flow-general・🌊

[silk hill]

yes i have a google AI studio link but none of the new images are there

[worldly yacht]

hmm maybe you dont have your things syncing to google drive ?

[silk hill]

if you have to do this manually somehow, i guess i do not

[worldly yacht]

Do you have the pro subscription ? i think you need that

[silk hill]

i have ultra

[worldly yacht]

o sync your Google Drive with Google AI Studio, you simply need to connect the two services. Since both are part of the same Google ecosystem, the integration is designed to be straightforward.

Here's how to do it:

Connecting Your Drive
Access Your Account: Log in to Google AI Studio with the same Google account you use for Google Drive.

Go to the Library: On the left-hand side menu in Google AI Studio, navigate to the Library section.

Allow Access: You should see an option to connect or allow access to your Google Drive. Click this option.

Authorize the Connection: A pop-up will appear, prompting you to select your Google account and grant the necessary permissions for Google AI Studio to access your Drive. Confirm this to proceed.

Once connected, your Google AI Studio projects and prompts will automatically be saved in a dedicated "Google AI Studio" folder within your Google Drive. This allows you to easily access, share, and manage your work from either platform.

Autosaving and File Access
When you create your first prompt, you can enable the autosave feature. This will ensure your work is continuously saved to your connected Google Drive folder.

To quickly access the folder in your Drive, look for a Google Drive folder icon in the upper right-hand side of the Google AI Studio interface. Clicking it will take you directly to the synced folder in a new tab.

This integration streamlines your workflow, allowing you to use files and prompts from your cloud storage directly within Google AI Studio.

[hidden epoch]

does anyone actually monitor the bugs/feedback or am i just yelling into the wind?

[silk hill]

weird not seeing a libary here

[clear cipher]

🙃

[main saddle]

we have reps for Flow who respond to reports, but it's not immediate! thanks for submitting!

[worldly yacht]

@main saddle Could help with this one

[main saddle]

ty + @silk hill! can I ask you to submit this into #1319424187730755636 if this continues? ty 🙏 (FWIW, I have raised this internally)

[worldly yacht]

ohhhh oki my bad

[silk hill]

np

to be clear i'm speaking specifically about images generated with the new reference feature that got released, like, yesterday it think

[unreal viper]

please post images and videos in #flow-videos・📹 thanks!

[fading dirge]

hold on, let me get this straight, veo 3 fast is now unlimited with ultra? and this is not just like a one week thing?

@high fulcrum can you help me out with this question?

[kind bison]

Is the accent feature ever going to come back? I used to be able to get a variety of accents for characters, but now all I get is American or British.

[high fulcrum]

hey @fading dirge Mods are here to enforce the rules, please tag @ google staff for product inquiries 🙂

[sacred crypt]

I tried enhancing the last frame of my clip generated but it still did not work.
And in the text-to-video generation, I am not getting the face consistency of my characters.

[icy dew]

i know this is out of topic but i just noticed.. workspace has now the ability to accept image reference 😍 I was working on this earlier today and this option isn't here. and now... waaaaaaaaa i'm excited to try!

[kind bison]

LOL - just got the "no prominent people" flag...on picture of a woman that VEO 3 generated!! It added a character to a video and I liked it, so I saved the frame and used it for a scene extension. One version went through, then two more gens got denied. Fix this nonsense, Google!!

[sacred dawn]

Hi where in workspace is this?

[icy dew]

it's on the right side. there is tab for veo and the output is veo 3.

[sacred dawn]

wierd not seeing it here

i just did its in google vids

[fading dirge]

Ok thanks

[wild jetty]

This 0 credit for ultra users is one of the best changes Ive seen

Thank you people

[fading dirge]

so veo 3 fast doesnt use credits but veo 2 fast does?

[fringe ferry]

If i use nano banana, with google ai ultra, does it still make watermarks?

[fading dirge]

used a reference of a make believe person to make this image in whisk, brought it to flow to use... it gets blocked. bruh i dont get it

[idle compass]

Did you know that all images generated in Nano Banana (and guessing anything created within the Google Gemini AI ecosystem) are embedded with an unremovable SYNTHID watermark, so it can always be identified as AI-generated. presume this will help with governments, education and media companies to detect fake or generated content. Here is the info from GOOGLE about this. https://ai.google.dev/responsible/docs/safeguards/synthid

[idle compass]

what message do you get from FLOW?

[wicked sapphire]

Is there a way to make 9:16 films in Flow? Veo on Krea allows for this, but not directly in Flow

[fading dirge]

The real person message.

[fair bone]

send me your work

[formal pier]

If you build a scene in flow and always take the last fram to build on why does evrything gets brighter and brighter every other new scene. once yo are at scene 4 or 5 its just like +100 saturarion

[burnt oriole]

is that google vids?

[icy dew]

yup. i'm happy they added the image reference option there.

[burnt oriole]

I don't have it :((

Does it use Veo 3 fast or full power Veo 3?

[icy dew]

I'm guessing it's veo 3 fast.

[icy dew]

oh sorry about that.

[burnt oriole]

yesss because of the speed

Also I generated a video but it looks like it doesn't have a sound

[icy dew]

Before there was an option for veo 2 on google vids. Now it's veo 3 only. But I ain't mad about it. Veo 2 is available through google ai studio. What I am mad is the 9:16 frame option is removed from veo 2 🤣 I use this for shorts and reels. Now I need to look for an option.

[regal crystal]

hey everyone i just got google ai ultra but my credits havent arrived after 2 hours of waiting

any help on this?

[real radish]

hey

yeah where's my AI credit?

I haven't gotten any credits yet?

for the month after making my payment

[hoary stag]

I have been having trouble since last night with audio generation with Veo 3 Fast clips. the audio simply will not come out and keeps giving me an error

[wheat steppe]

I was able to upscale 2 clips this morning but then it stopped working saying try again later.

[patent citrus]

yeah upscale is also broken for me as of 2 hours ago

was working fine up until very recently

[dire anvil]

guys what is happening to flow? i can't upscale videos, the generation quality and sound quality has been so bad for the last 2 hours, they got some issues?

i got an ULTRA subscription

[dire anvil]

yeah an announcement when everything it's fine so i can keep working

[primal lodge]

Hi any plans for pro users or are we still stuck with the same old restrictions?

[wide marten]

I don't see that option. Could you take a screenshot of the UI for adding reference to google vids?

[dire anvil]

atm works just with veo2

so it's not really usefull

[meager edge]

How to make face consistency as during movement or talking in veo 3 using flow

As it changes totally the other person

Have been trying for couple of days Could some help here or provide me any resource

[tepid tundra]

Get the proper last frame of previous scene if they're together, it shouldn't be blurry
Also, try to not extend the scenes as every time it will look worse

[meager edge]

Ok I have tried all the detail prompting like given description of the subject in the image and then the action details with camera positioning but still it changes the face and even if it smiles or tile the face little to right or left

[meager edge]

Ok for perfect image generation what would you suggest will be best with same location

[icy dew]

it's actually workspace. the model is veo 3. i finished my project from there earlier today cause I could generate more on workspace than on gemini.

[icy dew]

you can try ingredients on flow for character consistency.

[meager edge]

That would be the parts of face is not visible like side profile

[meager edge]

yeah can try that one

[kindred needle]

So sorry if has been discussesd, i didnt see it i received a message saying veo 3 fast is 0 credits (i have ultra). Can you then call the api infinitelly?

@edgy holly but cant you just call it via the veo3 standalone api?

Like, you need. VErtex?

No,no , im just asking: isn't there a standalone api

I see

[wise dirge]

"Hi! Quick question—can I post my social media account and the reusable prompt from FlowTV?"

[kindred needle]

This is where i got confused

[meager edge]

I Dmed you the prompt and clip

as I am still new in this video Generation so need help to understand how the people is making consistent movement characters

[kindred needle]

@edgy holly it hallucinated

[sudden island]

I ve create a bot that automate flow creation . Who wants to try ?

[sudden island]

You write a pitch and than he create prompt . Copy paste to flow you got your videoq

[meager edge]

Yeah will make to cut short and to the point

[wise dirge]

Hi! Quick question—can I post my social media account and the reusable prompt from FlowTV? Can it be monetized?"

[kindred needle]

Thanks for clarifying it.

[meager edge]

nice tip mate thank you

Sure I will try that also to get better consistency

[gentle harbor]

Why!? 🙁

[wise dirge]

Hi! Quick question—can I post my social media account and the reusable prompt from FlowTV? Can it be monetized?" answer please?

[gentle harbor]

I know @edgy holly ... but it happens only we have to finish something 😄
Already done.. thanks anyway

[odd gull]

Hi guys, Flow working in Russia?

[patent hedge]

Hi Guy : ) If you have à visual exemple we could have a better understanding of your proposition. Like you explain above, we do not really project thé workflow?

This is the Picture you ask Dear one

[sudden condor]

What does ingredients do?

🙏

Ty

[main wren]

Hey I was interested in getting the ultra plan bc it says zero credit usage for veo3 fast and id like unlimited generations on flow for veo3 fast but im curious if flow gives a monthly rate limit. I found some stuff online that says flow has a rate limit of 625 veo3 fast generations per month regardless of if you have ultra plan or not. Is that accurate?

Oh great. Okay thank you. Do you know who the admins are in here who would know the per day / hour / month limits ?

Ahhh is there a chat that would have the devs where I could ask a question like that?

[mortal wing]

Why whenever I create a video with 2 characters talking it messes up and only 1 talks for both of them?

How do u fix that?

[marsh tapir]

Two fighting roosters in a Pixar-style boxing ring, one rooster delivering a powerful knockout punch to the other, dynamic action pose, feathers flying, intense expressions, boxing gloves, cinematic perspective inspired by Rocky III, dramatic lighting, cheering crowd in the background, 3D cartoon rendering, ultra-detailed, vibrant colors, heroic atmosphere, victorious rooster standing strong while the other falls back stunned.”

[formal pier]

if i use a generated drame in veo 3 flow the characters don't speak is this a known bug?

[analog wharf]

Hey everyone, how are you? Well, I've been testing VEO3 for a few days, and I'm really interested in VeoTV. I wanted to know, how do you create your prompts? Is there a standard? Do you focus on detailing well?

[hidden epoch]

hey everyone beware that in the flow bugs feedback people are commenting to submite a support ticket and then having you run a command on your PC to steal data

mods should be doing a better job

[junior horizon]

more like: more mods would be recommended

or an AI agent that flags scammy stuff

[hidden epoch]

so since the bug/feedback section gets you no where, I guess I will throw it out here. I am a sports card collectro and I use Veo 3 to help come up with some reveals for my YouTube channel. I have made a bunch of videos so far and this week when I went to start to make one, I tried uploading the image of the card and it keeps saying this image violates our policy, this is kind of a bummer since i relied heavilt on veo 3 to help create the content for my channel. Is there a fix coming for this or am I just kind of screwed?

[spark delta]

Please use English only in this server

[mortal sorrel]

Hello everyone.

I'm writing again about a problem I'm having trying to connect a Dialogflow CX / Vertex AI chatbot to the Gemini model. I have followed the documentation, but I'm still unable to establish the connection and I'm a bit stuck.

I know everyone's time is valuable, but I was wondering if there would be any experienced volunteer who speaks Spanish and might have a few minutes to guide me. I am fully willing to share my screen on a Discord call to show my configuration and follow the steps live. I believe we could solve it much faster this way.

Thank you in advance for your time and consideration.

[shell heron]

Hey Guys, I have a problem with flow. I can't upscale my videos at all.. It's been over 13hours since it occured..

