flow-general・🌊

[kind bison]

I feel like there was a dip in quality for while tonight - generated a whole bunch of videos over the last several hours and the resolution seemed lower than normal (I used both 3.1 Fast and Quality). I generated some videos earlier in the day that looked much better. Was also getting a lot of random "policy violation" errors for prompts that went through without issue just a few minutes earlier (and then would go through a few minutes later). Another issue: characters were changing a lot throughout the scene (I was using input images, first and last frame) - like people even changing race!

[proper drum]

Hi, just asking is there a way to turn portraits video into landscape?

[tardy idol]

Hey guys, when I use Start-End frame, Veo 3.1 adds a 3rd image in between and ruins the video. Does anyone else have this issue?

[tardy idol]

Photoshop content-aware scale

[proper drum]

Yes, that works if the video is static, which is not the case. I have camera movement in the scene.

[tardy idol]

turn the static image to landscape first, then animate it

[proper drum]

I understand, but I don’t want to convert a portrait image to landscape. I want to convert a portrait video to landscape.

[warped wind]

I once again encounter "You are sending requests to fast", is this a new update or a bug or what I can't do my work all week

[warped wind]

@tacit halo is there any update about how much generations we can do now?

[digital yoke]

Hey

[turbid lintel]

wow

[cinder ingot]

Did the guy who lost 4000 flow credits ever get his credits back lol.

[slate geode]

I’m really excited to explore Flow for cinematic storytelling.
Here’s the type of scene I’d love to generate: poetic, bright autumn morning inside a cozy shabby chic French home decorated for the fall season. A brunette woman in her 30s or 40s wears warm plaid pajamas and walks slowly down a wooden staircase, followed by her little cat. She prepares coffee with a vintage espresso machine, then sits on a beige sofa near a large bay window. She holds an autumn-colored mug with visible steam rising from it. The cat curls beside her as she gazes peacefully through the window, watching the golden autumn leaves outside. The room feels luminous and spacious, filled with fall décor — pumpkins, apples, candles, and a bouquet of fresh wildflowers. Gentle sunlight filters through the curtains, creating a warm, calm, and poetic atmosphere.👍

[stuck adder]

Oh they must have made it available for pro users too

[turbid lintel]

I got good expands , so far the generation is accurate

[last thicket]

Still not working? I haven't received my credits back

[twin mortar]

Has anyone on here been approached by a company called Kweeks? I’ve had a message saying they liked some of my work and want to have a chat. Can’t find any info on them though.

[warped wind]

Does anyone know how much generations we can send now? In the morning it was all ok now I can send a request per 4 generations every 5 min or so

[turbid lintel]

Veo 3.1 now can generate living dinasour

[queen lantern]

Is flow down? seems i cant access my library

[flint oyster]

How do you get flow videos to stay true to the input ingredients? ie if I upload images of an AI-enabled camera / sensor pod, how do i stop flow from morphing it into some other unreal camera that isn't my product?

[near gate]

A realistic scene of a dinosaur: a massive Tyrannosaurus rex roaring in a prehistoric jungle, mist rising from the ground, morning light filtering through the trees, cinematic lighting, detailed textures, hyper-realistic atmosphere.

[remote jetty]

any tips on controlling the intonation of the dialogue? I know one can just dub them later but wondering if there's prompt wording or markers for this

[queen lantern]

Is anyone else unable to sign into google labs? Has been an issue all day

Mobile works, but not pc - ultra user

[vale fern]

/lmr

village

[upbeat anvil]

is anyone else having problems with prompts they are inputing into google labs?

[wind saffron]

Working on PC for me, no issues

[tidal anvil]

I cannot get any camera movement while using a start frame with veo 3.1 fast. Dolly, crane shot, orbit. They are all ignored

[leaden mirage]

how does one use ingredinets to video? so I just add three ingredients, for example one background and other two characters. how do I indicate which ingredinet must do what?

[sharp hedge]

create a video footbal

[lucid iris]

do you guys know why the ingredient is all messed up when i try to add it to the video? like im trying to add just a page of website to the video but the content (words) of the webpage screenshot is all messed up in the generated video

left is my original ingredient (a ss), and right how it is in the generated video

[wispy forge]

Well flow is finally working it is generating 1 video at a time but it's better than nothing

[lucid iris]

thing is its kinda bad at websites as there are tons of words.. but im trying to use it promote websites, just need to figure out a way

[wispy forge]

I think it will take them some time to improve this feature.

[pallid kite]

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

[unreal belfry]

"Couldn't generate video. Please try again later." over and over and over 😫

[wispy forge]

Trying keeping the no of generations just '1', it has been working for me that way from past 4 hours no fails yet.

[autumn vapor]

hey guys, what is the differance in flow Pro vs ultra sub ?

[mellow crown]

Harsh singh is my son

[solid prairie]

O

Hello

[stuck adder]

lol

Hey folks

[full flicker]

Hello @keen hearth , as 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. If you are asking for advice or tips on your prompts do so in the appropriate channels.

[pulsar crescent]

Double check your prompt to make sure it's clear

Does anyone know if there's a way to move a Flow-generated video from one project to another within Flow?

[tiny geode]

Guys help me. Why doesnt audio play?

No audio generated

[pulsar crescent]

This happened to me a couple of times. A reboot of my computer fixed it.

[tiny geode]

Dude ive tried over days

[pulsar crescent]

So did I. The problem persisted until I rebooted. Did you reboot? Flow also won't generate sound if kids are in the video. That was the case for at least one of my instances. Good luck!

[unreal belfry]

After logging out then back in, it started working.

[misty stone]

Feature Request: On mobile (Android) hitting enter submits the prompt rather than adding a line break. Please have
Make a line break instead, please. Submit only when hitting the send button. (I've wasted so many credits with this muscle memory of normal typing behavior! I have to enter a wall of text: It's not possible to format a prompt within flow, because it's not possible to do a line break. If I want to for, I have to do it in an external editor and copy and paste.)

Can someone help me prompt this correctly? I've used up all my credits on this. It's video from an existing image of a Halloween mannequin with an inflated pumpkin head.

Every single type of prompt I've tried has resulted in the same inaccurate result: His head doesn't deflate, it simply drops off onto the ground.

The following is just ONE of many of the prompts I've tried....

1. This pumpkinhead man says in a distinctive character voice, "I'll bet you can't do this when
   you hiccup."

2. He pauses a moment in preparation, proudly standing up straight

3. He then hiccups which instantly deflates his head, sucking it into his torso though his neck. His torso, size, and position stay the same.

4. Wind and leaves settle afterwards as he continues to just stand there, motionless.

I'll share results in a thread...

[crisp basalt]

is this what u were hoping for?

oh wait i think this one is better

The subject says in a distinctive character voice: I'll bet you can't do this when you hiccup. The subject's body remains still but straightens proudly. A loud hiccup sound effect occurs. Instantly, the subject's inflated head rapidly deflates , sucking inward and vanishing completely into the torso's neck opening. The subject's torso and body remain perfectly still and motionless in the same position The sound of gentle wind and rustling leaves settles in the background

[placid blade]

looks real if u ask me

btwww can u only adjust video duration through the API ?

[crisp basalt]

yeah , im not sure if flow team planning to bring the functionality to generate how many sec of vid we want

i think they will

[misty stone]

Thanks @crisp basalt ! It's going down the right road now, following the instructions. I'm aiming for a real sense of the sudden in-breath of the hiccup violently sucking the inflated head instantly with a shreddy pop into the creature's lungs.

[placid blade]

u can technically extend it to 140secs if i remember correctly (read that in a reddit thred)

[crisp basalt]

i mean last time i was rooting for them to bring portrait mode to Flow specifically for veo 3 , and i saw they already enabled the portrait mode for veo 3 api back then

and few weeks later , portrait is on flow

[placid blade]

ohhh for some reason when i tried to use portrait it gave me an error (one week ago)

[crisp basalt]

the subject says in a distinctive character voice: I'll bet you can't do this when you hiccup. The subject's body remains still but straightens proudly. SFX: A sudden, sharp gasp of in-breath, immediately followed by a single, loud hiccup Timed exactly with the hiccup sound, the subject's inflated head instantly and violently deflates, collapsing inward as if it is being vacuumed into the torso's neck opening. The subject's torso and body remain perfectly still and motionless. The sound of gentle wind and rustling leaves settles in the background.

[misty stone]

Very cute. Funny it added the cartoon lines. I like it.

[wooden crow]

I’m not sure if this is feedback, a bug report, or my ignorance. Using flow and scenebuilder, I want to make a video of 3-4 minutes using several scenes, each of which has several clips. I may do this over the course of a week or more. I want to be able to replace, add, and rearrange clips in each scene until the whole video is done. What is the best workflow to do this?

As far as I can tell, there is no way to save a scene of several clips, close the computer. And the next day start a new scene; and also have the old scene available if I need to change it as I work on the new scene.

Is there a way to save it? If there is a way, how? If not, my feedback is: please save it! (And what is the best workflow? (Gemini suggests saving each clip externally on an editor like DaVinci – but that defeats the charm of scenebuilder.

[cinder ingot]

<@&1366831844502274180>

[copper mica]

Thank you

[eternal bronze]

I getting a lot of failed generations...how do I know if my credits are being charged? I know it takes a bit to the credit counter to update

[next jay]

i think they update instantly

[eternal bronze]

is there a way to look at my pior generations spending?

[frigid viper]

Oh no Momo ^^

[carmine forge]

Right now, I’m making a short film using Flow AI. I create my images in Leonardo AI, and then I use the Frame-to-Video feature in Flow.

However, when I try to upload my character image, I get this error:
“We do not allow uploads of minors at this time. Please try a different image or send feedback.”

The issue is that my script includes a child character, so I need to include a child in the video. I’m not sure how to generate or upload this kind of character in Flow without getting this error.

Can you please help me find a solution or suggest a safe way to include a child character in my Flow project?

[cinder ingot]

I'm not sure how to help you with that, sometimes I successfully generated a child in video without even asking.

What I find helping is adding certain keyword to integrate using child in the video. [I make children clothing videos so I will ask for a model and it'll bring a child model]

Other than that, you might need to rely on competitors.

[latent cypress]

does anyone know why veo 3.1 is not generating any sound or music in the video today both in flow or gemini? i have lost many credits because of this 😩

[lapis pine]

do you guys also get broken pixles across the screen when you upscale to 1080p directly in Flow? Its especially noticeable during bright scenes

these things

[stuck adder]

How can we shorten the time in flow? I don’t need all 8 seconds to wait for something to happen

I wanna use like 3-4 sec max

[crisp basalt]

idk

[zinc swallow]

Relatable. We always have "to wait" before the actual call to action happens in the vid 😆

[full flicker]

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. Thank you!

[kind bison]

It's such an amazingly stupid policy, it's drives me nuts!

[glossy thicket]

Hw

[shadow warren]

in general: Is it possible that the image upload modes (ingredients, frame to video) are generally more restrictive? I have so many more fails iwth those, even with stuff that the text to video created

[tardy idol]

I did, I even did it for 2 different scenes and it's adding in both of them

Is the Google Labs team even here? I wrote a bug in the channel, and no one has answered for days now, despite responding to other cases.

[shadow warren]

i mean if they responded to other cases, that means someone is here

[stuck adder]

real they need a 2-3 second option

[opaque minnow]

Hi, I’m trying to expand videos in scenebuilder with veo 3.1 on my IPad but I get an error that video can not generate try again later, do you know guys whats going on?

[turbid lintel]

Anyone here want to share custom expands?

[light plover]

Hi, I sometimes experience that too. It might be a network issue. In some cases, the video shows up again after refreshing the page after a while.

[wooden crow]

Suggestions: I'm trying to make a video several minutes long, over the course of a few days. One thing that would help is a persistent scene. Another thing that would help is if when Veo creates multiple versions of a clip, it would name them a/b/c ... A third thing would be a global library, to make it easy to use clips from multiple projects/days (in that persistent scenebuilder scene) or to combine deifferent scenes into a single video. A final thing that would be very very helpful for my video would be the ability to sync to an external voice track.

