flow-general・🌊

[valid turret]

but forget it lmao

[solid acorn]

flow is tweaking rn

[noble carbon]

generating too quickly anyone?

[edgy holly]

Can you give me specific examples, with the prompts and inputs?

[solid acorn]

flow isnt letting me login

[lament lotus]

Yes sir

[zealous cosmos]

hello guys

[lament lotus]

Helo bro

[zealous cosmos]

you everywhere bro ahaha

[lament lotus]

Hahaha

[pure venture]

You too man 🗿

[lament lotus]

I want to see your ai pictures bro

Very good, friend.

[pure venture]

Wow,how many hours you need to make this

[lament lotus]

OMG very good

[pure venture]

Haha ,you made some bro

[lament lotus]

I will try to make it later

[crystal talon]

it took me 11 days 15 hrs a day. i have nightmares about it still 🙂

if veo would do everything i tell right away, it would take only 3 days. but as you guys also know 99% of the generated videos are pure slop and Acid trips. still no update on physics. what it can do best is talking heads and walking. as long physical contact is involved. good f#ckin luck.

[tawdry burrow]

hello everyone

[slow glacier]

@tawdry burrow @near gate

[near gate]

hola

[slow glacier]

What is the author James Windirsandsulf know of?

What Nick Portsant did to be first in sales on the sci-fi novels of 2021

if veo would do everything i tell right away, it would take only 3 days. but as you guys also know 99% of the generated videos are pure slop and Acid trips. still no update on physics. what it can do best is talking heads and walking. as long physical contact is involved. good f#ckin luck.

[turbid lintel]

there is a way to reduce slop by 50% , use timestamp prompting .

timestamp give you more control

[supple wharf]

Hi everyone

[pure venture]

Thats a lot of work man

[tardy ridge]

Will we ever get ingredients in 9:16?

[crystal talon]

just some feedback if anybody who has something to say at googlelabs reads this: flow is infuriating. random failed generations. detecting minors when there are 2 adults in the beginning frame. no chance of making an action movie. I cant make a guy point a gun to woman for example. but a woman can point a gun to a man. You know there is a word for this behaviour :).

[crystal talon]

and physics basically doesnt exist. puncking kicking doesnt exist. actually everything involving physical contact is terrible. Kling is much better at this. Slowmotion doesnt exist. sometimes it randomly does slowmotion without prompting it. if there is a couple in the frame, a man and a woman, and i prompt the man says:"blabla", the woman says"blabla" then it often swaps the dialogue. it doesnt know who the man and who the woman is. we have to describe the person clearly like The man with red shirt and glasses says, or the woman with long blonde hair says.

[tacit jackal]

ummmm ok

[latent jetty]

OMG SNEAJER

I didnt know you were in this server aswell

[pure venture]

This is cool tho, how many prompt you got fro this?

[tribal perch]

Why is Flow not working on computer but working on phone?

[raven siren]

Hi there. I'm using flow on Windows 11 desktop (cable internet connection)but i have a problem with my project thumbnails not showing up. I mainly use it for image generation. However when I generate a video in the project it sometimes does give me a thumbnail. Inside the project everything works fine, except for maybe sometimes having to refresh the browser if a generation is not showing. The refresh isn't working for the project thumbnails. I've been trying to solve this with Gemini but no luck. It pointed to this error message in the developpers view of the browser (Chrome) Uncaught (in promise) Error: A listener indicated an asynchronous response by returning true, but the message channel closed before a response was received. Is this a known problem?

[tribal perch]

Please repair the flow, i cant enter the website, i dont have button "create with flow"

What is the problem with Flow?

[quasi mountain]

why flow DROPS the quality of video ,when using the extend feature .it drops the visuals quality .Anyone else noticed this problem.can anyone guide me what should i do?

[unreal viper]

please use #flow-videos・📹 to show your images and videos made in flow/veo

[patent arrow]

Won't use my ingredients either. Keeps generating a different image. Been like this since yesterday.

[edgy holly]

Ingredients don't work like image to video.

[patent arrow]

Oh. I meant to create images. It was working fine for me yesterday morning.

Keeps showing like this.

[edgy holly]

I saw that a minute ago (working on a scene right now), but the image still loaded.

[patent arrow]

It does. But, for me, it's not using the characters I'm working on.

[edgy holly]

What reference images are you using?

[jaunty stratus]

Does anyone know how long scene builder can be used

Like how often is it possible to extend

is it limited to 20,50, 100 or more ?

[iron quest]

When does the NBP daily limit tick over? I was using NBP Pro in Flow for the first time last night no issues. But now today, it says I’ve hit my daily limit. What is the limit?

[patent arrow]

~100 to 1k, but not all the time. Reset time also varies, based on my experience, at least.

[iron quest]

Thank you, was really random that last night hadn’t hit any limit and then the next day apparently hit some limit and hadn’t even generated anything this morning

[patent arrow]

It resets at a certain time, with no roll-overs. I think the limit depends on the downloaded quality, too.

[iron quest]

I usually use another api for NBP but decided to try it direct through Google Flow last night, usually only generating vids

[patent arrow]

How do you do it? The API. Sorry, I'm not very technical.

[iron quest]

Oh through another website I’ve got a subscription too

[patent arrow]

Oh ok. Tyty.

[zealous fractal]

Is there an app version of this coming out?

[maiden talon]

hello everybody? How do you generate video with a 9:16 aspect ratio?

[ebon snow]

Has anybody had difficulty with finding download locations?

Download locations for VeoFlow.

[delicate ridge]

So if you are using an image just make sure your image is in portrait. Otherwise if you will be editing with capcut then you can specify

[delicate ridge]

Your promt should say the characters should be clearly visible from head to feet at the end of the clip/scene. No closeup etc at the end. That's the magic. Hopefully next version will address this

[tepid kiln]

Omg how do you even use this app XD

[compact basin]

Hi, Is there a limit to build on google Labs, i got limitation message. how to unlock it?

[pure venture]

where i can find flow link website?

[patent arrow]

There is for NBP. Not sure for the others.

[kind wren]

Why can it be generated, but not downloaded?

[pure venture]

I really want to try flow,but still not available in my country

[wintry pine]

Unfortunately, you'll need to wait until it's available in your country. Are you able to use Google Vids?

[zealous cosmos]

Until when we gonna wait sir?

[elder briar]

Hi! I want to create a video from images. How do I start with Flow?

[upbeat sierra]

hi @elder briar you can do this via Frames to Video available in the prompt box dropdown

[elder briar]

thanks but the system is paused and I dont see those options

[woeful void]

I spent a whole week generating videos, and now they're gone.
The issue with extremely slow page loading and lag is still not fixed.

[sick sparrow]

is there any ongoing work atm for generating images? I'm trying to generate images with prompt > 200 characters. This was never a problem, but after clicking generate I get a failed error. Strange this is that when I use 3 characters the image is created.

[analog lava]

Hi @unborn ocean, 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!

[patent widget]

You have to go to https://labs.google/fx/tools/flow

[raven siren]

Nobody?

[wheat radish]

Just tried uploading a few product images

[upbeat meteor]

I can't upload my photos. Is the app experiencing problems?

The photo is perfectly normal, but it violates the policy.

[wide ether]

Have people been successful putting NOT clauses in your prompts to filter out unwanted things and concepts? Just curious

[crystal talon]

i feel like flow gets worse by day

[wheat radish]

Yeah same problem

[cinder pewter]

I just tried to upload even just a pic of an apple with new account, the same problem here

[upbeat meteor]

I need to finish the project by tomorrow, but it's malfunctioning today 🥹

[wheat radish]

Bruh same issue here

[upbeat meteor]

I used two computers and four accounts but still couldn't upload the images

[edgy holly]

More likely to backfire than help.

[teal island]

I can't upload anything.
Even the images I originally uploaded won't work.
Even a photo of just clouds in the sky won't upload.
I have an Ultra subscription, and the deadline is fast approaching. My life is ruined.

[wide ether]

That moment you realize that an image has too many biases backed into it that the model always wants to generate the video with the things/concepts in there that you DON'T want no matter how much explicitly positive desciptions I put in there to try to force it the other way, and even using the negative prompting too. Some images just have some biases baked into them when input into the model 🙁

[misty bison]

Hi everyone, does Flow have any API similar to Veo3 with GeminiAPI? I got Google AI Ultra to use for Flow, but now I realize I can't generate Gemini api-key with that subscription to use Veo API

[edgy holly]

There is an API with Veo (Vertex), but it's a separate service.

[misty bison]

do you have any suggestion for automating things with Flow?

[edgy holly]

I wouldn't do that. To get good results with any of the AI video models, you need a human who knows how to prompt well, with good visual ideas, good input imagery, etc.

[analog lava]

Hello, if you're encountering errors and /or bugs or would like to leave a general feedback about an of our Google Labs products please post in the new #1238621511401734254 channel with as much detail as you can☺️

[ionic halo]

does anyone know if there are plans to be able to render elements out with an alpha background instead of solid colors or otherwise?

[charred fern]

Hello guys , why veo3 when you want to create video the create but you see black screen only?

[supple wharf]

Happy weekend everyone

[deft patio]

thanks you too!

[sonic loom]

Is Google Flow ultra plan offering unlimited video generations? @ivory harness @vale lily @nimble mulch

[crystal talon]

yes, you have to choose the veo 3.1 fast(lower priority option. it does the same job

i still dont know what veo3.1 quality does better. they are not better at all. all the same. kinda a scam tbh.

[sonic loom]

That means quality and fast are same? No difference at all.

[crystal talon]

i use it since 4 months daily never saw any difference.

it just burns credits, nothing else

use fast(lower priority)

[tiny magnet]

I have the pro plan, but there's still a watermark in my videos why ?

[edgy holly]

No.

[sonic loom]

Okay for telling the truth 😁

[turbid umbra]

it does work id say 80% of the time it gives better generations but yea, I use it only when the fast model simply does not comply.

[bitter flame]

Do you can make video?

[exotic shell]

how many nano banana generation can you do as ultra user?

[jolly gorge]

How to keep the font clean during the hands on move?

Any suggestions guys?

[turbid lintel]

unlimited

plus unlimited fast generation

[elder briar]

can somebody create videos right now?

[dull goblet]

When I tried to download the images I created yesterday, an error occurred. However, when I created a new image today from the same project, I was able to download it.

[jade crest]

Why is the downloaded file saved with a .txt extension after the video is generated and upscaled to 1080p?

😹

[analog lava]

Hi @elder briar , as a 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. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[elder briar]

Thanks an sorry I am new with this

Do I need to pay something in labs.google

[jade crest]

This is a bug. If the prompt is in English, the filename is displayed correctly. If the prompt is in other languages, the file extension is not displayed correctly.😹

[vale fern]

@elder briar

@edgy holly

[ionic halo]

👍

[latent jetty]

Anyone know why they made the veo logo bigger as if I’m spending this money just to have a bigger watermark smh

[grizzled storm]

I use VPN, but the site still writes about unavailability in the region, does anyone know how to solve it?