[stuck adder]

Just use topaz

It’s way better anyway

[bleak sail]

Why can't the characters in the videos I generated with Flow make any sound?

🥺 pls help me

[regal crystal]

I payed for google ai ultra and after about 15 hours I still didn't get my credits

[tribal veldt]

"we do not allow uploads of photorealistic people in your country or territory at this time. please try a different image or send feedback" - This filter for people is insane. I have AI generated characters. 80% time it flags as photorealistic people. and then sometimes not.. The filter basically makes the tool unusable.

[tribal veldt]

It is in flow. When i upload image to Gemini then there is no errors. + most of the images are edited in nano banana.

[open beacon]

this flow credit ....wadjklwdjl sucks per se

runs out so quick vs chatgpt. 30$ unlimited pretty much

its so weird. uses like 40credits when i wanted 1vid generation.
Waste of credits....

n its expensive as heck... getting more credits. diabolical? dont see the justifications ish

nvm?

<@&1366831844502274180> theres corn....

[tribal veldt]

and there is nothing about photorealistic people in Google Ai policy: https://policies.google.com/terms/generative-ai/use-policy

[idle compass]

100% agree with this. It is more than frustrating. And the rule makes no sense, because Google allows uploading of photorealistic images in its recently released Gemini Image generator (Nano Banana) app. Such a stupid restriction, and makes like you say, using Veo almost unusable for people affected.

[silk nexus]

Two fighting roosters in a Pixar-style

[idle compass]

what is the policy violation message you get?

[silk nexus]

and there is nothing about

[silk nexus]

Yep and here in the UK we have laws to say what can and can't be done with uploaded images. I totally get that people might upload images of say a famous person and do something to them but we have laws that pertain to that anyway and if someone did that it wouldn't be Google getting in to trouble but the creator.
That's like prosecuting Amazon for selling kitchen knives because someone goes and uses one to harm someone.
I would have thought there are laws in the US about not being able to modify images of people you don't have permission to do anyway isn't there? And if there is then that makes it even sillier to prevent other countries from doing so.

Not telling us at all during any of the product pages / demonstrations or sign-up pages isn't on at all though.

[dire ore]

Hey, I need some help with Flow.
Whenever I click “Create with Flow”, it gets stuck on the screen saying “Flow – A new state of creation” and never loads the creation interface.

Location: Israel (supported country).

Using personal Gmail (not Workspace).

Tried Chrome (latest version) + Incognito.

Cleared cookies/cache for labs.google.com.

Still stuck after multiple tries.

👉 Is this a known issue? Anyone else experiencing the same problem?

[dire anvil]

u r not the only one

it's like that since yersterday

to fix that i just spam the same prompt over and over again

it feels like they have a lot of reqeusts and they fix giving bad results

to make the genereations faster

and maintain the high requests

Yeah that's why it's not aproblem for me to spam the same prompt over again

but i waste so much time

[main mantle]

are there any plans to bring image to video to veo 3

[dire anvil]

yeah

they said they are working on it

[main mantle]

i need it so bad

[main mantle]

oh did they?

[dire anvil]

yeah

[main mantle]

is it on the announcements or what?

[dire anvil]

it poped out 1 day when i loged in

i read theat they are working on it

[main mantle]

alr 👍

[dire anvil]

when tehy made it avalaible even for pro users

and not only ultra

[silk nexus]

I've personally not noticed any loss in audio quality and I'm creating videos every single day. The audio quality is pretty bad (sounds like it's bad MP3 encoding and underwater) but it's always been that way as the audio side of things is currently in beta. Like I say though, I've not noticed any difference in quality other than the usual bad quality it gives but that's to be expected.
I'm a bit of an audiophile as work in music and have done my entire life so I'd notice if there was a major difference right away so this may be only doing it to certain people maybe? For what reason that would be I'm unable to offer any ideas about though, sorry.

I'm not sure what to say then sorry 🙁
Like I say it's in beta at the moment and they have mentioned that on their product pages and elsewhere. I'm not sure how you were getting such good quality in that case. Can you possibly upload two videos here, one with the good quality and one with the bad quality so I can listen to them?

[crisp basalt]

yes

at this point im just gonna say yes to whatever speculation that you read from internet lol even tho they dont make sense

a huge thumbs up

https://tenor.com/view/spongebob-spongebobsquarepants-gif-9146238

i wasnt trying to shade u , i promise

😎

[hidden epoch]

It doesn’t say, it just says this image may violate our policy try a different image and try again

[silk nexus]

Can you not provide me with your own examples? I'd rather see the quality you're getting and then an example of one of your own videos where it isn't now the same quality. Most of the veo3 videos on YouTube as far as I've seen have bad audio. The only ones that have really good audio are not audio coming from Veo3 but they've taken it and put it through another app to change the audio to higher quality. I'd much rather see your own video examples to try and understand what is going on.

[crisp basalt]

exactly..

[compact bridge]

Anyone getting “flow is temporarily unavailable”? Been trying for the last half hour. Tried relogging, new projects and new browsers. Nothing.

[stuck adder]

This already exsist do you mean start frame and end frame?

[silk nexus]

If you can't provide us with a couple of your own videos as examples then I highly doubt there's anything anyone can do for you I'm afraid. That would be like going to a car mechanic and saying there's a problem with the brakes on your car and when they ask where is your car you say "Just pick any car on the road, they're all the same"!
Sorry but can't help if you can't offer up some examples. 🙁

[astral pollen]

you just have shitter prompts most likely

[modern wave]

Do you have any tips for writing a good prompt?
I’m using Gemini, which can now analyze videos (to describe each scene, setting, etc. in detail), and then I use “image to video” to try and recreate the exact same scene using Gemini’s prompt.
But the result looks completely different from the original video.

[versed peak]

I subscribed to Ultra literally a day before they announced the credit boost 😭 Is there any way I can get access to the new benefit without waiting a whole month?

[fringe nova]

hey does anyone know how long Ultra members get free Veo3 Fast generations for?

[fiery token]

at least the next couple of weeks, and maybe indefinitely

[fringe nova]

yo no way if thats the case im 100% keeping my ultra membership. was planning on canceling once it jumped up to $230 a month, but now im unsure lol

[fiery token]

yeah don't cancel. a lot more good stuff coming that will make ultra worthwhile

[velvet apex]

How to generate 4k output in flow?

[fringe nova]

you can only upscale to 1080p in flow iirc. however u can use other softwares to upscale the videos further

[velvet apex]

But on the website it shows 4k output

Z

See this

[fringe nova]

o this is news to me

[patent citrus]

So what generator does Runway ML use? Flow uses Veo right?

[velvet apex]

Yes but it actually generates 720p only and the upscaled version is very bad

[fringe nova]

agreed

[fringe nova]

runway uses Gen-4 and Gen-3 Alpha, their own models i believe

[velvet apex]

Why I should I waste 100 credit if VEO 3 Quality also gives same 720p results

[fringe nova]

while the resolution is the same, the content that is generated in "quality" is better from my experience. though depending on the project u might not need to use quality

[fallow mason]

Use this image to create a marketing ad.

[velvet apex]

I know I just need to understand where is the 4k output as promised on website <@&1173646484311384114>

[silk nexus]

But on the website it shows 4k output

[tawny sequoia]

I really thought Ingredients-to-video would have worked and I upgraded to AI Ultra to experience highest limits to VEO 3. It still does not work and forces me to use VEO 2, and that does not have Audio

<@&1173646484311384114>

[kind bison]

The audio absolutely degrades badly on certain days. On other days it's good enough (though obviously not high quality). I have been an audio engineer and sound designer for decades and have professional grade monitors and headphones (though it is even noticeable on my cheap TV speakers).

[kind bison]

Truth be told, this is not always the case. I have been noticing lately that many of my Quality videos are no better than my Fast videos.

[winged plinth]

Agree on this, today’s generations are out of touch.

[silk nexus]

Same here! Audio engineer professionally for around 30 years and have professional setup here at home also but I've never had anything better than audio that sounds like bad mp3 encoding here. Definitely never anything which is anywhere near professional though what does come out is pretty usable. Sound effects especially are quite good just not speech.

[kind bison]

Sometimes, for videos that require top notch audio, I will take the VEO 3 audio track, split it into stems on LALAL to separate the voices from the sound effects/music, run the voices through a voice swap on Elevenlabs, then take the new voices and match the EQ and reverb/ambience of the VEO dialog, and finally replace the audio when editing.

[kind bison]

Yeah, what I'm saying, though is that there are absolutely days in which the audio takes a nosedive, with really bad digital artifacting. I create several VEO 3 videos every single day of the week, and from time to time this gets so bad, I have to do surgery on the audio track.

[inland pier]

all controls are being updated to Veo 3, so you will have ingredients, first/last, etc, with audio. stay tuned!

[silk nexus]

Ah well it must be different for different people. Why I have no idea though. I make around 40 videos a day and the quality is absolutely consistent here. Consistently not great that is!! 😄 Still usable though.

[kind bison]

I'd prefer the "no prominent people" bug gets fixed before any of that. I'm a professional AI filmmaker and it is seriously messing up my workflow and causing embarrassment for me with my clients (some of whom are well known producers and directors here in Hollywood).

[fast summit]

been having a super hard time getting text to work right all week in veo3

it always feels like a crapshoot but this week its just not cooperating at all.

[kind bison]

Yeah, I have noticed in the bugs/feeback channel people reporting problems that I have never experienced - these issues don't seem to be global for all users.

[silk nexus]

I don't think it's a 'bug' unfortunately. It's by design. They're protecting themselves from people uploading say an image of a famous person and then making a video with them in it. We here in the UK have laws about that anyway so there's no need to just blanket stop people from uploading images they have the rights to use. If they use them illegally then we have laws that will deal with it. This is just silly.

[inland pier]

yes a very valid complaint. our standards and policy are constantly shifting. we are doing our best to make the experience as seamless as possible, without punting.

[winged plinth]

I give it designed audio prompt like that and it helps greatly

VISUAL: Macro on goblet rim; smoke curls form non-readable sigils; at lift, coaster runes flare—an arrival micro-beat. 85 mm tight, locked.
AUDIO: MUSIC airy shimmer; SFX flare ping + soft hiss; MIX ping +2 dB, bed ducks −2 dB.

[vapid pollen]

Guys, do we have free credits every day?

[opaque sand]

Hello, everyone. After yesterday's launch of “unlimited veo-3 fast,” have you encountered any problems with it? Has the video quality deteriorated, and I am particularly interested in whether the sound quality of the videos has gotten worse?

[winged plinth]

It’s what we’re talking about right now:)

[main mantle]

no just images not frames (ingredients)

[main mantle]

damn

[stuck adder]

Image to video is frame to video is it not?

[main mantle]

no its ingredients

frames is to mix two images

[inland pier]

there are two ways of using images at the moment. Start/End frames (i.e. interpolation) and using images as ingredients (i.e. reference characters or styles)

[main mantle]

i know but ingredients is much more helpfull

frames glitches

[stuck adder]

You’re talking about the ingredient option

They do have image to video Veo 3 only start frame only thought

I wish they had Starr and end frame VEO 3 tho

[main mantle]

google released in google video image to video but no one can see it ??

which team do i ping

[main mantle]

<@&1173646484311384114>

https://www.youtube.com/watch?v=jG31dMulimg

[kind bison]

No, it's 100% a bug. I can upload an image and it render a video. Then I try again with the same (already uploaded) image, and it gets blocked. These even happens with images that Google itself creates on Flow! There is no legal justification for it whatsoever - plenty of other plaforms have no such restrictions.