--- Unrelated: Is there a "suggestions-bug" subchannel in the Flow section? It's not showing up right now, but I thought I've seen it before.

[crisp basalt]

#1319424187730755636

[wooden crow]

--- I'm not sure this is the right place for this post. I'm going to repost it in flow-bugs-feedback

[crystal olive]

hello! I just started using Flow. I created a scene extending multiple videos to complete a 36-second clip. My issue is the voice of the character. It is different in each clip. Any tips on how to get consistent voice OR any advice on how to change the voice over for the entire clip the same?

[eternal bronze]

ive gotten 4 videos out of 6 with no sound. Prompted correctly for the subject to speak to camera

credits flying

unfair innit?

Can i get my credits back?

[crisp basalt]

drop the prompt here so we can help refine it so u dont have to make same mistake again

[eternal bronze]

Man speaks with confidence to the camera in an American accent: "It's what a commercial looks like now" Finishes with a smirk .

same prompt woirks some times...what is causing the audio not to show?

Ive been using this prompt structure in other UGC videos and had no issues for 2 days

[crisp basalt]

was it image to vid?

[eternal bronze]

yes...now i have my subjects reapeating some words that are not in the prompt

[crisp basalt]

can you dm me the image so i can try it ?

[eternal bronze]

same image is working once every 3 times

bottom wporked...top didnt just now..top is silent

this is a bug

[crisp basalt]

lmao i got it too

[eternal bronze]

its crazy unprofessional...there must be a way to get those credits regunded

its obviously a bug...had to go to weavy and use credits there to keep my workflow going

[crisp basalt]

so 4 out of 4 vid with no audio

@eternal bronze

theres two succesful vid tho

both did not finishes with a smirk, it finishes with gibberish talk

yea either veo hates the prompt or

i dont wanna say it hates the image LMAO

[eternal bronze]

that looks good...yeah..its not enough text so the engine filles uop with jibberish

[crisp basalt]

wait no yeah its not the image

its the prompt

imma try this "A confident male barista, medium-dark skin tone and short, curly black hair, stands in a brightly lit, modern coffee shop, the camera slightly below him in a selfie-style, wide-angle close-up. He is wearing a fitted grey collared shirt and light-colored pants. He actively gestures with one hand while maintaining strong eye contact, speaking with conviction and an American accent as the camera subtly drifts back a few inches. In the background, stainless steel espresso machines gleam, and a blurred cityscape is visible through large front windows. The man concludes his statement with a smooth, knowing smirk, his eyes narrowing slightly as he delivers the final line, "It's what a commercial looks like now.""

DAMN NO AUDIO TOO

[eternal bronze]

i really think it a bug

thanks for your time

[crisp basalt]

im generating again

dont worry

[eternal bronze]

ive been promting similar ti this a few days, and using this prompt (not saying its the best, just getting audio) in third party sites and all good

[crisp basalt]

im gonna flag it , its best for u to flag those failed generation with no audio btw,

[eternal bronze]

what for?

[crisp basalt]

as inaccurate, since theres no other choices

im gonna refine the prompt again

im gonna remove the american accent

[eternal bronze]

its all good...ill have to use the bits that are useful...i just dont like credits being processed if the audio is not generated...bit unfair since oibviously audio is prompted

[crisp basalt]

yeah 8 vids total , no audio generated at all. my god thats annoying

[eternal bronze]

yep...pain in the ass...i have a few I can rescue

need to edit but should be fine

pain in the ass...lost all my remaining credits this morining on these "simple" vids

[crisp basalt]

have u tried using vids.google.com and gemini.google.com? both got image to video feature btw , it doeant use ur ai credit, only quota of idk how many vids per day for vids.google but gemini can only do max 3 vids per day

[eternal bronze]

no i havent tried there

[carmine forge]

Thank you for your support. Is there any way to contact the Flow team regarding this issue? I have already sent feedback about it, but I haven’t received any update yet. Do you have an official email address for them?

[carmine forge]

When I create a text-to-video, I get the video, but I can’t maintain character consistency. However, I’m getting perfect character consistency in Leonardo AI.

[cinder ingot]

There's a feedback button inside the web on the 3 tick I believe.

[cinder ingot]

Oh sorry, for email I believe there is? But I'm not sure I remember what it is..

I believe it's around here somewhere though, someone might can help.

But even if you asked for help, I believe they can't help you with generate child in video. Since it's not allowed in their policy.

Some people usually got it accidentally, which means the AI can generate it but google don't allow it officially.

[carmine forge]

Thanks for the info!
I’m actually making a short film — it’s just a filmmaking project, not anything inappropriate. I’m using Flow as a creative tool, so it’s a bit frustrating that it doesn’t allow child characters at all.
Do you know how I can contact Flow’s support team directly?

[hidden latch]

I'm getting a knocking sound in audio in every single video clip, i even tried Veo 3 Quality a few times to see and the same knocking... It's not a large knocking but, its distracting

I didn't notice it last night, but this morning when i went to fine tune my Youtube short before uploading.. it's in them all...

[cinder ingot]

You can send it into #1319424187730755636, inside the feedback section. Sometime the google staff will answer there. I try to find the email but I didn't see it anymore. ^^

Anyway, Veo 3.1 fast have been really stable for me. I'm already planning on getting Ultra if this keeps up. Good job, the past week was messy to say the least.

[sharp quest]

My Bigfoot videos look terrible now. What’s the reason? I’m using the exact same prompt, but I’m getting ridiculously bad results...

[burnt oriole]

I tried to make one video with the "Ingredients to Video" feature of flow for the first time and I was so shocked with the outputs. I don't know how it's so good and accurate

Guess I'll have to learn the other Flow features! I love it so much

[lucid iris]

how would you guys promote a website using veo

ill tried so many ways to include the website as an ingredient but over and over the texts are messed up

[fair gust]

Audio not working

[strong quest]

Epic cinematic scene — tourists screaming as a massive python wraps around safari jeep, helicopter rescue overhead, photorealistic jungle background, action-movie cinematography.

[jaunty stratus]

Guys, any prompt expert here about scene builder (extend) function?

🙏

[queen lantern]

Hey all is scene builder supported on mobile?

[burnt oriole]

you should try gemini for flow prompting

i use it and I like the results very much

[exotic shell]

i also use same prompt as in veo 3.0 and receive worse results. worse audio, worse sync and 80% the speaker seems to have a stroke doing weird noises or randomly moving mouth.

need like 50 generations to get one decent result. 😄

[exotic shell]

Does anybody know how to get rid of microphone "feeping"? if a person talks into a microphone, in 90% of the generations theres annoying microphone feeping in the background.

[jaunty stratus]

Do you have good example prompts?

[burnt oriole]

i use a gem

[light plover]

I’m a Gemini Heavy user, but not gonna lie, ChatGPT actually makes better prompts 😂

[warped wind]

yep, gemini it's horrible for this matter...

[light plover]

but here’s the pro tip — just do what yashjit said and use gem lol 😎

[snow lion]

yeah same prompt, very bad result

[royal trench]

hello everyone

Who can teach me how to create an animated video that loops endlessly?

[terse lodge]

Damn, I didn't even know i can put labels on my characters for ingredients to video and the AI would know who I'm talking about. That feature is a godsend; no more describing my character as subject in the studded leather jacket!

[stuck adder]

What you mean?

[terse lodge]

When I upload the reference image, I create a collage of my cast of characters and use it as image reference 1 (when I do this it also seems to have less morphing), you can edit the reference image within the platform and type in the name of your character on the reference image and draw an arrow pointing to them. The AI then seems to know which character you're talking about when you simply refer to the name you tagged for them when you type your prompt.

My second image reference, at least the way I do it, is for non-speaking characters and props, while my third reference image is for the scene's setting.

Now if only Veo learned that there are more camera movements than a dolly in and out shot.

[crisp basalt]

use Google lol

[stuck adder]

Ahhhhhhhh nice

[stuck adder]

Show me what you did with me

😂😂😂😂

What u been creating lately wrig?

[crisp basalt]

i recently got Ultra plan and i been

playin with ingredients to video

[stuck adder]

Oh wow I know you been killing that 0 credit fast

That’s just a game changer…

[stuck adder]

You ever found yourself using text to video ever? Lol

[crisp basalt]

I’m taking a short break from text-to-video for now Veo 3.1’s new prompt adherence kinda throws off my usual prompts, making them generate way different results than expected 😓

Veo 3.1’s prompt adherence is literally ....literal

it’ll follow your words to the letter. But that’s kinda the catch too

ometimes it locks onto a single word and goes off in a totally different direction. I spent quite a lot of time tweaking my prompt, and turns out just removing one or two words fixed the issue i guess

[stuck adder]

Ahhhhh yeah I get what you’re saying

I usually use start to end frame when I want something very specific

[dapper bronze]

Right! However, I don't think it's fair that I pay for a plan and it comes with a watermark, especially since I paid for it and it's not free.

For example, I use the "Google AI Pro" plan for a client, will I be required to show the VEO watermark???

[crisp basalt]

get the ultra plan then

[dapper bronze]

It doesn't work that way. I'd like to understand it better because the guidelines aren't clear. I just want to understand if I'm obligated to keep the watermark or not. For example, can I enlarge the video and make the watermark disappear?

[crisp basalt]

you can do whatever u want with the generated video lol

u dont have to ask permission from.. anyone just to remove the watermark

[dapper bronze]

where is this written friend????

[crisp basalt]

dang really have to feed u everything for u huh

this icon should answer your question

[dapper bronze]

I didn't ask you to answer, friend, you're the one who answered. At no point did I mention your name. I'm being very polite. If you don't want to help, just don't answer, okay! Anyway, thank you very much.

[crisp basalt]

LOL you asked so i answered, and u mentioned me in the reply why else would your message be highlighted for me, lmao.

weirdo

[dapper bronze]

If you answered, please remain polite and friendly to those who have questions. If you don't have patience, please don't answer people.

[crisp basalt]

🪞

have a great day tio weirdo

[dapper bronze]

You too! But be a lighter human being

[crisp basalt]

https://tenor.com/view/laughing-meme-talthazhra-gif-18828742

wait till you find out Veo runs your prompt through Gemini first before generating the video

[icy drum]

What I generate has no match

[brave crag]

Hello, I just thought I would share my latest project with you all. Would really appreciate any feedback - likes / shares please.

“I’M DONE”

This short film is based on a real breakup letter I wrote, reimagined through artificial intelligence, however... there's nothing artificial about it. I used ‪VEO 3.1 ‪‬ to create, animate and storytelling.

My focus was performance... could something 100% generated capture how it really felt. Every frame, gesture, and breath was created by me... I wanted to capture that raw, vulnerable moment when I finally learned to let go of the pain.

https://www.linkedin.com/posts/robpearson2001_im-done-a-breakup-letter-activity-7388714376742174720-Fw1C?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAjBo1QBSzDJ2hKONUrEtYh-JzvE9t5HISg

[versed marsh]

Hi everyone, question.

Is anyone having issues when generates videos using VEO 3.1 Fast, and the results has no voice...?

[vale juniper]

Hello, Is there a way to keep using the veo3.0 model ?

[kind bison]

This is how insane Google's censorship has gotten: tried to make a Scooby-Doo style video in which a character complains about "those meddling kids". Video blocked for policy violation. No minors are seen in the input image and no minors are prompted to appear in the scene...and by "kids" the character means older teens/young 20-somethings anyway. I changed the dialog to "those meddling fools", then it goes through. So sick of this crap!!

[crisp basalt]

So anyway

https://tenor.com/view/sips-tea-gif-10558867

[zenith latch]

Hello!

[torpid rain]

Hi

[dark perch]

[echo escarp]

[subtle heath]

dude why cant i use flow

it keeps saying i dont have access can anyone help

[woven garnet]

huh

[unreal viper]

Hello, as 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. If asking for advice or tips on your prompts do so in the appropriate channels.

[supple python]

Hello, I think in the UK, for frames to video, you couldn't upload realistic characters. Is this still the case?