[tepid iron]

To what extent is this app free to use?

[fickle fossil]

I Can Get More Credits For Flow

[hidden latch]

There's nothing free in this life. Pay to play!

[hidden latch]

Sounds like you need 2 VPNs that'll solve it

You have to set up the second in an unblocked country first, then transport it anywhere in the world

They always check the set up origin..that's why the fails

All vpns are geo tracked

You can manually program the pings also, but this is a deep knowledge.. chat gbt 5.2 can assist

But it's programmed not to help with that so you must trick it.

Bottom line, in a years time, it will no longer be blocked

[solid pelican]

Hello everyone, I am new to AI and I am not a native English speaker. I recently used picture materials to make myself a character. I have difficulty generating the prompt words for the video. I would like to ask for help.

1. Is my use of prompt correct? I'm not sure my description is oversimplified, or does it still need to be more specific?
2. I still can’t get the feeling of Spider-Man passing over my head using such prompt words, and then the camera has no time to capture the speed of Spider-Man passing over my head.

Please give me some advice, or if there is any place where I can get these teachings~

A dynamic, immersive cinematic action sequence in a golden hour metropolis. A male protagonist in a tailored dark suit swings powerfully on a web-line. Mid-air, he executes a tight backward flip while his momentum carries him rapidly forward.

The camera performs a continuous, sweeping spatial movement: starting low looking forward at his approach, then tilting straight up dynamically to track him as he flies directly overhead (filling the frame with the underside of his moving suit and creating an intense rush of speed), and continuing the rotation all the way over to find him landing silently behind the camera's original position.

The camera movement settles on his back as he calmly adjusts his cuffs, then pushes in for a tight close-up as he slowly turns his head over his shoulder to smirk confidently at the lens.

[edgy holly]

https://one.google.com/ai/credits?g1_landing_page=75

[edgy holly]

Not.

[hallow river]

Yoo wassup with the Audio generation ? It just goes in and out when it wants too?

[jolly gorge]

Precise mode is currently not available in your region

[narrow sigil]

are you guys getting views by making video in vo3

[crystal talon]

when will flow get a tool where i can create a characters, give them names, save them and all have consistence voice.

every damn line he says has a different voice. so annoying. i dont want to work with elevenlabs. too tedious.

[reef mural]

Well, if I wasn’t pissed off before because of constant generation failures and no sound issues, I’ve stood firm with Google and have had Ultra for four months now. And what happened today? My paid Ultra account was suspended. No reason was given except “suspicious activity.”
That account has two-step verification enabled, and no suspicious login attempts have been recorded. I’m fed up with all these fing problems. This costs me a lot of money, and I’m seriously considering switching to Higgsfield.
This is bollocks. What the f is happening? I expect an apology and compensation for this. Grrrr.

[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.

[crystal talon]

flow is super woke. i cant even make an image of a man pointing a gun to a woman. how am i supposed to do an action shortfilm. ultra woke.

it detects two adults as minors and doesnt generate. what the f is going on. i payed 130euros a month.

am i supposed to make kid friendly movies or what? oh wait, it doesnt generate kid actors too 😄

so i guess i make a stu#id love movie then.

if this goes on, china will win the AI competition.

if i just prompt a police officer walks down the street, 90% of the videos are either indians, asians or blacks. if I am lucky 1 video has a white actor if you dont specify the actors race. speak about being racist. flow is super antiwhite. try for yourself guys. so ridiculous.

[hidden latch]

When my prompts get flagged, I change them over to Json prompts and it passes more of the time..

[tired nova]

What's a json prompt?

[hidden latch]

It's like Java script.. what better way to tell AI what to do with code.. just ask chatgbt to change your prompt to Json and it will

[quartz drift]

Text to video

[jade crest]

My prompt itself is fine and works for video generation, but enabling this feature frequently triggers policy violations. What is the underlying reason for this behavior?

[edgy holly]

It's a type of structured prompt that uses a code structure used in Javascript. Overly complicated, and you can get the same results with the same text broken up in to organized sections, without all the code bits.

[crystal talon]

flow is a joke at this point. text to video: "a women in bikini looks at herself in the mirror" works fine. then when i extend the shot flow tells me that violates the policy for sexual content. really behindert at this point. who programs this crap?

[jolly bay]

chat im a swifte

[kind bison]

They just don't care, man - Google does not believe in freedom of expression or the rights of artists to use their own judgement in what they create.

[crystal talon]

it is crazy really, it is so obvious. this needs to stop. art is expressing, inspiring, polarizing. google is like a teacher who gives the class only red color and tells them to draw green trees. i already canceled my ultra sub, next time my money goes to china.

[fierce stag]

Hello, Gemini gives me 3 free video generations with Veo 3.1 per day with my Google AI Pro subscription.

Can these some how be used on Flow through labs.google? When I generate it charges me credits.

[full flicker]

Hi @snow sparrow ,videos aren't created in the discord server. In order to make a video you'll need to go to https://labs.google/ pick the tool you want to work with then go from there. If you need help with prompts please ask in the appropriate channels such as #flow-tips-n-tricks・✏️ for advice. Thank you!

[kind bison]

Yeah, it's truly bizarre that Chinese models offer less censorship than "American" ones (I don't know how many actual Americans even work at Google anymore).

[crystal talon]

none i guess. it is india bro.

[jade crest]

Is there any way to keep voice consistency for speakers on the Flow platform?

[rocky prairie]

is there any way to makle 1 min videos on flow?

[full flicker]

Hello @clever remnant , as stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now. Thank you!

[rocky prairie]

can you generate 1 min vids on flow

[rocky prairie]

?

[crystal talon]

no it sucks

[supple wharf]

Happy weekend everyone

[ashen mirage]

hi

tomorrow i need to go back work

😭

i am very nervous now

[crystal talon]

yo google. i do text to video an asian woman infront of a mirror. and it always gives me the same woman. what is going on. i want a different face

is only one asian female actress working for flow? 😄

[turbid umbra]

Honestly everyone is looking for consistency with the extend to feature, while it's obviously not perfect and I have to do a couple rolls until I get something satisfactory I learned to embrace the ridiculousness of the ai glitches, obviously the niche I'm into allows for it but yea, I don't really mind a comedic glitch here and there between generations.

[lusty oracle]

the most important IMO is that the generation gets lip sync right. If consistency is required, you could always replace dialogue track with a voice clone from Elevenlabs or something.

[turbid umbra]

I feel like Veo4 will for sure address this in some way.

[lusty oracle]

I agree, just how close to actual consistency? I am not holding my breath, instead trying to work with the model, and accept that (serious) movie making still is a lot of work.

[edgy holly]

No.

[edgy holly]

You make 8 second clips, and extend them in Scenebuilder until you get to a minute. I made a few around a minute yesterday.

[lusty oracle]

are you generating something photorealistic? How is the scene consistency with multiple extensions?

[edgy holly]

Dreamy semi-realistic. If you want consistency, you have to do cuts regularly, starting over with reference imagery.

[late turtle]

hi! newbie here, just want to know if 8seconds is the maximum per video output or can I extend it more if I type in text?

[ashen mirage]

when I use flow to generate image，it often goes “something went error...”

still don't know the reason

😭

[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.

[edgy holly]

Each clip is 8 seconds. If you go into Scenebuilder you can extend a clip, making a 15 second clip (each 8 seconds, with around 1 second overlap).

[hallow plume]

Spot publicitario unidad educativa Jose Mariano Serrano Mañana, ubicada en el centro de la ciudad de Yacuiba, calles Comercio y Jacinto Delfin, tiene cupos disponibles para nivel inicial y primario, pronto el nivel inicial con su propia infraestructura, este año por promocionar el nivel inicial la Unidad Educativa regalará: uniformes y material educativo a los niños que se inscriban al nivel inicial

👍

[hidden latch]

You need to add details. Using the same detailed prompt gives same result. its great for character reference looking the same, you must be descriptive when prompting. Tell it what you want to see.. good luck!

[hidden latch]

Try not Using the extend..download in editor, get the last frame screen shot, change your prompt to Jason and chances improve 300 percent

Just don't let AI win, you can find work arounds

But I understand, what it produces sometime doesn't let you extend..

[lusty oracle]

frames to video, close up of a face, that is to be lighted with some kind of flickering and flashing orange light, where the light should come from behind the camera. As if the character is watching a burning building. Know what I mean? Whatever prompt I try it usually ends up with the characters head literally catching fire. Anyone?

[jagged marsh]

Same thing here. We're all just cattle guinea pigs to the AI overlords at Google, we eat when they say we can eat

[dire cypress]

hi everyone i am new to discord .

[crystal talon]

but this is new. it wasnt like this 2 days ago. with the same prompt it used to give me different faces everytime.

[hidden latch]

Well I'm been using veo3 for a couple years and my same detailed character prompt remains consistent... I change one detail and it changes everything

[crystal talon]

but i just prompt an asian woman. no specifics at all. this is weird man. it wasnt like this at my end really.

it is was like this, i would never have any character inconsistencies

[hidden latch]

Hmmmmm.. I do understand

[hidden latch]

Try the scene prompt: his face radiates with an orange glow from 10,000 watt orange painted flood light... it'll work

[tawny sequoia]

Is Flow capable of generating videos of people like for example, I have seen circulating AI videos of the Stranger Things cast. Are those videos generated through different AI Video engine/apps?

[zinc pewter]

Guys i really need help
I can't use nano banana
My athor accounts i can but this one is not
Even without loging in i was able to use it whats the deal?
What i do?

[zinc pewter]

Same man what i do?

[turbid lintel]

no news about what coming next ?

veo 3.5? veo 4 ?

[onyx linden]

Hello Google Flow Support Team,

I am using Google Flow with an Ultra account and the Create Picture – Nano Banana Pro feature.

Previously, my Ultra account allowed me to run up to 3 image tasks in a single execution. However, currently I am only able to create 1 image task, and I have not seen any notification about a plan change or new limitation.

Could you please clarify the following:

Has there been a recent change in task limits for Nano Banana Pro?

What is the current image task limit for Ultra accounts?

Is this behavior due to a temporary bug, system update, or a new policy?

Is there any additional setting or add-on required to enable multiple image tasks again?

Thank you for your support. I look forward to your clarification.

Best regards,

[edgy holly]

I think its 100 a day?

[onyx linden]

nope, my question is, the task same time

Previously, it was possible to run 3 image generation tasks simultaneously (each task generating 4 images).
However, now it only allows 1 task at a time (still 4 images per task).

[vast current]

Hi can someone tell me how many generations of nano banana pro do i have if I have PRO subscription?

[sick sparrow]

is NBP downloads down for anyone else?

[naive temple]

Any tips on how to make a realistic image to video haircut prompt? only thing i havent tried is json but i have a refernce image so kind of difficult

[lyric yacht]

When is it going to be available in Spain?

