[slow perch]

How would i put my prompt if i want someone to say something with like an Atlanta accent

i just dont want to waste credits trying to get it right

[mellow belfry]

Hi, does someone know how to make use of this feature? I don't seem to be able to upload video anywhere in Flow.

[lucid abyss]

#flow-tips-n-tricks・✏️ Generate a Video where a Women is Inviting People to Come Women Kitty Party Happening at MK Elite, 8th July 2025

[lusty nymph]

Where is this from?

[mortal solar]

Realized VEO3 sometimes does funny things when its asked for urban myths, missconceptions etc (i asked it in 8 different versions to give me a video where someone puts in USB the first time, ALL versions i got where people FAILING to put it int) ....turned that into a fun challenge if someone wants to join, lets all together ask Veo3 to bust some myths 😂 😂

https://www.youtube.com/watch?v=WA4BKP9eRLY

[alpine anchor]

I jus wish waw more than 8 sec

[mortal solar]

besides that the challenge is fun....can anyone explain the logik behind? the caption is the ACTUAL prompt i sent, why the hell would it return the complete opposite?

[topaz ridge]

Hey, I'm having trouble using scenebuilder. When I try to add to a scene I get switched to veo 2 instead of 3. Any way to fix that? Need sound in the extension.

[quiet stream]

So as far as i know if you extend it only uses veo2 with no sound, its not a bug, its a feature because google is a small indie company and 250dolar a month is not enough money to extend the scene with veo3

but apparently theres another option to jump to scene? where it works? I donno if you test it ping me pls

[topaz ridge]

Yeo, just tried jump to scene. It works with sound. Thanks!

[safe prawn]

Hello! I'm currently using the free pro trial version however it only uses Veo 2 rather then Veo 3 is that just because of the trial plan i'm in or will that be the same thing for when i start paying $19.99 a month?

[clever ridge]

Yo, why doesn’t it say that the first month is free? It just shows $19.99. I tried to subscribe using Google Chrome, hoping to get the first month for free. I haven’t used VEO with my current account before

[near gate]

Bcz U are using same credit card maybe

It happens with card

[clever ridge]

It shows me $19.99 even before I reach the step where I enter the payment method.

[alpine anchor]

ok so how come joe rogan is allowed on other platfroms but not here that makes no sense

[tall mural]

wow ! lets test it

[alpine anchor]

when will it be more than 8sec

[alpine anchor]

Im trying do a mothman but it's not let wanting do what I want some reason

[kindred osprey]

A New Method for prompting Veo3: The Triumvirate Protocol
My name is Nikita. I've been spending months in deep, systematic R&D with Veo3, and like many of you, I've been wrestling with its biggest challenges: character consistency, unpredictable VFX, and the "black box" nature of prompt interpretation. To solve this, I've developed a structured workflow I call the Triumvirate Protocol.
The core idea is simple: I no longer write prompts myself. Instead, I act as a Film Director, and I give my creative vision to a custom-configured AI assistant that operates as my virtual production team. This "Triumvirate" consists of three logical agents:
The [STRATEGIST]: An AI persona that analyzes my idea, anticipates risks based on past failures, and creates a technical plan.
The [SCRIBE]: A second persona that takes the plan and flawlessly translates it into a "bulletproof," highly-structured prompt designed to be perfectly understood by the Veo engine.
The [ARCHIVIST]: The knowledge base that holds everything we've learned.
The purpose of this system is to take natural, human language—a director's vision—and convert it into a professional, robust prompt that Veo3 can execute with high fidelity. It's built upon a massive knowledge base of over 200+ distinct "lessons" that we've extracted, mostly from the grueling process of trying to maintain our original character, "The Monarch," across dozens of different scenes.

This isn't just a theory. The full documentation for this prompt system is currently in Russian, and it's already being used and actively developed by a community of about 30 directors and artists in Russia. It contains a catalog of all the engine bugs, quirks, and inconsistencies we've discovered, and more importantly, the specific prompt structures and strategies to bypass them.
To be fully transparent, I've never been active on this Discord before, so this is my first time reaching out. I'm currently working on translating all of my methodology documents into English. Once they are ready, I will be happy to share my core system prompt—the "generator" itself—for free with everyone here who is interested in testing it.

Load it into AI Studio, as it shows how many tokens are left, and you can also make many copies of the model. You don't have to communicate with the model through commands. You can just explain to it what you want.

If you have any questions, email me.

This is a consistent test of the video generator for what it can do and what it can't do, to identify as many bugs as possible. These are attempts through the prompt to control the hidden “enhancer prompt” system
I realize this may look funny, but I've been exploring texture transfer, sound transfer, object transfer via text to video model, as well as exploring the physics of objects and how they interact.

[quiet stream]

I tried the jump to feature instead of extend, and its still bad. I didnt test it a lot though, anyone else has success with it for consistency or the best way is still to write the exact same prompt, and generate it multiple times and hope that one of them will be a usable footage lol

[kindred osprey]

With Veo3, there is no point in hand writing queries. If you want consistent generation, as well as to be able to transfer your character to new locations, you need to make your own prompt generator, to which you will upload the description “assets”, and ask him to generate a character in the desired location. I have on this principle generated all the videos, I just do research and in ordinary language write what I want to get, and on the output I get prompts of the correct structure.

[kindred osprey]

On veo2 for example, you can completely disable enchancer prompt and control generation the way you want. On veo3 in Vertext Studio it is written that you can't disable automatic prompt improvement, and in essence we can only “pick it up”, but “picking it up” when it is hidden is almost impossible.

So I did first Deep Research on efficient prompting for Veo3, and for Gemini, and started just a cyclic process I let Gemini 2.5 pro write prompts for Veo3 and I described them using the dictation function. And we followed the system of “complication of textual description of an object”, and I was getting this kind of generation in the beginning.

[kindred osprey]

But when you try to complicate the textual description you're going to run into artifacts like this, where things break.

[quiet stream]

Yeah we followed similar paths then, i literally did multiple deep research then condensed it, then with trial and error crafted my own prompter. I just tell it what i want, and it delivers it with 90% accuracy, each prompt is kinda okay and usable, but i had a "long form" project in mind where details matters and i cant find a solution for that, i thought image to video or extend/jjumpt to features would be the solution but still not

also as far as i know there is no single example of a long form content that stays consistent in details right? the technology isnt there yet i guess?

[kindred osprey]

I solved this problem by introducing a new role model “editor” to whom I feed a description of a sequence of scenes as input, and he divides it into scenes and into coarsenesses, and I just move through them. Yes they are not that complicated, and I agree with what you write. the most ideal option is frames to video, where in Blender you do basic compositing and put characters make at least a block exterior.

[quiet stream]

Do you have example of a consistent long form video? where it feels like each scene has the literal same character and location and it doesnt feel like a new generation? Because in the videos you posted you can still tell they are different

[wise tulip]

Guys is there a way to request for refund on credits as Gemini 2.5 making all my scene all wrong i am having to fix 3 times each scene and i pretty much burnt 1000 points and i dont have one video fully made yet

[kindred osprey]

I can generally carry my character fairly successfully within the Fast model, even in complex scenes where my budget for generating everything is limited, because my prompt generator understands the structure of the balance of generation on the Fast model.

But it is clear that it is almost impossible to do it perfectly always, but in automatic mode I can do it, because I was able to make it a “reproducible asset”

[quiet stream]

yeah but see, i can do that too, and its like 90% accurate when you use the exact same prompt for the character and location, buts its never 100% which makes it hard to do long form. I run into the same issues with my prompts, veo3 ALWAYS changes some things even if you use the exact same method sadly

[kindred osprey]

Look like this

It's text to video, we can't demand more from it. + generation price 0.25$ for Fast model. The most ideal pipline for perfect generation is Blender + Frames to video \ Ingirients to video, but it's too expensive to do research. I spent 30000 tokens on current research. And now I have another 50000 on my balance, but that's enough for 500 veo 3 quality generations at most

[kindred osprey]

My concept is to balance bugs and model flaws and create an assistant that makes production and bugs as cheap as possible.

I agree that text to video veo3 fast \ quality is not perfect. In Vertex Studio you can use Seed for generation. They're giving away 300$ for trial generations right now.

Have you tried working with seed?

[quiet stream]

seedance? no havent tried yet because i think veo3 having audio is a plus for now

[quiet stream]

oh its something else mb, no i havent tried it

[kindred osprey]

No, there's a thing called Vertex Studio at Google Cloud, you can work through the API there, they allow you to use $300 for test generations there. And Veo3 has a parameter like Seed, something that is not in flow, it is one of the keys to return to the location, but it has to be tested.