[crisp basalt]

have u tried it urself

[supple python]

No I wanted to find out if there was still a restriction before subscribing

[crisp basalt]

i mean uploading the image doesnt require subscribe

plus free user gets 100 credit monthl;y for free

[dry frost]

There's other free options via google movies which would still use your local IP to determine

[supple python]

It worked for me on the free versions actually, just looking for someone in the UK who uses it regularly before I drop £125 on the Ultra plan, and then find out it is restricted after all.. I read before the restriction isn't applied based on IP address, because people just VPN past that, but by your credit card instead. So the fact it works on the free versions doesn't prove it for sure, as I didn't enter my crefit card. Anyone in the UK here on the ultra plan and able to upload realistic characters?

[limpid valve]

where can i get link for flow

[misty nova]

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

[limpid valve]

are the credits fixed. how will they be added to my account

[swift merlin]

Anyone have any idea with veo 3.1 will have longer run vid...(like 15 seconds) not including the extend feature. with sora 2 ,15 seconds really opens up a lot more seamless consistent dialogue especially with continuous uninterrupted dialog shots

[noble warren]

hey,
Just a little question. As beginner on flow, someone can help me to deal with this issue :
I created 1 video on flow, then i wanted to link another video of 8 seconds also, to have a final video of 16 seconds. But the sounds on the 1st and 2nd video is not the same... how to keep only one sound ?and how to keep the behind soung (a melody) but to cut the voice of the people speaking ?
Thanks !

[swift merlin]

Yeah that is an example of what I was just asking with Dialog. very frustrating . thats why we need longer shots not including extend,,,so it works like sora

[noble warren]

ok thanks

[exotic shell]

maybe doing the clips without background music first and then adding it later would be a way to go.
but no clue about how to keep other things like persons or speakers voices consistent

[noble warren]

right

[cinder ingot]

Is flow down rn?

Can't make any new videos, thought it's my prompt problem.

[neon aspen]

do someone have to pay before they can remove the veo watermark?

[turbid lintel]

This is veo issue since beginning , the only workflow to fix this is using capcut. Extract the audio from all your 8 second clip and change the subject voice .

[wraith sequoia]

how do i get my API key idk how to access Dev mode

[stuck adder]

why do people want veo 2? just asking

[near gate]

so they can use the free version? Idk, I wouldn't go back.

[stone summit]

Hi. I just did the Ultra Plan, there is no 1080p option in https://labs.google/fx/tools/flow/, can that be?

[lilac quest]

Hey I make the coolest ai cartoons

https://youtu.be/Uot58Qsopwk?si=7lC8hJjOB-7ZfO0E

Made this with google flow few months ago … The series is about a kid getting put in a situation to do credit card fraud and a lot of things go bad for him ..ai animation

[stuck adder]

can veo do transparent images?

nevermind it saves in mp4 so no

[hidden latch]

I've been using flow for about 3 months, sound fluctuates, however, I've found if you prompt the voice for qualities, ie, deep, or high Pitched, ect. It tends to be close

[harsh cedar]

Hi everyone, question.

Hello, Is there a way to keep using the

[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. Thank You!

[dusky notch]

Where do we generate?

[still sentinel]

I am here wondering what has happened to the quality. I have been able to generate, but all the mouths are pixelated. Hoping its going to change soon. Have you been able to make new videos yet ?

[still sentinel]

This has changed. 🙂

[supple python]

It’s now allowed? I didn’t see it announced. But that’s great news. Thanks for confirming!!

[eternal bronze]

hi everyone...stuck at 99%

is this still going on?

[proud vapor]

Flow creates really good videos, but there's one big problem that really bothers me, and that's that it doesn't record words correctly in the videos, with lots of spelling errors, and even if I quote what it says, it still doesn't work every time I try.

Does anyone know how to overcome this?

[proud vapor]

ho and its stuck at 99% again

[wind saffron]

an ongoing problem for every ai vid gen is spelling...u just have to keep trying over and over

[tranquil canyon]

Anyone help me 🥲

[tidal anvil]

It’s a common AI problem. As surprising as it sounds, those model are very bad with words. Not sure if there are workarounds as of now

[tranquil canyon]

Veo 3 one year ultra plan price ?

[exotic shell]

Is there any trick to:

• avoid speaking people having a stroke in 95% of the generations
• microphone "feeping" (if somebody talks into a microphone)

[stuck adder]

Oh yeah it’s like it’s talking in Russian or one Forgien alien language sometimes

But when it works it’s very good

[stuck adder]

Best to just create photo transitions

[tranquil canyon]

Bro veo 3 ultra price?

[uncut geyser]

$249.99 / month

[tranquil canyon]

Can we buy credit after exhausting 25000 credit?

[potent swan]

yes

[tranquil canyon]

Credits price?

[potent swan]

US$ 25 → 2,500 AI credits
US$ 50 → 5,000 AI credits
US$ 200 → 20,000 AI credits

[tranquil canyon]

But my friend gives me only 100usd - 45000 ai credits

[round dove]

Its cheaper to sub to Gemini Pro: 1000 credits = 50 fast videos, plus 3 per day in the app = 90 fast videos, total 140 videos compared to 125 videos if you get the $25 extra credits.

[tranquil canyon]

200$ - 90,000 ai credits

[opaque sand]

Why? Create a new account and pay $150/month again for 3 months.

[tranquil canyon]

Credits?

[opaque sand]

Ultra subscribers can use Veo 3 Fast without spending their bonuses.

Compared to unlimited plans, it still doesn't make sense.

[tranquil canyon]

Lol

[sacred kettle]

is it me or now veo 3.1 cannot nro simple nor complex prompts where you super explicitly specify "subject A says balblabla , then subject B says xxxx, then subject A ends with lolololo". and all of a sudden, what worked ALWAYS IN VEO 3, now in veo 3.1 is just ....sora 2 like. you dont know who is gonna say what. the dialogue is randomly apply to the characters in image to video. i got 40 failed generations , tested it on labs, higsfield and minimax for the api call. improved the prompt with gemni 2.5 pro deeo thinking, improved the prompt with the txt to video expander and then used that as a guide for the img to vid. yet, no matter, the dialgoue is just....random. anyone else experienced this? workarounds, solutions?

[devout gate]

where to find my previous generated videos?

[stuck adder]

125 discount

[hazy acorn]

Anybody having issues with Veo 3.1 Frames to Video - all of a sudden all my generations are failing and I'm doing pretty basic stuff (no copyright or questionable content)

[glacial panther]

Same here. It also seems to effect some (but not all) source images in "ingredients" mode.

[hazy acorn]

Veo 2 with the same prompt works fine (but the output is far inferior of course)

Of course, but it isn't working for me right now 😥 - is frame to video working for you?

[glacial panther]

Sounds like someone adjusted the safety filter overnight.

[hazy acorn]

My prompt and image are sooooooo vanilla - see my post in flow-bugs-feedback

[kind bison]

Yep! Trying to re-run a dialog scene with an image that I have already run a dozen times - now I keep getting "policy violation" errors. There are zero policy violations!

[kind bison]

Yeah, I have been running into that a lot in 3.1.

[sacred kettle]

thanks for the confirmation grind. at least if we are two, that could be a potential sign. bcause really... you now i kept workng...and pushing....and well....i made the most insanely revised prompt, obvously after so many iterations, making the safest picture ever (as it the one i already had wasnt... IT WAS AN OFFICE WITH A WELCOME FRIENDLY ENVIRONMENT) . so...still generations are a little randome and dialogue goes to whoever veo decides. you could mitigate it, but the randomeness will be there. i googled it, and it says that 3 characterds is already probematic....ok . i didnt tough THAT WAS TOO MUCH. and a characters is mute on my scene.... the talking between the other tweo? 2 lines. one talks. only one. but the randomness spreads that to the 3 characters. and the Jon line says: "blabala"... well... iwas even more spcific than that, everytime. nothjng. randomeness always. you know what i had to do, give up, because the effort to get there and then get a little bit of randomenss that really kills my story, is not worth my time nor my money. i went with a really bad geeneration from kling, which even in 2.1 master is bad, i tried 2.1 pro, 2.5, hailuo 2.3 and wan 2.5 . the chinese mannersim never were a problem. now all of a sudden to salute a person, i find a strange thing we only do here on the western side on the military or in medielva times.... so.... bad gemerations with kling and then lipsincin is time consuming, but is less consuming than this thing with veo 3.1 . see if you can get past thru it. i thought that 3 characters, quite differen tin appareance was ok, wiht one only delivering 2 lines. nope. this didnt happen on veo 3. there was an update, but they brought a lot of nasty bugs like a ny software. just n case you wanna save some time. or perhaps if you find somethng better,,,, please share. thanks

[crisp basalt]

https://tenor.com/view/jimmy-butler-confused-jimmy-butler-confused-eye-squint-gif-12144635464811342257

its too early for this let me just ask gemini to summarize whatever u just said

well

what

[peak lark]

Hi! again there is an update of proeminent people? I get erors now because of that. before was working very well. Can anyone confirm this?

[crisp basalt]

@hazy acorn So , i tried using your image in Opal just to see if it manage to generate without any issues or not, and apparently

So yeah its the prominent people issues all over again

the image in question

[peak lark]

omg thats not good.. before was so good. Now its bad again. There are not even proeminent people.. even with basic clearly AI character is giving the error

[crisp basalt]

Yeah its.. weird, also this prominent error seems to go accross the API too apparently?

    "errorMessage": "{\"code\": \"INVALID_ARGUMENT\", \"message\": \"No videos generated: Veo could not generate videos because the input image violates Vertex AI's usage guidelines. If you think this was an error, send feedback. Support codes: 15236754.\", \"details\": \"Veo could not generate videos because the input image violates Vertex AI's usage guidelines. If you think this was an error, send feedback. Support codes: 15236754\"}"
}```

[peak lark]

yes. its everywhere..

this is not good. 🙁 How we can talk to someone to take a look and fix this because its horrible. I had clients for this and now I have to use Wan2.5 which is good but not like veo3.1 . @main saddle ? can your team check this new update and fix it?

[crisp basalt]

not even gemini veo 3.1 can

[peak lark]

they better fix it because its very important for so many people. .

[sacred kettle]

k

[sacred kettle]

i still have some but less than 1 week before. i know they did make an statement that they put a fix on place for this specifically leaving a little bit of leniance towards face detection

[peak lark]

this needs to be fixed asap..

[main saddle]

hey all - thanks for reporting here and in #1319424187730755636 - we've deployed a fix: #flow-announcements・📣 message

[crisp basalt]

thanks for keeping us updated @main saddle !

[main saddle]

thanks for reporting!!

[peak lark]

thank you @main saddle !

[upper cosmos]

Back to Baghdad (PS5, Xbox Series X/S, Nintendo Switch 2 and PC) coming on 2026

[hazy acorn]

Hey I was afk - thanks for digging in! yeah that's just some random ai generated image of a guy - not a prominent person whatsoever

[near robin]

Is there a flow app or anything ?

[native barn]

LOL VEO 3 your update is getting worst prompt is not accurate now!! hope sora will have unlimited soon, so i can switch and stop my subscription on your non sense app

[manic bone]

hello sombody can help me

I have Gemini through the university plan and it said it included unlimited flow, but when I try to create something it says I don't have enough credits.

@pastel surge

[meager edge]

does anyone has tried new fix inside google flow for F2V Generation failure

[cunning hedge]

"A cinematic video starting from the man's full-frame shot, slowly panning up to show the towering, layered rock walls and dense jungle canopy above him. The water should have subtle ripples, and the leaves should be swaying gently. Style: Adventure, high-saturation, 4K

[manic bone]

I have Gemini through the university plan and it said it included unlimited flow, but when I try to create something it says I don't have enough credits.

[bright patrol]

The day Veo3 solves the issue of character consistency, I would be very happy.

For now, it's still not as good

[burnt oriole]

is it me or does ingredients to video not work for the 3.1 quality model

[burnt oriole]

unlimited what?