[rotund elm]

I noticed that when upscaling the download of Google Flow videos 3.1 to 1080, the quality of the image looks a bit more wrinkled on people's skins and other details than when you see the 720 version. Have people raised this before?

[teal tapir]

Every generation-is greeting failed in flow video and img both? Whyy? Happening with anyone else?

[visual quiver]

I'm getting a lot of "something went wrong" error on Flow - Nano banana pro last days.
What is problem?

[uneven garnet]

Hello! I’m UPZZY, a professional video editor helping creators and brands transform raw footage into high performing content that increases engagement, retention, and views.

[limber junco]

Hello!!
I created a project in Flow and generated about 40 videos. When I tried to put them into the Scene Builder, I got a temporary error. I refreshed the page, but then it said the project didn't exist, and when I went back, it was completely gone from the list... Does anyone know how to recover it?

[serene juniper]

Veo has become horrible lately, I give very specific prompts, just for it to spit out trash. This is on Veo 3.1 Quality. Literally wasted $50 trying to create one scene and ultimately unable to. I was not having this experience 6 months ago. Is it just me or has it gotten worse at actually following the prompt. So much clipping through objects and things floating.

[hazy acorn]

yeah, lots more failures on Veo in the last couple days, even on very basic stuff

[full flicker]

Hello, please post this in a thread in #1238621511401734254 . Staff regularly checks those threads and will responds there. These channels are not regularly monitored and your question may get lost in the chatter. Thanks.

[onyx linden]

i canot post in there, i dont find the box to coment

[toxic hound]

Hello, could someone please tell me how to make this kind of video?
https://vm.tiktok.com/ZNRh55SUp/

[onyx linden]

please post for me

[full flicker]

At the top of the page there is a box where you can search or create a new post. The blue box as shown in the image allows you to create a new post. Hope that helps.

[full flicker]

You are not in the correct channel. You need to be in #1238621511401734254 .

[onyx linden]

yeah, got it, thanks u

[tawny sequoia]

Anyone?

[turbid umbra]

Is anyone else having issues with the export?

[umbral thunder]

Hi @tawny sequoia!

Thanks for reaching out.

Users should not create videos in Google Flow using images of prominent people.

[mossy drift]

Same problem

[turbid umbra]

worked in the end for me

[mossy drift]

Did you do anything?

[tawny sequoia]

Too bad Flow has too much standards set. Some things get flagged easily. If I wanted to do a video of an ice cream melting, you would have to workaround the prompt so it won't violate policies.

[foggy frost]

I apologize ahead of time if this has been answered already - did the unlimited generations recently go away for the Ultra plan?

[turbid umbra]

no just tried a bit later

[mossy drift]

Thank you

[turbid umbra]

the low prio ones are still 0 credits

[autumn loom]

personnaly i've just been limited to my downloads one time for 10 mins maybe. Iv generated for 12Hours straight

Does someone have the prompcipee to make logos looks exactly the same as provided ?

[earnest walrus]

same as you

[pure venture]

Is there any limit to generate some image on flow?

[earnest walrus]

im having a lot of problems to make images with nano banana pro in flow

[pure venture]

Why?

Any specific problem?

[earnest walrus]

all the time

[crystal talon]

flow is horrible when it comes to their damn policies. everything is forbidden in there.

[crystal talon]

mostly trigger words. share ur prompt, i had this alot

[crystal talon]

you have to choose veo3.1 fast(lower priority) , are you ultra member?

[merry summit]

I am encountering an issue where trying to download my video just results in an "Export failed" error with no other explanation. Is this a server issue, or is this something on my end?

I seem to be able to download individual clips, but not the version from the Scenebuilder tab

[foggy frost]

Cool thanks for the response. I've tried the low priority prompts and they keep failing on me but i'm not clear why.

[foggy frost]

Hi thanks for the response - yes I am an ultra member. I mentioned to someone else that the low priority iterations have been failing on me.

[versed radish]

Hey friends 👋 I need a little help 🙏

Imagine I have one character image 🖼️
I want to make multiple short videos (about 2 min each) with her 🎥

What I really care about is:

Same character every time

Same voice every time 🎙️

Ability to keep building on the videos

Possibly different languages later 🌍

Bonus: her writing on a whiteboard while teaching ✍️

I’m using / considering Veo 3 + Flow and wondering:
👉 Should I use ElevenLabs or Google Studio voices for voice consistency?

Would love any advice 💡🔥

[crystal talon]

failing how? you mean slop or doesnt generate at all?

[cloud falcon]

so i generated 50 images with the notion that if you put no music only asrm that their would be no voice dialouge and now 50 prompts are waisted, voice was added even with the prompts , is like my ex wife does not listen!!!!!

[dusk magnet]

I keep getting this then it deletes the project what the hell is up with this 😭

[elfin lark]

Have you received recurring errors today that prevent video export?

[elfin lark]

same to me bro, recurring errors today that prevent video export?

[strange forum]

What’s with generating images today… it keeps declining me lol

[rose vigil]

I still have my subscription. what the hell is this?

[wide ether]

I see a new update just pushed out ... and it's absolutely AWFUL, just TERRIBLE! I had a set of prompts that were perfect for the types of videos I wanted, and now they've completely jumped the shark as far as the videos generated from them!! Almost 100% of the videos generated from the same prompts are now junk! 🤬

Completely trained the wrong way!

Looks like I have to spend all this extra time to craft different prompts to accomplish the videos I want generated again 🙁

I just wish you could put in negative prompting as part of your prompts because there's certain concepts that I absolutely do NOT want, and with this update it's putting that crap in everything.

I've been using the flag button --> inaccurate generation ..... for like every video generated ... does that help at all? Does that flag feature have any weight?

referring to this update: Commit v-release-candidate-01-08-26-ca5dd5527-260112170819

[fleet terrace]

The flow asset operation cannot be downloaded.
Is the server a problem?

Someone please answer quickly

Specifically, Scenebuilder cannot be downloaded

[wide ether]

refresh the page ... I had that happen as well, I refreshed the page and it worked

[fleet terrace]

I'll try, thank you.

The server is unstable today, right?

[fleet terrace]

The flow asset operation cannot be downloaded.
Is the server a problem?

i'm so busy

[fleet terrace]

???????????????????????

[turbid lintel]

what is this?

[analog radish]

Where is flow-announcements channel I miss it 😢

[crude fjord]

Why can't I export the scene?

Why can't I export the content from scene builder?

[unreal viper]

hey! id reccomend submiting a bug report in #1238621511401734254 !

[turbid umbra]

Again export failing for me too, I'm assuming its some peak hours that's why, would be great to have like a feature where we can see if the demand is high or not.

[turbid umbra]

Ok export is completely not working for me in EU, is something down?

[wide ether]

That's the new version number of the new release they just put out there. Completely broke all my prompts which now generate complete garbage! They've training certain words which are prominent in my prompts the wrong way IMO! I have to come up with completely new prompts to try to generate the types of videos I want.

[crisp basalt]

hello

the upsampled video feels very

or maybe placebo im having but it felt different

in a good way 🙂

[arctic kelp]

Does anyone have any information about Veo 4.0?

Hi everyone, does anyone know when Veo 4.0 will be released?

[digital parrot]

Good day, Google team. I just wanted to share this concern with you. I'm testing your resources with Flow and Veo 2 Fast. It generated the first two clips without any problems, but when I try to create the third clip using basically the same instructions, only changing the scene and actions, it asks me to switch to Veo 3 Quality. This consumes more credits, and I assume it's to test for better results, which I don't need at this stage of testing. I don't know if this is a technical or strategic issue, a way to test your new features, or to deplete credits faster and force a subscription to a higher plan. My current plan is sufficient since I had 180 credits accumulated, and using Veo 2 Fast gave incredible results. However, being forced to switch to Veo 3 prevents me from continuing with the project.

[obtuse vector]

hello, i noticed that almost all of my 1000 points were gone and i went to see my library and there was 40 fully black 8sec videos that i havent made. How can i get my missing points back?

[unreal viper]

hey, please use #1238621511401734254 to submit any bugs and feedback you have for the team!

[umbral thunder]

Hi @pure venture!
There is no limit on the number of images you can generate in Flow, but it is better to check how much credit you still have.
You find more information about credit management in this article.

[novel tendon]

flow doesn't seem to work rn

[digital parrot]

i can't post there anything because the summit feedback is disable for me 😭

[unreal viper]

hmm have you added a tag?

[digital parrot]

nou..!?

[wide ether]

all - regarding the export in scene builder, refresh that page first, then you can export.

[upbeat sierra]

please take a look at this thread for instructions on how to file bugs, the information makes it easier for us to investigate. thank you! #1452796616271597681 message

[digital parrot]

thanks

[arctic kelp]

Please help reply to my message.

[untold aspen]

Whenever I try to extend a video I get this message (Veo 3.1): Video could not be created. Try again later.

Does anyone have a solution?

[upbeat sierra]

hi @untold aspen sorry for the difficulty, we are looking into this, please use #1238621511401734254 to report bugs & feedback as this chat is not actively monitored.

[untold aspen]

Okay

[upbeat sierra]

thank you! we are doing our best to get everything up and running as soon as we can. appreciate your patience and feedback!

[upbeat sierra]

hi @arctic kelp which message are you referring to?

if it's in reference to your question about Veo 4, we have no information or updates at this time

[obtuse vector]

@upbeat sierra what can i do?

[dusk magnet]

Why does generating images on Flow work on my phone but not on my pc?

[crisp basalt]

4K ?!!

[zealous tusk]

video upscalling is not the strongest point on veo 3.1

you get overall extreme sharpness and loss of detail

[dusk magnet]

Can anyone explain why when I generate images on Flow through my laptop it says "something went wrong" and then "project deleted" after but If I try to generate through my phone it's fine?

[arctic kelp]

Thanks for the clarification! Good to know that the Veo 4.0 release today isn’t true.

[upbeat sierra]

hi @obtuse vector please use ⁠labs-bugs-feedback・📥 to report bugs & feedback

[crisp basalt]

4K Upscale is actually nicer than i thought dang

[paper ermine]

Bruh Nice, how do i get it tho?

[crisp basalt]

ultra users should be able to do the 4k upscale today

might need to wait for their announcement so stay tuned 👀

another try w/ 4k upscaler

[tawdry vapor]

Hello anyone here is having problem to generate videos in veo 3.1, every time shows error

[fiery wedge]

Me, had to try with a different account and browser and worked, but still can't export scenebuilder

[pure venture]

How to get our credit back to 100?

[tepid drum]

Is it flow update today for new VEO model?

[sand zealot]

For Flow, is "Get more AI credits "only in the US? I cannot see the option from Japan. 😭

[arctic kelp]

no

[tawny flare]

Hi, I can't export video on Flow

[jaunty stratus]

Flow Update ?

[turbid lintel]

new update!!!

[edgy holly]

Hopefully this new upscaling is a real upscaler.

[edgy holly]

no

[turbid umbra]

omg, this is bonkers, topazlabs is done

[edgy holly]

Try leaving and coming back into Scenebuilder. It's super buggy, but sometimes just leaving and coming back clears out some of the bugs.