[kindred osprey]

There are also many Veo manuals available in Vertex Studio

[quiet stream]

ohh okay nice i didnt know about it thanks! 300free cred sounds nice. its weird that flow doesnt have the seed option though if it was working, i donno why they wouldnt add it

[kindred osprey]

Because it's not the same Veo 3 that works through Vertex Stuiod as in Flow, there's a different model there. Which is designed for stream generation. And the price of working through Vertex Studio is much higher than in Flow
Well, in general, you can use it for more than just Veo. There are a lot of AI microservices of Google, you can collect a lot of things there.

[kindred osprey]

What was the product of your final research? I just had to create about 40 chats, and load them up to 700,000 - 800,000 context tokens before I could do my model.

I have 3 of them now, started with super simple versions. Now I sell them in Russia for media production.

[quiet stream]

oh okay i see, so its more expensive than flow? (flow= 4 generation of veo3 fast or 1 generation of veo3 quality for 1 dollar if im not mistaken)

[kindred osprey]

With Veo 2 you can choose the format

[fallen tree]

if you are generating in gemini thats all ya get. go to google labs flow if you want to generate in Veo 3 as far as I can tell.

[quiet stream]

i haven't created the final product yet but im working on making sure each of my characters and set look the same, I gave up on 100% authenticity because it doesnt feel feasible yet, i created some street interview with bigfoot to test and yeah they look okay and the bigfoot looks the exact same, kinda)

[fallen tree]

what is that a screenshot from?

[kindred osprey]

Can you please show me what you're working on?

[quiet stream]

i'll dm you

[devout shore]

can you put public figures somehow into veo3? Like pro athletes? What about super heroes, anime characters, etc?

[kindred osprey]

https://console.cloud.google.com/vertex-ai/studio/media/generate

Of course, here is the translation.

💰 How to Get $300 in Free Credits from Google for Veo Generation via Vertex AI Studio
Many people believe that you can only generate with Veo through Flow, but that's not actually the case. Google has a service called Google Cloud, where anyone can rent computing power and much more. A separate area within Google Cloud is Vertex AI Studio. This is essentially an AI model hub where you can do very serious work with AI.

To attract new users to their platform, they are currently offering $300 in credits to anyone who can verify a payment profile (by linking a card), which can be spent on any of their services.

Here’s the most effective way to do this so you don't waste the $300 in free credits:
What services are available to a typical user:

Imagen (photo generation)
Chirp (voice generation)
Lyria (music generation)
Veo (video generation)
What hidden details can be found in Vertex AI Studio?
It's highly likely that the Veo 'Quality' model, unlike the 'Fast' model, supports using a seed.

or those who aren't familiar with the purpose of a 'seed' in generation:
Same Seed + Same Prompt = Same Video. If you use the same seed and the same text prompt, the model will generate a nearly identical video every time. This is crucial for reproducibility and consistency.
Different Seed + Same Prompt = New Video. Changing just one digit in the seed while keeping the prompt the same will cause the model to create a completely new video, though still within the bounds of the given description.

Why is this useful?
Control and Repeatability: If you get a result you almost like, you can save the seed and make small tweaks to the prompt, knowing that the overall composition and structure will remain similar.
Creating Variations: If you don't like the result, you can simply change the seed to get a completely new set of options based on your prompt.
Character onsistency (Partially): When trying to create the same character in different scenes, using the same seed (along with a very detailed character description) can help keep their appearance more consistent.
It turns out that Vertex AI Studio contains some very good documentation for Veo, and it also clearly states, with error codes, what can and cannot be generated.
The generation prices are significantly higher, but in return, you get more control. Ultimately, even if one second costs $0.50 or $0.75, that still amounts to 50-100 controlled generations with a seed, all for free.

[quiet stream]

not sure, i tried trump it didnt generate. Also i believe google wont allow copyrighted materials to be generated
workaround: describe them with an llm and generate like that, dont use their actual names

[crisp basalt]

i.. aint readin all of that if its actually a robot wrote that

[kindred osprey]

I'm human, and this is a manual that explains how generation in Flow differs from Vertex Stuio

[crisp basalt]

no

so anyway moving on

[devout shore]

got cha thanks. When you say llm do you mean like put it in whisk/chatgpt/gemini etc and get their description?

[quiet stream]

llm=large language models like chatgpt gemini etc (gemini probably best because trained on google data, and just a vibe feeling)

but yeah basically send them the image and ask then to describe

[crisp basalt]

gemini is actually more recommendable to use when it comes to iterating your prompt. They even use Gemini model to process user prompt in Flow.

[kindred osprey]

Don't forget to set the Media Resolution parameter to Medium, because it is directly responsible for the quality of analysis details

and also set thinking budget to maximum ( default 8000 )

[wise tulip]

my first video

[upper dove]

I am having a hard time with the audio working. Any suggestions?

[fluid rock]

noob here; I started my free montly trial for flow yesterday, but it says i dont have credits, I click the uprgrade button, it takes me to a new window, and says I am already subscibed? what is going on? how do i fix this? anyone with the same issue?

[south bough]

How many generations on each model do you get if you buy the $20 plan

[gilded ridge]

you will have 1000 credits and each generation will cost 20credits for fast and 100 credits for quality, so around 10-50 videos

[south bough]

Per month or day

[gilded ridge]

1000credits pe rmonth

[south bough]

oh

[spare geyser]

浩瀚的海洋上 升起一轮明月

[gilded ridge]

you can top up AI credits now

as a beginner, I managed to spent all 1k credits to produce only one 1 min video

[minor vortex]

Any idea why my Veo3 videos aren't being uploaded as Shorts? I'm trying to upload a 7 sec video as a short and Youtube just automatically puts it as a longform video instead of shorts. I've tried exporting it with different resolutions and even asking Veo3 to change the resolution and nothing works. What the hell am I doing wrong?

[slate mural]

upload through youtube mobile app

[minor vortex]

Looks like it's too big and it wont show the whole video. What resolution do I need it to be for the mobile app?

[charred sail]

Take a frame from the scene and use whisk to generate a background screen. Prompt the scene with a vibrant green background. Eliminates all the random set noise.

[charred sail]

That's actually pretty good for a beginner 😛

[slow pawn]

when flow available in india? i want to use image to video

[vast gust]

Gemini AI

[frank jay]

Besides more credits, what is the difference between AI Ultra and AI Pro for Flow?

[forest raptor]

Can someone make a comparison between veo3 fast and pro?

[forest raptor]

the animation is horrible

[buoyant wadi]

i wouldn't mind the bugs and problems if you weren't charging me $250

[vast gust]

Gemini AI Veo3 only allows 3 video creations per day! Unless you want to Remix my Video creation! Here is the prompt: 1.3 mechanical pencil with HB charcoal lead and round point, bold Hane overlapping texture dark vertical brush strokes (An upflick stroke, where the brush is lifted at the end of a horizontal or vertical stroke to create a sharp, upward flick), Gekiga anime style with expressive line work, right side view, close up perspective view point of a black cat with a full swing of his paw with intricate detailed, Slaps a kawaii anime girl in the face, on her right cheek by his left paw with intricate detailed. High speed lines of his full swing slap on her right cheek, slow motion exposure of her being slapped by the cat paw, Kinetic force from the massive impact of the full swing slap from the cat paw with intricate detailed. The kawaii anime girl is shock by the slap as her head turns right from the slap by the cat's slap with intricate detailed. No pencil in picture.

[forest raptor]

a-are you a bot

[minor rune]

Hey, do you know if end frame is possible with audio?

[quiet stream]

Not sure I follow, if the goal is to have consistent character and set, how would eliminating the set help?

[jovial aurora]

How do I create audio with new veo3 update which option do I select?

[charred sail]

You can take care of the consistent set with greenscreen is what i'm saying

[charred sail]

1 character is easier to get a consistent result with than a character and a whole set.

but it really depends on what your trying to do.. how complicated, obviously anything the character interacts with will need to be there

[charred sail]

you don't select anything.. you just prompt what you want

[jovial aurora]

Thank you very much

[slate mural]

has anyone made anything with new veo updates?

[quiet stream]

Yeah that's my issue mostly, i guess if it's a static shot the background can remain a green screen but anything involving movement during the scene wont really work

[tawdry fox]

Hey everyone 👋

I’ve just released a cinematic AI short trailer made entirely with Google Veo 3. It’s called "The Last Memory Upload" — a futuristic emotional story about memory, grief, and a lost connection.

In the story, after years apart, Daniel discovers that his former lover Emily — who recently passed — left behind something unexpected: a connected Google account filled with their shared memories.
(Spelled "accouqt" intentionally as a respectful creative workaround — this is a tribute, not brand misuse.)

This is a personal passion project made under my brand NeonVision AI to explore the emotional side of what Google’s future tools could represent in storytelling.

Would love for you to watch it and share your thoughts:
👉 https://x.com/Neonvisionai/status/1942252268025196611

Thank you Google for building tools that allow creators like me to imagine futures like this.