Veo 3.1 quality? I don't think they ever gave that to any plan

oh yes that's what I was talking about

it's limited

only fast model is unlimited for ultra subs ig

niceee

nope I have some credits for my generations

but ingredients to video doesn't support the quality model

which is what I was sad about haha

[tranquil canyon]

Support buddy 🌚

[lapis pine]

Veo not working again, great

what are we paying for

[crisp basalt]

Your m

[tight kite]

Do y'all know what type of content can i do with AI to grow followers faster?

[tight kite]

On tiktok

[eternal bronze]

Cna someone tell me what use cases veo2 is better for?

[heavy tinsel]

Google is the only company I know that will censor and false flag the hell out of adult creators, while leaving their watermark on your creations yet still charge you more than premium. They used to be a good reputable company until the bait and switch on creators and film makers. So glad China is catching up.

[heavy tinsel]

Creating a bugsbunny style cartoon with bad physics and morphing issues.

[eternal bronze]

im having issues "jumping to" a complete camera positiuon change in the scene builder...can someone help?

[heavy tinsel]

Yes. anything you need give me specific details on you vision and ill turn you into an AI Jedi in no time!

[eternal bronze]

oh ok

so im using the scene builder to create different scenes of the same subject in a gym setting

he is performing different exercises, so i need charachter consistency as well as lighting mood etc...so far so good with the character ( more or less. can solve with editing) but i keep prompting the camera to be in a specific place( camera overhead. Shot from above with an 85mm lens) and i keep getting a profile

profile is the previous scene, so thats why im using "jump to" instead of extend

[crisp basalt]

do u have clear image of the character

[heavy tinsel]

You have to prioritize the camera position i. Your shots by anchoring it at the beginning, middle and end of your prompt in details. See how that works. Camera angle is not a veo 2 strong suit. Let me know.

[crisp basalt]

if u do, then u can actually remove the image in jump to section

and replace it with another image of your character

[eternal bronze]

dont understand this

[crisp basalt]

wait are u trying to jump to into a new scene or what

[eternal bronze]

ok...will try this

[crisp basalt]

jump to usage is for when you want to use your character in A NEW SCENE

[eternal bronze]

yes...comp,letely new senec

[crisp basalt]

and u said the jump to feature uses a .. profile side view of your character, right?

im saying this based on my understanding of what you wrote btw

[heavy tinsel]

You should give Wan and Hailuo and especially seedance a shot. They ste all masters with cam angles and great shots.

[crisp basalt]

so anyway

[fading inlet]

guys just want yalls opinion so i generated this add kinda thing with veo 3.1 if you saw this while scrolling would it grab your atention

[eternal bronze]

yes...but i should create a good first frame for thios right? Will they keepo mood and lighting consistency between shots?

[heavy tinsel]

Seeing your face grabbed mines

[eternal bronze]

yes...it obviously uses the previous frame to start, bt as AI TONE said, not getting the right camera placement probably has to do with not anchoring camera placement at the begining of my prompt

[fading inlet]

you can run it localy on comfy ui(oyu need like a not insane but decely powerfull gpu) and yeah you can do a lot of tweeks to make it do some insane consistance wan 2.2 btw

[heavy tinsel]

@eternal bronze of course, as long as you anchor the lighting conditions and everything with your prompts.

[crisp basalt]

just change the jump to frame to a clearer image of ur character thats all

and then do some small trimming and done

[eternal bronze]

ok...makes sense..thank you

[crisp basalt]

no problem

[eternal bronze]

even if its done in a different "snebuilder"

love this discord 🙂

[crisp basalt]

ok

[eternal bronze]

meanning i can just open a new project to get the scene that i want using a good frame of my charachter as starting poiunt

[crisp basalt]

no

did u not watch the video carefully

did u see me start a new project

u can just get the good frame of ur character from other generated video in the project that youre doing , just do "save as asset"

[eternal bronze]

yes..i dint express correctly i guess

[crisp basalt]

you really did

i dont blame u for that, english hard as hell

also

[eternal bronze]

i see what you did and I fully understand the process 🙂

[crisp basalt]

you can create a "character sheet" of your character that you hav using Flow image editing

i actually used nano banana for that character sheet

let me record video again oops

nvm ill do screenshot

so take this frame . save as asset

[eternal bronze]

yes i understand

[crisp basalt]

once it saved, check your assets gallery/images asset, hover ur mouse and click edit button

[eternal bronze]

this is helpful

[crisp basalt]

ok first attempt was not what i wanted but we can iterate

[eternal bronze]

https://tenor.com/view/big-daddy-hellyes-gif-26450264

[crisp basalt]

much better

LMAO even tho no design at all but hey it stays true to the appearance of your character, as for my case mine was ,,,,, silhoutte... uhhh shadow man from the corner of your eyes

and then save as new ingredients , and boom youre done ✅

[eternal bronze]

perferct...rellay had issues with camera placement but now Im gt this also sorted

awesome

[crisp basalt]

im glad!

[eternal bronze]

https://tenor.com/view/oh-yeah-gif-23378792

[crisp basalt]

You can also create a brand-new start frame scene,, like maybe you dont want to waste your credits on the Ingredients-to-Video feature, scared that the generated video would not look what you had in mind.
Instead, just use Flow’s “Generate Image” feature to craft a fresh start frame scene that already includes your character.

click this,

[eternal bronze]

yep

[crisp basalt]

and then upload/select image of your character

[eternal bronze]

yeah...ingresient doesnt support 9:16 does it?

[crisp basalt]

hmmm i forgot \

and boom, you got a starter frame of your character in a scene you imagined.

[quick rapids]

wish the 99% stuck stayed fixed, was a whole week of flow working properly

[rough halo]

just more and more nonsensical content policy violations lately

[eternal bronze]

yeah...ive been trying to upload some halloween characters and i get flagged coinstntly

not cool google....after all, you are aware of everything i do online and then some

give me a break

[misty nova]

In my opinion while I understand yes hur-dur it’s a corporation, once competition releases better models that are just as and or better than veo alot of people are just going to stop paying for veo at this point because of how sensitive the filter is

I have the ultra plan i personally wonder how many of those people are going to stick around once better options show up with more flexibility and less egregious censorship especially for how much they demand for the usage of the platform and tool

[quick rapids]

waiting 5-10 minutes to be able to queue another generation while this crap happens is such a good ux

[misty nova]

It doesn’t seem like what their doing going to be very viable unless they’re entire sales strategy is just at getting enterprise to be financial whales for them and having the consumers just as a afterthought

[clever axle]

Flow is not working? It is complain about any prompt

[meager edge]

tried with will smith but still violates the guidlines I thought it can work

new fix Frame to video

[errant grove]

any work arounds/solution to the content violation error? it flags litrally anything atp

[digital yoke]

Swedish language need more love

[hazy acorn]

Yeah I have a bug open on flow-bugs-feedback since yesterday. Still can't generate any frame to video without getting flagged for "This generation might violate our policies. Please try a different prompt or send feedback" - The text in the prompt doesn't matter, it is the image. For some reason it suddenly stopped liking the image provided even though it is an image from the end frame of a previously generated Veo 3.1 video 🙃

[errant grove]

it's frustrating man

<@&1173646484311384114>

[main saddle]

hi everyone, we issued a fix for this late yesterday (see here #flow-announcements・📣 message) but helpful to see that this issue persists for some. will connect with the team for an update. thanks all

[errant grove]

Awesome, thanks

Ingredients is such a cool feature it sucks not being able to use it haha

[quick rapids]

does anyone know the actual hard limits for ultra and veo 3.1 fast? it's "unlimited" but there's always gonna be rate limits built in to keep individual users from blowing the cost margin for everyone. I'm guessing RPM and RPD are separate limits, one of the google guys (ian) mentioned a 24 RPM limit in a bug thread earlier. what's the next level of throttling after 24 requests per minute?

[zealous elk]

Hello everyone,please while creating the same characters across different scenes,how can I make the characters consistent?

[crisp basalt]

@zealous elk use jump to feature

[zealous elk]

Sorry,but I don't really understand...please just break it down

[crisp basalt]

watch the video lol

just follow what i did in the video,

i wont go through writiing down steps by step until i actually see whats the actual issues youre facing

@zealous elk did u watch the video

Its a bit frustrating when I explain things step by step and people still say they don’t understand 😅
So now, I just focus on figuring out exactly what part they’re having trouble with makes it easier for me to understand and help properly,,

[zealous elk]

My main problem is,I want a character to appear in about different scenes(about 5 scenes) but the character end up being different characters. I want the character to be consistent across all scenes.

[crisp basalt]

and did you use scenebuilder @zealous elk

to generate those 5 scenes

[zealous elk]

Nope

[crisp basalt]

so you didnt use that, did you use ingredients to video?

[zealous elk]

I just prompted each scenes.

[crisp basalt]

yeah that wont work buddy

[zealous elk]

Nope

But you understand my problem now right?

[crisp basalt]

If you want to keep your character consistent, you’ll need to use either Ingredients to Video or SceneBuilder.

yeah i understand and you did it wrong

[zealous elk]

Okay...

[crisp basalt]

aight dm me the image of your character

or anything so i can try go through with you

[wooden crow]

I had remarkably good (not perfect, if that's what you need) results by beginning every single prompt with the same description as a header for the prompts. I had Gemini format the prompts. (I haven't tried it, but have you tried using a reference picture as part of the prompt in one of the ways Flow permits?)

[scarlet mica]

When using a photorealistic female character in ITV, above a certain age it creates a policy violation message, no matter what the context or scenario. If I use a different younger character it’s ok. I’ve had to stop a whole storyline in a project, because characters that previously worked are now being blocked. It does seem to be because they are older and female, or their clothing lacks, let’s say, fantasy details or embellishments. It could be I haven’t tested this enough but the way it stands I cant’t continue with a project I’ve been working on for months now. I don’t know if I develop a story, and then have to abandon it because characters I’ve created are then blocked.

[scarlet mica]

Just a note, I think I’ve found the issue. It appears the filter will not tolerate a closeup macro shot as part of the triptych for older woman. Which is clearly ridiculous but what we’ve come to expect.

[gilded shell]

How do I get free flow?

[pallid violet]

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.

[keen token]

Hey everyone!

I'm working on a comedy-drama short film in Veo, and I'm hitting a wall with character consistency whenever I use the Jump To function (switching angles/making a hard cut).

The issue is this: I'll generate a beautiful Top-Down Wide Shot with my custom-designed robots (the blue/white head and the Assembler bot, visually defined by my prompt and image references). But when I use Jump To to switch to a Medium Shot of the Assembler, the robots' appearance completely mutates. They lose the specific textures, colours, and shapes from the original frame.

It seems like Veo drops the visual context of the Source Frame and just uses a generic interpretation of "robot" from the text prompt for the new shot.

My Question: Does anyone have a secret phrase, technique, or best practice for forcing Veo to maintain the absolute visual identity of a key object (like the protagonist's head) when moving from a wide shot to a close-up via a Jump Cut?

How do you guys ensure your characters look the same across different angles? Any advice on "locking in" the visual context would be hugely appreciated! Thanks!

Would you like to prepare the next scene description now, or do you have any other requests?

[odd cape]

https://youtube.com/shorts/jv8OhsMEcuI?si=WZT7olE2PELmTJRy

[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. Thank you.

[twin inlet]

ahh its a shame that we can't use veo 2 ingedients to video anymore :/ its only veo 3.1 fast now, its not giving the same results

[twin inlet]

please bring back veo 2 quality ingredients to video , not sure why you guys are retiring models customers are still using for their projects

[crisp basalt]

veo 2 model still works for me for ingredients to video

[twin inlet]

what region are you?

This iswhat i get when i try to do veo 2 quality ingredients to video, are you using veo 2 quality or fast?

[crisp basalt]

oh its not avaialble fro me too

the status says model deprecated . oh well rip

[twin inlet]

very rip, not sure why they retired it without a comparable or better model for ingredients to video

[stuck adder]

Isn’t 3.1 better ?

[twin inlet]

If we could use 3.1 quality.. but we can only use 3.1 fast, so all the output is pretty inconsistent, can barely get anything good out of it with ingredient to video

[night badge]

hi everybody

[eternal bronze]

hey frems...so upsacaling to 1080p on Veo is a no no?

my skins now look like shhhh..

