flow-general・🌊

[thorny dagger]

@sleek vale Explain more? API format?

[sleek vale]

Using the Veo 3 or Veo 3 fast api via Gemini or Vertex. I can’t seem to find a way to “cut to” or “extend” clips the same way you can in Flow. Didn’t know if it was just a flow feature or also possible in the API

[crisp basalt]

I mean you can use Extend in Api , the model is called Veo 2 exp model in vertex

as for jumpto, thats Flow stuff for now

[sleek vale]

Good to know. Thanks. Can’t really find documentation anywhere.

[crisp basalt]

i think JumpTo is basically,.. ingredients to video

[thorny dagger]

I can find the exact API call for you

[sleek vale]

If you don’t mind.

[thorny dagger]

Just hang tight and give me a minute, I'll have to whip something up and then check the network log

[sleek vale]

No rush.

[thorny dagger]

I'll provide the request in JSON is that ok?

[sleek vale]

Perfect

[crisp basalt]

https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/vision/getting-started/veo2_advanced_controls.ipynb

actually you can just use it from the ai media studio lol, idk why i sent this

also veo 2 inpaint/outpaint is on colab/github/api but it seems to be in experimental

https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/vision/getting-started/veo2_editing.ipynb

[thorny dagger]

Ok, awesome. You should DM me so I can ask you some dumb questions about the platform, so maybe I can grab you some nice request payloads, other than the splicing/splitting

[sleek vale]

Thanks

[crisp basalt]

no prob

[graceful grotto]

Hello I'm trying to launch Flow following my free subscription on Google Pro AI but I can't and I have this message. I should point out that I'm over 18 and I live in France.

[crisp basalt]

im confused

[crisp basalt]

https://labs.google/fx/tools/flow try this

[crisp basalt]

since youre on iphone.. try open Flow Website through Safari Browser

[graceful grotto]

I just tried again but it doesn't work, I'm asking what could be the source of the problem but thanks for the quick response👍

[thorny dagger]

@graceful grotto https://myaccount.google.com/age-verification?pli=1&utm_source=p0

[graceful grotto]

My age has been verified, but I can't locate the problem.

[analog lava]

Hello, as a friendly reminder images or videos cannot be generated directly in the discord server thank you!!

[crisp basalt]

uhh are u on pro plan or not

[graceful grotto]

I want to make AI videos but I'm not on a pro plan. I took the subscription at 0 euros for the first month

[crisp basalt]

first time user will get 100 credit for free

[quick rapids]

the pro plan is their basic paid plan that the intro deal covers, its not like they'd offer a $0 discount on a plan that's already free, so..

should include the standard 2500 credits, but maybe there's fineprint on the offer

2500? 5000? I forget

[graceful grotto]

Okay, so they don't offer anything and it's just a marketing technique to attract customers? In the end, we don't benefit from 1000 free credits

[quick rapids]

it should come with everything the pro plan includes

wait ai pro is only 1k credits thats so low lol

[crisp basalt]

did u spent all 1000 credit

[graceful grotto]

Okay, so I don't understand why I don't have access to my 1000 free credits for the first month or the Flow app

[crisp basalt]

are u on pro plan ornot??

if u subbed to pro plan, you will get the 1000 credits

[graceful grotto]

Yes, I took Google Pro for a month (it's the free month)

[crisp basalt]

ok , when did u subscribed?

[quick rapids]

you should be able to see what plan you're on by looking at your user icon in flow. if you're not seeing "Pro" next to this then it's probably a billing cycle thing since credits don't roll over, if you are seeing Pro then it's a bug and I'd hit up support

[crisp basalt]

u can also click on the credit balance and it will open up page where it shows the details of your ai credit

[subtle plume]

does flow generate the sound and music , I created a simple prompt but no music or audio getting generate

[ruby roost]

I get it to produce voice from a character consistently, and sometimes sound effects (that generally are not quality and don't sync up to the event), i don't attempt to have it make music. If you are stitching your 8 second shots together there's a nil chance you will get the music to sync from shot to shot.

[graceful grotto]

It is marked that I have 1000 € of monthly credit which will certainly come on October 19. However, I paid for the free month in which I was supposed to have 1000 credits to make my videos. I have no additional credits and I do not understand why access to Flo is impossible for me and why I do not have the credits is it a time problem

Not € but credits *

[graceful grotto]

⬆️

[crisp basalt]

When did you pay for the free month

Based on what you said, you’ll. Have to wait until October 19 then for credit to be replenished

[graceful grotto]

I took the free month three days ago. I was told that there should be 1000 credits there that will allow me to make videos with Flo. However, I have not had access to these 1000 credits or to Flo. I contacted them on Google One, representatives who will certainly answer me tomorrow, I thought that with the first month's subscription at zero euros instead of 21 € I would never access the application and the 1000

[tranquil apex]

Anyone pro at extending their video with frame-image? Anyone know how to make a really consistent character for a lon 2-3 minutes scene?

[subtle plume]

Yes Veo 3 is not supported in Flow for extending the scene , it can be done only with Veo 2 but the trade off with Veo 2 is it wont generate voice

I want to generate a character for 2 to 3 minutes scene

[sacred kettle]

this consumes moeny from your credit card directly right? is not from the subcription? i m still confused. i get a lot of different sources that tell that calls to vertex may or may not use your credit card but rather your google one subcription credits, others say vertex by bein another platform consumes from your credit card bacuase is not the same platform , but you can actually achieve this for gemini , veo and labs with whte gemini app token, while many others tell is not possible. gemini itself? .... tells me both thngs at the same time. can i ask you where the api calls to vertex are being charged to you? credit cards or ai credits from google one? any experience trying out same on gemini app studio? thanks wrigley in advance

[crisp basalt]

Google one AI credits aren’t connected to Google Cloud Vertex at all, so use them responsibly

[sacred kettle]

yeah i was inclined to get such conclusion. thanks for the feedback. gemini app studio is the one that i get the most rather contradicting answers. in internet and by gemini itself. and perplxecity, claude, etc.....

all with deepresearch and internet access activated in their most powerful models i must say

[crisp basalt]

I also got that too haha Gemini couldn’t answer my inquiries about vertex stuff

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

[analog lava]

Hello, as a reminder we cannot create images and videos in the server! Also as stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now. Thank you!!

[last thicket]

"Can't generate video. Try again later". But it used 400 credits and haven't gotten that back yet

A minute ago happened

Thats very sad i just lost them lol

[young vigil]

I keep running into "Export failed!" errors once a project gets up to around 7 minutes long. Is this a known limitation?

[night hinge]

guys has anyone create scenes with rain in flow? im having a hard time

[errant tangle]

I can't seem to create a music for an entire video?

Use case is that I have a few scenes and now I would like to have a single music throughout, with a single style. It's just not possible to generate 4 videos that have a same style audio

[crisp basalt]

nope

(in my opinion)

[errant tangle]

a bit sad

[tepid tundra]

You'd have to create the music separately

[full flicker]

Hello, a friendly reminder. As stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion/comtent in this server will largely remain in English for now.” Thank you.

[silk pelican]

why cant i get in flow it says flow television?

[crisp stag]

is flow working for anyone?

[limber breach]

yes bro

[crisp stag]

it wont load for me

for few days now

[limber breach]

ohh okay i make use of klaviyo flow

[limber breach]

did you have store that you wanna run it for bruh?

[limber breach]

i can help you to run flows for it

[onyx pier]

Hi all! Does anyone know if veo (flow or other) lets you upload a video to do ai edits on it? video-to-video? or only images allowed

[elfin fulcrum]

Nice shot , bad dialogues delivery

[crisp basalt]

Flow doesnt have that functionality yet. But the api (veo 2 inpaint/outpaint) is available to use i think

[supple pendant]

this is salvageable - generate new audio and replace the audio on this with audio you prefer. they are far away enough that you can get away with a lot if the dialogue doesn't line up

[polar shore]

where can I find the scene builder?

[hot shale]

With ultra, fast generation is free.
Does this is same with video through API??/

[crisp basalt]

Nope

[chilly kite]

How do you make a scene extend beyond 8 seconds? I know you can save the last frame and tell the next video to generate from frame, but it doesn't retain the context from the previous video. For example, a character might have a completely different voice.

[tepid tundra]

The base of character consistency is generating an specific prompt for your character, how do you want him to be, what's his personality? how his voice should sound like?

all that can be included in a prompt, you can even give Veo3 a prompt such as Characters: 1 - Name 2 - Name
Scene:
Style
Composition
Etc

Name: Elias
Appearance: An African-American man in his mid-30s. He has a lean, wiry build, shaped by constant movement and scarce resources. His face is defined by kind but deeply weary eyes that have seen too much. He maintains a short, neatly-trimmed beard—a small, disciplined act of clinging to his former self. A faded, thin scar cuts diagonally through his left eyebrow. He wears functional, dark-colored cargo pants, a worn thermal shirt under a tactical vest, and sturdy boots.
Demeanor & Internal Struggle: On the surface, Elias is calm, observant, and methodical, a remnant of his emergency response training. He moves with a quiet purpose and speaks only when necessary. Internally, he is tormented by a profound sense of failure and survivor's guilt and he is always under influence of psicodelic mushrooms.

example of one of my recurrent characters

[autumn vapor]

hello guys, when will we have extend and ingredients feature with veo3 ? anybody knows ?

<@&1173646484311384114>

[chilly kite]

Does the description of the scene, style, character, etc. need to be repeated in each subsequent prompt?

[tepid tundra]

In most cases I only mention the character description of the person who will do the talking in the scene

I try to mix it up with the first frame as well

[analog lava]

REMINDER!

We cannot generate images or videos in the server! please do not just paste your prompts into the server as it will not generate, unless you are asking for advice!

[last thicket]

I used to do : Subject: Main focus (person, animal, object, landscape), including detailed characteristics.
Scene: Location (indoor, city street, forest, beach).
Action: What the subject is doing (walking, jumping, talking on phone).
Style: Visual aesthetic (realistic cinematic, anime, stop-motion).
Camera Movement: How the camera moves (following, rotating, panning, dolly zoom).
Composition: Shot type and framing (close-up, wide shot).
Atmosphere: Mood and lighting (warm tones, cool lighting, night scene)

[azure torrent]

greeting

[buoyant token]

‘Don’t worry, my friend.

[unreal viper]

please post in #flow-videos・📹 ! thank youu

[cunning crater]

k

[iron matrix]

Has anyone noticed anything different about Flow recently? We may have just added some cool things quietly!

[winged plinth]

Expander👌

[worldly yacht]

what you added ?:D

[kind bison]

Please fix the Prominent People bug. It's killing my AI video production business.

[tepid tundra]

Yeah I mean, as long as it works one can use either option, the important thing it's to keep doing what is working haha

[fierce nimbus]

Anyone having fun with expanders yet?

Gemini directly in your prompt box. Any prompting format you love can just be used directly. Want it to always format your ideas in JSON. Go ahead

https://x.com/flowbygoogle/status/1970609918454776288?s=46

FlowbyGoogle (@FlowbyGoogle)

Hungry for more? Create your own custom Expander that you can reuse across a variety of clips, giving you even more control and consistency across your story. For example:

— Describe a specific style or genre, like a live-action musical, and Flow will help maintain it as you

X · Sep 23, 2025, 10:03 PM

[winged plinth]

System level instruction for every shot🔥

[crisp basalt]

You bet I do nick

[winged plinth]

Have you tried the mixboard?

[cobalt swallow]

Has anyone had lucky generating Spanish audio?

[stuck adder]

Anyone else getting alot of fail generations?

[cobalt swallow]

Mine are erroneous

[molten urchin]

First time messing with Flow. It's really neat what you can create.

[molten urchin]

Try Grok it's like that on every generation.

[analog lava]

REMINDER!

Hello, a friendly reminder as stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion/comtent in this server will largely remain in English for now.” Also, we cannot generate images or videos in the server! please do not just paste your prompts into the server as it will not generate, unless you are asking for advice. Thank You!