[edgy holly]

Haven't tried it yet. Good?

[tardy ridge]

Does green screen prompting still work I saw the update in Gemini

[turbid umbra]

the upscaler from topazlabs is amazing, but costs like 20 dollars per 5 min video

[turbid lintel]

veo follow sora route

[edgy holly]

No, I'm asking if this new Veo upscaler is good.

[turbid umbra]

oh, havent tried yet, was just excited

[modest lotus]

this update is soooo good

thanks google

[pine radish]

hi! are you having issues generating image to video in flow?

[craggy tiger]

Can i use the AI with gemini subscription?

Like the Video AI? Im confused which one is the right one etc. 🥀😂

[loud panther]

When will all these updates be shown on the website, I just tried but we did not see these updates.

[crisp basalt]

@edgy holly 👀

[fair gull]

I'll have to see how the 4k upscaler compares to Topaz Video (desktop). Thus far, I've been downloading my videos at 720P and upscaling them in the Topaz Application because 1080p looks bad.

[pine radish]

I am having this error with every video today... "Audio generation failed. Please try a different prompt or send feedback. You have not been charged for this generation"

Someone have the same issue?

[edgy holly]

Does this mean the ingredients will stick between generations with ingredients now when extending in Scenebuilder, or is this in development and coming later?

[crisp basalt]

I think that’s already been a thing , is it?

[loud panther]

Yes
Flow takes no credit for this.

[fair gull]

It means the dialogue you are trying to enter is too short for Flow to work with.

[loud panther]

When will all these updates be shown on the website, I just tried but we did not see these updates.

[loud panther]

Right 👍

[edgy holly]

I meant in Scenebuilder when extending. (edited my question for clarity)

[crisp basalt]

Oh got it

[pine radish]

Oh! I gonna try. I though simple prompt or dialogue should be fine for Flow. Thanks

[edgy holly]

Can you share the prompt?

[edgy holly]

Rollouts tend to take a day, sometimes two.

[main saddle]

this is currently rolling out, and you should have access soon if you don't already!

[main saddle]

if you continue to have troubles generating, feel free to add a bug report to #1238621511401734254 or share your prompt here!

[atomic plume]

You should be able to see all the new features now! Let us know if you can't see them still.

[main saddle]

try now 🙂 we're fully rolled out 🔥

[wide ether]

@main saddle with this latest update, has there been ongoing training with the veo 3.1 model such that prompts that used to generate a certain concept are changed due to extra training to bake in biases towards other concepts?

[modest lotus]

hey dont you think 50 credits for 4k is to mush ?

@main saddle

[tepid geyser]

Previous 1080p versions always looked way too forced and sharp, so I’d been sticking to 720p and doing my own upscaling. This new 4K is finally a viable alternative.

I've also noticed that the ingred generation quality feels a bit more refined lately

even if it's still labeled as version 3.1, something definitely feels improved.

[modest lotus]

as someone who make veo 3 videos full time i can say yes a lot has changed

[turbid lintel]

this latest update improve the physic

[proper drum]

This update include better consistency between start and end frame in frames to video?

[turbid lintel]

the sloppiness still there but this is really good

big improvement

[turbid umbra]

I can't from Romania

got the ultra

did you guys have to rrelogin to get the new features?

[turbid lintel]

it already implement

everyone can use

[turbid umbra]

I cannot see anything new 😭

[gloomy verge]

Hello. I first try Flow. When I have image with 2 characters and I am writing, what I want on video, (Example: characters are talking), can I writ, what characters will be talking? How words? Or it is random?

[prisma ferry]

is it possible to generate in 1080p/4K could this be a future update to avoid the upscaling look? great new addition though seems in line with Topaz and other video upscalers on first look

[atomic plume]

Try refreshing to see if that helps!

[edgy holly]

Not sure if I reloaded, but it wasn't there earlier, then I tried again about an hour ago and it was.

[turbid umbra]

it worked thanks!

[edgy holly]

Every time one of these big updates happen, they take time to roll out to everyone.

Feels like they're getting faster, though. I think some early last year took like 3 days. Now it's usually one day.

[upbeat sierra]

Our 1080p Upscaler has also been updated with improved quality!

[turbid umbra]

any plans on lower the credits for veo 3.1 quali?

people say there's no difference but there is a clear one, I love 3.1 quality

[zealous tusk]

Is the new upscaller better than davinci resolve Ai superscale? ayone has tried this?

too much extra sharpness with the upscalling

it would be nice to control the sharpness and noise reduction

[edgy holly]

I've only done a couple tests, but it looks like the upscaling does a good job of smoothing out light pixelation I get a lot in 720p, and doesn't do that harsh highlight thing it did before. Pretty much all I wanted, for free.

[shut elbow]

Hi, does anyone also keep getting "export failed" errors when exporting to video tonight? I have a project that was half made before and half made after today's update, and seemingly I keep failing to export to video. Any ideas how to fix?

[barren mulch]

Yes, just say character x says happily in English: „hello“ to which character y responds confused in German: „ Guten Tag!“ or whatever

[barren mulch]

Really? Quality consistently gives me much worse results than fast. I never use it unless I am forced to for „jump to“ for example

[gloomy verge]

Thank you.

[turbid umbra]

One question though: If I have a full scene built up, can I upscale the whole scene to say 1080? or do I have to go to each individual video and do it and then build the scene again?

[celest mortar]

Hello! New and looking for patch notes? When I went into Flow today I saw an update but I accidentally clicked off before I could read it!

Was wondering if there was anywhere here I could see them? Or if anyone knew where to see them on the site?

[edgy holly]

Scroll up to earlier today. Wrigley shared a screenshot.

I tried to link to it, but links aren't kosher in here, even if they're to other messages in here, haha

[celest mortar]

Thanks so much!

[foggy frost]

I'll switch from fast to slower priority and it fails to generate. If I switch back to fast it works just fine.

[rough halo]

whats the quickest way to download all my frames that ive created /uploaded to flow? Whisk has the download all images option. Please tell me I dont have to donwload each one individually.

fit thats the only way it'll mean ill have to spend hours and hours doing it

I'm trying to organize my local library of frames, since Flow doesnt provide a centralized interface where all my created/uploaded images can be found. No unified album to my knowledge. And to download an image from the prompter is also a logistical nightmare because one I exit out of the image I want to download, the prompter resets and I have to reopen the images and scroll and scroll till i find where i was and go from there.

I'd like to think that something this basic is available but I just dont know how to find it. OpenAI and grok both have this. Whisk also... Is there a reason why something like this isnt available>

[uneven matrix]

You all got to be kidding me with this new update!!! you all took away venom she venom and other comic book characters!

where is support?

[stuck adder]

Really??

[uneven matrix]

and you also stole everything every comic book person ever used on the app!

[rough halo]

Does anyone else notice how the model itself has gotten dumber in the past 2 months or so?

[uneven matrix]

did u notice now they have gaurdrails all of a sudden after they made 100's of millions of dollars now they dont care about people who used characters like venom like myself

[edgy holly]

Disney/Marvel signed a deal with Open AI, so Marvel characters aren't supposed to be generated with other models.

[rough halo]

yeah but the model itself is also very poor. Im done with google. im processing a refund. even gemini 3 has become to bad

going to use fal.ai and grok

[stuck adder]

The 1080p upscaler seem ok, but it’s still way too sharp

What back end is it using? Topaz?

[analog lava]

Hello @violet inlet , 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.

[uneven matrix]

when 6 months ago or yesterday? because we been using them for almost a year now

[edgy holly]

Last month.

[uneven matrix]

and they just implied it today

because i been using them up until today

[edgy holly]

These models aren't simple machines. This is the first major update since.

[uneven matrix]

such trash man i spent a mess of money over 6 months doing this

[edgy holly]

Veo has it's issues, but this is on Disney. If you want to make stuff with Marvel characters, you're stuck with Sora, or probably some of the open source models.

[exotic shell]

is there a way to specify a font when trying to add text to a banana image generation? i can just add text but the font is badly? or randomly picked?

[rotund elm]

I just did a test. The 1080 upscale option is still looking terrible. It over-sharps people skins making them look wrinkled. However, it looks like the 4k option gives better results.

[paper ermine]

The noises in the video is insane

[rotund elm]

I am not seeing a 1080 (New) option. The 1080 upscaler is still looking bad.

[regal raptor]

Holly crap 1080p upscale no longer looks like complete dogsht, great job Veo team! Its still not great at like 6.5 or 7/10 but massiveeee improvement over the absolute crap we were getting before

[pallid aspen]

anyone facing downloading issue or only for me?

[pallid aspen]

this works beter?

[ruby roost]

4k is a nice gift 🙂 thank you

[pure venture]

It's a good update

[stuck adder]

How is the 4k upscale

Since by that one costs credits

[exotic gulch]

yes i am getting it as well. this is so annoying

[alpine gull]

Anyone tried the portrait mode for ads yet or something people speaking content in flow. will try afteroon but very curious

[inland saffron]

یہ لڑکی ہنسنے لگی

[surreal phoenix]

has anybody heard about this platform: https://veogen.studio/
I found out about it while doing some research and it looks like it is still under development, but if it does offer whatever is promised on the site, it would be a very good cheap alternative to google flows since it offers cheap video generations with veo 3 model

[topaz haven]

Please, who else is still having issues with flow generating videos.
Mine is not generating, it only counts to 99% and stops there when I tell it to generate videos.
Please, is there a way I can pass this, or what could be the possible solution.

[patent narwhal]

Hey everyone.. Are you guys also facing issues generating videos with prominent person image?

[fleet terrace]

Export failed. Same error has been occurring for 3 days. You are not receiving service for the product you paid for.

👎

Does anyone have the same symptoms as me?

It's not cheap, why is that?

I even have two accounts.
It's really not good. There's no answer

[patent narwhal]

I have been facing this problem since yesterday.
Is this a temporary restriction? Or will it be permanent?

[split glacier]

Ugh

do we know anything about Veo 4?

[warped wind]

Same here!

[knotty viper]

same

[turbid umbra]

getting as well many export fails it works eventually but yea its kind of annoying

[pulsar cliff]

what do you think... how can i make it more realistic?

[spiral python]

Too many bugs

Scenebuilder not working

[fleet terrace]

same

[edgy holly]

Copyrighted characters and famous people arent legal. The filters arent perfect, but theyve been getting better

[serene cradle]

Hi is Nano Banana in Flow down ? Try to make Image but its not working

[patent arrow]

Same.

[surreal ruin]

Flow 1080p Upscale is not taking place after clicking. Nothing is happening after clicking that option.

[patent arrow]

I add "Use Comic Sans or Comic font in my prompt. Comic Sans is consistent. But Comic font will give you varied results.

[quiet kindle]

How am I supposed to download this in 1080p? I clicked the option, but nothing happens. Has anyone actually managed to download it?

[sick sparrow]

one was made in 4k flow. Can you guess which one 👀?