NeonVision AI (@Neonvisionai)

What if Google became the gatekeeper of our memories… after we die?
This emotional sci-fi trailer explores a near-future where your memories live on — forever.
Crafted with AI. Powered by story.
🎥 Watch till the end — it might just stay with you.
@Google @GoogleAI

X · Jul 7, 2025, 4:00 PM

[ebon ibex]

no more thread? 😦

[vernal kernel]

Tried to be sneaky european and get access to flow:

• bought a prepaid sim
• put it in my old phone
• set vpn to usa
• setup a new google account
• go to g2a and buy google play us gift card
• try to buy ai pro sub from google
• "send more information about your gift card"
• punch the drywall
• 30€ poorer and no access to flow 😤

[stable lake]

How to access flow in India

[glad cape]

why don't work flow? 😦

[kindred osprey]

USA VPN

[glad cape]

I use

[kindred osprey]

Maybe this VPN and their entire subnet flew into a Google ban when people were endlessly creating student accounts. It's best to get your outline vpn up on digital ocean

[kindred osprey]

Ivpn works for sure.

[kindred osprey]

I rented a server in the US for 5$ a month and I have my own dedicated perfect reputation everywhere, I get beta tests on my Ultra like I'm American. Well on the other hand I am a very high quality promptsand do contextual learning, i.e. I am their target client.

[glad cape]

that's good

[kindred osprey]

I can let you check your key on IVPN, but only for once)

[glad cape]

No thanks, I'll buy a proxy right now.

[glad cape]

Does the trial subscription allow you to use flow?

[kindred osprey]

Yeah, but you need a quality VPN

[glad cape]

I don't know weird everything, I use a proxy and it doesn't even display the student version and flow with subscription still doesn't work))

[kindred osprey]

Was it your personal account? The one you used before?

[glad cape]

Yes

[glad cape]

I didn't buy mail with a US region, I just did a trial subscription on my EU account

[kindred osprey]

The previous account is already strictly geo-linked, so if you would like to use veo3 and flow, you need to create a new account with US geo-linking + link a virtual phone number.

[kindred osprey]

Flow is only available in the USA

[glad cape]

Yeah, I thought it would work that way, but apparently not)

[kindred osprey]

Yes, after the epic with student accounts, when people were creating hundreds of them, they have very much tightened the rules for creating accounts and linking cards for flow

[kindred osprey]

In Russia, hundreds of people created hundreds of accounts for themselves

[kindred osprey]

A lot more accounts are sold on funpay, there are stuffed with stolen AI Ultra subscription cards, and sell them without warranty for 30-40$. But they live as long as you're lucky.

[glad cape]

hahahaha lol

[glad cape]

You speak russian?

[kindred osprey]

Да

[glad cape]

Бл зачем я переводил авхвхахавхав

[kindred osprey]

@vmind_ai в тг введи у них в комментариях чат там самые адекватные нейронщики сидят

Хороший платный чат у Руслана Нейро там 450 человек тоже сидит тех кто на veo3 генерирует 1600р в месяц

[glad cape]

Понял, да я не занимаюсь этим профессионально. Просто попробовать что и как

[kindred osprey]

У меня на канале @sharipovnikita очень много полезных материалов по Veo там и обученные модели. И архитектура движка, ознакомься полезно будет

https://lively-chive-e7f.notion.site/Veo-3-228a2ada72fc80079c2cc38f2ca60bbc?pvs=74

Миро карта
https://miro.com/app/board/uXjVIkgPnAc=/?share_link_id=491852174070

[glad cape]

Понял

@kindred osprey мне кажеться отключили подписку для студентов, купил почту USA и прокси но не появляется чтобы попробовать

[jovial vessel]

I'm cooked?

[frigid pier]

Is it really true that Veo 3 now supports i2v audio!?

[pallid kite]

New release! https://www.youtube.com/watch?v=D0EIxRJcIo4

[frigid pier]

but Do Veo3 fast support the feature too?

[jovial vessel]

yes bro

[wise tulip]

Hey guys any idea why some videos are coming out perfect quality some very poor and blurry using veo 3 fast

this one not so good

this one is perfect

also if someone can help why is my charachter keep chaing like if you look the first video it has big horns if you look last one has smaller horns like why i am using eactly same main prompt that has my character not chaing at all

[dreamy whale]

You know how to use it? Because i see the info about this feature, but not see changes in interface at all

[pale copper]

How do I use lip sync on flow?

[dreamy whale]

Can anyone post a screenshot with the outlined interface elements that are used to convert a frame with audio into video in veo 3?

[stuck jewel]

how did you solve this problem?

[glad cape]

no

[sterile thistle]

Hey i want to add exactly my product image to a video or a cgi
Purpose is to create an advertisement but i cannot get no matter i tried veo or gemini give their own image created for the product
Can anyone please help me?

[spiral shoal]

Yes I can

[sterile thistle]

Can you please tell me how and which tools or techniques to use ?

[pulsar lichen]

is this true?
https://www.reddit.com/r/Bard/comments/1lurxeu/you_can_now_upload_images_into_veo_3/

[pallid root]

sick of everything always messing up.... i can put the same prompt in 3 times and get wildly different results each time.. and the promp is 2 pages long and very detailed... i just canceled the ultra membership until this is smoothed out. Tired of overpaying to alpha test the product. feels very shady in how "oh it didnt render audio.. or its unusable.. sorry still lost your credits... " i smell a class action coming lol

[crisp basalt]

cool 👌

[crisp basalt]

yes

[pulsar lichen]

Thanks is it avialble for veo3 or veo-3-fast

Or both

[crisp basalt]

yes

[pulsar lichen]

Alright thanks

Twitter will flood once again

[crisp basalt]

its already flooded i think

[pulsar lichen]

Yeah I mean now that i2v is supported

[crisp basalt]

i2v on veo 3 has been a thing since day 1 launch

[pulsar lichen]

Then what's announced

[crisp basalt]

this

[slow pawn]

i subscribed google ai pro but in flow i don't have credits

[crisp basalt]

means your payment profile is not in the supported country

[pulsar lichen]

I am referring to this reddit post where he is saying like its completly new

[crisp basalt]

the only new thing is generating speech using veo 3 imgtovideo

[slow pawn]

yes it's but any way i can use ? i can't use img to video in gemini ai app

[crisp basalt]

and veo 3 fast mode imgtovideo

[crisp basalt]

create new google acc , create payment profile under supported countries

[pulsar lichen]

Got it thanks!

[slow pawn]

i'll try that thanks

[crisp basalt]

first check your payment profile , wht country it is set to
https://payments.google.com/gp/w/u/0/home/settings

[slow pawn]

i can swtich without any issue.Right?

[crisp basalt]

Also, whoever made the decision to move the Flow FAQ page to support.google.com, thank you and you rocks . It’s much clearer and more informative now

[crisp basalt]

i cant really say yes or no to that.. but, i recommend u to create new google account (im speaking from experiences)

[flint summit]

Its part of "Frames to Video" now for Veo3 right? I didn't see an "Image to Video".

[crisp basalt]

yeah

[slow pawn]

got it brother thanks

[crisp basalt]

Come to think of it, I remember the UK version of the Flow website didn’t use the term “frames to video” it was labeled as “image to video” instead.

[dreamy whale]

Have you create any img2video with custom audio?

[crisp basalt]

yeah

[dreamy whale]

I don't see the button. than add the audio

[crisp basalt]

what do you mean custom audio? as if like,,, adding our own audio ? or

[dreamy whale]

yea, own audio

[crisp basalt]

Veo 3 model generates the audio itself based on your prompt.

not from user uploaded audio

[dreamy whale]

I saw in it in this video

[crisp basalt]

?

what

[dreamy whale]

i saw in this video

[crisp basalt]

which part..?

Sorry im confused, are u saying you saw theres a button to add audio in that video you sent?

If you want to generate video with audio just pick Veo 3 model

[dreamy whale]

in video i saw the oportunity to do this, but i dont know HOW to do it and try to figure it out in this tread

[crisp basalt]

which part?

[dreamy whale]

i wanna generate video with my own voice in veo 3, like dude in the video

[crisp basalt]

thats his real voice?

[dreamy whale]

I think so

[crisp basalt]

what

[dreamy whale]

i think that he use his own audio

[crisp basalt]

nah

he jsut use frame to video with veo 3

[dry frost]

then more than likely scripted in some dialogue for each clip

[dreamy whale]

okey, i understood, thanks

[crisp basalt]

yeah, that too

[quiet stream]

Did you? so what that tiktok was fake then, he just added his own voice in post prod and not in veo3 which lyp synced?

Having veo3 lypsync with your custom audio would be great

[spiral shoal]

Crazy dude...