Crazy thing is, I am currently working on a series of videos for a celebrity. Have only had one image of him get blocked, then rest went through just fine. The whole thing is a farce and Google needs to abandon it if the platform is going to continue to exist.

And, of course, the bug mainly occurs on Flow. An image that gets blocked there can 9 times out of 10 be used without issue on 3rd party aggregator sites with VEO 3 access.

[silk nexus]

Yep I agree but Google and their representatives all keep on saying it's for legal reasons. Getting any of them to tell us what those legal reasons are is proving to be impossible though so there's definitely more going on here than meets the eye.

[vapid pollen]

Can you generate veo 3 videos for free?

[kind bison]

Google suffers from the same problem as all giant corporations: it has become a byzantine bureaucracy and moves at a snail's pace. It is the same with Microsoft, for whom I just completed a large AI video project. Decisions on basic things required committee approval, legal department approval, etc, and just turned the whole project into a quagmire that took months longer than it should have. Small AI companies are able to deliver changes and updates so much quicker.

[kind bison]

Only VEO 3 Fast (on Flow), since I have an Ultra plan.

[silk nexus]

And if you ask anyone that works there what the reasoning is behind it they suddenly go exceptionally quiet and nothing but tumbleweed!!

At least if they were told what to say that would be something but to say it's for legal reasons and then not even point us to what it is in the UK that legally prevents us from using an image that we have the rights to or even to upload an AI image or use an AI image that Google themselves created is beyond silly!

[crisp basalt]

Looks like Veo 3 Portrait Mode is already on API 👀 fingers crossed Flow Team brings it into Flow too!

[silk nexus]

WOW! I seriously hope so!

[silk nexus]

You really do like to argue don't you? 😉 They may not be lawyers or big decision makers so they really shouldn't be making things up then should they?
If they don't know then they should say so. Being honest and saying you don't know the reason is a much better legal position to put yourself in than making stuff up or is that okay with you?

[past tundra]

i need help

I can't get the 3-month offer for $125.

[silk nexus]

How are you trying to get it and what is happening? Is it a problem with your card or the process or something else?

[past tundra]

I can't get it directly, they make me pay $250 a month.

I was able to get the 0% one-month trial.

but I want the ultra

[silk nexus]

Are you seeing it say 124.99 for 3 months or are you seeing something else?

i.e. Are you on this page?
https://one.google.com/about/google-ai-plans/

[past tundra]

yes

but when ii click on the offer dissapears

[silk nexus]

That's strange! What happens if you try in an incognito window?

I get this though had to use a different email to get that as have already subscribed using the account I was logged in with so did this in an Incognito window and a different Gmail address.

[past tundra]

I am trying it now.

same G

no sense

@silk nexus

[silk nexus]

Very odd 🙁
You may not want to sign up with a new or different Gmail address but can you try going through the process with a different email logged in to see if that's the issue? Don't actually purchase using that email (unless you want to of course) but just to see if the problem lies with the account or something else.
Also which country are you in? Maybe Google are changing things in certain countries and they haven't update the subscription process properly yet.

[past tundra]

United States

I have restarted the router and even created five new accounts.

[silk nexus]

I'm not too sure then. Hopefully someone from the US or one of the Google support might be able to help you out with this one. Sorry I wasn't able to help but at least it's nothing that you're doing. Or at least it doesn't seem like it would be. I hope you get it sorted soon.

[past tundra]

And if you do it for me?

@silk nexus

[silk nexus]

What do you mean? Do what for you?

[past tundra]

Create a new Gmail account, check if have the offer, send it to me, and I'll see if it lets me buy it.

If don't let me, we'll buy it from your computer.

I work with this tool and I would like to fix it.

[silk nexus]

Creating a new Gmail account wouldn't affect whether the offer shows at that point though as the two are unrelated.
Do you ever see it saying $125 at any point in the process? If you do then this is something else going on and my creating a Gmail account and sending the login details of that email wouldn't really make any difference. Not that I can see anyway.

[silk nexus]

I'm afraid I wouldn't be prepared to buy it from my computer and I must ask you not to let anyone else do that for you either. There is no way you want to be giving your payment details to random strangers. That's a really bad idea 🙁

[past tundra]

okey G

[silk nexus]

Sorry 🙁
Hopefully one of the Google support people in here will be able to offer some advice or look into this for you or perhaps you can get support from online although I note that it's very difficult to get any support from Google online easily.

[stuck adder]

Which frame to frame gen is better veo or kling

[peak hatch]

Does anyone know when last frame or ingredients will be coming for veo3?

[stuck adder]

They need that feature badly

[wicked valley]

Anyone else noticing the excessive amount of poor or failed generations with 'Veo3 Fast' ever since it went free for Ultra users? Veo3 Quality is also suffering (today, especially).

[plain iris]

Why veo3 flags my clearly robot as a human? Is there a way to get around it and use my character?

[stuck adder]

Not really

Mine is fine

I want them to do video to video really bad like runways aleph I think they’d lead with that Toto

[gloomy helm]

Try having my screen name lol

I get flagged constantly and have to reiterate that I am actually a groomer

[cedar ridge]

ok here a question is flow used to replicate characters between clips? how long can the scenes be in total in flow? with the same characters?

how does that work in the prompt?

so dont use extend just make a new clip with the same photo to video?

oh thank you

[dawn grove]

hey is there any way that I can create 9:16 videos on flow??

[silk nexus]

Not currently no but vertical video is supposedly coming to the API so hopefully it may get added to Flow at some point in the future too.

[unkempt plinth]

What's the official link to open VEO3

Anyone who know how to do give proper prompt to VEO 3?

[crisp basalt]

i love this

[opaque sand]

Either try rotating the video 90 degrees (use the initial frame that is rotated) or use third-party platforms to “extend” the video at the top and bottom.

[foggy rapids]

Anyone here actually paid for ai ultra

[foggy rapids]

Ig do json promting using any text based ai

[crisp basalt]

Well yeah there are

[willow summit]

Hello everyone

I wanna bring my photo to live using flow but don't know how

[crisp basalt]

check dm

[willow summit]

Ok 👍

[hazy void]

cinematic shot, 35mm lens, cinematic softness, shallow depth of field. A neutral adult in soft, minimalistic, cozy neutral-toned clothing stands in an infinite white void, carefully flipping through an oversized, luminous glowing IKEA catalog that emits a subtle ambient light. Each page flip triggers a glowing, organic portal to bloom and expand, seamlessly blending the edges of the void into a new hyper-realistic Scandinavian interior. The camera performs one continuous, smooth slider pan, gently reframing from one portal to the next without any cuts. Gentle portal breeze slightly lifts hair and fabric. Subtle floating dust and light motes are drawn towards the portals. Elegant realism, imaginative, calm, and dreamlike. No text, no logos, no branding.

[long gyro]

None of my videos on Flow loads, any idea on how to fix this?

[shell heron]

It seems that the fast mode of veo3 now has lower quality... since it've became free..

[stuck adder]

It prolly isn’t nowhere hear aleph

[stuck adder]

Rotating to 90 degree is a huge miss and barely hits cause the model thinks it’s 16:9 so sometime it added the elements that’s side ways in relation to the vertical video

It’s really crazy that they haven’t added 9:16 though very crazy

[crisp basalt]

now that you mention it , yeah veo 2 extend model is basically video to video

[crisp basalt]

I mean, they’re already working on integrating 9:16 for the Veo 3 model in Flow, since the API already supports Veo 3 portrait mode

[stuck adder]

Oh so you can use that else where then???

[stuck adder]

How does it stack against aleph?

[crisp basalt]

Else where?

[crisp basalt]

What

[hollow hemlock]

Hi

[crisp basalt]

I’m not really sure what you mean. my point was just that the extended Veo model works more like video-to-video. I wasnt trying to compare it to anythin

[stuck adder]

As in another platform that uses the API but also has vertical integration

[crisp basalt]

Hmm I’m not sure cuz as for now the 9:16 aspect ratio support seems to be only in Google Vertex Ai Veo

M

[tacit roost]

A skinny young man standing alone in front of a mirror inside a dimly lit bedroom.
He looks weak and disappointed while touching his thin arms.
Moody cinematic lighting, realistic environment.
Camera: Slow dolly-in towards the mirror reflection, focusing on his sad expression.

[winged plinth]

Hi @tacit roostYou'll need to use the website to generate videos https://labs.google/fx/tools/flow

[mortal swift]

My prompt, which included the line "The woman is happily applying drywall filler with a spackle":, got flagged for possible sexual content? Is there something about applying drywall filler I should be concerned about?

[sudden condor]

@digital escarp what prompts did you use for your nyc vid?

[digital escarp]

It’s a ton of prompts honestly

Couldn’t conceivably drop all of them. If there’s a specific shot you love I can share

[mortal swift]

I get that. The balance of the prompt was that the wall falls down and reveals a fantasy forest, and I'm wondering if "fantasy" was the trigger word? Nothing I changed in the prompt worked. The pic was of a woman drywalling, nothing at all salacious about it or meant to be!

[meager edge]

I need help to understand Frame to video as I have tried experimenting with my character which I generated using IDEOGRAM but my character does not coming up to the desired result , sometimes the face morphs or sometimes it changes the color on face like bit yellowish which change the texture of the Subject in the frame

Could some help me Here

Can I post it here

@edgy holly

This the the Video made using Text to video

This is video made using Frame to video with image reference

Medium waist-up shot, static camera at eye-level. A South Asian man in his late 20s with short dark wavy hair, neatly groomed beard, and medium warm-brown skin is seated on a sofa, wearing a blue polo shirt and light denim jeans. He sits casually with both arms resting on the sofa’s armrests, knees bent in the foreground, and looks directly at the camera with a composed, confident expression. He says, 'If you’re trying to make a high-performing advert, it needs 3 things. It needs a hook, it needs a value proposition, and a call to action.' Behind him is a minimal modern indoor setting with a beige sofa, a light-colored wall, and a shelf with plants. The room is softly lit with natural daylight. Cinematic clarity, static framing, and perfectly consistent lighting across all frames — no flicker, no glowing, no changes in skin tone or face texture. The subject remains sharp in focus while the background is softly blurred. The mood is professional, conversational, and approachable. Quiet indoor ambience is present with no distractions, while his voice is steady, clear, and synced to his lips.

Prompt for both is same

As you can observer in first video its natural and clearly visible no texture changes

but in the second video it change's the light on the subject face you can feel the difference if you pause at 00 the image color is different from when the video starts playing

Look at 01 or 02 seconds then may you can observe the differnece

@edgy holly

[crisp basalt]

testing veo 3 portrait mode from vertex

[meager edge]

Let me try it once

[sand pond]

So it is free to try now?!

[versed shale]

Yo

How do I delete a scene in scenebuilder?

[subtle citrus]

How come that I can't access Flow Ai despite being in the supported region

[crisp basalt]

https://tenor.com/view/angry-cute-girl-baby-pink-gif-5384585426126614778

[stuck adder]

I’ve used both Veo 3 is great a motion graphics and other things but it still has that slow motion feel, which when you use aleph you’re using your uploaded fps.

However with nano released I do hope that a dynamic version of it is resealed soon to fully support video to video

[stuck adder]

How did you do it?

[stuck adder]

Understood but anything other than dialogue it seems like slow motion

If I say “run, jump or punch soemthing” from frame to video it seems slow still

[crisp basalt]

Could be prompt issues or the model itself, but Veo 3 image-to-video often shows a that uhhh idk I call it anticipation/delay before the scene starts

Which I kinda don’t like about it

[stuck adder]

Yeah I love it though def have helped me improve a the production level of projects

[crisp basalt]

ok one last test , this one is image-to-video veo 3 portrait mode.