[karmic kindle]

Any idea when or if veo 4 is on the way?

[stuck adder]

We need native 1080p first lol

[pseudo owl]

When will VEO3 support keyframe and endframe functionality?

[wispy widget]

Hello there, don't know if this is the right place. Just update my account to pro to have access to FLOW (france member here) but it seems the tool don't want me to allow me to use it anyway!

[crisp basalt]

Soon tm

[bronze whale]

I have a real client need to be able to upload minors - we produce products for children, have photography but cannot use VEO/Flow as we cannot upload base images with minors - is there a way to override/request access?

[crisp basalt]

oh its gone thanks mod ❤️

[thick oriole]

Nothing right now, sorry. We are looking to support this in general, but it comes with a lot of safety concerns as you can imagine.

[covert valley]

https://youtube.com/shorts/Ui85OSeIYRA?si=h0NJ9-EoVNjox0d4
Any one piece fan here, emeth making his move guys

[hallow comet]

where can one find a guide to use flow for advanced users?

[bronze whale]

Of course, fully understand - I would have thought AI image detection and prompt analysis would be capable of dealing with bad actors though by now 🤔

[crisp basalt]

official guide?

[hallow comet]

not official necessarily, a community made would be better

[crisp basalt]

yeah theres like thousands of them at this point

[hallow comet]

which channel will I find them in? Can you point me in that direction?

[crisp basalt]

oh i thought you mean like outside of discord server but yeah the channel would be #flow-tips-n-tricks・✏️

[hallow comet]

im guessing its a similar story for the rest of the labs fx tools as well? I can't find a channel for whisk guides actually, cause I've been having a little problem trying to convert my whisk creations to flow seamlessly into flow

[ornate ether]

@fierce valve I need help restoring my old Google/YouTube account. I know the account email and password, but I cannot log in because it is asking for a recovery code sent to a recovery email that I do not recognize.

[crisp basalt]

oh i think you can.. actually do that within Flow, were you using whisk for normal image generation stuff , or use it for image editing?

click on generate image and the create image should load ,

from here you can add your own images too to generate an image based on your uploaded reference image

[hallow comet]

this is an image i generated using whisk, now I'm trying to convert it into a set of 4 scenes where this cute orangutan will be talking to the camera, vlog style, about a product

the issue is that the video its taking inspiration from has like multiple reference points that can be only explained in like 3-4 images. only veo 2 supports it

if you see in the 2nd image, its locking onto veo 2 cause apparently ingredients to video doesn't work with veo 3

[crisp basalt]

yeah you can just use Flow’s generate image feature to make a “start frame” of the scene you want, then run it through frame to video with veo 3 model.

[hallow comet]

actually I've been looking for an explanation on what are these "ingredients" mean exactly. there's only frames to video and text to video left after this

[crisp basalt]

its a "temporary" workaround for now if u want to use veo 3 model for ingredients to video

[crisp basalt]

those “ingredients” are just your own uploaded images. they just use that term as their unique naming style.

[hallow comet]

that's the kind of shrewd tricks I've been looking for! thanks dude. I'm guessing I can also do an end frame similar to this as well?

[crisp basalt]

for end frame, only veo 2 works rn but i’m sure they’re already cooking up a veo 3 version as we speak 👀

[charred light]

Hii

[crisp basalt]

they said months ago that veo 3 ingrd to video is in the works, so yeah just stay tuned for any future update 👀

[hallow comet]

im going to try this frames to video feature, I think that's what I've been looking for

also, the veo 2 was generating text captions even when i told it not to

bruh...

i did create like a whole video with this monkey character, it was just lacking audio, i was trying to solve that problem

yesterday

[crisp basalt]

yeah veo 2 doesnt do audio generation, your prompt shouldnt include those dialogue line as veo 2 model will interpret it as "subtitle/caption" in video

[crisp basalt]

you dont have to use that image of the character as your start frame if you want to generate a scene that uses that character

ill demonstrate something hold on

[charred light]

How can I have 100 points in an account that has waited for more that 1 month?

@crisp basalt

[fair bone]

flow is not working everythhing is black

anyone else facing this issue

[crisp basalt]

i generated this "start frame" image using the generate image feature in flow btw

and then use it as start frame in frame to video, and youre good to go

[hallow comet]

@crisp basalt let's say I decide to create the same video of the orangutan walking around a magical mushroom forest, how do I make sure the character is consistent in each scene? and like i feed the model same kind of context and reference as I do with ingredients to video and frames to video?

[hallow comet]

can you run it through veo 3 like this?

[crisp basalt]

yeah im generating the video rn

[hallow comet]

i have 2 images here, was that the problem?

[crisp basalt]

you can use create image feature to generate an entirely different scene/location but still uses your character

[hallow comet]

i use gpt 5 to generate all prompt AI. whether it be image prompts, content prompts or whatever of any kind, I just mind dump a voice input of like 3-5 mins in it and it gives me an output prompt that gets what i want just right 80% of the time. what do you use? what would you suggest I do to make the prompts better

yesterday i fed a blogpost to gpt which was a guide to veo 3 prompting and after feeding it the guide, i got prompts from it

[hallow comet]

ayooo that's littt

[crisp basalt]

hmmm actually you dont really have to.. think much when it comes to prompting, as for me i refer to any existing veo 3 prompting guide from google themselves.

this should be a good beginner guide for anyone who want to learn how to prompt for veo 2/3 https://deepmind.google/models/veo/prompt-guide/

[hallow comet]

i haven't read any of those, do you have links to them? i'd like to read them and burn thro all the credits learning how to prompt accurate stuff

[hallow comet]

thanks!

[crisp basalt]

if you want to read more advanced guide, You might wanna check this out
https://medium.com/google-cloud/veo-3-a-detailed-prompting-guide-867985b46018

[hallow comet]

yep, i fed gpt this exact guide lol

[crisp basalt]

in conclusion,,,

[hallow comet]

im crying lolll

[crisp basalt]

Oh shoot i forgot

@hallow comet you can check out @FlowByGoogle on X. they always drop tips and guides there, worth a look too 😄

https://x.com/flowbygoogle

FlowbyGoogle (@FlowbyGoogle) on X

Hi👋 We're Flow, an AI filmmaking tool from Google, powered by Veo. Follow along for product news and creative tips to bring your cinematic stories to life.

X

[hallow comet]

I'll definitely check 'em out, thanks @crisp basalt!

[verbal reef]

Hi

[crisp basalt]

uhh 100 credit is for new registered users only

[thick oriole]

Child safety is particularly difficult. Any 'miss' is a serious incident, and collecting training data, doing evaluations etc in the usual way is much more difficult in this domain. But, we're working on it!

[crisp basalt]

whaaaat 🥺

at the same time im hyped for other stuff (extend, start→end frame, etc) to be compatible with veo 3 soon 🥹

but im gonna miss using veo 2 so muchhh 😭

[tepid tundra]

No need to add the last frame tho

[crisp basalt]

stop

[misty wraith]

@crisp basalt Veo 3

[main saddle]

hey @misty wraith - videos will not generate anywhere within this server, neither can Wrigley do it for you! thanks!

[crisp basalt]

https://tenor.com/view/trying-to-calm-down-faye-valentine-daniella-pineda-cowboy-bebop-darkside-tango-gif-24124730

me the moment i got the forth ping

[tidal condor]

Is there any plan to allow for upscaled vertical vids?

[forest light]

When is the plan to use vertical videos in a scene / sequence?

[bronze whale]

@thick oriole With Veo3 becoming the default, can we get an audio on/off toggle to keep the credit usage then same?

[silk hill]

bummer, veo 2 is still better than veo 3 at a lot of things

[gritty ruin]

Guys help me for prompt .. can use for map can have character and have soldiers can move it like 3d

[sonic cedar]

im not sure about that. I find the details of hair better in veo 3 than 2.

[sonic cedar]

Talking abuot credits. Not long ago i received an email that the api use has become cheaper. Rethorical question: why are they not cheaper in flow too?

[silk hill]

veo 3 is not as good at doing specific types of documentary style footage

[sonic cedar]

I create cinematic docu style education videos. Can i see your prompt?

[silk hill]

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

[sonic cedar]

Stunning video m8. Very creative mind.

[silk hill]

ty

i had to revert to veo 2 to get some shots recently because veo 3 just wasnt getting them at all, and veo 2 did exactly what i asked, anyway its all loosely in that style

i'm not saying veo 3 isnt much better for many things but its not always better at everything

[sonic cedar]

To be honest, I was a gpt aka sora / kling user. Just 30 min ago i ended all my subscriptions and go full Google. But I noticed waiving hair with the wind looks awefull in veo 2. Maybe you know how to download generated images from flow? and how to make flow videos a loop video?

[silk hill]

i actually had to just train a "hair being blow by wind" lora for wan 2.2 ahaha

because the speedup lightX lora killed hair motion on kinetic shots

[mental robin]

hey guys I need help pls, on flow this video looks really cool and realisitc but when i download it (upscales) quiality looks very bad. any tips pls?

it does not look like that on flow

im tryng to do a comercial for an app but when i try to make flow show a notification pop up or a phone screen it looks really bad, what can i do?

[stuck adder]

Atleast we willl have start to end frame yay

[crisp basalt]

Oh oops I meant to forward this to my own DM

[kind bison]

Looks like normal VEO 3 quality to me - don't see any issues with it.

[mental robin]

on veo 3 it looks much better than when i download it

[kind bison]

Looks fine to me.

[crisp basalt]

ah did u delete the image ? i hope u dont mind i used it 😅

[full flicker]

Hello, a friendly reminder. As stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Thank you.

[wind saffron]

I have the same problem with the 1080 upscaling, always comes out too sharp, I just download the original at 720 and enhance it using other apps

[wind saffron]

Flow is one of the cheapest apps for vid gen (Veo 2) 10 credits per gen works out (for me) at approx $1 per min, which is by far cheaper than most if not all the apps out there, I hope this isnt going to change when Veo 3 takes the realm.

[loud raptor]

hi

[solid thorn]

The upscaler on veo is auful. I've found the process of keeping it at 720p and using something like topaz astra to upscale it. Does a really nice job, especially with skin tones

[jolly sparrow]

.

[lilac ermine]

Hey all, Ive just joined and noew to all this. Im learning FLOW. I cant figure out how to make the video length longer. I have prompted "Make this a 30 second video" etc and nothing changes vidoes comes out to like 5 to 8 seconds. Can anyone point me in tne right direction please. Thanks in advance

[crisp basalt]

Veo 3 can only generate videos up to 8 seconds long, which is its maximum duration.

You can “extend” an 8-second video in Scenebuilder, but keep in mind it’s also capped at 8 seconds. The extend feature still relies on the Veo 2 model, so it won’t generate audio. If you’re planning to use the extend feature on a Veo 3 video with audio, you may want to reconsider

Check the Flow support page for a quick-start guide on generating videos with Flow here : https://support.google.com/labs/answer/16353334?hl=en&dark=1&sjid=4433278100678971876-NC

…reading back what I wrote above, it really sounds like ChatGPT ahh type respond now 💀

[river bone]

Hello everyone, and thanks in advance for your help.
I’m a new Flow user and I’d love some guidance on filmmaking with Flow. My goal is to create a short film entirely with AI. To avoid inconsistencies in the character between scenes, I’d like to build a training model for Flow, so that whenever I need to use the character, I can just mention this model and keep it consistent.

Has anyone here tried something like this before? If so, how did you do it? Also, I’d really appreciate any extra tips on how to make a complete, error-free film.

[patent hedge]

Hi guys and Girls. In First frame mode, veo3 is changing the color of my firsth frame. Even if i add " muted colors". Is somebody find the some prompt to fixe that? Here is my firsth frame made in midjourney and the résult.

[proven island]

when will they put continues scenebuilder on flow on 9:16 aspect ratio like the landscape one?