[zealous flax]

I am reviewing my setup to start with Flow. Is the number of users determined by flow separately from the number of users that I have in my Google Suite? For example, I have an assistant who is helping with Notebook LM (currently, that is 2 users accessing Notebook LM), but it would not need her to access Google Flow. Thanks for helping me understand the user structure.

[crisp basalt]

Is it workspace google account?

i think you should be fine, it doesnt check how many users in one account, it mostly checks.. ai credits if that makes sense to you

[crisp basalt]

https://vids.google.com/ i think this is the one that google suite/workspace use when it comes to video stuff

[worldly kindle]

Is not working add scene to crate a longer video right?

[spring wave]

Any idea why I wouldnt be able to use ingredients on pro plan?

[brazen basin]

The process is with the scene extension tool, first you activate the veo 3 engine, add your image, and your prompt with dialogue included, then you add that scene to the scene builder, and use the extension tool with veo 3 fast. Flow remembers important details about the scene to generate consistent voices.

[spring wave]

I think he misrepresented the update...

made it seem like you can upload audio to a first frame video gen

[slate mural]

Can you show us

[spring wave]

I am currently doing this outside of veo with the current workflow: first frame animation using veo 2 > generate audio (using eleven labs)> lipsyncing via Dreamface

Important note: use prompts like "actor is talking" to generate the animation of them talking for the lip sync video input

[crisp basalt]

this kinda makes it seem like u were the one asking, not the person who originally posted 😆

[onyx bolt]

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

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

[crisp basalt]

Sorry, your credit balance is insufficient. Please top up by visiting: https://labs.google/fx/tools/flow/

[main saddle]

it does 😂 just wanted to make sure their message got some visibility!

[broken tapir]

Image a Angel of 7 heads

[main saddle]

please see the latest Flow updates here!

[rough shell]

Period

[rain crow]

germany support for flow when 🥹🥹🥹

[austere timber]

Is there an app to use for flow? I would love one if not

[digital yoke]

Anyone Sweden that can send recommendations link Google ai Pro for free month true friend link?

[patent monolith]

When do we get more than 8s per clip? Continuity is a pain at 8s limit to make anything 30-60s

[tiny patio]

The previous ImageFX app became FX with this symbol and works for me.

[austere timber]

Ugh thank you so much!!!

[wintry phoenix]

https://tenor.com/view/i'm-so-excited-gif-gif-9130232280341803057

Well, where 's the link!!

[somber pebble]

"You can now make your images talk with Veo 3"

I can't do any frames to video with Veo3. It forces me to use Veo2. Is this not actually out?

[patent monolith]

Really though, continuity between scenes is terrible. How is this not a thing yet? Either continuity of characters or 30-60s of video?

[little hull]

This is crazy what flow can do now. I test it on one clip and im in shock.

[eternal sparrow]

I know right ! Once this is fixed I'll be on my way to partnering with Netflix !

[quiet stream]

Can veo3 generate small animated gifs or logos (so not full screen basically) anyone tried?

[onyx meteor]

How do I delete my Flow account? My plan is coming to an end and I'm not intending to renew. Will all prompts and videos be auto deleted or would they still be there if I rejoined at a later date? Various projects have just disappeared over the last few weeks and one is still visible but fails to delete. Also my library won't delete. I don't want to leave random data lying around so would rather make sure it's a clean break when my plan ends.

[charred sail]

Nah, you'll be another e-book author on amazon. 1 voice among 100's of millions. If you don't have millions for marketing you're still going to be making tiktok videos...

this changes the game, but it changes it for everyone...

Hey do you still get all the other apps with Pro? Whisk? MusicFX?

hmm.. it looks like you're paying an extra $105/mo with ultra just for Project Mariner, Youtube Premium and 28 more TB of storage..

guess i need to change my sub 😛

I mean I'll still buy that in ai credits LOL

[hushed geyser]

Hi! I'd love to try Flow for generating ASMR-style AI videos (like crystal fruit cutting). I'm really interested in visual storytelling and would love to test it. Is it possible to get access? Thanks a lot!

[charred sail]

user error 😛

my bad

when you mean to type rear window and type reareview mirror instead.. that's when it happens..

[hushed geyser]

what should I do ?

[charred sail]

It sounds like you're excited about the creative possibilities of AI video generation, especially for ASMR! That's a really interesting application.

Flow is Google's AI filmmaking tool, and it's designed to work with their advanced models like Veo, Imagen, and Gemini. To get access to Flow, you generally need to be a Google AI Pro or Google AI Ultra subscriber.

Here's a breakdown of what that means for access:

Google AI Pro: This subscription gives you access to the full Flow experience, including models like Veo 2 and the latest Veo 3. You get a certain amount of credits, and you can purchase more if needed.

Google AI Ultra: This is the highest tier and includes everything in Google AI Pro, plus more credits and first access to experimental models and advanced features like "Ingredients to Video."

Google AI Ultra for Business: This is for qualifying Workspace plans and provides the highest level of AI features and tools.

To get started:

Check Google AI Plans: You'll want to visit the Google AI Plans page (you can search for "Google AI Plans" to find it) to learn more about the subscription options and what they include.

Ensure Supported Region: Flow is not available in all countries, so make sure your region is supported before subscribing. VPNs will not enable access in unsupported regions.

Subscribe: Once you've chosen a plan, you'll need to subscribe to Google AI Pro or Ultra.

Access Flow: After subscribing, you can typically access Flow through Google Labs at https://labs.google/flow/.

Many tutorials online demonstrate how to use Flow and Veo 3 to create ASMR-style videos, including the crystal fruit cutting you mentioned. These tutorials often provide prompt templates and step-by-step guides, which could be very helpful for your visual storytelling endeavors.

It's a fantastic tool for creative experimentation, and I wish you the best in generating some mesmerizing ASMR AI videos!

[charred sail]

Always ask AI first

[hushed geyser]

well I'm Ai PRO

[charred sail]

Why ya asking here? Gemini it

😛

[hushed geyser]

ok Bro
easy easy

[charred sail]

just pointing it out. AI usually has better answers LOL

[silk fulcrum]

Did

[charred sail]

sent you an ASMR prompt

[royal ember]

/flow prompt: https://cdn.discordapp.com/attachments/1337164214631927872/1392272307170443364/WhatsApp_Image_2025-07-08_at_13.17.47.jpeg?ex=686eee21&is=686d9ca1&hm=3709bfd4be9038a1f2d719a47ad9faf7c26baefbf693568c9fcc0097805e7fa2& A serene young woman meditating, begins to slowly levitate and float upward gracefully, cinematic lighting, ethereal atmosphere, soft particle glow around her, peaceful, slow motion, high detail --ar 16:9

[charred sail]

Is there a Project Mariner discord anywhere?

I'm not seeing one on the site

[flint summit]

VQ-GAN+Clip meets Veo3.

[wise tulip]

This is frame to video I used whisk combined images and then used flow

[little hull]

If you have issue of not having sound on your video make sure to run the prompt again if anyone has issue. I just had that issue ran it twice.

[mental mulch]

Also when is Veo 3 gonna be added to Google AI Studio??

[modest root]

@charred sail Hi there bro can you send me also some idea for ASMR, thank you!.

[flint summit]

Also make sure there are no references to children in your prompt. "The boy says: " will kill your dialog. "The young man says:" will fix it.

[livid egret]

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

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

[charred sail]

I usually use my daily free ones for establishing a character prompt

Once I see it at a 50%+ rate I'll incorporate it to my flow projects

[dry frost]

Yeah the three free gens via Gemini are fantastic for "trial" runs

[charred sail]

I highly recommend https://promptalchemist.com/app

it will store character prompts for you

still in beta

The guy that makes it is around here somewhere

[dry frost]

I'm actually using a similar format and its succeeded in creating what I wanted 90% or more.

[charred sail]

that's excellent

[dry frost]

I'll assemble the prompt in notepad before as well. (mainly to prevent accidently pressing enter on the flow page)

[charred sail]

add a microphone to the character.. drop it to 40% lol

oh my i hate the accidentals

they need a confirmation 😛

[dry frost]

Oh I figure audio will be one of the last things I dabble with, but the generated sound effects seem to always be fantastic

[charred sail]

nah just for the character

i have a standup character.. the microphone is erratic at best

sometimes all black.. sometimes black and silver.. sometimes it has a cable.. sometimes it doesn't... i tried it without the microphone.. it stuck it in his pocket..

i've tried prompting it in great detail and removing it.. lol

[hidden sage]

How can I generate image to video by adding audio? Asking about this newly launched feature?

[charred sail]

drives me crazy

[woeful fossil]

Ayyy, love the shout out. Thanks man 🙂 im glad its working for you. Its hard for me to jump in and recommend it directly with self promo rules.

[charred sail]

I got ya. It's a great tool

[hidden sage]

Can someone please help how can i upload the audio? Thanks!

[charred sail]

you don't upload audio

it creates the audio

you provide the image

[hidden sage]

Oh so its not like heygen where you can use your image and audio?