[twin inlet]

be great to get an update from the team on all the above

[soft kettle]

In the grimy restroom of a roadside diner at midnight, a burly man in a tshirt is talking to himself in the mirror above a chipped sink. His voice cracks quietly to nervously: You don’t need the mask anymore. Spread love.. handheld 35mm camera and a 50mm lens, creating a sense of intimacy and candidness. Low contrast. ( Whats the issue with my prompt ? )

[hazy acorn]

Yes, upscaling creates weird artifacting.

I'm not finding that 3.1 Quality is that much better than 3.1 Fast - am I missing something?

[kind bison]

I wish that Google would just partner with Topaz and integrate Astra as their upscaler. If Adobe can do it, why not Google?

[rare shard]

So i opened the free flow account Oct 31st.. I was hoping the credits would re-boot since it is November 1st... but I am not seeing that. So does that mean - its a whole month cycle until the credits re-load?

[ancient blaze]

Just click on the that credit, you will redirect to a page, that tell you, when the monthly credits will reset. It might be on November 30th.

[kind bison]

The stupid "might violate our policies" bug is acting up again - totally random, too - a prompt with images (which are totally safe, by the way!!) will work one generation, then fail on a re-roll.

[wispy forge]

@keen token you could try to generate an image of the final jump cut using an image generator,then maybe use it as end frame. I am facing similar kind of problems, the first frame,end frame feature generally never cuts through frame it generally just slowly morphs the first frame to the last fame, I hope there was some way we could provide references images to generator regarding the character other than the images we need to animate

[ruby cypress]

8-second cinematic realistic video showing the full process of online shopping for “Asankharid”.
Scene 1: Close-up of a man using his smartphone to order a product on the Asankharid app.
Scene 2: Quick cut to a warehouse where a worker packs the item in a brown box with the Asankharid logo.
Scene 3: A Tipax delivery van drives through the city streets.
Scene 4: The courier delivers the package at the customer’s door; the customer smiles and opens it.
Realistic lighting, smooth transitions, commercial style, shallow depth of field, 4K quality, no text on screen.

[pseudo orbit]

is this real <@&1366831844502274180>

[pallid violet]

It's not, thank you for reporting!

[full flicker]

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.

[oblique condor]

How to generate videos in 1080p?

[light plover]

After generating the videos, you can select 1080p. It only supports landscape mode in Flow.

[unreal breach]

I'm no expert, but you should post this is the Flow bugs feedback channel #1319424187730755636
. I've had to completely change my workflow to another platform because I'm getting so many conetent moderation triggers inexplicably. But I don't know that Mods are monitoring this channel -- but hopefully the bugs channel.

[oblique condor]

Where? I've been actively using it for two weeks now and I can't figure it out.

[ripe yacht]

Hello, as 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. If asking for advice or tips on your prompts do so in the appropriate channels.

[light plover]

Is it different in other countries? In my case, there’s a dropdown on the download button.

[ancient blaze]

That feature actually exists.

[burnt stone]

Doesn't Flow support camera controls anymore? now all the models Veo2 or 3 have no option for that. Is it deprecated? And seems like the models don't follow the camera movement prompts either.

[crisp basalt]

yea its deprecated

[burnt stone]

Why? what's the thinking behind that? now all the clips I generated remained in static shot, even with explicit camera control prompts, like orbit, dolly in, way to go!!

[crisp basalt]

idk , imo its useless and only useful when it comes to image to video, plus it doesnt do audio as it uses veo 2 model, as for your issues, 3 words : Prompt Skill Issues

imma just drop it here and not gonna say furthermore, its free to read ,

https://medium.com/google-cloud/veo-3-a-detailed-prompting-guide-867985b46018

https://cloud.google.com/blog/products/ai-machine-learning/ultimate-prompting-guide-for-veo-3-1

[burnt stone]

Thanks, love you, I'll check it out

[crisp basalt]

If you wanna drop your prompt and the video result here, go for it the peeps here can help you refine it so you won’t run into the same issues again. (Plus, it helps the Flow team spot what’s going on too since they kinda lowkey lurking and reading our messages 👀) @burnt stone

[burnt stone]

Thanks for your tips, I'm OK with it for now, but I'm trying to guess it's Flow's policy to push us to use more Veo 3 instead of staying at version 2, since my prompts work pretty well on 3.1, no solid evidence but the results said it.

[wooden crow]

When you click to download, THEN it asks if you want to upscale to 1080. It will notify you when the upscale is complete, and the video is ready to download. In other words upscaling to 1080 is part of the download process and generation is only at the lower resolution.

[unreal viper]

Reminder we cannot generate images in the server! Please do not use non English languages as we cannot moderate these and messages and videos will be removed! thank you!

[cyan cypress]

can some one tel why i was charged with 100 credit for video genartion?

[ancient blaze]

Hi buddy, there are 2 types of models: "Fast" & "Quality".
The Quality model costs 100 credits for one generation video. So check your preferences or settings before starting to generate videos.

[shadow warren]

yeah or you end up like me and accidentally leave it on quality an go wild....my whole Ultra month gone in like half an hour or so
thankfully fast is free, so i can still use that 😅

[crisp basalt]

how did u not check the info shown on the model settings tho

[shadow warren]

I mean you have to click on settings and it has a small text in the popup. its not like its in your face or visible if you dont click on settings to start with

[crisp basalt]

i mean , did u not know theres other veo model thats available to choose?

at that time i mea

[shadow warren]

I did know, i just didn't know that i switched it to quality at some point

[crisp basalt]

ah thats fair

no cuz

nevermind , human makes mistake idk where i was going with that

[shadow warren]

todays struggles against the filter make my decision easy if i want to add another month of ultra or not though

[shadow warren]

Next time they ask for the monthly sub payment, i send them this:

[near gate]

sora 2 is greater than veo 3 and has longer length and doesnt cost a bunch of money

[crisp basalt]

I mean yeah go ahead , nobody really cares tbh

[shadow warren]

as much as i cared about your response tbh

[crisp basalt]

what

[shadow warren]

cant figure out DMs, so i just write here

I think i misunderstood you. Thought it was a snappy kind of like "Yeah, theres the door, dont let it hit you on the way out", but now i believe you mean that google wouldn't really care, to which i can only agree

[analog lava]

Hello @digital pagoda , a friendly reminder we cannot generate images or videos in the server! If you are looking for help or guidance please visit the #server-directory・☎️ and use the appropriate channel to do so. Thank you!

[light plover]

Sora2 isn’t sustainable. It’s great for users now, sure — but the gradual move toward monetization is obvious.

I’m just enjoying it. For now, consumers are the winners.

[proper drum]

Is anyone else having issues where they can’t generate? I’m an Ultra subscriber (Frames to video)

[real gazelle]

Hello everyone,

I’m facing a small issue whenever I create a video on Flow, my character keeps changing automatically. Every new video generates with a different character, even though I want to keep the same one.

Can anyone please guide me on how to fix this or maintain the same character across all videos?

Thanks in advance!

[crisp basalt]

Use ingredients to video

[real gazelle]

Thanks its done

[dark badger]

hi

[idle eagle]

Hello, everyone. I'm using Flow for the first time, and I'm really enjoying it! I have a question. If you choose to generate four outputs per prompt, will it cost four times as many AI credits to produce the generation?

Oh don't worry I found out myself lol

[near gate]

does anyone know how to change veo 3.1 time length to 4sec

[mental apex]

Hello everyone, I run a page with over 600,000 followers across Instagram and Tiktok. Making videos with Flow. 200 million plus views in 4 months. I am also an Ultra subscriber.

I am running into an issue with Flow, it keeps muting all my videos right now, started yesterday. All the dialogue and sounds are muted in every video.

[mental apex]

Did they release some anti curse word filter?? That would ruin it all

@main saddle

[winged nova]

Am I allowed to upload on YouTube material created in flow and assembled to another art work on video editor from pieces? Who knows this, seems no one can summarize these. 🤷‍♂️

Or is this like asking "who owns this 10 acres of land" in 1800s USA?

[proper drum]

<@&1173646484311384114>

I am an Ultra plan subscriber. For some strange reason, when the history is disabled in the “Google Labs” library, it does not allow me to start video generations using “Frames to video.” Enabling the history allows me to use “Frames to video,” but this didn’t happen before.

I have also noticed that the “Frames to video” generations no longer have consistency in the character. For example, in the initial frame the character was not included, but in the final one it was. A few days ago, the character consistency in the video was good, but now it generates a different character, and only in the last frames does the character start to appear.

[shadow warren]

battling the filters when you dont even prompt anything NSFW is so damn annoying

[plush stratus]

Anyone got good suggestions on how to come up with funny ideas consistently? Also When I write my promts sometimes it doesn't follow it like the voices will switch with who is suppose to be talking (I want to use these for youtube shorts)

[mental apex]

Now they added filters for curse words so annoying, ruining flow and veo

[shadow warren]

Since when? Yesterday i had several, but i also notice a large amount of generations being muted, which might mean that its not a 100% accurate filter

its really stupid. i wish americans would learn that curse words and nudity do not hurt you (and i say americans, because that has a big influence on media and the mindset of curse words and nudity is originating from there in media. In most european countries (germany, france) we don't have issues with nudity on tv)

[mental apex]

Since last night, I noticed all my videos got muted and I’ve made over a hundred. Then when I removed a curse word it worked
It sucks because I own one of the biggest AI influencer pages in the world and use VEO 3 primarily, the curse words are the best for shock factor and make the videos super funny, everyone loves it.
Now that they banned the curse words, everything will be boring. I also know other big marketing companies that use curse words in their ads and they won’t be able to either. PJ Ace etc.
so no more shock or funny factor of a grandma saying the F word etc.
Pretty dumb to limit curse words in my opinion and a big loss for veo.
Hope they realize and fix it

[shadow warren]

Yeah, i mean they want to promote creativity and then bring out the castration devices. and that in the year 2025, where that stuff is everywhere in media, either for authenticity or comedic value

but now that i think about it, we are talking about the same community that demonitizes videos if you curse in the beginning of a video

[dusky wraith]

Not sure if this helps you. But I just started an animated series. And all of my voices for the characters are actually recorded by me and my friends. So when it comes to cuss words, I don’t have to worry about it. I still use the audio from Veo so that I can get an idea for how to pace myself when recording voice overs. If you are good with editing, then you should be able to layer the VO with the video

[kind bison]

The censorship has to end. I can't believe that in 2025, Chinese video models are relatively uncensored compared to American companies' models - it is absolutely insane!

[drifting burrow]

Question. Downloaded upscaled videos have substantial compression artifacts. Any ideas on how to improve their quality?

[fierce nimbus]

Hey @mental apex definitely not intended. Can you DM me the prompts used?

[muted matrix]

So now, they just announced that there is a new Camera adjustment feature for Veo 3.1 which sounds really awesome. However, it's only for Ultra Users. Is it going to be available to everyone down the track?

[potent swan]

Flow is active right now?

[kind bison]

The camera adjustment feature is not good. The changes in camera position are either subtle or non-existent and the image quality noticeably degrades - everything looks way more plastic and artificial. Also, lip movements get totally messed up - looks awful.

[potent swan]

https://labs.google/fx/tools/flow
this link is correct?
i can't create video here
anyone can guide me?

[kind bison]

The camera adjustment feature should offer interesting angles, not just slightly left/right, up/down, or further/closer. Give us birdseye view, reverse angle over the shoulder shots, low angle looking up, Dutch angle, et al.

And I have not been able to get "further" to even work - it just repeats the same framing from the original clip.

[sonic loom]

@ruby lily @sly kite Does Google Flow removed the 0 credits feature in - Fast mode under the Ultra plan?

[alpine oar]

Hi, is anyone else having trouble exporting their video from the Scenebuilder? It keeps telling me export failed - I need to get this video exported for postprocessing for a client

[stiff harness]

Uhm, I have ultra, havent generated anything this moths, what the hell??