[stuck adder]

How are you doing this? lol

This looks incredible!!!!!!!

720? Or you upscaled

[silk nexus]

I'd not tried the Scene Builder until last night but am I right in thinking you only get one Scene per project or is there a way to have more than one per project? Thanks b

[crisp basalt]

i think it was upscaled automatically it seems

nvm its still 720p but wow it looks very high quality

[turbid lintel]

when veo4

[crisp basalt]

i used vertex media studio to generate that

[rocky tangle]

Hello everyone!
A quick question, please.
Is the "Veo3 Fast" model really free for "Ultra" subscribers?
For how long, exactly, did Google reveal?

[rocky tangle]

Hey.. Model.
Thank you.

Yes, I read that too. But is it only for a short time, or is it intended to be this way for a longer period?

I know it's only available to Ultra customers.

??? What do you mean?

[lofty crag]

hey I'm looking for AI artists to collaborate, would love to invite you as a speaker for our online event at AI/ML API

[long gyro]

None of my videos are loading on Flow. Been like this for 2 days. Anyone have any idea on how to fix this?

[stuck adder]

I just need to go and load it up there?

[stuck adder]

Btw is this their GitHub? For veo3?

[cobalt junco]

google cloud

[fair gale]

how come you have 9:16 ratio in flow 🤔

[vapid radish]

"Start with a realistic open school gate view. The camera smoothly moves through the gate into the campus. After entering, the next scene opens into a modern classroom. Inside the class, boys and girls are sitting at desks working on laptops. All boys wear white shirts, black pants, black ties, black coats (buttoned), and black shoes. All girls wear the same uniform with neat black hijabs. The camera then moves further inside the classroom, showing students focused on their work in a professional, modern AI learning environment."

[cobalt junco]

vertex test like wrigley

[vapid radish]

#flow-general・🌊 create a video Start with a realistic open school gate view. The camera smoothly moves through the gate into the campus. After entering, the next scene opens into a modern classroom. Inside the class, boys and girls are sitting at desks working on laptops. All boys wear white shirts, black pants, black ties, black coats (buttoned), and black shoes. All girls wear the same uniform with neat black hijabs. The camera then moves further inside the classroom, showing students focused on their work in a professional, modern AI learning environment.

[cobalt junco]

videos are not made here

[fair gale]

how do i get ar tho ?

[cobalt junco]

use vertex AI via google cloud

[fair gale]

ohhh

[vapid radish]

Which AI will make the video?

[cobalt junco]

the Gemini app or the Flow website

[vapid radish]

thanks

[fair gale]

do i have to pay for it ?

I have gemini subscription but if i gotta pay for vertex ai its different thing

[cobalt junco]

Google Cloud is separate from your gemini sub

[fair gale]

I see

[fair gale]

cant I do it in flow ?

[cobalt junco]

not in a 9:16 currently

[fair gale]

ohh

[crisp basalt]

i think flow team are working on integrating the 9:16 for veo 3 as we speaking 😄

[cobalt junco]

i have to assume soon 🙂

[fair gale]

that would be good

[stuck adder]

You have to have an acc with that etc right

They give u like $300 credits?

I wonder why it’s not integrated in flow yet

[gaunt crypt]

I am happy to jump on to the Google Ultra Plan from the Google Pro plan. Why? The announcement that: The "Veo3 Fast" model is free for "Ultra" subscribers?
I need to be careful with this.
Is the "Veo3 Fast" model free for "Ultra" subscribers in the UK?
Thanks

[trail coyote]

Is there a way to maintain the generated voice?

For example, since my video is only 8s i would like the next video of mine to have the same voice of my previous video

[true lava]

Why can't I get a free trial period on Flo (I live in Poland and I've already bought a trial period twice with my card, I don't know if it matters) I want to buy it on a new account

[wicked valley]

Audio has failed for me 5 times in a row (tried both VEO3 Quality and VEO3 Fast). Tried making the prompt as short and basic as possible and that didn't fix it either. It's 3 words. I don't understand why it's failing over and over right now. Also tried it in Gemini and it failed as well.

[junior horizon]

waittt I only see veo2 portrait mode on vertex, am I missing sth?

[empty saffron]

Yes. It is

[keen mesa]

Veo never does what you ask of it, always has bugs, it's tiring

[crisp basalt]

sorry to hear that lol

[crisp basalt]

It should be in the setting section

[stuck adder]

It’s only so much it can do bro lol

[tawny sequoia]

There are general things that google needs to consider and improve: To provide any mobile device (not just on a PC device) the option to use frames-to-video when creating a video. Also, I subscribed to AI Ultra plan to use Google VEO's highest limits. Unfortunately, VEO 3, whether FAST version or QUALITY version, does not provide "Frames-To-Video" and "Ingredients-to-video" and only "Text-to-video". If attempting to use the features other than "Text-to-video", a notification forces the user to switch Format to VEO 2 with No Audio generation.
At the moment, it seems that a user must master good prompts to generate a VEO 3 format video because the user is cannot use other tools than "Text-to-video". <@&1173646484311384114> I have been testing on video making with VEO FLOW. Those are needed general adjustments so the user can create with ease.

As far as Whisk, the first time I tried it, it was on mobile device. And Automatically it is in French into which the mobile browser has to translate the Whisk page. Whisk provides Ingredients-to-video by giving the user a page to try it out and input pictures that needed such as picture(s) for

1. Subject
   2)Scene/Setting
   3)Style

[crisp basalt]

I’m on mobile and I can use frame to video just fine

What mobile device are you using

[tawny sequoia]

But, the main language for Whisk is French and so clicking on the settings, the options are in French and also delivery is step after step

[crisp basalt]

What

Main Language is English tho

[tawny sequoia]

I just upgraded to the most recent Galaxy Series, the S25

[crisp basalt]

And what browser?

[tawny sequoia]

The famous Google Chrome

[crisp basalt]

And what makes you say that you can’t use frame to video at all when using phone?

[tawny sequoia]

I will provide screenshots..allow me

[crisp basalt]

You need to be more specific and clear when explaining about your issues

[crisp basalt]

Please screen record

Its much better than screenshot

[crisp basalt]

But whisk main language is English

Are you from France ?

[crisp basalt]

That’s not frame to video, that’s ingredients to video

Only Veo 2 is available for Ingredients-to-Video

[crisp basalt]

I don’t think you are using proper chrome browser here ..?

[tawny sequoia]

@crisp basalt Wait..my mistake. I really don't know why I am now seeing English after doing a new page on Chrome.

This was the other page

[crisp basalt]

Well if you still have trouble with the language : here’s proper link that uses en-gb English language https://labs.google/fx/en-gb/tools/flow

[crisp basalt]

Yeah that’s the France link

https://labs.google/fx/en-gb/tools/whisk here

[tawny sequoia]

Thank you Wrigley, as far a VEO 3, the frames-to-video is also not supported

[crisp basalt]

It’s supported tho

You are not on the correct uh dropdown from that screenshot

They’re working on ingredients to video
Veo 3 model right now btw

[tawny sequoia]

I switched to frames-to-video and there is an "extend" rectangular button but as I tap on it on my video in Mobile, no action is provided

[crisp basalt]

Uhh you don’t have to worry about that “extend thingy “ btw I think it’s a glitch , it happened after you refresh the page

But yea that is the right one for frames to video

Wait what are you trying to do again?

[crisp basalt]

I reread what u wrote and now im confused

Were you trying to .. use that particular generated video for frame to video?

[tawny sequoia]

@crisp basalt I inserted two pictures just to fill in as the frames. And I tried to tap arrow button so I can test if it can generate for VEO3, same error message of "this feature is mot supported on VEO 3

[crisp basalt]

Oh the start and end frame feature is only available for veo 2 only

Specifically fast model

[tawny sequoia]

Yes, I am testing what can function for VEO 3, so far only text to video

[crisp basalt]

veo 3 can do text to video and frame to video

Veo 3 model doesn’t do : Extend , Start to End Frame , Camera Control , Ingredients to Video ( the flow team actually working on bringing ingredients to video for Veo 3 soon

[crisp basalt]

This one you’re showing here , you were trying to use the start and end frame feature which veo 3 is not supported yet

[tawny sequoia]

Exactly, I am a tad disappointed thinking that if I upgrade my google subscription to Ultra Plan so I can gain access to "Highest limits to VEO 3", I woukd be able to use ingredients to video and frames to video for VEO3

[crisp basalt]

Did you not read the whole uhhh faq

[tawny sequoia]

But it would have to switch to VEO 2 w/no audio generation

[crisp basalt]

Veo 3 can do frames to video

I said this many times and I will say it again lol

[tawny sequoia]

Only for VEO 2 on Flow

[crisp basalt]

Veo 3 can

[tawny sequoia]

I can't

[crisp basalt]

You can , just upload one single image into the start frame and that’s it

No need to add image on the end frame ,

Tbh I think they should use image to video term lmao

I mean I like the uniqueness tho

[tawny sequoia]

It's just, experimenting with VEO 3, I would love to extend the video after that 8 second clip.

It confuses me..pretty overwhelming to be honest

[crisp basalt]

Ingredients to video term definitely the most confusing one tbh

I forgot the whole labs Google used to be ai test kitchen

It’s fun and unique but also confusing for people who are not familiar with the whole theme of ai test kitchen

[tawny sequoia]

I just read it, addition to VEO 3 Fast with Frames to Video

The update information

[dusty igloo]

Create a cute 3D animated scene of Lord Ganesha sitting with a bright divine glow. The background should look festive with flowers, diyas, and soft sparkling lights. Ganesha is smiling and gently moving his hand in blessing. Add smooth camera focus on Ganesha while he speaks. Use child baby voice-over for the dialogue in Hindi."

Dialogue (Hindi – Child Baby Voice):
"Doston, video ko pura dekhna… Aaj tumhara din achcha hi jaega!
Comment mein likho – Har Har Mahadev!
Aur channel ko subscribe bhi karo!"

Create a cute 3D animated scene of Lord Ganesha sitting with a bright divine glow. The background sh

[tawny sequoia]

Flow help center > stay up-to-date with changes in flow > July 7, 2025 Update : We're adding VEO 3 Fast option for Frames to Video

Enlighten me please

[kind bison]

Have a deadline to turn in a video. Using an image of the client. Flow and every 3rd party platform is blocking over "prominent people" bug. Have tried every single workaround. I could lost my job with the client over this. Thanks a lot, Google!

[tawny sequoia]

It generates but only in VEO 2

[dry frost]

Heck a google search might help too

Last frame and/or camera control in veo2 only. Starting frame will work in Veo3, as I've done that for my own project.

[cobalt junco]

https://tenor.com/view/doctor-who-doctor-who-the-interstellar-song-contest-the-interstellar-song-contest-ncuti-gatwa-15th-doctor-gif-18353590251090671795

[tawny sequoia]

Blue is starting frame, and red is end frame?

[dry frost]

You're still using camera control which isn't supported. What you've circled is correct.

[tawny sequoia]

@dry frost I see..

[crisp basalt]

https://tenor.com/view/spongebob-tired-sleep-deprived-exhausted-sleep-deprivation-gif-3654455578281021969

[tawny sequoia]

Y'all need to understand, for the very new user, I am a dummy in this. Ignorant. I know it's easy for ones to understand but we all have a starting point. I don't mind being mocked about it, because in the end, I have mastery over it.