[tawny flare]

The error persists even after I've logged out and logged back in multiple times. Has anyone else experienced this problem?

[quiet kindle]

fam, 1080p working??

not working for me, How can I download it in 1080p?

[marsh finch]

yes ive been unable to do an export from flow today so have to go back and download the videos individualy which isnt that much of a hassle as when you scroll over you can see whats been added to the scene

[unique abyss]

I thought it might be helpful for some people to able to compare the new 4K upscaling to what we could do before this new feature drop. So I uploading threes shots that I made last week, with the old download and the new 4K outputs. You can find them in the flow-gallery.

Interested to hear and see what other people's experiences are.

[marsh finch]

make sure not to close the window after you do it and dont try to generate anything else or download anythign else in the couple min it takes. if its still not workign try a different browser

[marsh finch]

looks much better in 4k, thanks for the share

[surreal ruin]

It's not working on today's generated videos.

Someone please inform google to fix the 1080p upscaling issue. Can't upscale and it's creating delay in my workflow.

[chrome imp]

Yes, it just takes some time, but is reall quick.

[turbid umbra]

Can you upscale all the clips from a scenebuilder and export the whole thing together?

[pseudo fossil]

I'm learning video editing and I'm having trouble syncing characters. Can anyone help me solve this?

[surreal ruin]

Any support admin here?

[exotic shell]

But then you already make the AI add text by itself?

I noticed that the image itself many times is quite good but the text is positioned badly or the font is trash.
But when i do not add the text generation into the prompt and press edit instead, where i can add text, the font is random.
Or do you just add the fact to use specific font but do not add text?

[tepid geyser]

Since the recent update, I’ve seen a slight dip in generation speeds for Nano Banana Pro. To help manage the load, perhaps the team could consider a tiered priority system? I’d be happy to use credits for faster turnarounds while keeping a zero credit option for lower priority generations, similar to how Veo handles its queue.

[glad condor]

Is 4k upscale on the scene builder? I haven't had a chance to look just curious

[tough otter]

anyone can generate image or videos？

[tough otter]

Broke again？

[rose gorge]

@tough otter At first, I did not notice this issue. After checking more carefully, I realized that I had attempted to generate videos approximately 10 times, all of which failed, yet the credits were deducted each time without any refund

[ashen sequoia]

How to make videos in free

[tepid geyser]

Image generation is being pretty hit or miss for me right now (mostly miss!), but strangely enough, video generation is working fine.

[tough otter]

Yeah. Me too. Cant generate images but video is ok

[upbeat sierra]

@surreal ruin sorry to hear, going forward best practice is to post bugs/errors in #1238621511401734254 as this chat isn't actively monitored. however, in this case, can you share your error message / where you are trying to upscale from?

[edgy holly]

not yet, no

[patent arrow]

I'm not very good at making prompts, but I can somehow make it work for me. 😂 I just do, Use Comic Sans, (where I want the text to show up): "Hello there!"

[patent arrow]

This is what I wrote >> Plain white image. Use Comic Sans, bottom right: "Hello there!"

[stuck adder]

The left one

[marsh finch]

Overlaying the dead center is large, heavy, bold, white sans-serif text that reads exactly: "PROVEN". 4k broadcast quality, elegant business atmosphere.....

ive used that prompt in a lot of videos and its alwasys simialr

[sick sparrow]

That’s correct 💯

[exotic shell]

In that case you would have to fully describe the generated picture in detail. else important objects / persons might be right underneath the text. i mostly do only tell the prompt to generate xyz in the picture, not how its arranged.

[patent arrow]

Hmm. I've never done that. I just tell the AI where I want the text to be. If it's a picture of a horse, I'd just say I want the text above the horse's head, blah blah, and it would appear there.

[surreal ruin]

@upbeat sierra This is how I am trying to upscale. But the 1080p one is not working. After clicking no response.

[marsh finch]

and you tried clearing your cache? if it worked yesterday only clear it for 24 hours and see if it works

if not no idea, spent 3months using the ultra service which is half off first 3m. with all the issues it still makes sense at that price point but in a couple hours it renews at double the price... that does not make sense for all the issues i experienced. so ive cancelled and will find a more affordable service as mods/support can not offer any sort of retention to keep me here. the last 3 months has been hit and miss and a lot of time wasted and not offering any sort of compensation or at least one more month at the 50% off seems like bad business. anyway last comment here, wish you all the best in your generations.

[edgy holly]

No idea if this is the case, but we saw the same issue yesterday as the upscale for individual clips was rolling out... I'm wishing/hoping this is them shutting it down so they can update it so the upscaler works with multi-clip scenes in Scenebuilder.

[exotic shell]

okay, in 70% veo also does a good job in putting it where i want. but sometimes it still places it above the object or in an unaesthetic position. even if its not covering the object, it could still look ugly. on a thumbnail for example aestetics are important and i wouldnt want the text in the left bottom corner

[upbeat sierra]

I saw the thread you started in #1238621511401734254 followed up there!

[calm hollow]

its just me or high quality 100 points worth are actually worse quality than when I ask gemini to create video

[valid drift]

I created a 1:50 video in Google Flow using SceneBuilder, but exporting or downloading the final timeline gives an error. Individual clips exist, but the combined SceneBuilder export fails.
Has anyone encountered this or found a workaround?

[unreal viper]

please dm @real wyvern thanks

[icy sleet]

I have exactly the same problem, it won't let me download the video

[pulsar ferry]

@upbeat sierra My flow is not loading the videos and the images it stops at 99% when loading idk why.

[hard oasis]

When creating a video in Flow, I get the error "Something went wrong." What should I do?

[calm hollow]

can someone help me

[edgy holly]

I don't find either to be always better or worse, but there does seem to be a difference in how Gemini and Flow interpret prompts. Must be using different prompt expanders/interpreters, or some such.

[calm hollow]

like compare these two

[edgy holly]

Two generations don't tell me anything. Every generation has an element of randomness. That's how AI works. To find a pattern, you'd have to do some sort of A/B test where you run the exact same prompt hundreds of times on both and look for patterns.

[calm hollow]

but which one is better for you

they both used the same prompt

[edgy holly]

If I had to really guess, and this really is a stretch, I'd say that Gemini does more hand holding in prompting. I usually get better results in Flow with in-depth prompts.

[edgy holly]

Use the same prompt in both, and each generation will be different.

[calm hollow]

I wish I could ask google staff about it

[upbeat sierra]

Hi Chicken! Please report your bug in #1238621511401734254 ! This channel is not actively monitored

[unreal viper]

please use #flow-videos・📹 for all videos made with flow!

[wicked shadow]

thank you

[scarlet saffron]

Hi guys did you see that horrible fake 1080p upscale??

Why is it still so hard to get proper 1080p AI video features?
Veo 3.1 has the best understanding and somehow the worst visual quality.
Expensive but unusable in practice.
@main saddle @tacit halo

[dull haven]

Hey everyone, is anyone else having issues with Flow right now?
I’ve been trying to generate videos for the past 2–3 hours and nothing works. Sometimes I get an audio generation failure, and other times a video generation failure.
I’ve tried different prompts, video-only prompts, logging out and back in, but it’s still not generating anything.
Just wanted to check if this is happening to others too or if it’s only on my side.

[regal raptor]

Capcut has an absolutely amazing ai upscaler to 1080 on their pro plan, but it only runs on 10 second vids max and takes like 10 min a generation (on their cloud servers, your comp hardware wont help). I do 15 min videos and was tempted to upscale every slide using it but its just time prohibitive for me vs flows upscaler since their update is like 6.5-7/10 quality its just usable enough to keep me from adding a second workflow...just

[scarlet saffron]

You cannot create artsy short movies with this visual quality.

[regal raptor]

ill dm you and example between the two, you can decide if its worth the extra workflow...just looking at it now and im back on the fence lol

[molten grotto]

Since the update, when using the “Ingredients to Video” tool, the model does not properly follow the reference images anymore. No matter what reference images are uploaded, the generated video changes the characters and environments almost completely, ignoring important visual details.

Before the update, the tool respected the reference images much better.
Now, even when uploading high-quality 4K images with clear details, the output does not match the provided references.

This makes it very difficult to maintain character consistency and visual continuity in projects.

I believe this behavior started specifically after the most recent update, since it was not happening before.

[full flicker]

Hi there, this is best to be reported in #1238621511401734254 . Those threads are regularly monitored and won't get lost in the chat here.

[exotic gulch]

is anyone having trouble exporting video files?

[exotic stump]

When creating a video in Flow, I get the

[mental wigeon]

Hello

[obsidian brook]

Yes! Came here for this!

trying to download a scene from the scenebuilder

[fickle glade]

Hello, Does anyone know how to get Flow to download created videos? I am trying all day long and it is not happening

[sand zealot]

same here. I cannot download since last night... Hope they fix asap

[smoky gull]

really? only 3 video a day on flow after paying 300$ x year? O.o

[noble carbon]

same cant export

[full flicker]

A friendly reminder for anyone having technical difficulties with flow, please open a thread in #1238621511401734254. It is less likely for your questions to get missed and staff monitors those threads. Please be as specific as possible with your issues, including screenshots where possible. Thanks.

[hazy acorn]

hey - upscaling is fixed now - nice job @tacit halo !!! 👏

[fleet terrace]

I am using 2 accounts.

[hushed geyser]

Hi Google Labs team,

I think I’ve run into a bug with the third-party content guardrails.

No matter what prompt I use, the system always shows this message:

“You cannot generate a response to this prompt due to Google’s guardrails related to third-party content.”

The image I uploaded is fully created by me and does not include or reference any copyrighted characters, brands, or third-party IP. I also tested multiple neutral prompts, but the result is always the same.

This feels like a false positive, where original user-generated content is being incorrectly flagged as third-party content.

Thanks for looking into this!

[fleet terrace]

No one is looking. Very slow to respond and not accurate.

[tacit aspen]

I know about the upscale update but audio/dialog and music seems to be amped up. Anyone notice this on 3.1?

[patent oxide]

Are there any users from China?

[austere quarry]

hi, always something went wrong

can someone help?

[fleet terrace]

It's just a flow error.
Give it up. No one's helping.

[austere quarry]

🙁

[fleet terrace]

true.

[carmine bolt]

Hi guys. I'm facing the same issue exporting the video, do you know if the credits will be refund? If I go out from the project the media generated will lost 🙁

[boreal canyon]

I am having the same issue with frame to video

[marble delta]

I don't have permission to add a topic to the flow-feedback tag. I keep receiving calls from you via email and Discord messages (which I find very annoying), but I don't have access to the actual channels. Feedback Labs, for example, and before any expert community volunteer explains, the statement was rhetorical. Warmly FF

[cursive dome]

cutting it make vidio animate

[solid cradle]

Anyone having the export problem on the Extend feature?

It's been like this for at least 2 days

[surreal ruin]

Is google going to fix 1080p upscale option or we just wasting our money and credits? Tired with this.