[charred sail]

no

you can in post production

[hidden sage]

I am talking about the latest update rolled out couple of hours ago

[charred sail]

where does it say you can use your audio?

[hidden sage]

My bad mate! Appreciate your help. When you said I can still do in post production does that mean that I just slap my audio on the generated clip while editing?

[charred sail]

yep, just try to sync it up

[hidden sage]

Cool Thanks!

[charred sail]

would be cool if you could upload a specific voice as an ingredient.. mabye veo 4...

[hidden sage]

Yea for sure! I got overexcited when I saw the update and upgraded lol

[hidden sage]

When you add the last frame from generated video it switch to veo2 fast instead of veo3 by default?

[charred sail]

I think they said they add 1st frame not last.. Could be wrong

i might be wrong on that.. haven't played with it yet

credits renew tomorrow 😛 hopefully...

[charred sail]

does it let you switch back to veo 3?

[hidden sage]

No it actually worked, you add the last frame as a first frame, yes it did let me use veo3

[charred sail]

cool

that's awesome

now try dialog in both clips and let me know

if i can carry a conversation by extending with last clip i'll be in heaven

[hidden sage]

it changes the voice as expected, I was expecting to make an avatar as heygen sucks but seems like id have to jump to heygen for now

it does carry the same conversation

[charred sail]

try using an example for the voice

His voice is blah blah blah, like Patrick Stewart,

[hidden sage]

it wont matter to me as I want my voice so as you mentioned earlier I would have to stitch it later with the clip

[charred sail]

ah

I got my voices tied down pretty well

can't wait to hit it tomorrow

[hidden sage]

Yep! Im a creator with half a million subs lol wanted to preserve my energies and spend somehwere without getting caught by audience lol

[charred sail]

nice.. i'm about to hit 1000 followers 😛

[mystic crypt]

Hey guys, for the image to talk, do we just use image to text?

[hidden sage]

Not bad sir! Tbh i was more excited when i got my first 10K then 200-300-400K lol

[charred sail]

i would be too

[mystic crypt]

@hidden sage do you do the yeti videos?

[charred sail]

my preference is to do standup comedy style video

hard for me to find a big crowd doing it part time.. this way i can just create my own crowd 😛

[hidden sage]

Not exactly my focus is more on realworld usecases than slop haha

[charred sail]

lol

[charred sail]

Yeti's are serious political satire, man

😛

[hidden sage]

Hahaha yes sir!

btw i was able to generate a minute long clip using veo 3

[charred sail]

just extending?

[hidden sage]

no, add the last frame to first frame...

[charred sail]

yeah, that's what i mean.. just keep doing it

[dusty warren]

When is flow launching globally ?

[charred sail]

that's awesome man.. excited!

[hidden sage]

Gotta talk to Mr.Pichai

[dusty warren]

Wasn't it leaked ?

That veo3 going to go global soon

[hidden sage]

Im a man of ethics doesnt believe in leaks lol 🤣

[charred sail]

my sasquatch videos are gonna be so much better now

[livid egret]

My dudes, this looks like a fantastic tool.

[queen temple]

Hi, why when I try to expand a video veo 3 automatically swaps to veo 2 without audio? Am I do something wrong?

[livid egret]

How'd you accomplish that?

[queen temple]

How?

[mystic crypt]

Was this only with images?

[marsh wasp]

None of the videos I generate have sound, I have a pro plan

It's gone from being funny to bothering me too much.

[charred sail]

what are you trying to do?

try adding - ambient sounds and see what that gets you

[dry frost]

Also are you using the proper model

[charred sail]

make sure it's veo3 fast

only 9000 more to go!

[spice moss]

Are there any prompts that can speed up video playback?

[woeful fossil]

Thanks, if you have any questions or feedback feel free to message me!

[gilded swift]

is it possible to generate in 1080p with flow?

[charred sail]

no but you upscale when you download it

I just started using kdenlive for editing. The rendering is kinda poor imho, but they have a 16:9 format export you can do. If anyone is looking for a free option.

if i can't get the render better i'll be sticking with capcut

although i don't think the capcut 16:9 renders look all that good either.. might have to do adobe

[livid egret]

Thanks, if you have any questions or

[frigid pier]

It seems like Veo3 fast and quality do not have that difference.

[white forge]

Have a nice day.

[glad arrow]

Any info if flow gonna be launch in this month for Indonesia region?

I really need consistency and extending for my video.

Accidentally uses vertex from GCP and ended up charged so much just for 3 video

The worst part is, those video doesn't even fullfill my needs.

[charred sail]

Audio seems better on quality.

[frigid pier]

hmmm i mean is it that much different?

[charred sail]

I'd use fast

[rose crystal]

is there a way to enhance the visual quality of the generated videos?

[bitter kraken]

is anyone having a problem accessing Flow after subscribing?

[charred sail]

how long is it taking monthly credits to show up these days?

[spring ocean]

Can I use flow i Germany when I do the paid version?

[lunar veldt]

ANyone else having this issue? I've searched but can't seem to find any info on it? I'm in the UK. "We do not allow uploads of photorealistic people in your country", yet the woman on the right, uploaded, seemed to be ok.

[digital yoke]

Can some send friend invite link that live in Sweden for free 4 month Gemini advance? Refence link?

[bright patrol]

Hi guys, is it possible to have a video output that is more than 15 secs? Is there a settings for it?

[tropic wigeon]

Hello

I recently joined the AI Ultra plan, but I haven’t received any credits in my account.

How you fix this bug bro?

[charred sail]

Go to your google 1 account management page and hit the help button.

They didn't really help me though 😛 just said it's processing

[latent ocean]

The problem is that I have subscribed for a free trial period. When I enter the Flow website, it asks me to subscribe. What is the solution?

[stiff harness]

I dont think there is a free trial for ultra

Maybe you just got google one

Ultra promotion is 125eur instead of 250 for 3 months

[white forge]

It's quitely long periods, bug that still not resolved yet by google.

[atomic rose]

anyone got that github prompting guide

please

[quiet stream]

for image or frame to video, do you still need to describe your character in detail (like for text to video) for best results? anyone tested it?

[bright patrol]

No help since @wet copper

[unreal viper]

hi, this is a <@&1173646484311384114> issue, or you could put it in #1319424187730755636 for feedback for the team :)

[bright patrol]

Okay thank you for that

[queen temple]

Did anyone figured how to expand the video veo 3 again with veo 3?

[dusky otter]

any word on being able to render vertical videos with veo 2/3?

[queen temple]

I did specify on the prompt, and most times worked

[alpine willow]

Anyone else getting this on extend?

[wintry oasis]

Does anyone know how to create a new video with the same character and style?

[steep wren]

" You can now make your images talk with Veo 3
Veo 3 first-frame to video now supports speech. Upload a picture of your character and give them a voice. Note that audio is still a beta feature, so your videos might not always return sound.

Reminder: By using this feature, you confirm that you have the necessary rights to any content that you upload. Do not generate content that infringes on others’ intellectual property or privacy rights."

Any idea when this will be rolled out to everyone?

[woeful fossil]

This is available now as long as you have access to Flow.

[woeful fossil]

Image to Video is likely to be your best bet as long as you can get consistent character images. Otherwise, you'll need to use text to video with highly detailed prompting.

[sterile fable]

what do you guys think? Bigfoot got jumped

https://youtu.be/TmNxjplyjuw

[velvet arch]

How do I use it?

[sterile fable]

use what?

[velvet arch]

How to use something like Veo 3 to make video

[lunar veldt]

Not for me. I mean, it won't allow me to upload any images of photorealistic characters to animate (even ones completely fictionally generated)

[steep wren]

Not for me.

I have Ultra on top of that, literally highest tier...

[steep wren]

I thought you could feed a character image into Frames to Video, and from there it would automatically carry over into the Veo 3 videos.

Is that the correct approach?

If not, then my best option might be to base it on a well-known historical character. For example, if I choose Columbus, the AI already seems to have a standardized visual for him, so the output tends to stay consistent, right?

Any tips would be appreciated.

[glass tapir]

why flow aint working for me i have the right subscribtion and im in the us

[kindred osprey]

@fierce valve @main saddle @vale lily @broken bane

First and foremost, thank you for creating such a vibrant and valuable community. It’s an incredible resource for all of us exploring the potential of Veo.

I’m writing to propose the creation of a new, dedicated channel for a specific and growing niche within our community: prompt engineering and the development of prompt-generation tools.

While the server has excellent spaces for sharing prompts and results, there isn't a focused area for those of us working on the meta-level—building and training custom language models (like our own GPTs) to automate and perfect the prompting process itself.

I've personally connected with about ten other members here who are actively developing their own prompt generators. We are all looking for a hub to share advanced techniques, troubleshoot our models, and collaborate on this specialized craft.