[crisp basalt]

Oops wrong link

https://x.com/FlowbyGoogle/status/1967675420905705814 Flow Team by Google posted this maybe this is relevant to your question about keeping character consistency. Open the link to read the whole thread

FlowbyGoogle (@FlowbyGoogle)

(1/7) As an AI filmmaking tool built with and for creatives, we wanted to put our features to the test. So, we created our own short film!

Follow this thread for a behind-the-scenes look at the creative process, and see how Flow can help you build stories with consistent

X · Sep 15, 2025, 7:42 PM

[crisp basalt]

Fast mode?

[patent hedge]

In scenebuilder, With prompt write to have directly the new caméra angle directly. without continuing the framing and angle of the previous shot

[crisp basalt]

I don’t think prompt will fix it cuz I got that kind of result too when I used veo 3 image to video model

The oversaturated color

[crisp basalt]

Using Veo 3 fast mode or quality?

[patent hedge]

Yes I am in Fast mode

[patent vigil]

/vídeo

[crisp basalt]

Ah yeah I’m not sure if this oversaturated generation is a bug or part of the veo 3 fast model thingy cuz almost everyone got it too

But

[crisp basalt]

Let me try something

[patent hedge]

Ok, You have the First frame

[crisp basalt]

From what I see , it misinterpreted the fish as goldfish lmao that’s why it suddenly changes appearance

[crisp basalt]

Do you just want it to be like just normal swimming video?

Or do you have specific prompt

[patent vigil]

/vídeo

[patent hedge]

As usual, the fish swims and dansing elegantly and its tail unfurls gracefully and fluidly.

[patent hedge]

nice .... : ) Please how did you fixe the colors?

[crisp basalt]

Prompt was The camera slowly pans right, following the graceful movement of three luminous, glass-like fish swimming through a swirling, sun-dappled underwater environment. The largest fish, positioned closest to the choppy water surface, is a large, translucent pink and white goldfish with shimmering, detailed scales and fins, casting soft internal light. Beneath it, a medium-sized goldfish with mottled orange and white patterns is carried along by a current of crystal-clear water, trailing effervescent bubbles. A tiny, iridescent fish with delicate, fan-like fins, displaying vibrant hues of blue and pink, swims in the foreground, catching the refracted sunlight. The water surface above is agitated, with bright, star-like sun flares visible, creating a high-contrast, mystical lighting effect and a strong bokeh in the surrounding water. The overall aesthetic is hyper-realistic and ethereal, capturing the delicate flow of water and the glow of the luminous fish.

I used Gemini to craft this prompt btw lol . I have a very bad English skills

[patent hedge]

and what was your request on Gemini please?

[crisp basalt]

Let me grab the Gems

[patent hedge]

It was a Gems that you Configure it yourself?

[crisp basalt]

kinda! well gemini did most of the config , i just lead the "instruction"

[crisp basalt]

ill dmyou the gem

[patent hedge]

Ok thanks

[uncut geyser]

flow currently has a fixed maximum video duration(around5-8 sec), so prompting for longer videos won't extend the length

you'd need to stitch clips externally if you want a full 30 second output

[severe sedge]

I tried to make 3d animation cartoon video for education and fun but audio generation faild again and again showing that audio generation not supported for minor. Even I can generate Educational and funny cartoon videos for children ?

[sonic cedar]

i will show my gf what you did to her 🤣

[hallow comet]

Can you also share it with me? I want to learn how to use gemini for veo 3 prompts like you do

[silk pelican]

how do i get so the bigfoot looks realistic not some goofy cartoony

[noble carbon]

anyone try the new expander feature yet

what do you think of it

[tepid tundra]

rather wait for someone to do a full lenght video tutorial on how the feature actually works

[noble carbon]

it's pretty self explanatory tbh

[tepid tundra]

Go ahead and achieve character consistency then

[noble carbon]

idk if it helps with character consistency

i think the style is what it's trying to help with

like if u want a certain design throughout all of ur scenes it'll do that. cus sometimes i run into problems and everything randomly turns into an animated video from realistic

but for specific character consistency what i do is just write my prompts to be the same for a character and pray

[tepid tundra]

yep, that's what we all do whenever we can't start with a first frame

https://x.com/flowbygoogle/status/1970609918454776288?s=46

FlowbyGoogle (@FlowbyGoogle)

Hungry for more? Create your own custom Expander that you can reuse across a variety of clips, giving you even more control and consistency across your story. For example:

— Describe a specific style or genre, like a live-action musical, and Flow will help maintain it as you

X · Sep 23, 2025, 10:03 PM

I'm still waiting to see if someone does create a youtube tutorial about it tho haha

[crisp basalt]

Exactly lol

[crisp basalt]

Ok I’ll dm

[turbid lintel]

This expand function to advance for me , need simple version lol

[remote jetty]

Hi, unused credits don't roll over? I lost almost 10000 credits

[thick oriole]

https://support.google.com/labs/answer/16526234?hl=en has full info on what should roll over and what should not.

[dim oyster]

On ultra plan, I can generate unlimited videos using veo 3 fast in India?

[cosmic cave]

yo what is going on this: You're requesting generations too quickly. Please wait a moment and try again. / this is driving me up the wall n back

max i can run at a time is 1 now?

Who broke the limiter >>; Who was it?

[crisp basalt]

https://tenor.com/view/you-youre-your-yourself-straight-face-gif-26118421

[tepid lotus]

helli

hello *

how do i make videos using this flow

[mental mulch]

I cant wait for Veo 3 to get the Veo 2 formula

So excited

[analog lava]

Hello @long lark just a friendly reminder we cannot generate images or videos in the server! Please do not post your prompts in the channels as it will not generate, unless you are asking for advice or tips. Thank you!

[mental mulch]

I hope they put Sound Effects as optional

like stock cartoon sounds

[sour silo]

Just downgraded from "Google AI Ultra" to the new "Google AI Pro" plan, and unfortunately the Pro plan does NOT include unlimited Veo3 Fast generations like the Ultra plan does. Wish this was specified in the features list when subscribing. The only difference specified is "Flow with limited access to Veo3" vs "Flow with highest access to Veo3" and doesn't specify exactly what "limited" or "highest" actually gives you. Upon logging in, all functions are the same (Text to Video, Frames to Video, Ingredients to Video) - the only difference is unlimited fast generations isn't available. The 1000 credits included literally only gives you 10x Quality generations and 50x Fast generations. It can sometimes take dozens of attempts just to get one scene right and the credits are burned extremely quickly, which doesn't really feel like a "Pro" plan to me.

[hazy needle]

Hi everyone! I’m new to Flow. Could someone guide me (or point me to a resource) on how to use Flow to stitch multiple 8-second clips together? Thanks in advance! 🙌

[sour silo]

After you generate a video, top left click "Add to Scene"

[cobalt valve]

Hi, i tried generating a video today. The genration consumed 60 credits instead of 20, was stuck ad 99% and in the end didn't generate the video. When I refreshed the page it's not there, like the generation never happened. Veo 3 fast. Is there an email or a support to contact to refund the credits?

[tepid tundra]

⁠flow-bugs-feedback・📥 that channel there and the "Send feedback to Google" in Flow itself I think are the only channels known for support

[tepid lotus]

@long lark
you

i want to make these videos

[golden nymph]

I really wanted to learn what you've been guys. making videos, especially in FLOW, how will I start?

[livid creek]

Hi everyone

[tacit shadow]

anyone knows why veo is 3 not generating videos ?

[opal sierra]

Having the same issue. Any video using a newly uploaded image will fail to generate, not even getting an error message. Hangs at 99% complete and disappears. When retrying existing prompts/images they will go through. Seems to be only effecting any newly uploaded images.

[dense flume]

** any solution to change this interface to normal. it looks like very congested **

[dim oyster]

@sour silo hey, I am planning to upgrade on Ultra plan, can you tell me, veo 3 fast is truly unlimited or there is any catch like cooldown or slow speed generation?

[copper pelican]

I am trying to create a video that I can use for a conference and it starts wide with a number of people and zooms in and I am getting pretty good results and happy with the content, but when I upscale to 1080 and export, the MP4 I download is pretty blurry and I don't think the quality will work. any ideas on how to get a higher quality output?

[cobalt valve]

Check your credits. The same happens for me and the credits are consumed anyway.

[slate olive]

When is Veo 3 going to be 1080p standard output because 720p is just to low quality having 25000 credits is so amazing its just 720p isnt great

[slow iron]

How I start vedeo guys ?

[granite sage]

Hello, how can I make a video using the same character all along the scenes? I try the resources "expand" and "go to", in the scene edition tool, but it don't work and have a problem, because don't generate in Veo 3 Fast (what is a complete problem in the generation process).

[sand sparrow]

do you guys have problem generate vids with veo 3? my problem is it stuck at 99%

[opaque sand]

There may be an image or promt policy issue.

[tepid tundra]

Happens sometimes to me, I just refresh the browser and it gets generated... or not haha

[tepid tundra]

Frames to video, or including a descriptive section of your prompt with your character description

[opaque sand]

metadata:
base_style: "cinematic, photorealistic, vibrant, bright sunlight, slow-motion (120fps), macro details, 4K"
directorial_intent: "To capture a vibrant and serene moment of life at the water's surface. The shot is not a complex narrative, but a beautiful, looped observation of three fish playfully interacting with the light and water, bringing the provided reference image to life with gentle, fluid motion."
camera_setup: "A single, continuous shot from a split-level (over-under) perspective, with the camera lens positioned exactly at the waterline. The camera drifts and sways very gently with the water's natural movement, creating an immersive, organic feel. The focus remains on the group of three fish."
lighting_mood: "Brilliant and joyful. The scene is dominated by bright, warm, overhead sunlight that creates sparkling caustics on the scales of the fish underwater and glistens intensely on their wet bodies as they break the surface. The mood is peaceful, vibrant, and full of life."
key_dynamics:

• "The gentle, playful, and continuous swimming motion of the fish weaving around each other at the surface."
• "The water surface dynamically splashing and rippling, creating a cascade of bubbles."
• "Sunlight beautifully refracting through the water and sparkling on every droplet and scale."
  timeline:
  ​

sequence: 1
timestamp: "00:00-00:08"
action: "The three pearlescent fish swim playfully and gracefully in a tight group, their bodies constantly interacting with the water's surface. The largest fish's back and dorsal fin are almost always above the water, creating gentle splashes and clear ripples. Another fish weaves just below, its form beautifully distorted by the moving 'ceiling' of the surface. Small, shimmering bubbles, created by their movement, drift around them, catching the sun like tiny prisms. Their delicate, translucent fins sway and catch the light with every turn. The entire motion is fluid, serene, and feels continuous, as if it could loop forever."
audio: "A mix of clear and muffled sounds due to the split-level perspective. Above the water, we hear soft, clear, sparkling splashes and trickles. Below the water, these same sounds are heard as a muffled, bass-heavy 'thump'. A gentle, optimistic, and minimalist musical piece (perhaps a simple, looping piano or acoustic guitar melody) plays throughout."

[hollow warren]

There are a couple of different ways.

I'm not sure how complex your character is but here are some basic workflows.

Create a hero image, I use sora but whisk works.

A. Use that hero image in ingredients to video.

B. Use the hero image to generate an image that will be your first frame for image2video

C. If your image2video ends with your character on the last frame, screenshot that frame and use it as the starter frame of your next video.

[gusty kindle]

Hi everyone, few questions i was looking for help on..

1. Does anyone know if Flow has API access?
2. Does it come with Google Ai Ultra or is it through gemini API?
3. If so does anyone know costs?

[slow iron]

Hi guys can I make vedeo from here ?

[analog lava]

Hello, friendly reminder we cannot generate images or videos in the server! Please do not post your prompts in the channels as it will not generate, unless you are asking for advice or tips. Thank you!