I did not understand the terms used

[crisp basalt]

no worries

[tawny sequoia]

Thanks for your patience

[crisp basalt]

i think the ui itself are not self explanatory either so i ubderstand

[stuck adder]

Yeah I was confused too veo 3 only works for the first frame but not the end

I know they’re probably working on that now for veo 3 that would be a game changer

In the main time I think kling uses start and end frame

[prisma holly]

Can't flow create videos in 9:16 .?? Vertical

[crisp basalt]

Soon

[prisma holly]

Any other way to convert the flow creation in 9:16

[crisp basalt]

No

Cuz flow doesn’t do 9:16 as for now

If you want to generate a veo 3 video in portrait mode using veo 3 model, do it in vertex ai media studio api

[prisma holly]

Huh ! How

[crisp basalt]

https://console.cloud.google.com/vertex-ai/studio/media?hl=en

Just a heads up if you run into any issues or questions about Google Cloud Vertex Veo, the Flow team won’t be able to help since Vertex AI Media Studio is handled outside of Labs.google team

[severe sedge]

Audio is not coming in veo 3 fast anyone else are facing the issue ?
I face this issue today only

[deft helm]

same issue with quality model as well

[stuck adder]

What would you say kling is leading with?

[cold thunder]

Do you guys also have problems with upscaling?

for some videos it works & for others not

[long gyro]

Heyo,

I've been having the same problem for the past 3 days. None of my videos on Flow are loading - I don't have any problem with any other website.

I also can't upload a printscreens of it here, nor can I start a thread in the bug-discord. Don't know where to turn ples help

[long gyro]

ok now I managed to post about it in the bug feed, if it even is a bug i dunno

[wild jetty]

Guys anyone made any animations in flow? Or you suggest another tool for it?
Also how we can use Banana that was released recently

[valid zenith]

Hi I keep getting rule violation / can’t generate realistic people errors. I am in the US so shouldn’t it work?

[plucky imp]

Hello I have a question can I create ai ugc?

[prisma holly]

Di i have to buy the subscription of vertex or it counts under Google's

[crisp basalt]

Vertex Credit is not part of Google ai plan ,

If you’re a first time user in Vertex , they offer free 300$ credit for first timer (which I don’t recommend ,

[stuck adder]

Oh really

[prisma holly]

Thanks for the help

I wanted 9:16 for reels

[noble carbon]

i can not wait for the day ingredients to video has veo 3, that would be like christmas

[hollow warren]

Do ultra member get Veo3-, fast for free now?

[balmy basin]

Hi all!
I love Google Flow, but I cannot find an option to change aspect ratio.
I want to make advertisement reels for Instagram. Would anyone know how to change aspect ration to 9:16 please?

[gaunt crypt]

Just to be clear does this apply to the UK. I have asked previously and Theresa qualified this to be correct but I was just wanting a second opinion. I can see myself signing up for the Ultra and this not being available to UK users. Thanks in advance.

[kind bison]

VEO has now totally screwed me over. I am working for a famous person who needs a series of videos. I was able to get some created last week but now all are getting blocked. I have legal permission to use his image!! Why the hell can't we provide documentation to Google to support such situations so I can actually make a living! I'm going to lost my contract over this!

[hot silo]

I think it's like, er, shouting into the wind on this point, the Google people that pay attention here have answered as well as they can, and I thank them, and encourage them to communicate to their managers and stakeholders that it isn't going down well!

As for a way around your issue, I'd suggest trying one of the other sites that will generate on Veo3 for you, or, going direct to Google Cloud and signing up for a Cloud services account, going to Vertex AI, and using the Media studio there. Neither of these options are cheap, though Google Cloud was giving people $300 of usage if they sign up a while ago and may still be doing so. In the API it appears you are expected to take more responsibility for using generations legally and you can generate videos without the famous people restriction; although it would, of course, only be appropriate to do so as you presumably have permission from them to use their likeness!

[kind bison]

I have done all of that. No luck.

Well, I haven't tried my own API, though - I don't know how to do that.

[wild jetty]

i live in UK and it applies to UK users

[clear cipher]

Hello my friends
anyone need help?

[wintry fractal]

I need assistance creating an ai video

I also want it to be realistic

Any prompts that can assist me?

[mortal wing]

A newborn baby wearing a tiny racing suit and helmet is sitting in a small race car at the starting line. The engine revs loudly as the baby grips the steering wheel. A reporter leans in with a microphone and asks: “Why are you racing when you just learned to blink?” The reporter then stops talking and looks directly at the baby.
The baby smirks and replies in a clear baby voice while the reporter still looks at the baby without talking and listening to what he got to say: “Because crawling is too slow for my lifestyle.”

I do this and still the ai make the reporter talk instead of the baby

[clear cipher]

Do you want to just make the video? or make a stream

[crisp basalt]

it was kinda difficult but i kinda got it, and the prompt was actually quite long , give me like few mins to try distill this prompt..

[crisp basalt]

Actually, the length wasn’t that long oops. But yeah, this doesnt guarantee perfect results on the first try. I had to run two different batches with this prompt, and only this video out of the 8 generated outputs ended up with the baby actually speaking.

A close-up, photorealistic cinematic shot on a bright, sunny race day captures the intense atmosphere at the starting line. A tiny newborn baby, dressed in a custom-fitted, vibrant red racing suit adorned with sponsor patches and a miniature aerodynamic helmet with a clear visor, sits firmly in the cockpit of a sleek, scaled-down race car. Its tiny hands grip the steering wheel with surprising determination as the engine of the race car revs with a loud, throaty rumble, shaking the air. A sharp, professional reporter, holding a sleek silver microphone, leans into the frame, her voice clear and slightly incredulous as she asks, "Why are you racing when you just learned to blink?" She then pauses, her eyes locking onto the baby with a mixture of confusion and anticipation. The baby, with a surprising amount of gravitas, lifts its visor slightly with a confident smirk, its voice unexpectedly deep and resonant as it declares, "Because crawling is too slow for my lifestyle." The scene concludes with the baby's determined expression, a subtle anamorphic lens flare catching the sunlight.

another one , with different style of prompt structure (this one is quite long if u want to see the prompt dm me)

in conclusion,,,

https://tenor.com/view/fadding-baby-gif-16693579216185658153

he looks like a busted can of biscuits

https://tenor.com/view/baby-cry-busted-can-of-biscuits-gif-26977424

[sleek tundra]

Can we bring in videos from different projects into the scenebuilder/

[clear cipher]

what is this?

[limber bay]

How may I know this ?

[clear cipher]

me too

[brave crest]

sky with birds

//flow sky with birds

[gaunt crypt]

Thanks Arshia. You are using Google Ultra Plan and are getting VEO-3 fast for zero credits a shot - meaning VEO-3 fast is unlimited in the Ultra Plan. Thanks for clarifying this

[graceful arrow]

Hi, anyone can help. Why scene builder isnt showing up?

[wanton verge]

Anyone willing to help me explain why my videos are coming out like 360p 2D horrible animated videos with my prompts?

I see all these prompts with high quality cinematic styled themes and mine is like a 1970's animated cartoon with horrible quality

[wanton verge]

If anyone can help that would be greatly appreciated

[kind bison]

Google is targeting filmmakers with this tool and even has famous ones consulting for them. Obviously they are going to have to open it up to people (even if just a select group) so that they can use licensed/approved likenesses of known actors.

So I don't want to hear that ridiculous nonsense "JuSt UsE aNoThEr ToOl".

[astral pollen]

It’s not really nonsense your just not understanding what everyone’s been telling you it’s that simple use another tool or if you were actually worth paying find a work around

[crisp basalt]

what i think you should do is , to calm down

[crisp basalt]

exactly..

[mortal wing]

It’s just annoying u pay for smt and then get bad results

[kind bison]

LOL, another loudmouth Stan. Blocking you now.

[kind bison]

Easy for you to say - this is how I make a living and I'm working for a very high profile client that many people here would know.

Blocking you as well

[crisp basalt]

cool

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

[kind bison]

Too many unhelpful know-it-alls (who actually know nothing) in here.

[meager hearth]

Riley Reid?

[main saddle]

hey all! keep it nice please (#rules-and-guidelines・📋 #1) - if you've got feedback for Flow, please file it in #1319424187730755636. thanks!

[meager hearth]

do you work at the actual Google HQ or are you just an online person

like do you have access to the fancy buildings

[kind bison]

Funny enough, I was actually offered a job to do a series of SFW videos featuring her, but it wasn't really my thing, so I passed.

[meager hearth]

why did I get a warning

what the hell @main saddle

it was a question

[worldly yacht]

hello 😄 i wanted to ask you , i want to participate in the interview , but its asking me to switch account , i only have 1 google account

[grand musk]

https://labs.google/fx/tools/flow/project/daca9184-0aaf-41b2-9597-d288a78b1255

[gloomy helm]

Maybe

[crisp basalt]

yes

[meager hearth]

who are you

[crisp basalt]

Dwaine

[empty vine]

Anyone from Georgia

[cloud wadi]

got Failed Generation 5 times in a row. something down?

[mossy flare]

me too!! i was just about to ask in here.

[weak jewel]

Me too, they released VEO3 without needing credits for ULTRA users, it must be some overhead

[acoustic geode]

Same

[tropic night]

same

[amber bison]

Something is up, yeah. Getting the same.

[tropic night]

lots of failed generations today that weren't happening yesterday

[simple heath]

Hey guys, I keep getting this failed generation error while using Veo 3 (fast & quality) even though I'm working with multiple different prompts. How to fix this please?

[tropic night]

yup

we all having the same issue

you cant fix for now ,just have to wait for google to fix i think

[stuck adder]

Yeah I’m getting it too

[gentle harbor]

I think today we had a busy day...
Got this message many times.

[slim sparrow]

yep me too. a lot

[gray kestrel]

Answered my question. Thanks

[gray kestrel]

What's this

[stuck adder]

Have to be cause nano is spitting back the exact same image lol

[stoic oyster]

just got one through, maybe working again?

[vivid fossil]

Same issue, 1-2 coming through but never all 4 videos. I keep getting "Oops, something went wrong"

[brazen ice]

I came to comment the same thing, I thought it was just me

[deft helm]

Was there any official announcement regarding the no audio at all the past couple of days?
AI Ultra subscriber so tried both fast and quality modes

[crisp basalt]

Is flow having stroke

[lucid cobalt]

Is there any plan to enable Ultra users to upload reference images of people in more countries?

[sterile junco]

Is there a way around it? Using a VPN doesn’t work I think…

Maybe someone know a workaround for uploading “real” photos? Cause even genarated images of “people” don’t work.

It’s not breaking their rules tho, it’s more like breaking the “law”, they probably don’t care at all.

I hate the EU, they restrict so many things when it comes to AI..

Tell that to the European Union lol

Basically, it’s stupid, even Apple Intelligence (not that it’s useful or “intelligent”) is unavailable, including iPhone mirroring and so on… I’m really mad at the EU when it comes to AI..

I literally just switched to android because of AI, I’m still rocking my iPhone for the ecosystem but use my fold for work

[sterile junco]

I bought the 6 cause i didnt want to spend so much on a 7 since the only difference really is the camera, i have an iP16 Pro Max so i saved a few bucks and got a 6 and holy is it great for work

[hard dove]

Hey everybody. I'm new to the server. Just finished a project working almost exclusively with Google AI Studio and Flow. Super impressed with the creative workflow. It's very liberating coming from traditional production and post production pipelines.