[sand portal]

The "Further" one is in fact move further+ zoom in. You should see background getting bigger.

[stiff harness]

DOes google have actual support that responds?

[sour stirrup]

Happened once

[stiff harness]

Did you get it fixed?

[sour stirrup]

I had less work so no

[stiff harness]

Got it fixed via support

[real gazelle]

I have my ultra account but just open and see this what happend ?

[kind bison]

Yeah, eventually got that effect - but on a bunch of tries it did nothing (or the difference was too subtle to really notice a change).

[dark badger]

hi

[dark badger]

what happened

[real gazelle]

Actuallly it shows just 60 AI credits perhaps it was 45000 credits

[dark badger]

Mine is back to 1,000 after a month.

[copper olive]

Previously, flow return the credit used when the video creation failed. But now, that no longer happen. Is it the same with everyone else?

[crisp basalt]

I think my ultra got revoked

[shadow warren]

Mine still says 0 credit cost

[stiff harness]

Quick, make a full length movie

[stiff harness]

Had the same today, maybe your payment failed? I had to retry payment on play.google.com

[tough prawn]

Veo creation

[stiff harness]

#flow-videos・📹

[tough prawn]

Sorry

[light cedar]

the images are not generating anyone else having this problem?

[steep summit]

how to generate video

[burnt oriole]

create a project and type the prompt?

[burnt oriole]

I checked and it's working fine for me

[gaunt crypt]

Hi, I'm on the Google Ultra Plan but can't even generate a video with Santa Claus because he's a prominent person. What a steaming pile?

[analog radish]

Guys we need better quality, I love the new camera control, but we must be able to upscale the videos.

[burnt oriole]

I think you can upscale the video while downloading

[analog radish]

I am speaking about a boost better than 1920x1080

[meager edge]

is there anyway possible to use celebrity faces in VEO 3.1

[burnt oriole]

I wanna make some cooking videos with Flow

Does anyone have any ideas to share?

I wanna utilize my remaining credits

[turbid lintel]

So how new camera control ?

[kind bison]

Not on my end.

[robust wedge]

Does anyone else have issues accessing Flow? I got the subscription because it was available in my country and now I just get a message that it's not available for me

[torn tulip]

Hey guys, I am an Ultra Subscriber, but I don't have that new "Camera Control" feature, is there any way to enable it?

[proper drum]

They are in "edit" on the generated videos

[torn tulip]

Thanks! I found it 👍

[shadow warren]

Flow kind of can't count...i write "two aliens" it constantly makes three

i would even say consistently so

[proper drum]

in my case i have in prompt jump (dont show his legs) and it shows his legs 🫠

[shadow warren]

yeah, that might be a leftover from the early img gen days where mentioning something makes it appear extra reliantly, but in my case i want two, mention two, but get three ^^

[proper drum]

In my case, a piece of paper tears and the character pops out, but it stays behind the paper; only its head and arms show through the hole. It does it well, but when it jumps, it shows its legs even though I tell it not to.

[shadow warren]

i also notice that if it adds fantastical stuff (or sci fi in my case), then the overall quality goes down rapidly

the woman, which would be well generated usually, is suddenly having weird eyes and artifacts

[vivid hazel]

Can I create a 7 minutes movie with Flow or do you think it’s still can’t do this much work

[random jay]

can anyone tell me why when i try to generate a prompt in flow and i describe the jewelry i want her to wear it fails to generate?

[tardy ridge]

finally spending time with 3.1, text to video image quality took a hit, it's overprocessed now

where 3.0 looked alot more natural, it's alsmost as if they added an upscale pass after the gen to increase the amount of detail in the shot but now it looks much more artificial, bummer

first 1 3.1, second 3.0

[tardy ridge]

maybe it sjus tthis particulat shot

[maiden mantle]

Hi everyone, we have a question. What is the best way to stabilise the sound jn different scenes? The sound is not always consistent jn different scenes. We now download and edit in capcut but maybe it is better to use scenebuilder? Anyone got some tips?

[pallid violet]

@tawny osprey 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.

[pulsar ocean]

Hello, can someone explain to me like I'm 5 the pricing for veo 3 and how to use it ?
I need to use flow right, then its 125usd for 3 months, then 250 ?
And I get around 700 videos monthly ?
Thank you 🙂

[supple python]

Hi I've started using scene builder to extend clips. I notice each extended clip repeats the last second of the previous one. When I download the full scene, that duplication is corrected, which is fine. However each extension produces 4 options for me. How do I select one of the 4 options, that I want to be in the scene? When I return to the project to download the option I want, the 1 sec duplication isn't removed, so I think I need to do it in scenebuilder, but as I said want to know how to vary the extension from the default first clip it uses

[spiral lynx]

hello

[supple python]

Hi

[tough prawn]

Heloo

[real gazelle]

hi

[hazy acorn]

Anyone having more trouble getting sound to generate with Frames to Video? Or is it just me?

[real gazelle]

John, at which step are you facing this issue? I’ve encountered it before, but after refreshing or recreating the process, the problem was resolved.

[hazy acorn]

I just posted in flow-bugs-feedback with more info if you want to chat there

[real gazelle]

Ok got it

[hazy acorn]

Note: If anyone else runs into this, it looks like the problem was that the quote I had was too short - "I undertand" - if I make the prompt and the text longer, then the audio seems to work

[hollow tundra]

Anyone else stucked in loading screen? Already deleted cookies but after I logged out can't seem a way to log in back again. I was working on several projects with scenebuilder 🤔 Opened a ticket but don't know if it's a general issue

[digital pagoda]

https://youtube.com/shorts/YieANgBbVJM?si=Godb-XMhmHL6KhIp

[eternal bronze]

I have 2 charachter in one scen...only one has dialogue, but sometimes one is talkig to each other with the same voice
how do I fix this? Been running through credits

[hexed cobalt]

With the ultra plan, do I have unlimited veo generations?

[full flicker]

Hello @digital pagoda , 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. Thank you.

Hello @hard badge , 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. Thank you.

[wispy ingot]

Anyone else having issues with scenebuilder? For some reason, when I load my project, it will not play or display.

[stone totem]

it wont load anything for me

Can't see any of me previous work, ot just says " something went wrong fetching your media" :/

[wispy ingot]

For me its weird, it will load the project per se, I can see each addition, but it will not load the actual video and won't play.

[stone totem]

Think thier is a few people having various issues

[wispy ingot]

Gotcha.

[stone totem]

Its letting me create new clips, and its there when i refresh, but i cant see any of my past clips...

[wispy ingot]

It allows me to start a new project and it works just fine, but my previous project is not working. It adds clips and removes clips fine, but it will not play or load.

[stone totem]

Basicly same whats happening to me.

[solid thorn]

Camera angles working really well, getting some slight differences, for example shoes look slightly different. This tool is going to be brilliant

[proper drum]

Anyone know why if I gen a video and I use "insert" in the edit options it cost me 20 credits?

[quaint lake]

Anyone here using veo for Meta ads? Would love to learn how to get better at that.

[shrewd harbor]

if I do 2 videos per command it costs 20 or 40 credits?

[stuck adder]

Have you guys seen the camera thing yet

It’s actually really good

[vivid hazel]

40

That’s for fast veo3.1

[vivid hazel]

You upgraded it to AI Ultra?

[quick swallow]

Hi!
I'm learnig to use VEO3 and I need to generate a video for a commercial that will present a bottle filling with fluid. I've seen many tutorials online regarding it and I've found perfect json prompt, but all of them are used with well known brand that AI knows. I need to generate it for not yet released product so it wouldn't know what i'm asking for. I know i can use frame to frame, but then I'm limiting it's creativity + i have to generate them with something like nano-banana first. is it possible to give it photo of the product and mention it with a prompt or frame to frame is the only option in my case? Thank you for all help!

[crisp basalt]

Use ingredients to video

[quick swallow]

How to do it? Can i do it it straight inside FLOW or do I use some different website?

[crisp basalt]

Yea it’s in flow

[quick swallow]

It's this one?

How to I refer it in the prompt for best results? "bottle I attached"?

Single ingrediant or multiple angles for best results?

[crisp basalt]

You can just prompt it like .. create a commercial video with given image

Or use expand

[quick swallow]

Thank you!

[quick swallow]

It turned out great. @crisp basalt thank you for help

[crisp basalt]

ahh im glad to hear that!!

[icy sluice]

Guys, 99% loading problem is back?

[past pond]

Hey guys i am interesting where you geting inpiration , exmplae any commercial products where can get inspiration?

[crisp basalt]

these looks interesting ! wait since you were planning to use ,,, images and make it as video... have u tried using frames to video?

[rare otter]

looks interesting, I'll give it a go. Was using Veo originally inside the Gemini interface but then it was like j/k you only get 3 video generations per day, then I switched to Whisk to change my aspect ratios for each still etc etc

still learning my way around all this

[crisp basalt]

you can actually generate image like that in flow too 🙂

since whisk and Flow uses same imagen model

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

[noble sierra]

how fix limit quota?

[rare otter]

Used the following prompt:

"Anime style, monochrome video. A young mother, wearing a sweater and jeans, is sitting on a stone wall in a stark urban park at night, illuminated by a streetlamp. The video begins with her back to the camera (Image 01's perspective). The camera then executes a fluid, continuous, and dynamic orbital arc (crane sweep) that moves from the rear-facing position, smoothly transitioning to a sharp high-angle oblique perspective, focusing on the woman's left side. As the camera completes this arc and reaches the final high-angle oblique position (Image 03's perspective), the very edge of a vibrant blue blanket becomes visible in her arms. This blue blanket is the singular color in the entire frame at the end of the transition; everything else remains monochrome. The camera maintains a 16:9 landscape aspect ratio throughout the movement."

Any advice greatly appreciated. Don't want to waste all my credits on trial and error lol. I have 13 days to finish this 7 minute shortfilrm

I don't want the young mother to move at all in this generation. I don't want the ai to mistake camera lingo for actual cranes and arcs, and I certainly dont want the stony wall doing ...whatever that was

[proper drum]

Someone can suggest a video upscaler free??… I don’t feel that “upscale video” to 1080p on Veo enhance the quality

[noble sierra]

can we buy api for fix limit quota?

[north lichen]

i have a question why my videos is english how to change to my language polish?

[crisp basalt]

im also trying to iterate mine atm, it does not want to produce smooth seamless transition..

hold off from generating anyting for now

[north lichen]

please help how to change english to polish?

[south aspen]

hh

[stuck adder]

Long time

[stuck adder]

Let me try

[stuck adder]

I think this prompt is way too long for a start to end frame

You don’t need to it to be this detailed because the information is already in the images

I know what the problem is

[delicate kraken]

/

[burnt oriole]

looks so good

[snow edge]

Hi, does anyone know how to make good UGC in VEO3? Basically I’m trying to create a clip where a dirty surface ends up looking shiny after a single wipe. It could be a countertop. But despite refining the prompt a ton, VEO3 keeps giving me clips that show lots of scrubbing and end up making the surface dirtier instead of clean 🥲

[rare otter]

indeed 🙁

[rare otter]

like if I don't specify certain restrictions, the ai seems to go off the rails

like if I dont say "shes wearing jeans" Ai is like OK HERES A SKIRT

[crisp basalt]

actually , dont mention those at all

[rare otter]

If I don't specify that she doesn't move, then it moves her a bunch

[crisp basalt]

in your case, you only want the camera to move ,

Since the 3.1 update, the model’s adherence to user prompts has become noticeably stronger

Earlier I was trying to make it look like one of those smooth camera crane movements but 3.1 took it literally and spawned an actual crane driving past her 😭

[rare otter]

How would you word the orbital swing of the camera then?

[crisp basalt]

hmmm i should give that a try

but as for now , this is the only almost.. succesful one i guess.

prompt was A continuous, seamless camera shot.