[fleet terrace]

It's just very upsetting. I'm the same. I'm being damaged sales. I also have 2 accounts.

[sick gulch]

Just started using Flow with an Individual Google AI Pro subscription but all videos are being generated at 720p with only a 1080p upscale option.

How do I generate videos at native 1080p? I’m using Frames to Video, my source image is at least that
resolution or higher, my prompts specifically ask for 1080p, and options are set to Gemini 3.1 at high quality.

I already have access to a lot of AI creation tools through both Adobe and Envato. I joined Google AI Pro specifically because I thought that would unlock larger resolution video generations. Am I doing something wrong or is there another button or preference I must select somewhere?

Also: I seem unable to create a post in the Flow Bugs section saying I do not have permissions though It will allow me to create one in the general Google Labs bug section?

[pallid violet]

Hey folks, if you notice any issues with Flow, please open a bug report here: #1238621511401734254 with all the details on how to reproduce said issue!

[turbid umbra]

there's no native 1080p for now, only upscale version.

[sly sparrow]

Hi, Ive been using Flow for a month or so now and while im loving the general user experience one thing I would love is to be able to make bookmarks inside of a project as I go, so I can go back to that specific generation at a later date if I want to re use a prompt or download at a higher resolution.

Also being able to have a separate place to save the best prompts along with the output it creates. I have other places I store these but feels like a feature that would be nice to easily be able to add these to a space without having to copy and paste things over from one place to another...

[wraith hornet]

again!!!??

[sick gulch]

I guess Google AI hallucinated when I was researching this. It told me the following 😂

“Google's Veo 3.1 model, which powers video generation on some platforms, can generate videos in native 1080p resolution. 
Users can specify the desired resolution, including 720p, 1080p, or 4K, when generating videos. The 1080p resolution has been improved in the latest update to offer a sharper, cleaner video ideal for editing workflows and general use. “

Is Veo 3 part of Flow or a separate service? All these different models are quite confusing to me. But I just hope they get native 1080p available in Flow soon as for my work anything less just isn’t fully viable.

[sly sparrow]

I can download at upscaled 4K, maybe thats what its talking about? downloading these though does cost credits

[sly sparrow]

Also another piece of feedback... Veo accepts json prompts but when you look at a generations input text it will just format the whole prompt to one paragraph. Would be great if it could keep the paragraphing/structure of the original prompt

[surreal ruin]

For God's sake, any support admin here who can inform the engineers or technical team to fix the 1080p upscale issue.

[quiet kindle]

plz fix this issue

[upbeat sierra]

Sorry for the trouble, as mentioned above it's best to use #1238621511401734254 for reporting bugs / progress, as this channel is not actively monitored. We have an open thread here -> https://discord.com/channels/1172568727942860810/1461040683342368953

[rain crane]

image gen down? not working at all for me

[peak hatch]

yes

[unreal basalt]

is back

[rain crane]

k thx

[weary flume]

Ultra-realistic cinematic drone shot of Burj Khalifa at sunrise, camera slowly moving upward, soft golden light, Dubai skyline glowing, calm luxury atmosphere, smooth motion, 4K, vertical video

[upbeat sierra]

Very cool! Best practice is to share your creations in #flow-videos・📹

[unkempt tide]

Does Flow has API?

I want to automate creating images

[pulsar ferry]

Ok thank u!

[tacit aspen]

Just asking this one again for clarity -- did something change on the back end of the model. Veo 3 (Fast) is behaving more like Sora 2 adding dialog and ambiant music to finish scenes. Even if I ask for {no dialog, no music} the characters are still talking and not scripted. Thanks if anyone noticed or its just me?

[turbid lintel]

new update enhance the generation

[tacit aspen]

I'm trying to figure out how slightly enhance it down... very cool. Appreciate the response.

[sleek cargo]

Hi everyone! I’m working on a project using Google Labs video tools to create content for a trade show. The videos will be displayed on a large LED wall (between 3 to 5 meters).

My question is: What is the maximum export resolution available? Also, would the current export format/bitrate hold up for such a large scale without losing too much quality, or do you recommend any specific upscaling workflow?

[edgy holly]

Flow is an app, not a model. Veo and Nano Banana are a models. and are on the Vertex API.

[edgy holly]

It just got an update two days ago, yeah.

[unkempt tide]

I can choose like creating 4 images on Flow but how do I do it on Nano Banana?

[edgy holly]

Flow is an app. Nano Banana is a model.

[unkempt tide]

Yea I can read, not helping

[edgy holly]

...you can't do anything on Nano Banana. It's not an app. It's a model. You find an app that uses it that allows for 4 images and run it that way.

[atomic plume]

You can generate multiple images using Nano Banana on Flow using the "Create Image" dropdown option

[unkempt tide]

Yea but I want to try using API, can't find how to get multiple outputs

[hazy acorn]

Image generation stopped working for me today - is anyone else having issues?

[main saddle]

what kind of error message are you seeing?

[clever sparrow]

Yeah same, i keep getting this error since yesterday afternoon

[hazy acorn]

ok now it is working again. same prompts.

[slim sparrow]

flow not working

[edgy holly]

You can get as many as 14 generations at a time in the API.

[unkempt tide]

How

[edgy holly]

My bad. it's 10 per prompt. 14 is the input limit per.

How: it's a setting, like all the other settings.

Or parameter, more specifically.

[dusk patio]

Is it possible to specify the length of the video to be shorter than 8 seconds?

[dull haven]

Same for me, since yesterday afternoon,

[glad condor]

When I try upscale it says I need more credits but I have loads of credits 🤔

[exotic shell]

is there any trick with extending videos? it has never worked for me properly. havent used it in several months and gave it a try again. when i extend, the visuals seem to be mostly okay but the audio is extremely bad. in video 1 it sounds good, in video 2 (the extended video) it sounds like somebody is talking into a can. i produced 4 extensions, in all 4 the same bad audio.
i then copied the prompt from video 1 and just edited the text the person should say and it again has very bad audio.
saving a frame to assets and extending from that frame also didnt help.

[full flicker]

Hello @robust pawn as stated in #start-here・📍 , “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now. Thank you!

[violet temple]

Anyone knows how to make dance short video with flow, I try many prompts, but they don't work

[grand palm]

Hello, is there a prompt so that my hands don't triple, I don't get a tablet out of nowhere?

[obtuse marlin]

A realistic cinematic video
where the man from the first image starts walking forward naturally toward aparked luxury SUV, opens the car door and smoothly sitsinside
the driver seat

[grand palm]

I make a video with a product in my hand and it comes out very strange with music in the background, it plays slowly, third hands

[obtuse marlin]

A realistic cinematic videowhere the man from the first image starts walking forward naturally towa

[grand palm]

Hi! I’m trying to generate a product video in Flow, but the output keeps glitching. Extra hands appear (third hands), the product/tablet pops into the scene out of nowhere, and random objects show up that don’t make sense. I need the video to be consistent and realistic, with stable hands/objects and no sudden additions or deformations.

[upbeat sierra]

not at this time. what lengths would be useful for you?

[glossy agate]

I'm currently unable to generate videos using any prompts; it always fails. What could be the problem?

[upbeat sierra]

where are you located, you may be impacted by our recent outage. #labs-announcements・📣 message

[glossy agate]

New York

This problem has been going on for several days. Sometimes I can generate videos, but after generating twenty or thirty videos, I can't generate any more. It might recover after a day, or it might not.

[upbeat sierra]

sorry to hear, please post in #1238621511401734254 if possible share your prompts and steps to reproduce and the team will look into it

[violet temple]

Buddy, do u have any good prompt to make a dancing short video?

[upbeat sierra]

not off the top of my head, but Gemini is great at helping make prompts!

[dusk magnet]

Is there a way to fix quality? No matter what I do or how many prompts I tell to fix the image quality it feels like the quality reduces with each image generation.

[upbeat sierra]

are you editing images and regenerating them? or trying to generate new images? i'm not fully sure how to resolve your issue without more details. please post in #1238621511401734254

if you need help formulating prompts, as mentioned above, Gemini is quite good at helping you expand and think through more detailed prompts to get high quality outputs

[stuck adder]

Is veo updating anytime soon?

[violet temple]

Is it possible to build up a group which always provide dancing prompts?

[worthy flame]

#1319423780057251921

flow-general.

[arctic kelp]

We constantly see overload warnings, yet new features like 1K and 4K upscaling keep getting released.
This feels inconsistent — if the system is already under heavy load, prioritizing higher resource-intensive features doesn’t seem reasonable.

Instead, the system should be optimized first, before pushing it into overload and causing widespread failures.

[inland pier]

Currently we do not support that, but there is a possibility of doing that in the future (e.g. 4 sec, 6 sec)

[north wigeon]

I cannot get flow to generate anything using Veo 3.1 fast. Anyone else experiencing this? i have credits i have to use. Not being able to generate anything when im a ultra user feels terrible.

[magic notch]

Same, it's been three days. ultra user. terrible experience. Always "Something went wrong" for all models.

Probably gonna cancel subscription if this keeps happening

[fleet terrace]

I couldn't download it until day 3, and I can't expand the asset from today. Builders can't.

Also, the AI model seems to have changed. It seems to change the standard of the image when expanding the video.

It doesn't matter if the AI model changes. The technical part needs to be updated.
However, you should clearly let us know that basic functions such as download and feature extension are broken.

[sick sparrow]

if i remember correctly, 4 months ago i was able to drag and drop image into flow as long as i selected the prompt box. Was this feature removed?

[hardy dove]

Hi everyone, can someone guide me on how to get a subscription for Flow Ultra Veo 3?

[proven river]

Hi everyone, can someone guide me on how to get a subscription for Flow Ultra Veo 3?

[sick sparrow]

also ingredients to videos give a lot of strange hallucinations. output is totally different compared to input images.

[crystal talon]

hey google, can you please be less strict? i cant even upload a woman in a bikini. come on man.

and why cant i post sth on the flow-bugs-feedback channel?

[potent ether]

guys how can i get more creditsss i have student plann aand my flow credit finished

:(( i was making trailer for final term :(((

oww i cant take no more :/ i found it

i can t be a cristopher nolan no more

[full flicker]

Hi, all reports are now centralized in #1238621511401734254 . You can post there. Hope this helps!

[tribal solstice]

I've switched to https:://creaa.ai and can use veo3.1 and nanobanana pro without limits!

[potent ether]

thank you ms

[upbeat sierra]

this channel has been deprecated, per the pinned announcment. please file bugs at #1238621511401734254 going forward

[glass ginkgo]

Guys, is anyone else having this problem?

[lunar berry]

I have the same message, but everything came through just fine so far

[glass ginkgo]

Okay, thanks bro.

[prisma ferry]

Anyone else finding ingredients to video has reached a new level of quality? Different league, did I miss an announcement?

[icy grotto]

is this right 10 credit per veo 3.1 fast render?

[lilac frost]

Anyone else getting hit with this error on Veo3 today? It’s blocking almost all my prompts.