[undone briar]

Has anyone found a workaround around the restriction on "photorealistic people" yet ? 😑

[fair bone]

you need a US account

has anyone figured out a way to have a voice over in different character voice, for example when you have a female in the image you want to animate and want to have a male voice over it always creates a female voice over and doesn't follow the prompt instruction

[stray dune]

https://youtu.be/KaHkweCqT-4?si=269V4TJK2vGaGYXd
I've made this film with Flow Ai

[idle compass]

But why does Google allow uploading of people as reference inages in the new Gemini image generator (Nano Banana) Shouldn't the rules be the same for both apps in the UK/EU. And that came out after Veo 3. Makes no sense to me that UK/EU can't upload people as reference frames in VEO 3 when they can in Nano Banana.

[tawny sequoia]

Finally finished a video that took me about 4 hours to complete from using VEO 3 start frame to video. Its tricky trying to maintain to replicate the character with the same facial features each clip

I will post it in creation gallery. Criticize the video so I can better myself

[tawny sequoia]

It was created with VEO 3 Fast using Frame-to Video. The first 8 second clip, I had to input my ideas from Gemini, as at the time I had Gemini generate videos for trial and error. When I found the video that went along with my idea in Gemini, I asked Gemini to give me the best descriptive prompt so I can replicate the video into Google Flow.

As I did so, using Text to Video for VEO 3 quality, that first 8 second clip became my base clip for the project. As you can tell in my project, towards the end the quality was deteriorating and the lighting of each clip was becoming more "warmer".

At the end of the video, I wanted to use the same characters from the prior video but the transition for using frame to video would lose one of the characters and thus I had to improvise the clip as it is difficult to replicate the same character into another scene.

If that makes sense

[silk nexus]

But why does Google allow uploading of

[scarlet saffron]

hi guys how can use real human face with veo 3

[graceful arrow]

Hi anyone? Scene builder is not showing up

[crisp basalt]

works fine for me

scenebuilder only for Computer/Desktop browser, if youre on android/ios, the scenebuilder / add to scene button will not appear

[graceful arrow]

Got it

Thanks

[rustic temple]

Did they just force crop for portrait images? nooooooo

[peak hatch]

anyone getting a headache from nano bannana?

It just probably has limitations but damn they are wildly different on different topics

[past tundra]

i cant use google flow, saying i cant make hiperralistic pictures in my country but im from US

[stuck adder]

What are you trying to do? So it’s great at editing and generating things it where it falls short

[sage finch]

Did something change with Veo3? The outputs are all over the place, prompts that used to provide consistent results now appear more randomized. I get dialogue that is not even included in the prompt. A recent example is the dialogue was completely changed from the prompt to "it's time to make a change, remember to vote" the prompt had to do with avoiding traffic collisions with elk.....

[tropic night]

dawg we gotta arrest u for this vid fr

lock bro up 🤣

[halcyon crystal]

where do AI video crators share their work and portfolios?

[livid silo]

The crop is broken in Frames to video. Anyone?

[tawny sequoia]

😂😂

[sage finch]

6843 N Stempson Avenue, Glasgow IR, knock on the pub door 3 times and say AI to the waitress.

[tropic night]

lmao

[tropic night]

2 things down is "flow gallery"

[livid silo]

what is this ? it used to take the whole image.

crop box was working yesterday. now you have to do this little window and it cuts off the rest of the image.
even worse for vertical images.

it was working yesterday.

do i need to use this on a phone now or something?

guess its time to cancel ultra

[tropic night]

i understand why that would be annoying for vertical images but

isnt the size of the crop just the default aspect ratio anyway?

[livid silo]

it's broken.

it was working fine yesterday

[tropic night]

i just dont see why an image would have to be a different size than 4:3 or 3:2 or whatever

but i see your point that it shouldn't even force the crop to begin with

[blissful eagle]

#flow-videos・📹 genrate a spaceship flying from moon and lands on earth

[tardy ridge]

YEP, LOL broke my workflow for a paying project

[stuck adder]

What do you mean?

[crisp basalt]

I’m gonna assume they’re preparing something

9:16 crop option perhaps 🤔

[dull pelican]

Guys , I am not able to crop the vertical image on google flow , is there a way to crop the image to 9:16 ratio . it was working yesterday .

[livid silo]

same issue. posted it in bugs

[dull pelican]

okies

[tropic night]

yo

is there any way to make an extended video with audio?

it switches to veo2 with extended selected

[midnight spire]

@tropic night 3 ways you can do it that I know of. If there is no cut/shot change, the easiest way is to just stick your clip into scenebuilder, take the image you want, and generate with veo 3 from that image. If you need a character in a new shot or scene, take an image of that character, drop it into whisk and fiddle with it until you get what you want. Then put that back in veo 3. The 3rd way is to use veo 2 first until you get what you want, then take that image and you guessed it back to veo 3. If you want consistent vocal performances, generate them in google studio and drop them in on your timeline in whatever NLE you use (im a resolve man, personally). Best of luck!

Yeah its a few more steps but jesus its incredible

Ive got a 15 min short nearly in the can

Would've cost me literally 500k to shoot it

Yeah, not at the moment for sure

Nice. You in la too?

Haha ok. A private person. Gotcha 😉

All of my filmmaker buds are not into it. You interested in sharing and some mutual feedback? Its such a new tool

Yeah that is pretty much the exact middle heh

One of my writing partners is from Omaha for what its worth

Very cool, I absolutely love the style

Curious about the narrative though and how thats going

Ive got 11.5 minutes. Interested? Warning: its quite silly

Im a writer director myself

Have one feature in distribution. 2nd one was coming along nicely, had all my cast, locations, crew...then covid

Nope. Ill look into it

But im a big dark city fan

Reminds me a bit of that

Lets take the convo to a private chat? Can't be putting up links in jere

Plus we are driving everyone nuts im sure who has to scroll past this ha

Those are solid influences. Well id love to show it to you but it might be getting sold so if you are down to go private then we can cook

But cant post it in a public forum like this

[proud cipher]

Korean dialogue never comes out in the flow.

[tropic night]

i made a rasta rat name rasta ramone

he spits heat bro

[bright patrol]

I don't know if there is a way i can feed veo an image such that it remembers it in all prompt, just like chatgpt has memory for every chat

Especially if i want to refer to that image in the next prompt

[devout viper]

that would be very cool ngl

[harsh sparrow]

got the same problem, and they haven't made a single fix, bad dev

[lilac hearth]

Is there any simple way download all clips of a project from the flow page ? while the Scene Builder is nice for some shorter projects, For longer ones I prefer to work locally

[lilac hearth]

Thank you for your answer. It would be great to have a button at the top saying download everything as a zip or something like that. Thank you for the feedback

And prompt search Text field would be nice as well 🙈 ☺️

[devout grotto]

https://blog.google/technology/google-labs/flow-resident-filmmaker/ - How can I apply for this?

[tall ocean]

Hey guys, is there a bug on Google Flow?
I can only upload landscape images

[devout grotto]

Okay

[lilac hearth]

Thanks for the hint 🙈 If you look sometimes too much on things you do not see them 😂

Here it is "Den Wald vor lauter Bäumen nicht sehen" which means the same. Never mind and thank you very much for the fast feedback

[bright patrol]

But it would be nice to have that

[green hare]

hi i have a question i want to try flow but when i send my prompt i got this

[green hare]

error of generation

[sturdy zinc]

I'm not able to download more videos in flow kindly how to resolve this issue?

[mint magnet]

how to animate vertical 9:16 image in flow?

[potent rain]

I don't have access to flow

[sleek tundra]

Dear Google Flow Team,,

I am an instructor on Udemy teaching students how to use Google Flow through my course: Google Flow Masterclass.

https://www.udemy.com/course/google-flow-masterclass-make-ai-videos-from-prompts-and-images/?referralCode=A7CF34641231E1CAB855

My training relies on live demonstrations using Flow, and I am running out of credits.

I kindly request additional credits so I can continue delivering hands-on lessons, which will also help promote Flow to a wider audience of learners and professionals. My email is [e-mail]

Thank you for your support.

[crisp basalt]

or u can top up your credit

[uncut igloo]

Hello guys.
I’ve noticed something strange on Flow. Until recently I could upload vertical images without any problem. In fact, I have a bunch of vertical photos already uploaded, and even though sometimes the video stretches them into horizontal, at least the full vertical image was still there.

But now, it seems like I’m being forced to upload everything in horizontal ,whenever I try to upload a vertical image, it automatically gets cropped and I can’t keep the full picture anymore.

This hasn’t always been the case (it definitely worked just a couple of days ago). Has anyone else run into the same issue or knows what might have changed?

Thanks!

[acoustic sparrow]

Ojala el team flow arregle el problema de consistencia en los personajes

Es dificil maniobrar con personajes que esten en escenas mas alla de los 8 segundos

[cosmic glade]

Gemini chats deleted automatically

Gemini chats deleted automatically

[patent dagger]

the look is stunning my friend. sound and ambience and character physics ar very accurate. beautiful work bro!

[grave linden]

i would also like to know what's going on

[calm temple]

Same. Really actually quite distressed to find out I can't upload 9:16 images anymore.

[grave linden]

Edit in photoshop, make the image 16:9

[calm temple]

Yeah I know, still, having to do that for every image kills experimentation

[sacred dawn]

the one time i agree with you

[peak hatch]

How much does it cost to generate via vertex?

nvm found it myself

Not the cheapest toy around

they changed the way you have to crop 1:1 frames in flow, yesterday i could insert a whole 1:1 frame and it would give me blackbars, today i have to add blackbars manually to every image in order to get the whole image uploaded 🙁

[silk nexus]

I don't have access to flow

[upper eagle]

hi guy, wanna ask about Google Ai ultra still unlimited free generate for veo3 fast?

[noble carbon]

Yes

[sacred fulcrum]

Hii

[left ledge]

Anyone know when Veo4 is gonna hit? Veo3 is terrible, and wrought with way too many problems to be useable. Especially since it's not free.

[crisp basalt]

Tomorrow 👌

[crisp basalt]

in 1 hour from now, veo3.5 will be release

[sour kelp]

Terrible is a bit of an overstatement.

[crisp basalt]

if veo 3 is terrilbe according to @left ledge then why are there good veo 3 videos out there 🤔

are we using the same model?

same model but probably different mindset

but yes veo 4 hit tomorrow @left ledge

😄

[left ledge]

It's really not. It's so inconsistent. Never keeps the same characters even when I never change the descriptions. The physics are worthless. The voices are horrendous. The scene addition works on occasion. Stop defending mediocrity. If there are "good videos" out there, it's because they spent a truck load prompting over and over.

[crisp basalt]

so.. you expecting a good result just from , first try?

[left ledge]

No, i'm expecting a decent result from my 50th try

[crisp basalt]

and u decided to blame the model?

not your prompt?

[peak hatch]

really? 😄

[upper eagle]

Veo3 is next level of video generation. I am very satisfied with veo3 performance.

[crisp basalt]

yes its true :

[left ledge]

well seeing as I can get what I want from trying over and over without changing my prompt, yeah. There's no consistency.

[peak hatch]

U must have not used any video generators before this one

From my perspective it took me 1000 clips to get 70 usable ones previously

with veo it took me ~200

[left ledge]

are you people bots? Does someone not agreeing with you, challenge you somehow?

[peak hatch]

and the quality is 10x

[crisp basalt]

@left ledge yes we are

[left ledge]

Yeah, taht's pathetic.

[crisp basalt]

The server with most human are next door

[upper eagle]

i tought u were just kidding, u sure about that? veo3.5? lol

[peak hatch]

https://tenor.com/view/no-bugs-bunny-gif-18219884

[crisp basalt]

you are free to leave 🤖

[left ledge]

less than 10% average for a good return is trash

so are you. please leave.

[peak hatch]

Just buy the higher tier subscription and u have unlimited generations, then you dont have to worry about the consistency

Lower tier is hobby tier, higher tier is professional

[crisp basalt]

no i agree with you, i agree that , you decided to not take advantages of the power of ai to help you with your skill issues.