[crisp basalt]

Flow isn’t a video model, it’s just a tool that runs on Veo to generate videos.
If you’re looking for Veo API stuff : https://cloud.google.com/vertex-ai/generative-ai/docs/video/overview

Vertex AI ≠ Google One AI Pro/Ultra (your AI credits don’t count)
Google Cloud/Vertex AI bills straight to the credit card you add when setting up Google Cloud.

[trail flame]

I'm facing this issue from last day vide generation stuck on 99%

And Also the Quality of the potrait video is trash 720p isn't enought

[peak wind]

https://youtube.com/shorts/NC_MroHNVYk?feature=share

[near gate]

I created these with Flow.

[peak wind]

https://youtube.com/shorts/TLvnE5Bn2vE?feature=share

[last thicket]

Why some people can't access flow even tho they are from an available country and over 21?

[vale arch]

I tried to generate videos using VEO2/VEO3 in Google Flow, but I only encountered errors, and credits were deducted. Even in the library I can’t find any generated videos. Can anyone help?
@crisp basalt

[crisp basalt]

did u make sure you check in Flow Library section?

[vale arch]

Yes. I found nothing.

[crisp basalt]

have u tried refresh your project page? the page where u were using to generate video

And do you remember what error message it gave you?

if the generation encountered error and it didnt refund you credit at all, try refresh the project page, does it shows this kind of message

[vale arch]

This screen is my library. but nothing @crisp basalt

[crisp basalt]

huh its empty

have u used flow before?

[dry frost]

Url is to video fx

[crisp basalt]

actually thats the default one

idk why it uses videofx still ,

[dry frost]

Oh!

[crisp basalt]

do u have this enabled?

[vale arch]

No button for enabled.

[crisp basalt]

the button is in the library page btw,

[vale arch]

Enabled

[vale arch]

I had only logged in before and never used it. I refreshed the screen many times.

[crisp basalt]

huh so i guess thats why its empty but, your credit got deducted somehow

[vale arch]

Yeah. That's problem..

[crisp basalt]

Uhh, try logging out and back in, then generate a video using Veo 2 fast mode, it only costs 10 credits, so you wont waste much if something goes wrong.

and try catch the error message (preferrably screenshot it)

[vale arch]

I logged out and then logged back in, yet I still found nothing.

[crisp basalt]

have u tried generating one video ?

[vale arch]

wait a sec.

?????
This time there were no errors.
@crisp basalt

[crisp basalt]

ooh

i guess flow had hiccup i got error too just now

[vale arch]

just I logged out and back in.

@crisp basalt ok, Can I get a refund because of the errors?

[crisp basalt]

i think u might need to contact staff about this

cuz u have evidence sop

so*

[vale arch]

Ok. I got it.

[crisp basalt]

@lone island in the flow page. click new project

[lone island]

so i wanna get this image animated, the background has good waterfall with the surrounding water also having gentle waves and wind is making the plants/leaves shake

i dont want the angle to change tho or bottle to change, just remaining frontwards

[lone island]

alright

[crisp basalt]

then it should load this page

[lone island]

yep

[crisp basalt]

click on the "text to video" and click frames to video (this is image to video btw

then click on the first "+"

it should load up this for you

[lone island]

yeah i can see that now, but it seems like its cropping me out

[crisp basalt]

yeah it needs to be cropped to 16:9 for landscape no matter how

[lone island]

dang

[crisp basalt]

i think that framing is fine,

since the main focus is the product

do crop and save ,

[lone island]

alright

[lone island]

so what do i keep as the prompt now to achieve this

[crisp basalt]

then go to setting, choose veo 3 fast , and do 1 output

honestly

wait let me try rfirst then ill get back to you

[lone island]

alright, let me know if you get something interesting

[crisp basalt]

i literally copied what you wrote and paste it in prompt and now im generating lmao

cuz usually you dont have to., sweat much when it comes to type what the prompt on how u want the image to be animated,

[lone island]

bruh thats risky, i dont know how to feel knowing you wasting credits on my trash prompt...

[lone island]

whats output btw

[crisp basalt]

@lone island idk if this is what u were hoping for

[crisp basalt]

basically how many vids u want to generate, maximum is 4 video

[lone island]

oh, does the number to generate consume the credits?

[crisp basalt]

yeah

[lone island]

it is actually looking decent

[crisp basalt]

it should show this

[lone island]

btw how does one get rid of the VEO thing? should i get the plan or?

[crisp basalt]

ultra plan to .,. get rid of the watermark 😅

its the only way to do it

[lone island]

how much does that cost?

[crisp basalt]

Ultra plan perk = Veo 3 fast is 0 credits. Basically unlimited vids if you’re on Ultra and use Veo 3 fast

[crisp basalt]

😬 very expensive

[lone island]

do you think its possible to make it appear like the perfume bottle slowly levitates down smoothly by adding the empty image first?

should i prompt it normally?

[lone island]

where can i see the cost? if its couple of dollars i wouldnt mind really

[crisp basalt]

https://one.google.com/about/google-ai-plans/

[crisp basalt]

like slowly appearing..?

or just hovering

[lone island]

like from the top it appears outta nowhere and slowly comes down to the surface and stands there

[crisp basalt]

ah that, you'd have to use the ... start frame -> end frame which for now, only veo 2 model can be use for that , which means no audio

[lone island]

oh thats fine

[lone island]

btw thats highly costly..

[crisp basalt]

yeah 😅 tell me about it

[dusky summit]

can anyone help me to refine this prompt to get a more realistic output? POV shot inside a modern bathroom. A row of plastic shampoo bottles sits on a shower ledge. A faceless hand suddenly sweeps them off the ledge, bottles falling dramatically to the floor. Bright bathroom lighting, realistic textures, cinematic realism.”

Many thanks would be in order the output has been meh at best so far.

best I've got so far

[crisp basalt]

wait it looks realistic to me tho

[crisp basalt]

minus that two shamppoo bottle still standing

[crisp basalt]

btw

Veo 2 model

[lone island]

bruh mine is still loading

[dusky summit]

Any suggestions

[lone island]

also the liquid thing is weird lol

[crisp basalt]

idk only this out of 4 vid i generated looks fine . I used POV shot inside a modern bathroom. A row of plastic shampoo bottles sits on a shower ledge. A faceless hand suddenly sweeps them off the ledge,

the rest were having that two bottle still standingsyndrome

[crisp basalt]

yeah i didnt prompt that actually,

[dusky summit]

ok thank you

[lone island]

hold on, do i have to prompt that to ensure it doesnt mess up?

why is my thing at 99% for so long

[crisp basalt]

its better to not include what you dont want in the video in the prompt

[crisp basalt]

refresh the page

[lone island]

bruh i did, im back to square one

where is my project 😭

[crisp basalt]

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

u should see ur created project

[lone island]

its not there

[crisp basalt]

rip yeah idk how to help u from there tbh, uhhh try relogin ?

[lone island]

nah, i can see my credits hasnt changed

seems like it didnt go through, ill re-do it

[crisp basalt]

oh so it was failed generation then

[last thicket]

Flow is weird

Since the update

[lone island]

yeah, it says i dont have access to doing it....

damn im out of credits, i didnt even do a single video man

thats crazy bruh

[last thicket]

Me too, i did some, got rejected havent gotten back those

Lost in the limbo

Before the update when you used image 2 video, the ones that got rejected were inmediatetly refunded and canceled now it goes all the way up to 99% and wait there a few seconds before cancelling.

[winter trail]

How do you add a scene from another project to a different project? For example, if I have an intro that I would like to reuse without creating a new prompt, how do I do this?

[muted kestrel]

hii

[lone island]

actually i had the same issue, mine also was just uptil 99% and stopped

its even more stupid to see my credits have completely run out even though i have completely 0 projects

nice scam lol, sorry for the ping <@&1173646484311384114> but clearly there is an issue going on

[vagrant laurel]

All my generations are stuck in 99%

[calm temple]

anyone else getting "you're requesting generations too quickly" when literally just running one generation at a time. Waited like 20 mintes, generated, went through, generated again after that one finished and got the "too fast" error again.

[sand sparrow]

i have same problems

[last thicket]

This usually happens when you use AI to make the code lol

[quiet dew]

When did the problems with Flow start? Does they know if they have said anything about when the issue will be resolved?

[lyric cosmos]

paid basically 50 dollars for nothing to work this is great

[analog lava]

Hello if you're encountering errors and /or bugs or would like to leave a general feedback about Flow please don't hesitate in utilising the #1319424187730755636 channel with as much context as possible. It helps to keep the team organised and its also where all the technical issues are logged. Your post may get lost in this general channel so to ensure visibility please use that channel for all reports and feedbacks.

Thank you !!

[crisp basalt]

sorry to hear that

[wind saffron]

on the pro plan, anybody know if the credit price will stay the same from veo 2 fast (10 cred) to veo 3 fast (currently 20 cred), when they switch over.

[crisp basalt]

no

[lone island]

any idea on how to remove the VEO watermark lol

[copper mica]

Hello, a friendly reminder. As stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Thank you.

[gusty kindle]

Thanks for this!

[night kayak]

does anyone know this??
I made something with firebase and I wanna send that to my github to make it work because if I click publish so they asking payment approval which I don't wanna do

[autumn vapor]

crop

[tacit portal]

Can someone tell me why there seems to have been significant 'rate limits' applied to Ultra recently? Last night I was makingsome Fast Veo3 generations on my Ultra account and I waited well over two hours and still got the 'too fast' warning...

[autumn vapor]

when will we have ingredients to video on veo3 ?

[crisp basalt]

was watching a kpop m/v earlier, and from the first part i already had a feeling it was ai-generated. then at the end it straight up showed this lmao

erghh that oversaturated look… i’m guessing they used frame to video veo 3 fast model for most of those scenes

[peak wind]

https://youtube.com/shorts/0vyCRIhkYyU?feature=share

[flint summit]

Played around with my friend's ULTRA account and realized that even with unlimited Veo3 fast, its impossible to get it to render what you want. There's ALWAYS something wrong, and that's using my best know-how and experience. Garbled dialogue, characters speaking each other's lines, random crap introduced into a scene, there's ALWAYS something. Even repeatedly rendering 4 scenes at a time you'll never get exactly what you need. Also, Veo3 quality is no better - and often worse than Veo3 fast.

[sour silo]

Hey Techno, it's truly unlimited but you can only generate 5 at a time, which is usually sufficient!

[dim oyster]

Thanks for your response.

[dim oyster]

You mean, we can queue generations upto 5, right?

[sour silo]

Yes and they're less than 2 mins on Fast mode

[dim oyster]

ok

[nocturne echo]

hello

that's a common point of confusion! Pushing your code to GitHub is completely free, you're just backing up your source files. payment prompt from Firebase is for 'Hosting,' which is the service that takes your code and puts it on the internet for everyone to visit. They have a free tier, but it might require adding billing info. So, you can safely send your code to GitHub.

[visual sun]

Did something happen this weekend??

Flow is broken for me now

ok thx

that goes to another server?

Can I trust that link, everyone? Arame looks like a new person like me
(no offense, Arame... thank you for the help, just gotta make sure!)

yes, do you work for these guys?

Flow?

I see, but they should give you some badge or something on your profile

my bad, I see the roles now

Arame, you deleted your chat?

careful guys, user @arame is a scammer

[visual sun]

"This image might violate our policies." 👇

even tried with an empty background because some people mentioned the photos on the back wall being mistaken for something that would violate policy

Flow is cracking....

[supple pendant]

This might sound like a silly reason, and i might be totally off - but it might be reading parts of the drawings as "sexual content" due to patterns in the shading and rendering of the pics resembling something inappropriate. Like the fingers in this instance

[visual sun]

hmm, maybe... I'll test that

[lyric cosmos]

someone rly gotta sort the issue out of flow taking credits even though it isnt generating and failing most of the time

[visual sun]

it does look like the fingers triggered it, I cropped them out and it uploaded the image. But this is still a major lapse in operation
thanks for the suggestion!