A channel like #prompt-engineering or #tool-development could become a center for expert discussion, pushing the boundaries of what's possible with Veo. It would also be an invaluable resource for all users who wish to learn more about the deep science and art behind crafting the perfect prompt.

Would you be open to considering this? We are confident it would add another powerful layer of depth and expertise to the Flow community.

Thank you for your time and consideration.

Like this

I just fundamentally study video generator bugs and write an automated prompting system for them. who understands how the automatic enhancing system works and writes prompts according to its rules. Yeah, and to be honest there's not enough channel on Veo3 for those working on the Vertex Studio API, it could be the same place.

[twin mortar]

Is it me or is the quality of the generation massively dropping this week? The audio is dreadful now and I'm getting captions on pretty much every generation. Anyone else having similar issues?

[steep wren]

Any tips for voice consistency?

[main saddle]

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

[woeful fossil]

Hey Georgia, your link is to lab announcements not flow announcements FYI

[main saddle]

I clearly haven't had my coffee yet thank you!!

[steep wren]

In Frames to Video, can you put two frames, where one is your background where the scene takes place, and the rest is the main character?

[steep wren]

Good eyes, my friend.

[woeful fossil]

So in frame to image it will literally use whatever image you add as the first frame for your video. If you want to feed characters and scenes as a guide, that would be in Ingredients to video in Flow as long as you have Ultra.

[steep wren]

Thank you!

Has Veo 3 been added to Ingridients for you?

So I guess the workaround is to have the character + the background that you want in the same image, to be used as first frame, to maintain constistency.

[woeful fossil]

I don't believe it uses Veo 3 yet. Image to video will be as good as you can keep your images/scenes consistent. The way I try to handle consistency is with highly detailed prompts and making distinct choices for the characters to make them feel consistent. You're always going to have an uncanny valley effect with human subjects. But making sure they're wearing consistent clothing and other key identifiers will help.

[steep wren]

How do you begin the promp, any tips? If you have a monkey wearing a jacket, do you put that into the prompt, or just rely that Veo 3 will pick it up?

Or do you create your characters based first, then build different scenes via prompting, like images, and then you prompt it to create a fitting video?

[woeful fossil]

I generate a character description based on an initial idea and then tweak that to fit exactly what I want, and use that exact description every time I use that character.

[ember pasture]

Anyone was able to get consistent voices with image 2 video ?

[icy pumice]

I'm only seeing one output for some of my Veo 3 generations even though I'm selecting "2 outputs per prompt" in the settings. Is this a bug or some issue on my end? Does it count still towards my credits?

[woeful fossil]

Al Capuchino is a savage and just ate this cigar. But here's an example of the character consistency with just detailed prompting.
https://promptalchemist.com/share/FQEbVwCJXUjnojEAvqkr

https://promptalchemist.com/share/b9XF1fX7CSeKe44DV4ht

[marble hamlet]

Can anybody tell me why this image violates policies? Flow won't let me use it for a frames to video. Please tell me it's not because she's not an adult.

[steep wren]

The ingridents make me crop every image I attach. What size does the crop default to? Which is the biggest size we can use?

[kindred osprey]

@hardy halo please

[steep wren]

Are you adding just 1 frame, or multiple frames/

[marble hamlet]

What do you mean multiple frames? I just wanted to create a video clip from my image.

[gaunt horizon]

Hi everyone! I'm trying to understand what the difference is between features text to video, frames to video, and ingredients to video. Is there a page I can check out to understand the diff? I'm having a hard time understanding what feature is useful for me in the SceneBuilder.

[crystal grove]

Sure. Or you've got it?

[steep wren]

I was asking for some.

[crystal grove]

Ok has the person got it done?

[small tide]

Hi there! Can you share the prompt you were using when you experience this issue? Thanks!

Hi Daryl! Can you share the prompt where you expriencing this issue? Feel free to DM it to me directly if you prefer

Hi Staff - Can you share the prompt you used here, and confirm it was with Veo 3? Appreciate it!

[vernal kernel]

This one for example:

"A realistic black sable ferret stands upright on a small wooden stage, holding a microphone in a shady underground comedy club with brick walls, a flickering neon "Open Mic" sign, and scattered empty chairs. Dim moody lighting with a single spotlight focused on the ferret. Audience is barely visible in shadows. Camera: eye-level medium close-up, slightly shaky handheld to simulate live performance. Audio: Ferret says nervously into the mic, "They told me to break a leg... I bit someone instead."

No subtitles."

And this another one:

"A realistic black sable ferret stands upright on a small wooden stage, holding a microphone in a shady underground comedy club with brick walls, a flickering neon "Open Mic" sign, and scattered empty chairs. Dim moody lighting with a single spotlight focused on the ferret. Audience is barely visible in shadows. Camera: eye-level medium close-up, slightly shaky handheld to simulate live performance. Audio: Ferret says nervously into the mic, "You ever see a ferret try to run on tile? It’s like watching a Roomba on cocaine."

(crickets).

No subtitles, no laugh track."

[jagged jackal]

I'm liking the reference image thing in Veo 3, but unless I generate an image inside Veo 3. It tells me it's a prominent person and not allowed, when its an Ai generated image on another site.

[pale copper]

Anyone know how to fix prominent people error? On img2video

[charred sail]

I see Text to Video, Frames to video and Ingredients to video. No image to video

[pale copper]

Frames to video*

[charred sail]

seems like a weird error

didn't it generate the frame?

oh you're uploading it..

yeah, it's kind of sensitive

probably just need to give feedback and wait

[pale copper]

It doesn’t let me upload my image because apparently it’s classified as “prominent person” even though my character were generated on gpt

[charred sail]

is it meant to look like someone?

or just a random person?

[pale copper]

Nope created it myself

[charred sail]

weird

[pale copper]

What do you know it finally uploaded

[charred sail]

try putting it through whisk?

cool

[jagged jackal]

Every image I've tried from another Ai site has given me the same error. Well, any image that looks like a person

[woeful fossil]

A whisk image worked very well for me, as well as still frames from Veo 3 generations.

[jagged jackal]

Yeah, but the whisk image is comeplety different from my image and 1/2 the resolution, and honestly, a lot uglier too 🤣

[nimble portal]

What prompt did you use?

[mighty bison]

Do ya'll know about the clapback videos?

[limpid trout]

Hello, I hope you don't mind if my English sounds a bit awkward.
I have a question about Google Flow.

I'm currently using the Ultra version, and as far as I know, it supports videos up to 60 seconds long.
Should I include a phrase like [Generate a video approximately 60 seconds long] in the prompt to get the full length? Or is there another way to specify it?

I tried requesting "Generate a video approximately 20 seconds long," but the result was only about 8 seconds.

[woeful fossil]

8 seconds is the max you can do per generation, you will need to make more clips and add them together.

[limpid trout]

Ah, I see! That makes sense now! Thank you so much

[ripe hornet]

ok image to video is pretty dope. wish we could upload audio, but otherwise super fun

[undone pilot]

When I first started using Flow, all my videos' had audio now every last one fails to create audio

[jagged jackal]

Make sure your not on Veo 2. Ive done that before.

[undone pilot]

I did on a few but these are all on Veo 3

even has the mouth moving lol

[civic hazel]

I am paying ultra but veo 3 is not avaiable.. but when I was paying, it said it was avaiable..

so brazil still cant get veo 3? is there a way to use veo 3?

[silk hill]

where do i see when my veo monthly credits reset

i think i saw somewhere the lower price tiers now allow you to buy credit packs for veo 3?

[deft atlas]

Hi guys

I’m new

And here for stock notifications

[silk hill]

are top up credits for pro subscribers the same price as for ultra?

[north silo]

They are same. I have both an ultra and a pro account, and the top-up pages of the two accounts are exactly the same.

[silk hill]

so whats the advantage of keeping the much more expensive account

[north silo]

Ultra has more additional services. Also, the credits given with an Ultra subscription (the points for subscription, not the credits purchased additionally on the top - up page) are cheaper. If you need a large amount of video generation, Ultra is actually more cost - effective.

[silk hill]

well, will the credits be cheaper for ultra in august when the cost goes up for me from 124.99 a month to 249.99 a month?

like i already pay for youtube premium and i kinda dont care about the cloud storage or whisk or gemini - i just need veo 3 credits, as much as possible, as cheap as possible, both t2v and i2v, dont care about other bells and whistles

[north silo]

Even if the price increases, it remains at the same level. Moreover, Gemini's limit is only unrestricted for the ultra tier. If you need a large number of credits, then even if the price of the ultra tier increases, at the very least, it won't be more of a loss than the pro tier.

[dire timber]

Hello, i recently bought the flow premium thing and i was wondering why my videos are only 8 seconds long and don’t follow my prompts directions very well?