[turbid umbra]

to be expected I think, during weekend everyone wants to try it.

[analog radish]

Yes

[exotic shell]

is there any trick with extending videos? it has never worked for me properly. havent used it in several months and gave it a try again. when i extend, the visuals seem to be mostly okay but the audio is extremely bad. in video 1 it sounds good, in video 2 (the extended video) it sounds like somebody is talking into a can. i produced 4 extensions, in all 4 the same bad audio.
i then copied the prompt from video 1 and just edited the text the person should say and it again has very bad audio.
saving a frame to assets and extending from that frame also didnt help.

[tacit jackal]

e are making a movie using veo https://youtube.com/shorts/f2_72qXB-aE?si=dNOtpjhZLkLqfefL

[tacit jackal]

this is one of the easier AI watermarks to work around. at least its not moving

[dry goblet]

some of my generated video disappeard in my flow account after relauching Chrome, any idea of what is going on?

[ivory niche]

Hello All! Has anyone shared how to keep the voice consistent across several scenes/projects without using another tool such as ElevenLabs? Is there a specific prompt that can be used? I'm creating an AI Series with 10 episodes and want to keep the voices consistent. It would be great if Flow would allow users to upload a short audio clip as a hero asset when using "Ingredients to Video"

[lusty oracle]

Unfortunately, you cannot control voice consistency, perhaps a little if you instruct some accent or dialect. Also, if you go with the voices Veo gives you, your characters will sound pretty much the same as in all the other Veo generated videos out there.

[obtuse wyvern]

Hi

[candid relic]

I have to wonder why I am paying for ultimate when easily 50% of my requests fail.

[candid relic]

Interestingly enough, even eleven labls lacks tonal continuity. I often find the volume and processing to vary in between "scenes". Now, I am a bit of an audiophile and a former broadcaster and V/O actor and the only way I have found so far is by recording and overdubbing my processed audio. I have no doubt that technoplogy will improve over time though. Just a matter of waiting.

[iron quest]

Your best bet is to take the generated video and plug it into ElevenLabs to change the voice to whatever voice you want it to be, then replace it in a video editor

[lusty oracle]

Cool background! Yes, we need to wait for more realism and better consistency. IMO the art of generating something with Flow today, is to find some niche or concept with a generation success rate not terrible.

[tacit aspen]

This new enhance mode is making characters super excited/jumpy and moving around quite a bit. They are excited I guess. So, has anyone had any success prompting {calm delivery, natural movements] ?

[slender needle]

Managed to stick hundreds 8 seconds VEO 3.1 clips together to created longest veo 3.1 based feature film. Currently in Hindi with some english. Used real life live action actress with AI actors. Love triangle. Total length 2 hours 16 minutes. Continuous characters and locations maintained throughout with narrated voice overs and 7 songs. AI can do emotions. This is a love triangle. Probably the longest VEO 3.1 video ever. Have mastered the prompting structure for continuous locations including multi angle train, plane, cruise scenes

[wind fog]

I'm having trouble exporting in Flow. I've been stuck for six hours and it keeps giving me the same error. Does anyone know how to fix this?

[tacit aspen]

Thats pretty amazing -- had this new enhance update improved your workflow?

[slender needle]

I actually did this last month. Finished it by December 15th. I will start a much longer project now soon using new features

[tardy ridge]

yes, AI can do emotions

[surreal ruin]

If it is online, may I take a look?

@upbeat sierra The upscaling issue, is it gonna be fixed ever? Or shall we switch to other AI?

[slender needle]

Yes it’s online since last 3 days. are we allowed to share links here ?

[slender needle]

Here it is : Watch last 5 minutes first. Full 5 minutes : https://youtu.be/HXrbWxIvccU?si=nmAMeUVvQUViH_LZ

[violet temple]

If flow could generate this kind of dancing video that would be great https://x.com/Hailuo_AI/status/2012109059995169135?s=20

Hailuo AI (MiniMax) (@Hailuo_AI)

2026 deserves an upgrade. 🚀

FREE NOW! • Nano Banana Pro • Seedream 4.5 • GPT Image 1.5
HUGE DEAL! 56% OFF on Yearly Plans

Follow + Repost to win ULTRA Membership! 🏆

#Hailuo

X · Jan 16, 2026, 10:25 AM

[surreal ruin]

Last was nice. First was too. Awesome. Applying some advance editing tricks could make it far better. At 1:03 minutes, after the word "propose" word the hard cut didn't feel good. Cutting frame height could make the movie more cinematic widescreen. The male actor needed to be more realistic. However, overall as a AI movie and with present limitations, it's a great work.

[slender needle]

Agreed this was my first time using AI video making. Learning smart prompting. And this all was done when 4K wasn’t great and character consistency was still not there yet

[charred loom]

idk why i can't generate video anymore?

[charred loom]

Does anyone know how to fix this?

[neon leaf]

Bro why error so many re captcha? How fix

[subtle cargo]

Adnan

[pallid violet]

@wicked crowHello, 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.

[surreal ruin]

Who is being able to upscale video to 1080p? Let us know. Here the upscaling to 1080p is not working.

[gloomy verge]

Hello. I don´t know, where I can ask this, because question is partially about Whisk too, but I try it here. If it is bad, sorry. How difference is between video created in Whisk and in Flow? Are there some differences, or it is a same? Thank you.

[prisma ferry]

Any idea if we’ll ever be able to generate images at 4K in Flow like you can with higgsfield? Seems crazy it’s possible on other platforms but not on flow, the upscaling videos is less effective than being able to generate frames at 4k I have google ai studio but it’s slow. And only 1 image at a time. Wish they would have multi 4K option in flow for credits, makes work flow terrible and flow unusable if you want full quality

[proven light]

I need to create designs for swimwear products, but Flow's rules keep giving me error messages indicating that I'm violating the rules. How can I fix this?

[plain gulch]

Are the images likely being recorded for nude content?

[proven light]

Yes, it is a product image of a swimsuit.

[plain gulch]

If you're using a hosted version that has strict filters (like those on Cloudflare-based platforms or paid sites), consider switching to open-source Flux models (such as Flux.1 Dev or Flux.1 Schnell) and running them locally or on uncensored services like Hugging Face Spaces or tools like Pixel Dojo AI, which are more permissive for swimwear designs. Always make sure to follow legal and ethical guidelines.

[winged plinth]

Great question, needs someone more experienced to answer it. @edgy holly

[warm sparrow]

im on the ultra plan but its used up credits even for veo 3.1. fast generations. anyway to fix that?

[wintry scaffold]

O

[vernal patrol]

Hi, does anyone know a solution for importing celebrities?

[wide ether]

Yeah, currently, it appears their filters are way too sensitive when it comes to swimsuits, so you're probably hitting that filter and that's causing the generations to error out

[grand isle]

Anyone running into consistent "Something went wrong" across multiple projects in their account?

[gloomy verge]

Yes, I have this problem now in other generator. But I don´t know, how fix it.

[cosmic pike]

How to use AI flow for free without subscribe

[grand isle]

I imagine this is from server load, but it is frustrating

[gloomy verge]

Yes, it is.

[molten grotto]

I always use adjectives after referring to the character, like: "the Mexican rancher grandfather" or "the frightened young woman." These things help me to make the voices generally sound the same.

[vital verge]

The voices that are given inside the video. Every time a new video is made, the voices come out differently. Does anyone have any idea to keep the voices correct? Please help me with some idea so that the voice of each video is correct.

[edgy holly]

I think Whisk uses a version of Veo 3 Fast

[edgy holly]

Buy more or wait until the end of the billed month.

[edgy holly]

That's against the rules. Some people get away with it, but the filters are getting better at blocking it.

[edgy holly]

It's a roll of the dice each time. If you need consistency, then you'll have to use a third party app designed specifically for this.

[crystal talon]

Are there any legit ai filmfestivals soon.

[tough yacht]

Hey! Just making sure you know that Ultra subscribers can upscale images to 4K as long as the image was made with Nano Banana Pro after December 12.

[tough yacht]

There are two identical Veo 3.1 fast models; the [Lower Priority] one costs 0 credits, but gets throttled when we're running out of capacity. You can see them both in the model picker.

[lone kindle]

can anyone help me please?

[crystal talon]

Yes

[terse crystal]

Hi, can u help me about account without access to flow?

[wicked shadow]

is anyone experiencing weird flow generartion it was generating battle scenes well now it does not want too anymore keeps putting battling soldiers next to each other

[quiet kindle]

4K is not working?

why this message pop up for me, I have enough credit.

[vale copper]

Hello, is the service down? I can't load any of my projects or create new ones

[urban remnant]

When I upscale a video to 4k, is says it is downloaded when it is done but I don't see it anywhere, where does it download it to? I don't see it in the project view, or in my downloads folder.

[fringe onyx]

it says ready to download in that notification

you need to press download one more time. Google being what it is, UI eXpErTs

[prisma ferry]

I do know and thanks for your response! I have been doing that but then you have to re upload which is a bad work flow, which isn’t really on brand. Also it’s not usable for wide shots as you lose all character consistency and even get warped slop faces in the upscale process. You also can’t do 9 frame grids which is incredible for work flow. Editing iterations of a shot means you lose quality very quickly so need to download and reload each time. You can do all this in Higgsfield generating at 4K with no detail loss. I dont Understand customers paying $250 a month to use the native platform of the model and get much better options to get the best out of it on 3rd party platforms. why not just give the option to generate in 4K for 50 credits. this is a tool specifically for film making and quality in wide shots is essential. This seems like an easy fix would love to see it

[tropic fulcrum]

Absolutely brilliant work. Well done.

[fickle drum]

hi all. u can generate anything rn?

[surreal ruin]

You have to generate audio separately from other ai. Have to add the voice in video editing. I am using an ai for voice generation and that works fine for this purpose.

[amber ether]

is there any issue with flow? all my images are failign to download

[full copper]

I have problems too

[fair gull]

Create your own custom voices in Eleven Labs, and then use their voice-to-voice model to replace the Veo 3 voice with your custom voice.

[surreal ruin]

Yes, I use Eleven Labs.

[turbid umbra]

I feel like this will be a big step forward when veo is able to do it, to have consistent voice over the clips.

[upbeat sierra]

The team is still looking into this, but thus far has not been reproducible on our side. Cannot provide an exact date on resolution at this time, but we'll monitor #1238621511401734254 and provide updates there versus this channel

[frosty comet]

My ranking of safety filters for image generation models:

1- Google: very flexible for digital art/anime, stricter filters for photorealistic images. Hard block on any even slightly suggestive image involving minors in any style.

How it can improve: loosen filters more on digital arts involving adult characters, add harder filters on photorealistic images so you can't create deepfakes of real people eg. "put this woman in a bikini".

2- Openai: very bad and far from google, but better than the ones under it.
Paranoid. Blocks any content that is even slightly suggestive in any style. The filters are so bad that they affect the performance of neutral art, action manga/comics. Which affects the quality of the images everyone gets.