[supple pendant]

I'm glad it worked out! Yeah, they really need to work on fine tuning the filters. It's frustrating to run into issues like this constantly

[visual sun]

I put in a suggestion earlier.... but it's like, "This image MIGHT violate our policies."
Okay, well let us challenge it.... because it also might not violate your policies..... and it probably doesn't because you made us agree to avoid all that stuff when we signed up

[supple pendant]

it really stifles creativity, and it definitely feels like a waste of time and money to sign up for the service only to find out you can't even make your project work. The guardrails are often WAY too sensitive and don't take context into account.

I had to iterate a children's birthday party scene a while ago - back then, they had a policy where you couldn't even use the word "child" or "children." They keep adjusting things... hopefully they'll find the right balance

[visual sun]

wow

[crisp basalt]

Even when I took first glance of the pic, part of the drawing does resemblance something so Flow probably tripping too lmao

[visual sun]

you think? yeah, maybe those little sausage fingers are getting mistaken for something, but this is a pretty far stretch

[crisp basalt]

I mean you cropped it out and it worked out for you after all so that’s alright

[visual sun]

as a test, it worked, but I kinda need that guys hands in the shot..... I'm gonna redraw it so he's like holding his fast, maybe something like that

but this is a nice new little issue now, I came on here earlier with the "prominent person" issue.... if you've encountered that, where it thinks all my characters are celebrities

[crisp basalt]

The error didn’t mention “sexual” message thingy right?

[visual sun]

not by words

[crisp basalt]

I don’t think it’s that, I think flow safety filter misinterprets the whole image

[visual sun]

but it seemed like that's the case, cuz using flesh's suggestion, I lowered the guy down the frame, hands out of frame, and it worked like that

[crisp basalt]

But if the finger was the cause of the error .. then I think they need to recalibrate their safety filter tbh

[visual sun]

for sure

so everything's sexual and everyone is a celebrity, according to Veo

[crisp basalt]

it wasn’t really this strict when it was in early-access stage ( VideoFx )

[visual sun]

I've had it for about a month or so and it has blown me away.... I was using Runway, and I pretty much started redoing my scenes one by one, cuz they were coming out so much better with Veo.... true magic

[crisp basalt]

But it gradually became more strict as they moving to public releases and I think at some point they had to reconfigure everything due to prominent people issues and it’s still an ongoing issues unfortunately

[visual sun]

but all of a sudden, of course, mid-way thru my project...

yeah, I can imagine corraling this new business venue is rough seas

[crisp basalt]

If anything , it’s not really Veo model the causes, it’s Flow website that’s doing the guardrails thingy

[visual sun]

right

I think there's a nice happy medium that they have already established...... we agree to avoid all the bad stuff..... so they have to honor that and let us work

[crisp basalt]

Have you tried using that same image that you had issues with on Gemini Veo 3?

[visual sun]

ok I'll try that

[crisp basalt]

Alright , just want to see if it’s actually the image itself is the problem or it’s just Flow being weird at interpreting image

[visual sun]

ok lets see

looks like it's the image

Gemini Veo 3 produces this error 👆

a larger overall issue is Flow/Veo's own inconsistency with the same content
For example, I already have 3-4 minutes worth of animation in this short film.... all made by Flow, using this same charcoal art style
Take this scene 👇 , with full audio..... here's the GIF but it already made me lots of content with this office manager guy and these three characters

it didn't have a problem with his sausage fingers for the last few weeks

and it didn't think they were celebrities, either

[lyric cosmos]

I think it truly is just Flow, the same thing happened with me and Gemini is letting me generate what I want whereas Flow isn't. I've even used the same picture, though I'm not sure why Gemini isn't generating yours

[visual sun]

whatever error code 500 is

[tardy ridge]

what hte heck I list 10k credits overnight

[kind bison]

Yeah, all of that's true. Unless I get extremely lucky early on, it typically takes me 10-20 tries to get a usable clip (and often I have to cobble together bits and pieces from multiple clips to get a decent version). Even then, it's rarely a complete match for what I prompted.

[torpid kraken]

hey guys is there any way to test out flow without a subscription?

[crisp basalt]

You should get free 100 ai credit if you’re a first time user

but yeah just don’t use Veo 2/3 Quality mode to generate a video, that one eats 100 credits per run lol

[crisp basalt]

I think you got the wrong Gemini Veo app 😓

Gemini Veo is in the Gemini app/gemini.google.com

[raw oak]

Hi. I am getting many many videos generated in FLOW (VEO3) Fast that are coming out without audio. Even the higher quality renders that cost 100 credits each!!!

[visual sun]

Thank you for the clarification. Seemed to get stonewalled with Gemini, too

[vernal cedar]

Anyone having this problem? I tried different kinds of browser and it's stil like this.

[crisp basalt]

Refresh the page

[crisp basalt]

Can you send me the picture ? I’m curious why it get blocked

[vernal cedar]

Whenever I try to do that, all my prompts vanished.

[crisp basalt]

🤔 usually when nothing pops up, it means the gen didn’t even start. Can you check your AI credit balance? See if those 4 vids you tried to make got deducted or not.

[crisp basalt]

To clarify , the gen didn’t even start part, it’s actually kinda an ongoing issues for many people right now

[vernal cedar]

I'm using the veo3 fast, I still have 21k credits though.

[crisp basalt]

I mean check the ai credits balances details

[tawdry edge]

people are getting 45k credits in gemini ai ultra and im still having 25k, is there anything i need to do?

[crisp basalt]

Cuz it’s important to see if you got deducted or not for the non existent generation

[crisp basalt]

Flow or Gemini? These two are different thing

[tawdry edge]

flow*

[vernal cedar]

There will be no deduction as I'm using the veo3 fast.

[crisp basalt]

Oh you’re ultra user?

[vernal cedar]

Yes

[crisp basalt]

How much did you generate using Veo 3 fast for today , like estimate

[vernal cedar]

Almost 100 scenes

[crisp basalt]

And did you see any error message pop up That says” you’re requesting gen too quickly , please try again”

[vernal cedar]

Right now, I'm generating 5 scenes with 4 results per quest. I'm at the maximum.

[crisp basalt]

Cuz if you do then , I think you got hit with that on going issues other ultra users are having right now

[tawdry edge]

can i please know :) not in gemini, in FLOW

[crisp basalt]

for ultra users , starter would be 25k tho, where did u see the 40k?

[vernal cedar]

Yes, I've seen this earlier.

[crisp basalt]

I think you should also bump that thread too, so the staff/flow team can see em and will get it resolve

[tawdry edge]

alot of my friends shared it

also, i have a google workspace, and they also

[crisp basalt]

🤔

Weird cuz no announcement regarding about ultra users will get 45k

[tawdry edge]

yeps im also thinking the same

i thought maybe i didn't knew about it

[crisp basalt]

Is it Google workspace ultra

[tawdry edge]

yes

[crisp basalt]

Cuz I’m interested on getting it too lmao

[tawdry edge]

xD im on trial

[crisp basalt]

Wait shut up don’t tell me there’s ultra trial workspace

They actually have it?

[tawdry edge]

not ultra trial but google workspace trial

noo 😭

[crisp basalt]

Well still , workspace but ultra benefit?

[tawdry edge]

its paid but the google workspace have a 14 day trial yk

[vernal cedar]

What should I do now? Do they have a way to resolve the bug?

[crisp basalt]

No choice but to wait

[vernal cedar]

thank you so much bro

[crisp basalt]

Until further notice by flow team , it’s an ongoing issues since ..yesterday

And flow team seems not available on weekend

[crisp basalt]

Ah

[tropic kite]

Are you guys able to use this “Jump To” feature with both the first frame and last frame? Here it never shows the option for the last frame. How can I make a Veo 3 video with audio, using both a first and last frame? It says the only solution is this “Jump To”, but the first and last frame options don’t appear. Is anyone managing to get this working?

[abstract beacon]

Is there any veo 3 Free usage method for students in Pakistan?

[crisp basalt]

veo 3 start frame and last frame isnt available yet

[granite sage]

Yeah, thanks. Make a description of the character and use in every prompt worked to create the more possible equal person in the many scenes. The frame tool don't work here.

Some feedback:
But I noticed a problem that can't solve yet: the transitions betweeen scenes. If I use the tool "go to" in the scene construction, only use the "Veo 3 Quality" and makes a video without sound! If I don't use, impossible to make the exact transition... always was imprecise. The "Fast" is far better of "Quality", why not permisse use this in the continuous mode of creation?

And a second feedback:

• Sugestion: the IA should memorize the "context", as like when we talk with GPT and it remember a thing that we said... if we can select the mode "remember" or "not remember", maybe this is better to the IA create the videos from the memory.
• Sugestion: should have a "talk" tool, like GPT or Runway, to improve the generations. This way, that videos are "isolated" from another one, without context, and permissing only the prompt, not talk, make things more dificult to build.

[crisp basalt]

veo 3 quality can produce sound

Only Veo 2 cannot produce audio at all

If u used Veo 3 and no audio came out, it means the audio generation failed usually because of safety filters or similar factors

also as for the "context" feedback, im not sure what you mean by that? like a preset prompt..? i think thats what the prompt expander feature is for

well its not really entirely the same like what you described , but it can be doable

[granite sage]

But when I use the scene constructor and the tool "go to", it makes a video without sound...

[crisp basalt]

DId you make sure you choose veo 3 model

and not veo 2, do remember that only VEO 3 MODEL can produce sound

[tepid tundra]

Hey Felipe, totally agree with the first suggestion, I'm starting to think tho that soon enough the AI will be able to memorize the "context" of the project as long as it's defined in one specific Project folder. I think the "extend" feature intent it's to start with that.
Morover reg the GPT, Gemini is available, you can upload the knowledge of Veo3 intelligence to make it better for the tool to help you with prompt generation for Veo in specific.

Jump to > Only works with veo3 quality
Extend > veo2

Best thing for transitions is mixing the tools that google offer, for example I have shots of my characters, then I'll mix them into Whisk generating the scene background or location to where I want to place them, then you'd play around generating the right posture you want them to have when starting the scene.

[crisp basalt]

can you show / screenshot your project page where your video was generated,

under your generated video, there should be model indicator

[crisp basalt]

Or you can just use the image generation tool in the Flow itself , it works the same like Whisk lol

[tepid tundra]

true, in Whisk tho you can keep on refining the same image in case you want anything changed, you can do the same thing with nano banana

[crisp basalt]

u can refine it too in Flow , it uses nano banana

and u can keep refining the same image

[tepid tundra]

I'll check on it later on today and see how it works, I'm happy with Whisk tho

going back to the point @granite sage do you have Ultra sub?

[tepid tundra]

cause then you'll have spare credits to use in Quality veo3, I usually do 2x prompt, just in case one does note generate audio

[crisp basalt]

i mean im just telling you the feature that they recently announced like weeks ago lol , no pressure so

[granite sage]

Sure.

[crisp basalt]

tbh you couldve just use chatgpt/gemini to translate what you wrote into english but its fine lol

[tepid tundra]

Yeah, I get it. Still is part of the process in learning this stuff, my advice it's to use the "First frame to video" Feature, then you will have more control on how the scene starts

[crisp basalt]

but yeah i think i get your point there

[unreal viper]

Hello, we wanted to remind you of our server guidelines, stated in #start-here・📍 : “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Please keep in mind that non-English discussions and content may be removed.

[granite sage]

Apreciate your answer. Can you help with this, please?

[sonic shale]

Hey everyone, hope you’re doing well. Quick question: I created an ad with Flow today, but every time I try to download it, I keep getting the message “Export failed.”

I already tried different browsers, turned things on and off, but it still shows the same error. I’m working in the Scene Builder and have several ads chained together.

Has anyone run into this issue before? Any idea how to fix it? Would really appreciate any help. I’d love to start working with the ad, but right now it’s just not working.

Thanks a lot, peace ✌️

[crisp basalt]

i also got this issue , like today and so far i have yet to find resolution for this 😬

wait let me try download from scenebuilder again..