[left ledge]

I dont even care about the money. I dont have 100 years to sit and prompt over and over. Its grueling

[crisp basalt]

but yeah anyway 😄

Veo 3 is trash

when the user is trash at prompting

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

[peak hatch]

It really is a skill issue though, the more u prompt the faster u become, and its only a tool its not a magic wand that transfers ur imagination into video form

[upper eagle]

https://tenor.com/view/lilnouns-nouns-lil-nouns-dao-noggles-gif-27131043

[peak hatch]

ur right tho its frustrating to get bad results i get headaches every day

[left ledge]

I've been prompting for 3 months. I've made 4 trailers. Gone viral. The issue isn't always the user. Veo 3 genuinely needs massive improvements.

[crisp basalt]

well, anything is possible , 👀

tbh , its probably gonna be next year for veo 4

google i/o 2026 etc

[peak hatch]

is 3.5 gonna be first + last frame?

That would be huge for me

or ingredients

[left ledge]

Well that sucks. it's such a nightmare to work with, I might just ditch the whole thing until it improves. It's too buggy for me, and my sanity

[crisp basalt]

I think ingredientts to video first

based on the changelog announcement

[peak hatch]

either is huge tbh

[crisp basalt]

i saw kling start + last frame feature and im actually impressed

[upper eagle]

I’m not sure if Veo 3 can actually help me create projection mapping content. I haven’t seen anyone use it for this on YouTube.

if veo3 can be first+last frame, we can made a movie

[fierce valve]

thanks for your interest! Flow Sessions is currently invite-only. this server is a great place to share your explorations with Flow and help us discover next & future collaborators

[left ledge]

I've found that on multiple ocassions, an output will be about 90% what I want, but with one glaring issue that makes in unuseable, so I'll back and fix the camera angle or adjust the lighting, or something, and then on the next output, the video is radically different. People's head's are spinning. Cars are morphing from front to back. buildings suddenly dissapear mid shot. It needs some serious improvements. Not just quality of life adjustments.

[peak hatch]

@crisp basalt Im stunned now cant focus on my projects, waiting for release

https://tenor.com/view/shocked-surprised-omg-oh-my-god-gif-16260557

[stuck adder]

I’ve found using JSON promoting is much better than just writing plan text

[wary sparrow]

https://tenor.com/view/disappointed-disappointed-fan-seriously-what-are-you-doing-judging-you-gif-17485289

What's going on with the Male Voices Flow Team, working on a Daytime Talk Show here and his voice changed. I just upgraded to Ultra, come on now guys.

[upper eagle]

im able to generate 2 veo3 video on gemini everyday, but using flow is unable.

[left ledge]

I find JSON prompting fine for individuals with single action in basic settings, but it's too cumbersome a mechanic to prompt activity heavy scenes.

[kind bison]

For consistent voices, it's best to do a voice swap using Elevenlabs, then do EQ and ambience matching to the VEO version with audio software.

[kind bison]

Ignore and block the Stans in here - there are some folks who apparently live on this server 24/7 and act as if they are Google stakeholders. Even the mildest criticism of the platform and model, they act like you insulted their mother.

I love VEO and use it every single day - think it's amazing - but it needs a ton of improvements for professionals to be able to reliably use it. We should be free to express constructive criticism without simps jumping all over us.

[left ledge]

my credentials. A trailer I made using only Veo3. I'm aware of it's capabilities. I'm no tourist. but yes, it 100% needs a ton of improvement. https://x.com/KingCamelMoon/status/1943392777045446793

King Camel (@KingCamelMoon)

NEW Trailer! This time I wanted to focus on multi-cam, shot consistency and dialogue interaction.
'The Shadow & Dick Tracy: Devi's Due' Movie Trailer
Made with Veo3 (12500 credits) and CapCut
#aivideo #Veo3
@RAZ0RFIST @hcnoel @TheCriticalDri2 @venturetwins @FillPops

X · Jul 10, 2025, 7:31 PM

[stuck adder]

Finally lol

I see they have the crop different now it doesn’t use the back spaces anymore

[coarse tendon]

slay

[worldly yacht]

We love Batman

[tardy ridge]

Is it fixed?

[wicked ravine]

what is going on with Flow?!?!

1. Completely random ai generations getting constantly flagged as “prominent people”

2. Just in the last few days, I’ve noticed that no matter what you type for the character to speak it just completely goes into some random monologue that has nothing to do with anything.

is this known

because obviously there is no such thing as support on Flow 🙂

https://youtu.be/DQDzms_4d04?si=R4txw_GWc2W2mWO1

for context i have a lot of hours on Flow these are both new issues

[vale tulip]

Cereal of Keycaps

[meager wolf]

Hey i'm new here but flow getting worst and useless update. Now i can't generate image to video with ratio 9:16. Can you guys make update that helpfull?

[vale tulip]

You hit your limit in the app that’s why

[drowsy egret]

anyone know what is the best setting to keep your character consistent? i tried frame to video but it appears totally different from image in the video. any settings i am doing wrongly?

[crisp basalt]

Have you tried using image generation feature within flow itself , the one that lets you upload your own image/subjects and then generate it into a starter frame scene

[drowsy egret]

as in i have the. image i want but when i upload frame to image it comes out as another person

not at all like my character

[crisp basalt]

No no

[drowsy egret]

is there any settings to do to be able to keep my character consistent

[crisp basalt]

It’s this one

[drowsy egret]

as in i have the image of the character i want

i need to use a image generator again? but i dont need another character

[crisp basalt]

???

Bro calm down for once and read again

I know it’s frustrating but it’s more frustrating for both of us if you’re not on the same page as me

Have you tried using that to generate a new scene with your character??? @drowsy egret

[drowsy egret]

ahhh i get what you mean. i havent yet

let me try it right away

[crisp basalt]

Alright no worries take your time

[drowsy egret]

so the best way is to use image generator first then after wards use that to create a video?

[crisp basalt]

Just click “generate” and it should load up this feature

[crisp basalt]

This is just optional if the veo still stubborn to maintain the consistency of ur subject

[drowsy egret]

do you know where i can get to generate image?

[crisp basalt]

Click on frame to video

Then press on plus button , and it should bring up generate and upload button

Click on generate

Tbh, I really don’t like how this thing is like you have to dig through many things until you can find the “image generation “ feature

[nocturne echo]

Is anyone here building their own blockchain or a full-stack dApp and in need of a developer? I'm a full-stack blockchain and AI developer with experience building core protocols like DeFi, DEXs, DAOs, and NFTs across a wide range of environments including EVM chains, Solana, Near, and ecosystems built with Polkadot's Substrate and the Cosmos SDK. I'm particularly focused on integrating AI with smart contracts to create more intelligent and adaptive applications. Let me know if you could use my help on your project, thanks!

[crisp basalt]

You can also create your own “objects” or “scenes” by generating them within the image generation feature … which I think that’s .. hassle too. But yeah , You can upload any image you want, and when writing the prompt for the scene, make sure to reference the images you uploaded. For example, if the image contains a man, just mention “a man.”

This feature also still slightly “hit or miss” but the good thing is you can iterate and generate as much as you want without using up your AI credits.

[cobalt junco]

https://tenor.com/view/elf-will-ferrell-buddy-elevator-pressing-all-the-buttons-gif-3577196

[crisp basalt]

Literally me having to find the “image generation “ feature itself in Flow

[cobalt junco]

lol yep it took me a while for sure to figure out what is what

[crisp basalt]

Also pls bring back the ability to .. not having to crop your image to 16:9 when using image generation feature…

[drowsy egret]

thanks so much man! appreciate it. i think it improve alot. So the starting scene right now, it nailed it after i use the image generation but once it switches scene, the person face visually changed

do you know if there is anything else i need to do to help keep it consistent

[crisp basalt]

Did you use veo 3 fast or quality

[drowsy egret]

fast

[crisp basalt]

Ah that’s .. actually I’m still figuring that out but

[dry frost]

Holding a copy of the prompt you used for a character that got created tends to help a lot

[marsh dawn]

use whisk, generate a character, save whisk prompt of that character and make images in whisk. Upload them to flow to make frame to video.

[crisp basalt]

When you prompting a scene, make sure you include the word or description of the character from the image you want to animate or turn into video.

[dry frost]

I also like to give my characters names

[crisp basalt]

Yes this!

[crisp basalt]

Whisk is also nice way to generate new scene with your characters but that’d be a bit of hassle having to switch between flow tab and whisk tbh but yeah whisk is actually a very good tool too

[drowsy egret]

so i want to put "myself" into the scene but my looks change dramatically after first frame. haha

but for veo i can only upload 1 image right

meaning if i want it to change scene in the middle i cant upload another generated image

[crisp basalt]

You can just prompt it easily like “make my character walk towards etc etc etc”

[marsh dawn]

@crisp basalt i tried to use whisk character prompt to flow, but its not accurate. The best way i know is to make pics ready on whisk. If you use same seed, u never have problem.

[crisp basalt]

Uhhhh you can just use “frame to video “ again with a new scene image you generated

Are you using the scenebuilder ?

[crisp basalt]

Sometimes I think the whisk generated “caption” doesn’t reflect the image , at least based my own experience

That’s fair

[drowsy egret]

yes i am. i am trying to build a continuation but somehow failed within the first video when i tried to change scene

the character face visually changed

[crisp basalt]

Did you use JumpTo

While using veo 3 fast

[marsh dawn]

@crisp basalt did you upload the image on subject? what i do is write a prompt of a character, make a black background, save the seed, upload the image of character to subject and use the prompt in subject to make it consistent character. it's bit of a mission but it's worth it.

[drowsy egret]

as in from the 1 video. meaning first 2 second one scene and leads into another

[crisp basalt]

actually they do now, just turn on “precise reference” toggle

[crisp basalt]

Yeah ( I think? Sorry my English comprehension skill is poor atm

[crisp basalt]

Which it’s STILL a hit n miss but it can actually use the reference if you prompt it properly

Which I still find it hassle cuz even simple word that I use to describe the reference image , whisk still refuses to follow

[drowsy egret]

how about audio? how do you all maintain the same character voice throughout the video over mutiple scene

[crisp basalt]

actually , you can remove the background of image of subject by just upload the subject image , turn on precise reference then write “remove the background . Change to dark grey background “

[marsh dawn]

elevenlabs

[crisp basalt]

A lot of peeps still figuring this out but , describing the voice of character in your prompt kinda helps

[marsh dawn]

@crisp basalt oh ok, ill try that next time

[drowsy egret]

you export the video and dub over it?

[crisp basalt]

Yeah

Eleven lab actually uploaded official video on how to use their tool for your veo video

https://youtu.be/tPP7D0MIOks?si=aTg3oj8q82Sh3Tsi

[marsh dawn]

@drowsy egret nah, u upload your video and choose the voice u want to use, it generates it automatically. So useful.

[crisp basalt]

Sorry I meant to say no

LMAO

This one is easy and quiet fast method too 👍

[stuck adder]

No, they’re probably prepping for a update

[stuck adder]

It’s it just easy to just do ADR?

[crisp basalt]

I’m not familiar with that acronym or whatever ppl call it

[tardy ridge]

Hopefully soon I have job starting I found a hack but it adds time and not reliable

[stuck adder]

Do a voice over and sync it with the character

Automatic Dialogue Replacement

[brave coral]

hey all, is there any rough information about when minors will be allowed in videos?

[silk nexus]

Yeah

[queen temple]

I saw this on tik tok, and you have always the same character! How you managing to do it?