[rare otter]

yeah ew sorry I don't want them moving her at all lol

the point is that she's a single mother

and we are slowing revealing her child

[crisp basalt]

is she giving a single mother who works two job type of character

[rare otter]

the opening lines are:

My mother was 17

when she held me

in her arms thinking

"that this

is the only one in the world..

that loves me"

[crisp basalt]

jk

[rare otter]

the thinking occurs here

[crisp basalt]

i think almost all of the prompt ive tried, does not have any word that involves her to move. yet almost all of the vid that was generated has her moving a lot

[rare otter]

"that loves me" occurs at the end of the 14 frames

[crisp basalt]

wait can u send me like all images u generated

put it in googl drive or something in one folder

[rare otter]

I can DM them to you- the "helpful bot" dinged me for "spamming" and removed them all last night

[crisp basalt]

rip

yeah go ahead

[rare otter]

sent

[shadow warren]

I really hope they find a fix for multiple characters in a scene saying each others voiceline or switching in between

[stuck adder]

The issue is the close up doesnt have the same background accurately

However you might want to use kling for this their camera control is crazy

[rare otter]

what is kling? Where can I find it? Is it connected to the Google Gemini suite?

[rare citrus]

<@&1366831844502274180> @full scaffold is a less than day old account posting these scam/spam messages in every channel

Message was deleted, thanks

[ripe yacht]

thank you 🙂‍↕️

[analog lava]

Hello @tepid mantle friendly reminder videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there.

[wicked glen]

Been prompting in flow to edit out a man from the shot its not refusing to comply .. anything wrong with the platform? (Ultra sub here)

Also .. mid level shots, within flow completely destroy the eyes .. if any way around it, do share

[past pond]

https://www.youtube.com/watch?v=Sr63vhgPOyA

i need feedback

[remote crypt]

is there a good way to prompt a background sound that isn't visually appearing in the scene? I keep getting unwanted visual things appear when all I want is a sound.

[near gate]

hi

[rare otter]

@crisp basalt what do you guys think? I left out frame 02 and 03, just used frame to video with frames 01 and 04

Only issue is I hate what it does with the baby’s head initially

I don’t know if there is anything I can do about it 😢

[wicked glen]

The following is the prompt but flow refuses to generate. Quite ridiculous, as this level of censorship makes no sense at all!:

A static, cinematic shot, maintaining high visual fidelity and photorealism, begins as the entire bathroom is instantly and violently flooded with an intense, oversaturated crimson-red light. In the exact same instant the red light hits, the entire texture of the room visibly corrupts; thick, inky-black shadows burst from every corner, joint, and surface, writhing and pulsing across the walls and ceiling in a chaotic, impossibly fast pattern, while the two mirrors become perfect, shimmering black voids. The person in the white robe remains perfectly, hauntingly still, their back to the camera. The audio is not a build, but a single, deafening, low-frequency SHOCKWAVE sound that hits simultaneously with the red light.

Please <@&1366831844502274180> fix this

With this level of censorship, people are going to frequent other platforms a lot more.

[turbid lintel]

weird.. i can generate your prompt

[peak frigate]

hii

[crisp basalt]

How is pinging discord moderator gonna help you with this kind of issue , be for real 😭

[crisp basalt]

Same I can generate it too

[wicked glen]

Its an image to video prompt .. with a woman standing, facing mirrors, in a luxury washroom.

Strange because I've tried a million times, it just won't generate.

[unreal viper]

please only ping mods where there is a moderation issue that goes against our #rules-and-guidelines・📋

[stone totem]

have you verified your age?

[eternal bronze]

is it better to make "shorter" videos insetad of 8 seconds? will I have more "control"?

how do you control this?

[rancid ginkgo]

Someone help me, I signed up and it keeps showing up.
It looks like you don't have access to Flow. Check availability here and contact support if you have any problems. Someone please help me solve this.

[supple pendant]

There's a known bug where most imagery that shows a woman in a washroom or bathroom will get auto-flagged. Women aren't allowed in bathrooms apparently

[eternal bronze]

Ill be honest...this VEO 3 model does hatever TF it wants

https://tenor.com/view/side-eye-dog-suspicious-look-suspicious-doubt-dog-doubt-gif-23680990

so when you type VOICEOVER in the prompt...doest it mean voiceover?

Or is it like my wifes wording when certain words dont actually mean what the dictionary says

[crisp basalt]

maybe u should take a break and take the time to read their prompt guide

https://cloud.google.com/blog/products/ai-machine-learning/ultimate-prompting-guide-for-veo-3-1

[eternal bronze]

i have

[crisp basalt]

unless u drop the prompt that u wrote , then idk i cant make a judgement just from this

post the prompt + the problematic generated result video

[eternal bronze]

TITLE: "Night Run — Windshield Front (VO, 1940s Voice Emphasis)"
ASPECT RATIO: 16:9
DURATION: 12s (single scene)
STYLE: 1940s–1950s film noir; black-and-white; high-contrast; streetlight/neon reflections on glass; no music

SCENE: Exterior front-mounted view of a car at night, camera looking in through the windshield at the detective driving.

CAMERA: Static, hood-mounted, slight natural road vibration. No pans, no cuts.

BLOCKING / EYELINE (STRICT):
• Detective drives, eyes forward; brief glances to road and rearview only. No other characters.

PERFORMANCE LOCKS:
• Detective: cool, calm; minimal expression; steady hands on wheel; cigarette at lip acceptable.

AUDIO: engine low, tire-on-road hush, light cabin rattle. No music.

VOICE PROFILE (STRICT — 1940s TOUGH-GUY):
• Gravelly baritone, dry and sardonic; slight Mid-Atlantic/NY inflection.
• Clipped diction, tight proximity effect; ~110 wpm; measured pauses.
• Period slang and dropped g’s allowed (“watchin’,” “gonna”). No modern phrasing.

VOICEOVER: DETECTIVE 1940s cadence:
“All his A.I. talk´s been eating me. This town can’t count to ten without using its fingers.”

NEGATIVE PROMPT: color, modern dashboards, flashy effects, handheld shake, text overlays, watermark, additional speakers

[crisp basalt]

so we can try help and iterate

[eternal bronze]

thank you

[crisp basalt]

jesus christ ,

ok let me try it

[eternal bronze]

the scene is correct, the style is correct...just the man keeps talking

i clearly state voice over...its worked before for me

[eternal bronze]

not good?

[crisp basalt]

oh nah its good, i think this is actually clear and well organized

im refining it atm

wait so the problem was,, the guy was not talking?

[eternal bronze]

this one was better...but still no luck with voiceover

TITLE: "Night Run — Windshield Front (VO)"
ASPECT RATIO: 16:9
DURATION: 12s (single scene)
STYLE: 1940s–1950s film noir; black-and-white; high-contrast; streetlight/neon reflections on glass; restrained period performance; no music

SCENE: Exterior front-mounted view of a car at night, camera looking in through the windshield at the detective driving.

CAMERA: Static, hood-mounted, slight natural road vibration. No pans, no cuts, no reframes.

BLOCKING / EYELINE (STRICT):
• Detective drives, eyes forward; brief glances to the road and rearview only. No other characters in frame.

PERFORMANCE LOCKS:
• Detective: cool, calm; minimal expression; steady hands on wheel; cigarette at lip acceptable, no big gestures.

ACTION:
• Streetlights sweep across the windshield; occasional wiper pass acceptable.
• Detective keeps a measured pace, jaw set.

AUDIO: engine low, tire-on-road hush, light cabin rattle. No music.

DIALOGUE — DETECTIVE (VOICEOVER, rude 1940s tough-guy cadence):
"All this A.I talk´s been eating me. Artificial intelligence, huh? Save it. This town can’t count to ten without using its fingers”

NEGATIVE PROMPT: color, modern interiors, contemporary dashboards, flashy effects, handheld shake, text overlays, watermark, additional speakers or characters

[eternal bronze]

no...they guy IS talking...but not a voiceoiver...hes moving his lips

[crisp basalt]

wait ok so u want it as narration style..?

[eternal bronze]

ive gotten voiceovers here before...thats whyI dont understanf the issue...I dont see why iots confusing for FLOW to understand

[eternal bronze]

yep

also..im having issues when hes talking in a bar...other characters keep saying his lines...even tho im promting for "no other harachter in scene" etc

[eternal bronze]

I could just narrate with eleve labs, but then I cant match his voice in the nupcoming scenes

This is the first episode 🙂

[crisp basalt]

oh one of the generated i got was , him not talking

but its singing lmao?

[eternal bronze]

yeah see...no music is propted also...and in the negative as well

i guess its just a matter of burning credits?

you can confirm the promt is OK...i mean its not a lazy prompt

[crisp basalt]

well thankfully im on ultra plan right now so i dont mind iterating

yeah im still iterating and generating

[eternal bronze]

im actually trying different things within

[eternal bronze]

I guess this is going to be the best step for me innit.....If I want to spend the hours learning without the wallet in my head?

[crisp basalt]

yeah its.. very difficult when its burning the credit,,

3 generated vid , actually got the man not talking at all

but the narrator is singing , why is this so stubborn lmao

[eternal bronze]

yeah...thats what I mean...been trying, cant really see where I need to "retouch" my promting

[crisp basalt]

well im trying this one right now

STYLE: 1940s film noir; black-and-white; high-contrast; streetlight/neon reflections on glass; no music

[00:00-00:12] VISUAL:
A static, hood-mounted camera looks in through the windshield at a **silent detective** driving. The detective is cool, calm, and **does not speak**, his face showing minimal expression, his eyes forward, hands steady on the wheel. The car has slight natural road vibration.

[00:00-00:12] AUDIO:
Ambient sound of a low engine, the hush of tires on the road, and light cabin rattle. A **narrator speaks** in a gravelly, sardonic 1940s tough-guy voiceover: All his A.I. talk´s been eating me. This town can’t count to ten without using its fingers.

NEGATIVE PROMPT: singing, detective talking, detective lips moving, music, color, modern dashboard```

mm ok i think im getting there

[eternal bronze]

Do you always used timestamped promting?

[eternal bronze]

thank you...reading and trying this now 🙂

[crisp basalt]

not always, usually its my last resort if nothin works

ok uhhhh

A 1940s film noir style black-and-white, high-contrast scene. The camera is static, hood-mounted, looking in through the windshield at a detective driving. The detective is visibly silent and his lips do not move. The scene is quiet, with no people talking, only the low sound of an engine and the hush of tires on the road. Separately, a narrator with a deep, dry 1940s male voice says: All his A.I. talk´s been eating me. This town can’t count to ten without using its fingers. NEGATIVE PROMPT: singing, music, detective speaking, actor talking, lips moving

[eternal bronze]

I will use this in the scene builder. Do you think prompting "a narrator speaks "will give me issues when I want to use it as the detectives voice on the next scene?

[crisp basalt]

but it was only one out of 8 vid that i generated using that prompt,

the rest were ,, singing and the man's talking

[eternal bronze]

well..thank you anyways...

ill keep trying...dont really know what to touch

ive gotten good reults with voiceovers before:(

[crisp basalt]

oh shoot i think i got it

[eternal bronze]

I though i had it dialed in

[eternal bronze]

yeah...getting better!

[crisp basalt]

minus the ... different cut angle it did .. but yeah

[eternal bronze]

yep...i need to change hios voice tho 🙂

might just do it in leven labs...or start from the next scene, and come back to this one using extend

[crisp basalt]

yeah thats the only way i supposed

in the mean time im gonna go ahead and flag the failed output

hopefully the flow team can check it out

[crisp basalt]

i think 3,1 really messed something up with prompt adherence

[eternal bronze]

thank you

[eternal bronze]

shoul i test other veo?

[crisp basalt]

hmmm, well theres no other veo now to use in flow... they removed 3,0

the only way to use 3.0 is through api

or other website that still offers the 3.0

[crisp basalt]

quality actually ... generate the video differnetly using that prompt omg

quality mode veo 3.1

[crisp basalt]

"BOTEL"

[eternal bronze]

haha

[crisp basalt]

@eternal bronze have u tried "expand " your prompt using the prompt expander "film Noir" preset

[eternal bronze]

no i havent

but i feel the voice etc it just not right...i think i need to start from sratch

Im using frames to ideo, all it need to do is move the scene and create the cvoiiceover...the dude keeps talking. Ive prompted VOICEOVER, no lip movement etyc etc

certainluy not what I would call promt adherence

[eternal bronze]

just when I got it right it added incoherent subitles

[full flicker]

forwarded to #flow-videos・📹 , the appropriate channel to post your work. Friendly reminder, please only post your videos once. More than that is considered spam. Thank you.