[sonic shale]

tried it at least 20 times

[crisp basalt]

oh its working now

[sonic shale]

driving me crazy

[crisp basalt]

oh i think cuz im using different browser right now

[sonic shale]

glad for you! Lets try again.

[tepid tundra]

yeahh that's the way to go always haha

[crisp basalt]

chrome was not letting me do the scenebuilder export

[tepid tundra]

if not, you can always try cleaning cache, temp files and so on just in case

[granite sage]

I deleted all the generated scenes without sound. Only few have.

[crisp basalt]

ah thats alright then u dont have to show it for now,

[granite sage]

What Whisk do?

[granite sage]

YEAH

[crisp basalt]

to generate images with your own uploaded image

[tepid tundra]

Any way you can edit an image with your characters, you can create the "First frame" that's the first second of your new scene, so that gives you more control on how the scene location will be and the characters in it

[crisp basalt]

actually my brain farted , i think @tepid tundra can explain better

[granite sage]

Sorry. Don't understand.

[tepid tundra]

so basically you can get one picture of a scene where all your characters are present, then go to Gemini, Whisk or Flow itself and tell the AI basically: I want this 3 characters sitting on an empty cinema

[granite sage]

Sorry.

[tepid tundra]

not sure if the voicechat here can be used to explain things, but if not let me know if you want to have a call, it's easier talking than writing haha

[granite sage]

I'm sorry. Understand. Don't had read before. Agree and respect.

[pine yoke]

any idea why im getting export failed arrors?

reset cache and cookies, tried edge (ew) instead of chrome

[tepid tundra]

unsure, it's been working fine for me, there is an ongoing thread at flow-bugs-feedback if you want to chip in there with your issue

[sour kelp]

im definitely getting a bunch of "flow is unavailable" right now

[left hill]

Getting the same issue

[sour kelp]

wonder if an update is incoming then

[kind bison]

Today my videos are getting stuck at 99% - usually I can get them to finish up by going to the Flow homepage and refreshing but today they are just disappearing completely - no record of any of my generations - super annoying!!

[sour kelp]

i think its all the same problem, most likely.

[iron matrix]

Sorry were having some issues today, things will hopefully be fixed soon!

[stuck perch]

Any extending effort is stuck at 99%. I hope they'd return my wasted credits.

[main saddle]

+1, we're looking into this!

[modern thicket]

I'm getting an issue where I make a video, get charged for it, and then I don't get the actual video being generated. I don't even know if it was rejected or anything, it just takes my credits

[drowsy kraken]

i have the same probleme

[foggy glacier]

stuck 99%

[naive nebula]

Exactly, same here, no videos generated- credits taken. After refreshing whole projects dissappear.

[ripe hornet]

is flow down for anyone else rn? Image 2 video isn't working for me, its going to 99% and then gets stuck

[dapper tundra]

yup uploaded a handful of images (all innocuous well within google's strict content guidelines) and tried generating some prompts. Every single attempt has been 0-99% and then disappearing, not even an error message.

[tropic kite]

So, does his means that this table from googles help is wrong?

[lean narwhal]

One message removed from a suspended account.

[crisp basalt]

the table is right, you misinterpret it

[tropic kite]

thank you for the clarification! I was thinking that jump to should include first and last, but i was wrong!

[drowsy kraken]

ty georgia i lost many credits i hope they come back 🙂

[stuck perch]

I lost about 250 credits for all the failed attempts...

[crisp basalt]

@fickle thunder move here

[dapper tundra]

yup, just reset our monthly credit allowance please. i lost count how many videos i prompted that fell into the 99% void

[drowsy kraken]

i don't know how much... but just have to reset server just befor the crash

[crisp basalt]

https://one.google.com/ai/activity?utm_source=flow&utm_medium=web&utm_campaign=flow_ai_credits_page

you can check here to see how many credit that u spent

[stuck perch]

I see they refuned it at every failed attempt

[dapper tundra]

now im getting a "requesting too many videos in a short time error" my last generation attempt was 30 minutes ago. sigh

[crisp basalt]

😏 yep

let me just grab the faq page..

[drowsy kraken]

me to

[crisp basalt]

https://support.google.com/googleone/answer/16287445?hl=en&visit_id=638947747344464441-3734778763&p=g1_credit_management&rd=1

[crisp basalt]

https://one.google.com/ai/activity?utm_source=flow&utm_medium=web&utm_campaign=flow_ai_credits_page check here to see if u get the refund or not

failed generation instantly auto refund ur ai credits

so if anything, nobody here "wasted" their credit becuz of failed generation. its more of... wasted time 🙂 waiting for the generation to complete only to see em failed

[dapper tundra]

my question is not so much the wasted time but the reason why the 99% error is happening. Does it have to do with the image that was uploaded even though im not getting any sort of content violation warning (all images are extremely SFW)? It's frustrating uploading a dozen images only to have every single prompt attempt one as simple as "woman waves" failing to generate.

[crisp basalt]

the 99% error is not exclusive to you buddy

everyone including me , also facing it

so yeah , flow is having huge hiccup right now

no one can generate anything right now

[cloud burrow]

it would probably be helpful if the team added a little more info in that latest announcement so users know if the issue they are seeing is the one that is known..

[inland pier]

the latest issue was a serving issue, that affected Flow and some other Google products. If you were having trouble with generations a couple hours ago, you likely experienced that issue. I believe the issue is resolved now.

[naive nebula]

Unfortunately its not resolved yet. I still have same issues. Generations stuck on 99% and disappearing into the void.

[cloud burrow]

👆

[main saddle]

the announcement was addressing the mass uptick in reports that Flow isn't working - mainly the 99% failed gens and total blackout (as Ronny mentioned above) when we have updates, we will share them

[slow whale]

I'm stuck on 99% still also...

[cloud burrow]

thanks for the info 🙂 Adding more info like that can save you moderators some questions 🙂 well from the people who read announcements at least 😛

[main saddle]

appreciate the feedback!

[dapper tundra]

not sure why they arent making an official statement on this, but the 99% problem is most definitely connected to the type of image you are uploading regardless if it gets accepted by the system. For example, just now i uploaded an image of an old man waving along with video prompt , and it went through just fine. Meanwhile uploading a picture of a 30 something year old couple waving dissapears into the 99 void.

Try uploading a picture of a potted plant or a kitten and im positive the video will generate.

[quiet nexus]

Chiming in to say that I am still experiencing the 99% issue at 4:20pm PT

[quiet nexus]

Seems to be working again for me

[quiet vector]

hi

[stuck perch]

now audio is not generating.

[fading dirge]

how do you generate 1080p videos in flow/

[valid crescent]

I am not able to generate the audio in the flow video ,
how can i do it .

[blazing silo]

Guys i have a question. Are the 100 free credits you now get without a credit cars worse then if you had an account? I tried generating a video from an imgae and only wanted the camera to zoom out, i had the pormpt:"Make the camera zoom out" and it didnt work, as the camera stayed still and the person i had in the image started moving, does anyone also have this issue on paid accounts?

[karmic trench]

I’m encountering an issue where my Flow video gets flagged for “child safety concerns” — even though it contains no child-related content. My prompt is below:
“Colorful 3D animation of cartoon characters clapping hands on a grassy field. Rainbow sky. Cheerful tone. No kids or toddlers mentioned anywhere. Lyrics: Clap your hands.
This is being flagged for content, which seems incorrect. Please help review this — I believe the filter is being overly sensitive. Thanks!

[vernal cedar]

I’m one of the people who are experiencing this bug. The thing for me is, I use text to video, and get stuck at the 99% bug.

[true pumice]

do they always never tell us what the problem is?

Is this the video generating getting stuck at 99% problem?

[lyric cosmos]

yes

[hexed reef]

Hi friends, does anyone know why Flow no longer lets you make the characters’ dialogues in Spanish?

[jade spindle]

Why Veo never lets me do deep horror 😭 any tricks?

[honest hemlock]

I sometimes have a problem generating a video, but it only happened 3 times.

[sleek vale]

Not sure where to ask this. I downgraded my ultra plan, set to downgrade on October 1st. Changed my mind and there’s no way to cancel or revert the downgrade to keep ultra??

[crisp basalt]

i wish i have that much of money to easily change my mind to downgrade ultra plan

jk

[autumn vapor]

im having trouble to make a video from the last frame of previous clip, can anyone help ?

[muted tinsel]

Hey, do I still need to be paying for google ultra to use google flow VEO 3?

[serene juniper]

In Scenebuilder, using the Frames to Video feature, you'll drag that marker to the end of the clip, click on the + icon at the top, which will give you the option to save that last frame as asset (uploads it to your frames). Without using the Extend option, you want to go back down to the prompt box and choose that Frame as the starting point to the next prompt.

This is how you can make all the clips in the scene with Veo 3 Quality. Its not perfect, because with the Extend feature it is a smoother transition. I only have a Pro account, so the only option to use the Extend function is Veo 2 Fast. But I'm not sure if that's how it is for the top-teir plans? This is kind of a work around I use to create the scenes in the best quality and with audio. The transitions are not going to be great, especially if there is a lot of movement happening, because the movement of smoke or flames will change speed. But I get around this by using transitions effects in my video editor (CapCut). This can be a visual trick that makes enough of a distraction for your mind to be able to notice the differences. I like to use glitch transitions, because it's quick but disruptive enough to make it look the most natural.

Hope that helps! I did see them announce that they will begin to phase out Veo 2, which will be amazing, because I'm assuming that will allow to Extend the scene with Veo 3, so you don't drop quality.

[serene juniper]

No, I have the Pro Account and you can generate with VEO 3.

[muted tinsel]

Okay, I’m going to downgrade because $270 a month is crazy😭

[serene juniper]

It really is. I'm not sure how many Credits it takes to generate Veo 3 Quality, but with the Pro account, it's 100 credits.

[muted tinsel]

Same 100 credits, how many credits do you get tho?

[crisp basalt]

The AI credit costs are per generation, not per request

[muted tinsel]

Ohhhh nvmmm

[karmic trench]

Hello, am following up regarding an issue am encountering... can someplease help me my Flow video gets flagged for “child safety concerns” — even though it contains no child-related content. My prompt is below:
“Colorful 3D animation of cartoon characters clapping hands on a grassy field. Rainbow sky. Cheerful tone. No kids or toddlers mentioned anywhere. Lyrics: Clap your hands.
This is being flagged for content, which seems incorrect. Please help review this — I believe the filter is being overly sensitive. Thanks!

[serene juniper]

I would remove the word kid or toddler from the prompt all together.

[full flicker]

Hello, if you're encountering errors and /or bugs or would like to leave a general feedback about Flow please don't hesitate in utilizing the #1319424187730755636 channel with as much context as possible. It helps to keep the team organized and its also where all the technical issues are logged. Your post may get lost in this general channel so to ensure visibility please use that channel for all reports and feedback. Thank you.

[karmic trench]

ok thank you

[coral pumice]

how do you creat shorts using Flow? I've been trying to get it done all week. no progress

[red finch]

Hello grand rising I’m new here

[modern crescent]

Hello, I have a question why does it say that I cannot use Flow, even though it is supported in my region (romania)

[supple pendant]

remove "no kids or toddlers" - negative prompts don't work. it is likely misunderstanding this part of your prompt as "include kids and toddlers"

[last thicket]

well at least you didnt lose 800 credits with no videos generated

if you try to push with a limit filter-reaching image and you do 2 times 4 quality videos. you are most likely going to lose all the credits plus not getting any videos back

[crisp basalt]

did u contact google one staff for support and refund lol

i dont think complainin in here would change anything nor will get u refund, also the ai credit system will instantly auto refund if no videos generated lol

[last thicket]

You clearly havent experienced what i did

That is buggy

[crisp basalt]

yeah show us the screenshot of your ai credit details then

[last thicket]

If you get mixed "you are generating too fast and rejected prompts you lose all the credits and no videos generated

[crisp basalt]

did u check your library page

[last thicket]

As you can see i was spamming

[crisp basalt]

so u spamming out of anger?