[pulsar cedar]

Does anyone know why I can't generate any videos based on text ?

[dapper egret]

Hey everyone its about 3 days since I generate videos on veo 3 or flow as it is not generating video with audio. Can anyone having the same issue in India pls help

[silk nexus]

I know it might seem like a silly question but you do have it set to either Veo 3 Fast or Veo 3 Quality, yes? Just that Veo 2 doesn't create sound so just want to check first. If you are using one of the Veo 3 settings then you'd need to show us one of the prompts you are using and then someone could perhaps have a look at it to see if anything sticks out there.

[dapper egret]

I have set it on veo 3 fast with beta audio

[silk nexus]

Okay and the prompt? Any one will do. I'm not saying I will have an answer but without an example of a prompt that isn't working then I'm afraid nobody will be able to really help you out much 🙁

[smoky frigate]

Go to Chatgpt and convert your prompt into json prompt... then it will work fine

[near narwhal]

has any1 experienced that they get charged 40 bonuspoint/credit while using the veo3 fast 8 sec video generator instead of 20 that is discribed?

[silk nexus]

Not here no. Only way that 'should' happen is if you've got it set to create 2 outputs per prompt?

[loud kite]

Hello, for almost 10 days, I can no longer generate video, even the most basic. I have a "generation failure" message. I tried several browsers. Can someone help me? Thanks in advance

I even tried with a quick as stupid as: "a man walks on a beach". No way

I had the answer. When you write the prompt in French, it doesn't work anymore ... you have to write it in English!

[loud kite]

isn't it ?

[limber ember]

Create an ultra-detailed caricature in 8K, ultra resolution, ultra-realistic with cinematic textures and lighting, of the character "Mano Sarrada".

Keep the same exaggerated caricature appearance, facial features, humor, and personality as in the reference images. He is wearing luxury, flashy, and colorful hip hop clothes customized with his brand: oversized designer T-shirt with "Mano Sarrada" in bold graffiti-style print, luxury shorts, branded sneakers, chains, rings, watch, and a stylish cap with "Mano Sarrada" in graffiti letters. His outfit must look extravagant, eye-catching, and comically fashionable.

The scene takes place at the Sambódromo da Marquês de Sapucaí in Rio de Janeiro, but in a comical way: Mano Sarrada is parading alone in the middle of the avenue as if he were the main attraction, striking a funny and exaggerated pose, with confetti and streamers falling around him. Tourists and random people in the stands are laughing and filming with their phones.

On the wall behind him, add a huge graffiti/pixação-style phrase in Portuguese, very bold, colorful, dripping spray-paint effect, witty, funny, and rhymed:
"No samba da vida eu sou a atração, Mano Sarrada na Sapucaí rimando com emoção!"

The artwork must be vibrant, exaggerated, cinematic, colorful, with graffiti/pixação aesthetics dominating the scene and the phrase extremely visible and impactful.

[pulsar crescent]

I used to be able to generate portrait images in Flow if I started with a portrait image but now I cannot. Am I doing something wrong, or has something changed?

[silk nexus]

Create an ultra-detailed caricature in

[frigid pier]

i have been paying for veo months. I'm making some money from videos finally

[merry plaza]

Please stop forcing us to crop the videos , this completely messes up our content on TikTok.

[worldly yacht]

yeah spill the sauce

[mossy narwhal]

Hello Flow Support Team,

I have an active Ultra subscription that expires tomorrow. I have only used about 10% of my credits, so I would like to kindly ask if you could extend my subscription or provide a way for me to use the remaining credits.

Additionally, I was supposed to receive extra credits so that my total would be 25,000, but these additional credits were never added to my account. Could you please check this issue and add the missing credits?

Thank you in advance for your help!

[lilac walrus]

Create video

[ionic mortar]

Hi ! i got a Pro gemini Sub, i want to create with flow, but when i try flow said to me " fail to generate" do you have some idea why ?

[wooden monolith]

Additional credits are only to be given to those who made payment after the announcement. And no, they won't do any extension, any unused credits will be gone after your subscription ends.

[stuck adder]

Just already put a 16:9 cropped black bars in it, it’s a extra step but it still works for now

[stuck adder]

Facts

[swift merlin]

Help/question on static shot starting with Frames?..... Not sure where to post this, does not seem like there is a "prompt craft" help thread like in MidJourney Discord. But Anyway No matter what I tried in Veo3 fast or quality I can not get the camera and shot to stay static so there is no movement what so ever expect for the scene movement in the background windows (moving people, cars, rain etc) . Anyone have an idea how I can get this to work? I need the shot static because the windows will be composited into live action existing footage inside the club. Thanks for any help...

[silk nexus]

Anyone know why a (as far as I know) correctly formatted JSON prompt given to Flow will sometimes give the wrong character the wrong line. Two people talking to one another and they say each others line.
Place the same prompt back in again and render and it does it fine the next time. What exactly is going on there?

[worldly yacht]

Is it only us in europe that doesnt have frame to video / or image to video ?

[silk nexus]

Where in Europe are you? They just made it work today for EEA, UK and CH.

[worldly yacht]

In cyprus mate

[silk nexus]

Hmm strange then as it should be working there. Have you tried logging out and in again, see if that has any effect?

[worldly yacht]

ohhhhh they actually did ! i just saw it ! thanks man !

[silk nexus]

No worries. Glad it's working for you 🙂

[mossy narwhal]

Thank you, of course, but you haven't made me happy.

125 dollars down the drain

[woven yarrow]

guys

Why does it force me to crop when I create frames to videos?

[kind bison]

There's no difference in outcome with JSON vs natural language prompting. This is just a persistent VEO 3 bug, no matter how you prompt.

[silk nexus]

Okay thanks. Thought there may have been something obvious I might be missing. I'll just slog on. Thanks again!

[tough garnet]

I’m having problems with the speech in flow. The characters are mumbling gibberish before the dialogue and this is costing me dearly because I have to keep generating until I get the correct render. Please fix this problem.

[crisp basalt]

My speculation for next week : 9:16 mode aspect ratio is coming for both Veo 2 and 3.

[tired pier]

When is portrait mode 9:16 coming out for veo 3?

[crisp basalt]

It’s already out for veo 3 model itself. I’m not sure when is Flow gonna integrate the 9:16 aspect ratio but maybe Flow will get it soon

If you want to generate veo 3 video with 9:16 aspect ratio right now, use Veo 3 API model in Vertex Ai 👍

[tired pier]

Will the veo 3 fast still be 0 credits though?

[crisp basalt]

Wdym will, it’s always 0 from now on for Ultra Plan user

[tired pier]

sorry still new, is vertex ai free and officially by google as well?

[crisp basalt]

Oh no worries, also it’s not free but new users who sign up in vertex ai will get 300$ credit starter for free

Yeah its by Google , specifically Google Cloud Team or something

Meanwhile Flow is by Google Labs Team 😃

[tired pier]

So vertex ai is one time payment, and since I have ultra plan, my veo 3 fast will always be 0 right? Seems the $300 credit wont really apply to me. Unless I'm getting something wrong.

[crisp basalt]

Yeah, Vertex AI isn’t tied to the Google One AI Plan at all. The $300 credit can actually be used for more than just generating Veo videos, but honestly, I wouldnt recommend using vertex ai stuff it if you’re just a casual AI generation user

[tired pier]

Yeah I'm just casual veo 3 flow user. I don't really mess with anything besides the video aspect.

[cosmic cave]

Dropping my Movie Tonight 8pm est. This one came out so special: https://youtu.be/dOVjKkbsQ3c

[potent bear]

What’s the correct size for images when creating video?

[silk nexus]

Timmy The AI Road Trip Movie | Family Co...

[pulsar crescent]

Is anyone going to answer my question?

[bronze whale]

Any chance of changing orientation and ratio at some point?

[stuck adder]

You still haven’t told me how to do it

[nova oracle]

Hello everyone, I'm new, I'm a free plan user. Do tariff credits change daily or monthly or are they not replenished at all?

[crimson pawn]

I've shared a Brand Store on Amazon with you. https://www.amazon.com/stores/TinaCoCustomGifts/page/A1C9338E-1340-4C8B-AE2B-BFC7F985684F?lp_context_asin=B0B1YTVQ2C&store_ref=bl_ast_dp_brandLogo_sto&ref_=cm_sw_r_apann_ast_store_PYHRYXN4FV4JV7W1AZ1E&tag=mshop-android-tmo-all-all-phone-pre-us-20

[crisp basalt]

👍

[clever flint]

Flow is awesome
But I haven't found how to send it a json prompt. Is there a reference sheet for the json structure? Thanks

[green laurel]

Hello everyone, I’m just starting to learn how to create videos from images using Google Flow. I tried making a video on Veo 3 but it’s not supported, so I created the video on Veo 2 instead.

The video I made is in the format where image 1 smoothly transitions into image 2. However, when I used images with visible teeth, I ran into the issue where the teeth appear to move along the lips, or the mouth looks like it’s talking.

Could someone kindly show me how to fix this? Thank you very much!

😢 😢 😢

[fierce nimbus]

We are working on a fix. We rolled out an update to the crop tool that is stopping you from using portrait images. Update coming early next week. Sorry about this.

[fierce nimbus]

You can use JSON in Flow’s prompt box. JSON prompting is completely optional as we understand natural language. Some creators find JSON helpful as it encouraged them think about every detail of a generation (like a form).

[clever flint]

Thanks for the answer I guess there's nothing official for the schema or fields

[fierce nimbus]

Nope. Theres actually no statistical evidence that JSON is or is not better than natural language. Its just something some creators like.

[crisp basalt]

As long as your Veo 3 prompt is written clearly and descriptively, without being vague, it should work fine

[clever flint]

All I can say is this is an amazing tech

[crisp basalt]

The most simplest prompt that I love to use with veo 3 would be “create a video about (insert scene that you wanted) “ and submit

[crisp basalt]

That’s where Gemini steps in to “expand” your prompt behind the scenes when using Veo 3. As long as what you wrote is clear and understandable enough for you to be able to envision it yourself , then you’ll be fine

[serene heart]

Can I make ai vedio there I'm new in discord

[crisp basalt]

No you can’t

You create video in https://labs.google/fx/tools/flow

Speaking of iterating, I was watching this Google talk on filmmaking AI, and there was this one part that really stood out for me , I think everyone here should see and understand it

this part

https://youtu.be/1M6RqtYEca0?si=BWCidxmRMdYoEPmh&t=578

[crisp basalt]

It just goes to show that people need to realize not everything will give you a 100% perfect result on the first few tries. You have to keep iterating until u get what youre after. Ive burned through who knows how many credits just tweaking my prompts. The point is u cant expect perfection instantly, and u definitely cant assume the AI will follow exactly just because u saw a “perfect” Veo 3 video on YouTube or TikTok

Who knows how many iterations they had to grind through just to produce that one “viral Veo 3 video,” lol

oof god knows how much i need the ultra plan right now lmao

[balmy hill]

12-second ultra-realistic video of factory workers operating advanced machinery, sparks and lights in the background, cinematic wide-angle shot.

Me

[analog radish]

@main saddle is there a reason why 9:16 isn’t included in the flow? Since most videos nowadays are short-form and viewed on mobile devices, this format is preferred. Right now we’re spending extra effort using different tools and hacks to achieve it. are you considering adding support for it?

[fair gale]

I hope they are working on it, cuz people mostly consume content on ar 9:16 nowadays. Why its not the integrated had me perplexed.

[crisp basalt]

No they won’t!

[crisp basalt]

deserved

[fair gale]

😔 🙏

[noble carbon]

that would be so cool