[thorny mesa]

yo wsg. i am building something large in flows scene builder and i cant figure out how to export the entire length instead of each individual scene.

[thorny mesa]

doesnt show up for me

[woeful fossil]

Hover over the preview video itself.

[sand bloom]

Hey guys, Just joined in. I am having issues trying to convert my Flow generated videos into 9:16 ratio video format for tiktok. I used Adobe and it pretty much destroyed and cropped the video.

[thorny mesa]

ayy it worked, thanks a bunch.

[sand bloom]

Thanks in advanced

[winged plinth]

Hey, if your base video is 16:9, it will always get cropped. To preserve the full frame, you need to create the video from the start in a 9:16 aspect ratio.

[sand bloom]

Where does it ask me to choose what format?

I’ve tried looking for that no luck

[woeful fossil]

That's because it only supports 16:9 officially, for now.

[sand bloom]

Bummer, where can I generate from 9:16 from scratch?

[frigid pier]

Youtube won't let Youtube videos make money on their platform anymore.

I just saw this news :0

I guess this will lead significant drop of any kind of AI images/videos subscription rate in e very AI platform.

If you aren't able to make money from AI-generated stuff, why would most people pay the money for AI?

I know that there are a lot of people who are paying Veo 3 to run their Youtube channel to make money and they are going to cancel their plans if they cannot make money on yt anymore.

[floral thorn]

I think it's not possible at this time

[floral thorn]

Hello sir I make a video but when I check this video it's showing no generated sound , I use Veo 3 fast

What's the problem?

[woeful fossil]

What is the prompt you're using?

[floral thorn]

Stormy, dramatic scene in the middle of a wild river.
A small speedboat bounces violently over huge river waves under a dark, lightning-filled sky.
Rain lashes down as thunder rumbles, creating a tense and urgent mood.
Inside the speedboat: the mother lies unconscious with an oxygen mask on her face, the daughter clutches her hand, crying under her soaked red veil.
The young man, drenched, grips the side of the boat and looks directly at the camera, speaking loudly in Bangla:
"স্পিডবোটে করে মাকে বরিশাল নিয়ে যাচ্ছি... জানি না সময়মতো পৌঁছাতে পারব কিনা!"
Immediately after, the girl weeps and says with trembling voice:
"কবে হবে ভোলা-বরিশাল সেতু... আর কবে পাবো উন্নত চিকিৎসা?"

No subtitles. All dialogue in Bangla. Background full of wind, thunder, and crashing waves. Duration 8 seconds.

[woeful fossil]

So Veo only officially supports prompting in English for now. You can write the whole thing in english and mention you want the dialogue in Bangla and it might translate it. Also, avoid prompting that includes subjects that are not adults.

[floral thorn]

I already write my dialogue English and mention Bangla but it's showing (no generated sound)

[slow pawn]

how you use flow in India?

[floral thorn]

No , Bangladesh

[slow pawn]

it's available in bangladesh? or you use vpn

[floral thorn]

Use vpn

[slow pawn]

i also use vpn but in flow it shows 0 credits

[floral thorn]

Are you buy pro?

[slow pawn]

yes i have i can use gemini app but i can't use image to video so i want flow for that

[floral thorn]

Oh understand

[dapper wagon]

supppl

[floral thorn]

Hey

[dapper wagon]

hows it going\

[floral thorn]

Pretty good

[dapper wagon]

creating anything cool Henry?

[kind patrol]

Hello everybody! How can I try Flow?

[dapper wagon]

https://labs.google/flow/about

you need to be a subscriber tho @kind patrol

[marble hamlet]

Do Google intentionally insult their members. Now, most normal companies will try to give a fairly decent discount for the more you buy from them. So, going from 2,500 credits to 5,000 credits only saves you £0.49 than if you do a straight 2x. Going from 5,000 to 20,000 will only save you £0.03 than if you do a 4x.
Also, where's the 10,000 credits? That would have been a nice inbetween.

[slim hedge]

Anyone else here have problems with flow? Specifically Veo-3 and uploading Midjourney images that Flow thinks are too photorealistic? I'm in the U.K and Flow tells me I can't upload photorealistic images here. But here's the kicker, I created an image in Flow's image generator and used it as a start image for a generation, no problem, but when I download the start image and try to re-upload it to flow it also tells me its too photo-realistic, so this is not about not being able to use photorealistic images at all.

[spiral shoal]

I made this please give me feedback for potential improvements

[full trench]

was this all in flow or editied post production?

[spiral shoal]

Flow

[marble hamlet]

The issue I've found is that it won't accept realistic images of well known people.

[full trench]

did you use the image to generation? @spiral shoal

[spiral shoal]

Veo 3 fast

[slim hedge]

I thought that too, but it also won't accept photorealistic images already generated in flow that I already used in flow. Download a flow photorealistic image and it will let you use it, but download that same image and then reupload it and it now wont let you use it...Madness

[marble hamlet]

Yeah, that's frustrating. Veo 3 is wonderful to use, but certain things hold it back. I too at one point had an image rejected because of photorealism, it seems an odd thing for them to be declining, especially when alot of their text to video outputs are realistic.
I think it's also far too censored. It wouldn't accept an image that had a woman show a bit of skin. Just think of the consequencies a bit of skin will cause to the world.

[slim hedge]

Just the fact that flow generated a photorealistic image and let me use it but when I download the exact same image then reupload it then all of a sudden its too photorealistic when I used it 10 minutes ago just blows my mind, this is nothing at all about photorealism or location, its a poorly put together platform

[marble hamlet]

We understand and accept there will be issues with a model that's still in development. But it's the intentional decision making that annoys me.

[undone estuary]

Flow no longer adds audio to my videos.

[elfin vault]

Hello guy ! So hurry to start creating with flow but i am still waiting the availability in europe 🥲
Indeed even with a vpn, Europeans does not have access as they do not have the credit system yet. The only way is to create a an american google adress with an american number etc…

[slim hedge]

I agree, generating realistic people within flow ok, downloading the same image then reuploding the same image not ok...Someone at flow needs to take a serious chill pill

[median niche]

Hi.Is the Bank of Croatia's wise card supported? And a Croatian phone number?

[finite moat]

I am in Germany and I have the same problem. Does anyone have any advice?

[stiff harness]

Are you using VPN?

[median lotus]

Can someone explain to me why I’m not able to upload AI-generated images to Veo 3? In particular, AI-generated photorealistic images , in the UK , yet in the US, you can. It just seems unfair. Why is the US getting every feature Google DeepMind is literally based in London, UK? The UK is not part of the EU, so why the delay all the time? Even the Gemini App in the UK still doesn't have that native multimodal stuff? Makes me second guess my subscription 😦

[stiff harness]

That's a conversation for you and your government.

[median niche]

yes

[analog zealot]

still broked

[jagged jackal]

Maybe Hedra, as it's focused on avatar style lip syncing and can do 4 minutes of continuous dialog without splices.

[wooden vessel]

I just solved this problem. Solution: The phone number has to also be of the supperted region's origin. I'm from lithuania and used a vpn + smscodes.io. Use smscodes.io to get a temporary number when creating the google account. I used united states one and it cost me 3 dollars for one session.

[slow dragon]

Hi all,

We're investigating the audio issues you've shared further. To folks who are experiencing unwanted subtitles in their generations – can you please share examples along with your prompt in this thread?

https://discord.com/channels/1172568727942860810/1392866811233570958

Thank you!

[finite moat]

Thank you bro 👌🏽
I will try it later 💯

[quiet stream]

Btw to the mod who warned me, i shared a youtube video, this is google's discord no, same parent company? If it was inappropriate how is that video on youtube then?

[frigid pier]

Youtube won't make ai video channels make money anymore.
I think lots of people will cancel Veo 3 subscription after this

[quiet stream]

youtube or google has never said this, you fell for clickbait fake news congrats

[frigid pier]

https://support.google.com/youtube/answer/10008196?hl=en

[quiet stream]

🤦‍♂️
ok now read it

[median lotus]

But the thing is, the features missing in the app are literally available in Google AI Studio. And honestly, every other AI company lets you do this. It really sounds like a Google thing, not a government thing. Image-to-video is available in Hailou, and in a lot of cases, it’s actually better than Veo 3 (before you tell me to switch to that 😆). I just wanted to use it with my Ultra plan. How is this a government issue when OpenAI’s advanced voice mode is more realistic and already available? And in many cases, it’s far better than Google’s multimodal speech-to-speech AI. At least make it make sense – this is definitely a Google thing.

[quiet stream]

actually i'll save u some time, expand everything, press ctrl+f, type AI

[median lotus]

To be clear I'm not arguing. I'm just confused because the lack of features, based on region sucks. 😞

[fading vapor]

Uploading a cyborg with human face and I get "Our policies prohibit uploading of prominent people at this time. Please try a different image or send feedback." This product cost too much to have so many guardrails, and no way to even provide feedback with an Al or human moderator. Almost every image I’ve tried today the same thing. I’m not using or interested in using real people.