[last thicket]

When it gets stuck at 99 its a bit annoying

But still, that shouldnt "steal" your credits lol

[crisp basalt]

this screenshot evidence is good enough for you to show it to the google one staff support team or flow team

[last thicket]

yea will do thanks

[crisp basalt]

id be pissed too if that happens to me 😬

but yeah right now , i think flow website is .. really buggier than yesterday, i cant do export from scenebuilder , it gave me 500 internal server error

[last thicket]

Yeah since the update

I'll be careful with generating videos too fast lol

Thats why you dont code with AI yet 😭

[crisp basalt]

well this ongoing issue making me checking flow tv for no reason

are these new or am i tripping

[crisp basalt]

they pushed an update to Flow and now the scenebuilder export are working properly ! thanks team ❤️

[red finch]

grand rising how are you doing today?

[stuck perch]

is the voice generation okay now?

I tried about 3 times yesterday and it just consumed my credits nothing else

[stuck perch]

no voice again

huhuhuhu now i lost about 100 credits

[south harness]

do put words in it like kid, child, boy girl or anything in that nature

[stuck perch]

there's voice generation error message after video is generated

[stuck perch]

i did

i wrote dialog in the prompt

[south harness]

dont put words sorry typo

do you want a veo 3 chatgpt script writer

make it easy

https://chatgpt.com/g/g-683b3de2ec988191a10181cfa4851b25-veo3-movie-scene-creator

try this

make sure your not writing anything bad, also you cant write boy, child, kid, tennager etc

no one underage

[stuck perch]

i didn't intend or specify dialog to be of kids'...

it's an animal animation

[south harness]

just double check your wording

can say kill or dying

[stuck perch]

nothing like it

[south harness]

you want to share me your script

[stuck perch]

okay

[south harness]

I can help you

[stuck perch]

how do I dm you

[south harness]

yes dm me

[copper mica]

Hello, we wanted to remind you of our server guidelines, stated in #start-here・📍 : “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Please keep in mind that non-English discussions and content may be removed

[flint summit]

Very prelminary impression - Sora2 is NOT superior to Veo3. Veo3 seems better at giving realistic facial expressions, better voices, and seems to just understand the situational context better. And looks more realistic. But maybe I just need to prompt better.

[crisp basalt]

Overall sora 2 actually generates much variety output and not as “rigid” like most of my veo 3 output I’ve done

But yeah veo 3 still good in my book tbh

[iron matrix]

yay!

[kind bison]

The floodgates are opening, though. Soon all the top models will have dialog and VEO is going to have to come up with something really impressive for V4 (and release it soon!) in order to maintain its edge. People are not going to continue to shell out $250 a month for a platform that shamelessly censors and is slow to address bugs.

[muted tinsel]

Anyone having audio issues ?

[dapper tundra]

If they just relaxed guardrails a bit they’d remain on top easily. Wan 2.5 while inferior allows a lot more creative freedom than veo 3. From what I’ve heard and seen it seems even sora 2 has more flexible guardrails.

[fringe nova]

has there been any talk of extending the 8 second video limit?

[dapper tundra]

Yes

[foggy glacier]

Stuck at 99%

[flint summit]

Playing with Sora more I'm finding it is better at some things. Like inserting appropriate camera cuts and angles. Also its 10 secs vs 8 secs for Veo3.

[small elk]

Hello, we wanted to remind you of our server guidelines, stated in #start-here・📍 : “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Please keep in mind that non-English discussions and content may be removed.

[woven escarp]

99% hang is very much still an issue, with no update or information from the @tacit halo team. Unacceptable that people are paying for a premium service without the product they expect. Like others, I have clients waiting for projects, and to have Flow down for nearly 2 days is damaging to business. With the latest Sora 2 announcement, I imagine they will start to lose customers. I certainly will be one of them as soon as it becomes available in the UK.

[raw oak]

and Sora 2 is free right now

[plucky patio]

You know I'm genuinely interested on your thoughts I've never used flow so me saying that via was better with whisk then you know I sound pretty stupid but if you guys could teach me what you guys have been learning with flow that be much appreciated because I'm a little behind

[pallid kite]

https://www.youtube.com/watch?v=HsU94xsnKqE

[severe geode]

anyone get stuck at 99% and resolve the prob?

[delicate yew]

I tried flow as a gift because of testing veo and I tried it with a open mind, but I don’t think anyone would buy this except for maybe once before realising the problem. You can only create 3 videos a month, and if a generation fails or is censored for any reason you lose credits too, so you could potentially make 0 videos and lose all of your credits. It only gives 2 versions of the video instead of 4, so all of the 3 free videos I generated were bad. I think the app should also list how many credits each video type will use because it is unclear. Compare it to Midjourney video plan where you can have many relaxed video generations, and even 25 videos on a really cheap $8 plan. The videos also have a watermark on them unlike in veo 2 and 3.

🐌 wait why am i a snail this is so cute

[thick oriole]

Hi, sorry if you had a bad experience.

1. The cost per video is available in the "AI credit menu table" here: https://support.google.com/googleone/answer/16287445?sjid=11417802212445045145-NA. This information is also available in the UI before you hit 'generate'. Please let us know if that's insufficient or something seems wrong there.

2. Failed or blocked generations do not cost credits. There can be some small delay in seeing refunds (sorry, we're working on minimizing that, we know it's a point of frustration). See FAQs here: https://support.google.com/labs/answer/16353333?visit_id=638949345316065429-2309286377&p=flow&rd=1, and please let us know (or contact Google One support if you're a paying user) if there seems to be a problem there.

3. You say "you can only create 3 videos a month", I assume you're talking about generating high-cost videos without paying? Yes, unfortunately due to the high cost of these models we're limited in how much we can give for free, but always looking to improve there.

[thick oriole]

See https://discord.com/channels/1172568727942860810/1420895068994277487 for more info if that helps (and perhaps follow up there if not)

[thick oriole]

See updates in https://discord.com/channels/1172568727942860810/1420895068994277487 and let us know there if you want to follow up.

[delicate yew]

Yes I generated the videos without paying but it is the same AI pro tier and if someone later pays for the AI pro tier I think they would get the same amount of credits. and they will have to wait hours for credits to refund and the generations fail more often than successful. Also i made thousands of videos with veo3 with no watermark so its kind of underwhelming.

[nocturne dagger]

is the watermark a requirement, or is there an option to turn that off?

[thick oriole]

I think there are a couple of misconceptions here I can maybe help with, but we do always appreciate any inputs & requests.

1. Pro users would get 10x the number of credits (1,000 vs 100).
2. I believe refunds take a maximum of 10 minutes, often faster, let us know if you're seeing otherwise.
3. Yes some generations can fail, but I wouldn't expect the number of failed generations to be anywhere near over >50% unless you're trying one specific thing that's either against our policies, or is hitting a particular issue with our filters that we need to improve.
4. I'm not sure what you mean by your point on watermarks, but only Google One Ultra users can generate videos without watermarks: https://support.google.com/labs/answer/16353333?visit_id=638949383111786928-3682762016&p=flow&rd=1#zippy=%2Cwill-my-generations-contain-a-watermark

[steel birch]

did you use the veo2 fast?

[leaden mirage]

Veo 3 fast

[steel birch]

It looks fantastic with a simple edit. I was wondering how you made the finger look so real until I saw the Veo logo.😆

[leaden mirage]

That was the frustrating part, because the first tap is really me, and then the character was supposed to open the door himself but veo kept adding AI hand to open it, in the end I said thats not too bad. 😆

[steel birch]

you are using ingredients to videos?

[leaden mirage]

No just frames to video

How much do you think I should charge for this video? 🤔

[steel birch]

My price set for local church is around $400 per min since they don't need any actors. but I will say at least give yourself $50 per hour on time you spent.

[leaden mirage]

That sounds fair to me honestly. Although Veo really works magic, end product is still extensively worked and deserves proper compensation.

Maximum i will get paid for this video is perhaps 20 dollars. (Max). But at least I’ll add it to portfolio 🧐 🤷‍♂️

[steel birch]

It is really depends on who's paying

[leaden mirage]

Excellent work

Parallel to short form videos my colleague and I are working on kids animations to put out on youtube. So far 1 video but not much views.

[placid spire]

I really hope to see a Flow mobile app or direct support for phone screen soon. Right now, after I generate a video with Flow, I still have to figure out how to convert it for mobile screen. Using Sora yesterday was so convenient. I could just share the video straightforward to social media. I wish Flow would truly focus on user needs (mobile screens) instead of executive style presentations (wide screens). Hurry up and keep pace!

[leaden mirage]

I abandoned sora as soon as veo came out but I see your point there

[placid spire]

Sora 2 is impressive

And Sora 2 is a native short video social media as well like Tip Top. A smart approach.

[noble carbon]

it is, i am hoping that veo 4 will be as good as sora 2 because i love the way scenebuilder and creating videos is on veo 3 already

i'm glad sora 2 is impressive because the competition for best video generation will make veo 4 prob be as good if not better than sora 2 which would be insane

[placid spire]

Agree. As an always Google fan I hope with Veo 4 I can generate videos and directly publish on YouTube shorts.

[tepid tundra]

Yep, the only way to avoid this is by getting the ultra tier to have unlimited veo3 fast video generation, it's expensive tho, I'll be using it for few months only that's as much as I can pay lol

[granite sage]

I don't know what is the discussion here, but only to talk something about this tool: here don't work. Says that you can't do the generation in "vertical mode" and needs to change to "horizontal" 16:9, but it's already in 16:9 and just don't generate the video.

[ruby roost]

Share some of your prompts that failed

[frigid pier]

now the question is when veo 4?

sora 2 is outperforming veo3

insane

[timber mantle]

share some codes for sora 2 😄

[stuck adder]

No it’s not, not by a long shot

You have no control over anything you get what they give you

Their image to video on the app sucks, can’t do people or anything

Veo 3 is stilll far in the lead

[hallow comet]

Hi guys, especially @crisp basalt, I've created a CustomGPT that enhances your VEO 3 prompts by asking you questions and then exporting the result in a ready to paste prompt. I personally struggle VEO 3 prompting and I'm sure a lot of others do so as well. That's why this CustomGPT will ask additional questions about character, location, camera movement, zoom effects, sfx and much more!

I need your help to make this the best CustomGPT for VEO 3 prompting so we can all use it.

(if anyone wants to see its backend commands, just let me know, I'll be happy to share them with you.)

check it out and let me know what improvements you'd like to see!

https://chatgpt.com/g/g-68d7a8aeab4c81918fe7007f42a709ff-veo3-promptsmith-v0-4

did i mention it also supports start frames? just give it a start frame (in simple terms, an image) of the video you want made and it will fill in the info for you

[snow lion]

just wanted to know, if i buy gemini ai ultra, did i still get 0 credits for using veo 3 fast ? @main saddle

[raw oak]

Hi. Is there a trick to get the character to speak. It seems that 1 in 10 generations I actually get the person to speak. Often it appears just as a voice over.. not sure if I am the only person suffering from this. Thanks

[forest raptor]

Now we have Sora. Are you feeling the heat, Google?

[forest light]

Does anyone know when we'll be able to use vertical videos in SceneBuilder?!

[granite sage]

Also me, sometimes have this problem, including a bug with "sensitive content" even without any of this in the scene.

Well, helping you, I can say: put a clear command to a voice line. What they need to speak and how and when.

[raw oak]

Yes. I have that one too. I have tried both. I find that when the prompt is simple it does a better job but still fails. thanks. let's hope it get's fixed. It happens to me on Fast and on High Quality renders which is a bummer because those renders are very expensive.

[granite sage]

For me is the oposite: simple prompts don't do exactly what I want. The job is good, because you don't restrict the IA creation, but this escapes what you intent. So I'm working with big prompts, full of instructions. I start with a simple one: "a mother and a daughter walking in a vibrant field of wildflowers". Then, after the generation, I start to do little changes, part to part, time to time, one on one... it's a very patience work...