[keen aspen]

where is weekly five prompts? it is not in #flow-creative-challenges・🎨

[eternal bronze]

Anyone having issues on Flow interface

its jumpy...can go into scenebuilder

something is wrong 🙁

[twin inlet]

is there anyone here that is having issues using Ingredients to veo with veo 2 quality, and can only use veo 3.1 fast? either they've retired ingredients with veo 2 quality and not given us a solid model or mine is bugging out, appreicate some feedback if there are any peeps out there that actively use ingredients to video

[burnt oriole]

I'm in love with the image generation and editing section of flow

I think that is the best way I could ever use Nano banana. But I don't like the fact that it's only accessible from the ingredients.

I wish there was like a separate option in flow for image generation and image gen/edit only

like no other google products have something like that

[trail oriole]

hello.. whyy there is no ASI? (Super Ai)..

i know why , its beacuse he is extreme powerful, we cannot control it, it can copy it self into other computers, such in cloud..

Right?.

[wispy ingot]

Does anyone know how I can get some sort of support regarding Flow? I have posted a few things on here with no resonse. I am starting to get upset. considering I am an Ultra member and pay so much for it.

[trail oriole]

how much you need support?, isnt you talk to gemini about what you do?, he will does give you the support you need ..

isnt that Enough?

or you need M૦Ney? , thats not how its works..

[trail oriole]

if you are giving a things such as a research papers that could make world better, then you should not be ever waiting for M૦Ney.., beacuse you need a direct m૦nEy.. , you want a thing if you do this, you get this in exchange..., i suggest that you are someone .. , Right?

at today, i discoverd a thing about how we gonna control asi even we trained it or make it learn everything in the internet.. , even it tried to get out, by copy it self, it Cannot..

its a OS..

its an IDEA!

i have it.., i just waiting to someone who could help in it. to ask about it..

[dry oar]

what kinda support?

[spiral isle]

hi

i got some questions about flow

so i wanna start up an ai instagram cooking account and videos will prob be about 1 or 2 minutes do i need to use veo3.1 quality to make long videos does anyone know

[dry oar]

there isnt a change in length between the models

[spiral isle]

so it can only give 8 seconds of videos ?

[dry oar]

as of now yes, but in flow you can extend clips

[spiral isle]

how

[dry oar]

add the clip you make to your scenebuilder

[spiral isle]

ohh

[dry oar]

then youll see the option show up

[spiral isle]

i got it can i make it up to 2 minutes is it posibble ?

[dry oar]

ehhh

[spiral isle]

no?

[dry oar]

the consistency kinda fades away the longer it gets

[spiral isle]

oh thats bad

[dry oar]

Id recommend using ingredients to video

[spiral isle]

sure thx

[dry oar]

helps it stay a little more consistent

[spiral isle]

and

i got 1 more question

should i buy google ai pro

will it be enough to make instagram videos

[dry oar]

theres gonna be a lot of trial and error

[spiral isle]

oh than thats bad

[dry oar]

id honestly say ultra if u know u can make a profit from it

once u get the hang of flow though it gets easier

[spiral isle]

hm ok

[dry oar]

im using the free student pro plan they offer

works well enough for me

[spiral isle]

i go to highschool i cant use that

maybe next year i can try

[crisp basalt]

Best investment I’ve ever made tbh

[wispy ingot]

Hi! So, for some reason when I am trying to go into a project that I’ve been working on for awhile, the video never appears nor does it allow me to download it. I’ve tried everything from clearing cache and browser history to switching browsers altogether just in case if it was a setting or something. No luck. Any advice?

[plucky summit]

Is there a way to learn from each other ? Reduce the trial and error a bit?

I feel like as soon as you want something else then 1 or 2 people talking a dialog it just doesn’t do anything good at all. Not sure if my prompts are not good enough or if my needs are just to complex for now.

F.e. I want to make an interview between 2 people. First the reporter asks a question then I want to let the other person think what he would like to say and after actually say out loud the polite version of it.

But all my attempts to let the viewer hear the persons thoughts, the AI ignored it all and makes the person say it out loud.

[crisp basalt]

post the prompt and the result video here or #flow-tips-n-tricks・✏️ so people can see and help you refine the prompt

its hard to know when we dont see the prompt that u wrote + the video

[kind bison]

How I would do this: I would do a version of the video in which the character sits in silence for a bit, then answers (you will have to use time stamps in your prompt to get the timing correct). I would then do another version in which the person says the impolite version, extract the audio from that clip, then add the audio track to the other version in my editing software. I would either prompt for the dialog in the impolite version to be drenched in reverb (to give it that "in my head" effect), or just add reverb in my editing software (gives you more control).

[plucky summit]

Thank you very much, that is an awesome idea

[fast summit]

does anyone know why flow/veo is making me crop my 800x800 image in half? can i not upload square images?

[ancient blaze]

In Flow, images must fit with 16:9 or 9:16 aspect ratio. That's why images with a free aspect ratio must be cropped.

But you know, Cropping the image actually is a good purpose. Basically, Veo supports free-ratio image input.

[fast summit]

it baffles me how i have previous uploads that are square. but i cannot do it anymore.

ill try to extend my square image so its widescreen i guess. thnak you for the speedy reply.

high 5

[crisp basalt]

I think back then we could upload images in any aspect ratio due to Flow’s built-in crop feature would automatically pad or adjust them to fit the 16:9 aspect ratio

now its just fully strict to 16:9 or 9:16

[ancient blaze]

Okay i think only two options for now. Crop the image or Expanding the square image into landscape image (16:9) using nano banana.

[fast summit]

thats what ill try, making it widescreen..thanks for the help!

[full flicker]

Hello @ancient blaze , 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. Thank you.

[full flicker]

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. Thank you.

[ebon snow]

I am trying to sync a vocal stem with a character. Is there a way to upload the vocal stem so Flow can have the character use the stem for guidance?

[cedar dew]

Why does Flow tell me that “portrait mode isn’t compatible in vertical and that I need to switch the screen to horizontal” when I try to create a vertical video? I checked the prompt to make sure it doesn’t mention portrait mode, but the same message still appears, even though it still generates the video.

[dry oar]

ooo free credits, thanks

[main venture]

hello i have problem with subscription i cant use my card to renew

[quaint lake]

Does anyone know why video playback on flow is laggy or slow? Do i not have enough RAM? or is chrome or?

[rustic sand]

I was a subscriber of Flow during October as a Pro yser and i didn't receive any credits.

[urban patio]

Same here. @main saddle have all credits been delivered?

[proud imp]

I think this is the accurate track... however, is there a way to achieve this or a way about this whereby you are precise of your functionality and also something that can confirm authorizations?
Afterall, a week or two it may not be notable... however in a few years you may notice something of the directionality of this that may be questionable at the least

[proud imp]

I wonder how long has it been since this the problem?
And what was considered "normal"?

[round dove]

Oh, just noticed the extra 500 credits, woohoo!

[kind bison]

?? What are you trying to say??

[mental rapids]

I didn't receive the 500 credits for pro either

does the subscription means subscribe to flow only? or mean general subscription for google AI pro?

[daring sable]

I’ve been trying to generate a sequence of numbers in flow - ‘0, 1, 2, 3, 4, 5, 6, 7, 8, 9’. But no matter the prompt I use, I keep getting visual errors. This frame is from the best video I’ve gotten so far. I’m using a JSON prompt. Do you have any ideas how to fix this?

[celest tendon]

Maybe try a non json prompt?

[ancient blaze]

I think the Veo model is not designed for typographic accuracy.
So if this is just an experiment, here are the results. Veo isn't fully capable of this yet. But if this is a project, I suggest using a more efficient way.
Create a video manually for example, or use first & last frame mode and combined with prompt if you want to try it further.

[hallow sorrel]

i subscribed too early not knowing there would be a free refill in between 15-22 💔🥀

[proper drum]

Can edit functions be free for Ultra users? I mean that they continue free credits after leaving the “experimental” phase 🙃

[forest frigate]

Hello,

I'm really disappointed with VEO3. I know it's in beta, but the physics issues are driving me crazy. In complex scenes, it becomes a farce and a complete mess. It's impossible to be creative, a huge waste of time and credits for nothing. I have the Ultra subscription, but I won't be renewing it. I'll come back once the physics are more realistic. We're adults, not children. I keep getting constant "does not meet the conditions" messages, and then it generates them. I should have at least finished VEO3 and sold it afterward.

[unreal viper]

Hello, as 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. If asking for advice or tips on your prompts do so in the appropriate channels.

[supple pendant]

try making a still frame in Whisk and using frame to video in Veo to make the video

[slender haven]

i'm trying to learn more and more about Flow everyday and today i decided to learn about extend function.

My question is.. consistency. How do i keep consistency between clips?

I see that if i use Start to End Frame, it gets to the point, if i add another scene (which shoudl be following the previous) it gives me a dry cut that i need to edit in an external editor

How do i make flow understand it's a sequence and not another clip? Which features should i use?

I'm not sure which one works better

Clip 1 -> Extend + Prompt

or

Clip 1 -> Add final frame as start + prompt

or

Clip 1 -> Add a final frame as start and add a final frame + Prompt

[vernal quest]

hy

[main saddle]

I was a subscriber of Flow during

[proper drum]

nice, now we have "create image" and model options "imagen 4" and "nano banana"

[turbid lintel]

The new feature nano banana is nice touch , now the workflow become more easier

[crisp basalt]

Thank you so much for creating the images section tab now within project page , this is very needed and much easier for me to find stuff involving respective projects ❤️

[noble carbon]

what do you guys do for ingredients to video if the image doesnt have a thumbnail when looking for an image

i even used the new nano banana feature to remove background and downloaded it, uploaded it and i dont know what causes the thumbnails to pick and choose when it wants to show up

i figured it out, if the background has too much white no thumbnail pops up, but if you use nano banana and add a black background you will see thumbnail

[crisp basalt]

Shout out to Andrew 🔥 thank you 🙏

Also Andrew don’t be shy share some vids u made with us 👁️ Jkjk

[stuck adder]

Yes very much needed

Does it give a water mark?

Ah it doesn’t niceeeee

[crisp basalt]

Also i saw that you can generate image from text prompt using nano banana but all i got is internal server error 500..

oh nvm its working now

tho theres this weird glitch, the images keep auto "refresh" everytime i hover my mouse on any images

nvm its working fine now lmao

[summer rapids]

The new image tab seems good. But all the start frames are pixelated when I add the images to a prompt to use them as image to video. Tried it on my phone

The images are in 131 x 230 pixel

[crisp basalt]

its full res for me tho

are u talking about this small image ?

have u tried generate the video , is the image still lookin pixelated in the generated video @summer rapids

[summer rapids]

The problem is when editing the picture (with the pencil). The original image is in 768 x 1344 and when editing the picture, it's 333 x 583 (on pc)
On mobile, we can't directly add the picture to the prompt (no button) and when we edit the image, the resolution is 131 x 230 so pixelated

[silk peak]

what do I do?

this is mean this: It looks like you don't have access to Flow. Check availability here and contact support if you have any problems.

[tough rivet]

Hi

[summer rapids]

The problem is really visible in portrait mode (1M pixel to 194k pixel when editing so 19% of the original generated resolution on pc, even worse on mobile)

In paysage mode, the original image resolution when creating an image is 1344 x 768
When editing it, it's 1020 x 583, less pixelated but we lose resolution (58% of the original resolution)

[restive needle]

HI

[crisp basalt]

they lowered the pixel temporarily when editing the image,

[crisp basalt]

no button to add pic to prompt? what do you mean

i think its better if oyu show your screenshot on the issues

[summer rapids]

Seems ok if we go back to image to get the picture and not doing it in the editing view

[crisp basalt]

or record the video , reenact the step