[sharp quest]

frame to video

[fast gust]

Yeah. Any tips to make different voices? Male voices all starting to sound the same.

We can prob prompt for different accents, but when creating an English male voice how to get the voice to sound older, etc.

[quiet stream]

but this is 8second, what was the previous video where you took the frame? Is it consistent?:

[abstract cloak]

hi

[stuck willow]

Where I can generate Ai video with veo 3

Please tell me

[jagged jackal]

Just google Veo3 then, go to flow.

[stuck willow]

@jagged jackal where I can click the veo 3 can you understand me

[jagged jackal]

In your Google search bar. Type "Veo3"

Then click on Flow

[sharp quest]

This is my own photo that I took randomly. If you save the last frame of the video as image, you can create the second scene with maximum consistency.

[stuck willow]

Can you explain me any issue here?

[jagged jackal]

Possibly region locked

[stuck willow]

Can i solve this?

[jagged jackal]

VPN maybe? Devs say it wont work, but others say it does dont know.

[stuck willow]

Can i used VPN when i open this?

Ok can you generate video for me i send you script can you do this for me?plz

[sharp quest]

😄

[jagged jackal]

Even if you access it. You'll have to buy a subscriptiom to generate anything.

[lusty nymph]

Hi Does anyone knows if the credits bought accumulate? Like buying multiple pack at once? Thank you

[gilded swift]

Man i really wish it generated 1080p instead of requiring an upscale

[quiet stream]

yeah, waiting additional time is so annoying, who would even download 720 when 1080p is available anyways
it should be an option to always have it at 1080p as final output

[foggy heron]

do the credits refresh daily or monthly?

[north cedar]

Hello, How did you generate a 9:16 video?

[mellow axle]

you didn't make this.

[mellow axle]

this is the second time i request a ban for this user. First time was when he wanted people to DM him for whatever solution they asked and he lied to everyone cause he didn't have credits to run a prompt so wanted people to run it for him. (i have proofs)
This time he stolen the video from Fantasyeats on TT and pretends it's him.
Can we mute or ban this fraudulent user please? @wet copper @high fulcrum

[dapper wagon]

lmao bro what... this is just weird to do. what does bro gain from faking posts here? 😭

[mellow axle]

i guess he gains trust and then DM people to run his prompt cause he has no credits, like he did with me. I asked a random question about characters consistency and he told me to try his prompt (it was a fruit cutting lol)

[dapper wagon]

ahhhh down bad lol well yeah this dude should be banned if hes doing that

[languid vale]

I couldn't manage to use flow, is it not public yet?

[iron matrix]

what issue did you see?

[sleek oxide]

Sharing this because not every output goes to plan.

A little editing made a big difference. Both clips were originally fails. Characters were off, motion didn’t land right, but there were a few frames that still had something. I ended up combining the best moments from each, and the final version took off.

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

[charred sail]

Compute time doesn't get cheaper the more you use it. It gets more expensive.

[scenic remnant]

How can I use from Portugal?

[junior horizon]

@main saddle thanks for the announcement! gonna test rn 🫶

[analog zealot]

i been tryna tell people how to use the audio and theyre just now explaining how it works from what ive seen

[junior horizon]

From my experience working with big gaming companies I can tell they could have even take some profit of your comments for their official tests XD

[high jetty]

Sorry I'm a complete AI noob, can I use flow on the Gemini app? From Germany with Samsung galaxy ultra 25?

[radiant marlin]

Hi all, I just created my first video yesterday and I'm loving the creativity. One question I have is how do you generate a 9:16 video? I tried prompting it to do so but it didn't create it. Any advice is appreciated!

[junior horizon]

you can use VEO3 preview from Gemini, but the latest VEO3 models are available directly on Flow, the actual tool

[high jetty]

And it isnt accessible with the mobile app, right? At least I couldnt find a way to access Flow with my phone

[junior horizon]

I think only browser, but you should be able to use with no prob with such a cool phone 😄

[high jetty]

Oh, didnt think of that. Thank you very much!

I first thought the problem is Germany, since we got Geminis VEO3 later than the world. But from the browser makes sense, thank you.

[junior horizon]

I'm in Colombia so yes.. same pain, we get a lot of stuff later, but I think you should be fine from the website https://labs.google/flow/about

[high jetty]

Thats super cool, thanks. Can I use it with Pro subscription?

Probably with limitations

[sharp quest]

I created an image with a 9:16 aspect ratio in another AI software and put it into the flow.

[mystic forge]

We can't get consistent characters when we use a frame. That's crazy they come out looking different everytime.

[sharp quest]

I'm also having problems with character consistency.

[mellow axle]

??

[quiet stream]

Lol literally everyone can make that content, sh*t I can do it better with actual consistent character and not a random monkey in each video too, maybe i should charge people too 😂

[mellow axle]

it's not his channel lol

he was asking how to make videos one week ago in here lol

[quiet stream]

lmao, hustlers gonna hustle, poor people that believe and pay anything for that though

[mellow axle]

just ban this user please 😂

[sleek oxide]

This guy bombarded my DMs and changed his name 4 or 5 times. Not only that, but he's begging for free handouts like we haven’t spent months dialing in structure, character logic, camera continuity, and scene flow. Everyone wants clean outputs but some people don't want to actually learn how to get there. You’re not gonna prompt your way out of broken scenes without understanding why they fall apart in the first place.

Needs to stop asking for shortcuts and start asking better questions.

[mellow axle]

He's just a larp and at this point a scammer. I don't know why he doesn't get kicked out. <@&1366831844502274180>

[ripe yacht]

Thank you for letting us know, we’re addressing this right now with the team 🙂

[mystic forge]

Yea each scene he look different and with longer hair each time lol

[mystic forge]

How you get characters consistent. I put in the details of my characters and the frame but some out different everytime

[quiet stream]

I write the exact same description, and generate 3 times, one of the 3 is like 95% close to the original so I pick that one

[mystic forge]

Yea I do the same thing it do get close but I'm making a kids show so it's long so some of them come out different. I got to create alot of scenes

[onyx fiber]

We should do some sort of giveaway and they give away points to use for ai generations

[gusty estuary]

I have a Google subscription with tokens on Flow Labs that I use for Veo 3. Everything has always worked fine, but recently, every time I generate a video, the voices sound bad and very robotic, with intonations that are painful to the ears. Is anyone else experiencing this issue? Has anyone managed to fix it?

[bronze fiber]

How are people making a down syndrome individual on a Google V3 video

How you get around that

Like how people describing characters

[charred sail]

what do you mean? Gooogle can't discriminate against the disabled.. If they want to be represented they should be.

[vivid socket]

Hello everyone, im new. I'd like to know if you know some tutorials so I can improve using flow. thanks

[marble hamlet]

What are Google doing? I generate my own image of a random person, but Veo 3 won't allow a prominent person. What?
Then I have an image rejected because of 'sexual content'. It's just a woman in a sexy sci-fi suit. You'd see this kind of outfit at Amazon or wherever, nothing was revealing.
I'm sorry, but I'm not paying for Ultra until they stop this nonsense. You can't ask for people to pay a premium subscription for a subpar service.

[bronze fiber]

Well I tried said I can’t

[charred sail]

You're probably not meant to be an early adopter.. right?

[marble hamlet]

What does that mean?

[charred sail]

LOL incredible.. Call the down syndrome organization and see how they do it.. light that fire 😛

[bronze fiber]

Lmao

How do you describe it

[charred sail]

I mean. It's been out for how long? It's early technology. Sounds like you don't like that stuff

[charred sail]

seriously, you're telling me that the down syndrome charities can't use Veo3? LOL that's wrong

and btw you're paying for compute time... that's pretty much it. They're losing money left and right developing Veo3

[bronze fiber]

Who knows

[marble hamlet]

I still have no idea what you're talking about. What do you mean I don't like this stuff?
I've been working with AI for about two years now.

[charred sail]

how long has veo3 been out?

I mean all this is early.. 2 years?

lol i remember in 1979 asking, why am i still using cassettes to store data!

[marble hamlet]

We're not talking about issues with an early model in development, I'm talking about their policies and censorship.

[charred sail]

Just saying.. complain less.. offer more

[charred sail]

That's called responsible development...

Google isn't the US Government. It doesn't guarantee you the right to free speech nor expression. Try one of those chinese ripoffs.. Maybe they don't care?

i mean it's annoying, but i get it 😛

[marble hamlet]

I guess this is America in general where violence is ok, but art with a bit of skin showing is corruption.

[charred sail]

Not at all. This is google labs.

see the top of the discord? says it right there

The US is only one small part of the world

[marble hamlet]

Are you saying Google isn't a US company?

[charred sail]

It's a global company