[median rover]

With watermark and limited resolution. Pro version coming soon.

[raw oak]

Looking forward to the next update here of VEO 3- hope it comes soon..

[opaque sand]

Only if they surpass unlimited generations at approximately $400 for 3 months for each new profile...

[forest raptor]

they're talking about giving free gpt users access, so

I'd LOVE it to see the whale spenders regretting spending all those money on veo3

[floral kindle]

My video dont have audio. What wrong guy?

[opaque sand]

What can we say about Runway Gen4 users who bought a 12-month unlimited plan for $1,000 at the beginning of the year and have been sitting in “silence” and the “Stone Age” for half a year now?

[forest raptor]

exactly, same thing is going to happen with veo3 eventually

[opaque sand]

I don't care which platform I work on, as long as it gives me the best value for my money. Let them fight among themselves for my next subscription payment.

[forest raptor]

unless you're very rich, 200$ per month to have a limited use is definitely no value for anyone's money

heck, even if I were rich I'd kinda feel I were throwing away money

[opaque sand]

I'm trying to understand your logic.
Expensive compared to what and for what?
Playing director for fun when your real-life salary is $600 a month is definitely expensive.
Being a loser who tries to make money on content for his “new” channel with brainless or stupid template videos is also expensive...
$140 per month for unlimited generations. Just create a new account once every 3 months. What could be easier and cheaper?
The same runway, $90 per month unlimited, but there you need to do 30 iterations at a rate of 2 pieces/10 minutes (that is, a couple of hours minimum) to get something similar to what you get here in 3 minutes.

[forest raptor]

if you think 140$ per month to play with 8 seconds ai videos is fine, and you can make a new credit card every three months, then........ enjoy yourself?

Again. I'm on a high school teacher's salary, but even if I were on a CEO salary, I'd think twice

And believe me or not, 99% of the people think the same - which is why I hate when a company caters to a 1% of whales

[opaque sand]

Once again, bro.
I did NOT say it was cheap. I said that either you need it or you don't.
When gramophones were invented, no one talked about whether they were expensive or not. Either you bought one and listened to it, or you listened to it from your neighbor's window. It's the same with this technology.
And yes, in my country, the official salary for a teacher is $250. The average unofficial salary in the country (office work, movers) is $500-600.

[opaque sand]

You're not thinking correctly.
$140 to “play creator” is a lot. But make sure you get your money's worth (make a video ad for your school or the barbershop where you get your hair cut) and this month will be free for you.

[unreal viper]

reminder of rule 1: Be compassionate and treat others with kindness and respect.
All kinds of harassment, hate speech and impersonation will not be tolerated. Keep conversations civil, welcoming, and on-topic. Post messages and content in good faith.

[opaque sand]

I think everything is fine here.

[median rover]

In flow how is the negative prompt handled? I.e. no this and that ….. I rather preferred the dedicated box for them.

[misty merlin]

Hey guys, i created a new e-mail and Can’t get Access to flow. Can someone help me please?

[kind bison]

I feel triggered!! 🤣

[crisp basalt]

veo watermark +random in video watermark , pick a struggle

[stuck adder]

You do not have ultra?

[thick oriole]

Is that a hallucinated "Film Maker" watermark? That's sad, and is a core model failing. Sorry about that, perhaps some other prompting may help avoid it.

Yes, the Veo watermark is present unless a user has Ultra.

[thick oriole]

Hi, what error are you seeing? It's possible you need to complete age verification. We recently added that to our FAQ and will add it as an in-product reminder when we notice that's the issue soon too, it recently started cropping up as more of an issue. https://support.google.com/labs/answer/16353333

[misty merlin]

It says: You apparently don't have access to Flow. You can check availability here. Support is available if you have any problems.| i am over 18 and so is it in my account. I am from germany and used veo3 pro before on another Account. But my e mail got hacked and i am afraid of buying Ultra on this Account. So i created a new email but still can‘t use

[thick oriole]

Got it. With that message, https://myaccount.google.com/age-verification is my best guess. If that doesn't work then Google One support is the next best step to see what they can find specific to your account.

[misty merlin]

Thats it! Thank you very much

[peak hatch]

Anyone got a sora 2 invite code?

[fathom flax]

I created a character lora (on wan 2.2) based on a face 100% generated with ai, but veo3 says it's prominent people, any advice?

[snow lion]

lol no one answer

[granite sage]

If understand, Ultra signature provide 25000 credits. But Veo 3 Fast don't consume credits in Ultra signature.

https://support.google.com/labs/answer/16352836?hl=en&ref_topic=16355608&sjid=2054686089773086997-SA#:~:text=about experimental features-,Audio Generation,-Important%3A This

Look this.

[fathom flax]

Im using ultra plan, exactly no credit will be used for veo3 fast

[crisp basalt]

Maybe not here since this is flow channel

You might get better chance in OpenAI discord

[hollow warren]

Nahhh 😆 too many people are asking. I figured I would have better luck asking here since I'm sure like me, at least one person here uses both services. I just don't have an iPhone:(

[crisp basalt]

Mine all 4 invites code were finished yesterday sadly

[crisp basalt]

Oh there’s many here uses sora too

[hollow warren]

Waaaaaa 😭

Sora has the best image generator. Idc what anybody says.

Invalid.

[stuck adder]

Can’t wait for veo 3 start end frame

[warped surge]

I created a 3D cartoon-style video, with a frog, lion, snake, monkey, and children dancing happily in a colorful environment.

[stuck adder]

There’s a 100 limit on the videos

Should’ve kept those 4 invite and use different accs so you have 400 per 24/hr

[sterile junco]

What do you guys think of making ads for a business using flow and VEO 3? is it somehting that could work or is it something that would require more time for the generations to get better?

[crisp basalt]

@hollow warren

[hollow warren]

Thank you but it's invalid:(

[dapper tundra]

You can’t upload any custom images of people to sora 2. What’s all the hype about? For random text to video generations with zero consistency?

[crisp basalt]

thats one of the disadvantages of the sora tbh,

rip veo 2 soon

[crisp basalt]

There are still quite a few things Sora 2 can handle better than Veo 3, but honestly thats pretty debatable and depends a lot on personal perspective

[kind bison]

I shelled out for Sora 2 Pro - it's awful!! As wonky as VEO 3 is, it's way better for AI filmmakers than Sora 2 Pro.

[mental mulch]

Why is Veo 2 ending support

[past pond]

how removing logo VEO 3?

[lusty cradle]

Is there a cheaper way to use Ingredients than a 100 euro a month subscription?

[snow lion]

thank you so much

[cloud burrow]

Maybe to make room for VEO 4? 🤷‍♂️

[cloud burrow]

Ultra plan i believe

or crop it out in an editor..

[patent hedge]

Hi, I'm desperately looking for resources, tutorials, or explanations about the ‘Scene Builder’ feature. Any suggestion?

[stuck adder]

What is sora 2 pro

[stuck adder]

I doubt they have veo 4 coming soon

They need video to video now

[cloud burrow]

i have no idea myself but my guess is they will be removing it to make room for something

[stuck adder]

Once they do do video video with itl be crazy

[wooden helm]

Did you guys got access to sora 2?

[dapper tundra]

I did. Lost interest in it immediately when I found out you can’t upload custom images of anything resembling a human. Also there is an annoying floating water mark that jumps from corner to corner on every video you make.

[viscid topaz]

id still love to have a code

[viscid topaz]

its decent tho for lil post

[crisp basalt]

try ask in openai server, not here

[smoky mauve]

Hello everyone

[forest raptor]

your move, google....

sora 2 ^

I don't trust openai, but if they keep their word, we can finally see video ai models become a commodity for everyone and not just for whale spenders

[nimble bolt]

Today I feel Veo3 is not adhering to the prompts as it does a couple of days before. Can someone please tell us if there's an internal error that's causing the problem in the generations?

[icy lodge]

yo does anyone know why i am stuck at this screen?

[granite sage]

Is Gemini working normally there?

[zinc swallow]

Try Ctrl + F5

I haven't used Flow today yet

[kind bison]

The Pro version of Sora 2, LOL. It sucks - already canceled and got a refund. Looking forward to VEO 4 in December.

[zinc swallow]

Daym how did it get so bad?

[mint tiger]

Every time I try to generate a video with a start and end frame, it tells me that it violates the policies. Is this happening to anyone else?

[mint tiger]

This generation may violate our policies. Please try another request or send us your feedback.

[heavy oriole]

Hey everyone, quick question. Is anyone else having trouble generating videos? I keep getting 'policy violation' errors on literally every prompt, even super simple ones

[mint tiger]

Yes, the exact same thing happens to me.

And in the initial and final frame mode it is a mess, it does not follow the instructions correctly

[granite sage]

Similar here. I even already did a bug report in the chanel for bugs.

[stuck adder]

Damnnnn it was that bad

[heavy oriole]

Seems I'm not the only one with these issues

[quartz zinc]

Hi people, just got a question, does Ingredients 2 video only support Veo 2 - Fast model atm?

[mint tiger]

For now only for the Veo 2 model

[pliant raft]

Hey i left for a month and im back, did they change something regarding the policy, i litteraly kinda cant generate anything, i mean its almost not usable, if not usable at all, everything related to realistic humans especially womans, text to video, image to video, no matter what it is, will automatically violate the policy, even the most casual things

[flint summit]

Ok here I am crawling back to Veo3 after cheating with Sora2. But then Sora2 got nerfed today so I'm back. Can you see it in your heart to forgive me Veo?

[hazy glen]

Is flow down? I bought Ultra yesterday and generated around 8 videos and now when I do anymore it gets stuck at 99% for every setting?

Not really a good start to paying for Ultra lol - @signal geode Do you know anything?

[signal geode]

Hi - I don’t! Isn’t there a status page for Google somewhere?

Alternatively, try clearing web cache, signing out, restarting. If none of those work, then I’d assume it’s a Google-sided problem!:)

[icy lodge]

thanks bro

[hazy glen]

I have tried to look but cannot seem to find anything online about it being down, just wondering if I am the only one

[signal geode]

Give it a few hours and see if it’s still problematic

[hazy glen]

Its been down for me for a good 8 hours so would've thought there would be something stating it

@crude cloud Can you take a look?

[wet burrow]

Sora 2

[worthy veldt]

Please can someone help me with code

[sinful delta]

"I can't create any videos today. The system says I'm violating the policy, even though I haven't done anything wrong."

[proper ginkgo]

Could someone suggest a workflow for creating videos where humans/animals are talking and lipsynced? I got the basics but looking exact/good tools. Does VEO3 do the whole thing from start image? Example: https://www.instagram.com/p/DPWTl2cCNVz/

wearegoodpets

Don’t worry if you feel exhaustipated. We are here to make you smile 🐶😊
Share this with someone who needs a little cheer today

Likes

10660

Instagram

[vital quest]

guys is it possible to make AI characters speak from script if so how to do it

[keen quest]

Any mod active? im facing problem with my veo 3. my videos always get stuck on 99%. its happening from past few days

[forest raptor]

Folks? I heard rumours about Veo4?

is it real?

[devout nymph]

hey guys i have a quick question, for the selected number of outputs, does that use more credits? like 1 output = 100 credits, but if i selected 3, would it cost me 300 credits?

[crisp basalt]

yes

and since youre a pro user, do 1 or two output video, and choose fast model to save your credit,

[devout nymph]

ok thanks

by the way, im wanting to create a 30 second video, but of course, this means compiling multiple 8 second clips, do you have any tips for succesfully doing that? even if i reuse prompts from a previous video and simply change whats said, it can mess up and randomly change the persons accent for example, even after explicitly listing the accent within the prompt

also my videos are coming with veo watermark, how do i get rid ot taht

[stuck adder]

You couldn’t use sora 2 in any thing production level anyway lol

Veo is better for that stuff with an upscaler