3- Midjourney. Crazy schizophrenic filters. Block content randomly, sometimes allowing racy and gory content, and sometimes blocking neutral content.

4- Grok. The worst filters currently. Loosest filters. It became the go-to model for creating unconsensual deep fakes. Even accused of loose filters on content involving minors.

[sick gulch]

Using Flow, can one generate violent action scenes with shooting and blood aka John Woo / Sam Peckinpah and / or graphic Zombie flesh eating scenes?

[full flicker]

As a 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. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[near gate]

every time i try to animate 4 characters, flow keeps throwing in additional characters or characters with the combine characteristic of the ones i want, how do i prevent this?

[frosty comet]

I get what you're saying. But it's a tradoff. You don't get to do your scene, and in return thousands of women get saved from deepfakes. "User should take responsibility" is such a bad argument against filters.

You mean the AI that was literally creating CP? Yeah that was amazing wasn't it?

You can do whatever you eant with digital art

Yeah it's all dumb and funny until it happens to your mother or little sister or daughter.

No way. I feel like I'm arguing with a kid.

I didn't say anything that can harm anyone should be banned(this was your strawman).

We already know that: there are pedophiles and malicious people who want to create unconsensual deepfakes.

Which is why we implement safety filters.

This whole thing falls under risk based regulation of dual use technologies(which you're obviously and clearly totally uneducated on🤦‍♀️)

We apply this everywhere and you should thank god we do.

We allow knives, maybe guns with licensing, we do not allow civilians to own rocket launchers.💀

Bro wants AI systems to be able to generate photorealistic sexual content using real people’s faces and images, including image-editing and insertion tools, with no meaningful safeguards, and believes that the predictable abuse this enables should be tolerated because “users should be responsible.💀💀

[unreal viper]

please can all be reminded of our first rule: Be compassionate and treat others with kindness and respect.

let's move this conversation on now thanks all

[sick gulch]

Can masks automatically be created for generated video to facilitate comping in other programs? For instance, say you prompt for two characters in a specific environment. Could you have two videos generated - one being the visual content and another exact same video but with black and white mask to separate the characters from the environment, much in the way one might create a mask or Cryptomatte layer pass in 3D rendering?

[misty gazelle]

How to check the usage of my credits in flow?

[kind wren]

Why did I receive a warning that my content was against the rules when generating the video, but my content actually wasn't violating any rules? Is there anyone who can tell me

@misty gazelleClick on your profile picture to view it.

There is an option available for checking the points.

[surreal ruin]

Is there any way to buy Flow Credit only without any subscription plan?

[surreal ruin]

Anyone noticed, today's video quality is very poor? Both in conceptual and picture quality.

[broken mirage]

is it only me? everytime when I upscale the video, it becomes .txt, I have to change it to .mp4 manully to download every time.

[tepid geyser]

That's odd it should definitely be an mp4. I’ve never seen it switch to a text file on its own. It might be worth checking if a setting or even some malware on your system is renaming your downloads.

[broken mirage]

but when I download the original size, it's normal.
well it's a small problem

[tepid geyser]

So, after you hit upscale, the browser doesn't just quietly save it to your downloads?

[tepid geyser]

Having to manually pick a folder every time via that popup is such a mental tax.

[broken mirage]

nope, it's fine I need to save to different folders

[full flicker]

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.

[teal tapir]

Is flow down????
Nowadays it's always down it's so irritating

Paying for ultra and still getting this type of service

[junior hazel]

IS DOWN!

[tough otter]

down again

[sick sparrow]

Image generation is down for me only, video is working

[surreal ruin]

Lost a lot of credits. It's generating wrong video, and quality is very low.

[mighty rock]

1020 clips. OMG. How was your the editing process?

[versed garnet]

still down?

[gloomy verge]

Can I ask...when I am creating video from image, can the prompt be written so that the video is in slow motion? I must write in prompt: "Slow motion"?

[jaunty peak]

Yes ooo

[fiery panther]

Still down?

[teal tapir]

Yup

[analog lava]

Hello, if you're encountering errors and /or bugs or would like to leave a general feedback about any of our Google Labs products please post in the new #1238621511401734254 channel with as much detail as you can☺️

[versed garnet]

tried but can't post

[analog lava]

Hi @plush berry , as a 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. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[teal tapir]

But its not generating there as well

[analog lava]

We do understand and apologise for the inconvenience. The team is investigating so we ask for your patience as they work through all the requests. Please continue to utilise the #1238621511401734254 channel so that they can keep track of any new or ongoing issues.

Thank you☺️

[slender needle]

I treated this like a feature film. So there were 4 ACTs, each act has sub acts and each sub acts had multiple sequence. I separated the sequence and filmed each sequence at once. Within the sequence I segmented locations. And filmed based on locations and wardrobe. Eventually stitched only one sequence at a time and created a video. So for example a train song 7 minute long with 2 wardrobes, shot as a single video, edited and closed. A scene in Las Vegas chapel 1.5 minute long, a scene in Venice 7 minute long, each of them are their own videos. Eventually all these videos stitched together. The beauty is it’s a single continuous story with same characters and same home location in Dubai. So it turned out to be feature film length story. However our limitations is spoken dialogues, hence used narrative voiceover. With current technology this is what is possible but still impressive that a one man cinema is born ✅

[teal tapir]

But its everyday problem now

I will switch to some other model now, i can’t pay from ultra and expect these type of services

[unreal viper]

please can all be reminded of our first rule: Be compassionate and treat others with kindness and respect. we have #1319424187730755636 for feedback specific questions/ queries

[edgy holly]

None of the AI models are very good at that sort of thing yet.

[edgy holly]

Image to video or ingredients to video, and more structure in your prompts.

[edgy holly]

Entitlement to be smarmy creeps really is disturbingly widespread.

[edgy holly]

I think you have to have at least a Pro account to buy more credits.

[wicked summit]

Whats the best way to solve the text on video issues with Flow/Veo?

This is BASIC text and it cant handle it.

[unreal viper]

please use #flow-tips-n-tricks・✏️ for asking for tips for flow!

[knotty sigil]

does flow work for yall ?? cause personnaly i always have a failed generation message, could someone help me i tried everything

[fiery panther]

Today flow its troubleshooting a lot

[digital meteor]

I dont see aspect ratio 3:4

[frosty comet]

this is strange. both quality and fast seem to consume 20 credits. and it does not increase if i ask for 4 output per generation?

or do 4 outputs per generation consume 80credits?

maybe a visual bug?

[native barn]

BULLSH8t

[frosty comet]

at least they didn't charge you

[native barn]

LOL

[knotty sigil]

does someone know if the staff is working on flow?

[edgy holly]

correct

[edgy holly]

The staff in here are moderators, not tech support.

[ruby roost]

If I cancel my monthly and then buy annual will I get 50% off?

[barren cypress]

Is there rate limiting? My generations keep failing but I have the pro subscription

[edgy holly]

It'll tell you if you run out of credits, or hit a limit.

[barren cypress]

It's just the generic error message. I guess it's just an issue on their end then

[near gate]

Someone upped the censorship, I’m getting flagged for non sexual things again. Wit Gemini made prompts

[stiff ledge]

Anyone else getting a lot of policy violation errors?

Trying to create videos of my character as a robot but they keep failing with that error.

[near gate]

Hello everyone 👋

[whole hill]

hey, i use flow veo 3.1 to generate looped videos but for some reason they all have some kind of artifacts in it
do you have any tipps to get over with that?

[crystal talon]

Sorry for my sarcasm but is veo3.1 and nanobanana controlled by the taliban or sth? Everything is restricted. Wth is going on?

Woman in bikini is sexual, error. Kid actors minor issues, error. Even a short tshirt with free belly, nope, too sexual.

[frosty comet]

Google should add daily refresh credits for flow

[crystal talon]

Pointing a gun to a woman , nope, policy.

How am i supposed to make a damn movie with all these policies?

[stiff ledge]

Google should write a way for us to earn AI credit by policing prompts.

[turbid lintel]

🫣

[cinder saffron]

For the love of God, what is happening to Flow!? I can't generate absolutely anything, everything is a policy violation.

[stiff ledge]

Yeah something is going on.

[barren cypress]

Flow is not working at all for me all day.

[stiff ledge]

Same here. It is labs though..

[vocal radish]

I just created a 1 min video without issue , but there are many grammar error

[surreal ruin]

Video quality is very poor for last few days. Conceptual low quality, image low quality, and not following instruction properly. Just wasted a huge amount of credits because of last few days' low quality video generation.

[kind bison]

The acting of VEO 3.1 characters is so awful now, I don't know if I can take it anymore. Flat, boring, so little emotion (even when properly prompted), and the chronic sameness across projects. I'm praying 3.2 fixes this, or I'm ditching my Ultra plan.

[mighty rock]

With what do you compare with? Do you have examples @kind bison?

[tepid geyser]

Does generating an image take forever? A lot are failing right now, it's so annoying.

[wraith hornet]

what the
im waiting the 99% to be to completed then if failed, both veo3 and nano banana

[tepid geyser]

If they are upgrading the system, I get it, but doing it without prior notice is unacceptable. We rely on this platform daily for different projects. A professional company would warn users before servers go offline for adjustments. I wouldn't mind if the update took days, but give us a warning so we can plan around it. With us paying $250 monthly + extra credit cost, this is frustrating. There should be a proper alert system.

[worldly willow]

Yes images are not generating, please do the needful please

[opaque wagon]

Anybody save the world?

[turbid umbra]

Can confirm Nano Banana is not working for me either

[tepid geyser]

Nah, we’re all doomed. My screen looks exactly the same.

Looks like the Flow team is silent, so I gotta renew Higgsfield.

I honestly hate their platform and all that aggressive marketing nonsense, but I have no choice with this deadline. Just have to deal with it until Flow comes back to life.

[turbid umbra]

it sucks, but i think stuff like this is to be expected, its still a labs thing, hopefully it one day it will get out of the labs phase.

I also had to deliver a vid in the next 3 hours btw :(.

[worldly willow]

same here... need to make a video within 2.5hours

[rapid idol]

same problem (((

[tepid geyser]

Downtime happens, I don't mind that part. But lab phase or not, a quick warning isn't too much to ask. Just put a banner up beforehand so we aren't left guessing.

[opaque wagon]

1:1 size, drag photos, 4k download with pro, please update first.

[tepid geyser]

Hopefully, the devs see this and just drop a quick announcement next time.

[near gate]

by the way_
Tha'ts my output.

[fluid abyss]

what is happening with nano banana pro generations. literally every single generation has been failing

[rapid idol]

so few reviews, maybe the problem is local?

[minor badger]

is flow completely unusable again this morning? When will the devs finally fix its lack of robustness. It's mad. Especially as part of the 275$ Ultra plan. Very frustrating.

[gray sluice]

Hey, I’m Zammy.
I’m a 20-year-old graduate with a background in video editing. Right now I’m working on creating a psychological web series using AI generated video focusing on serious story driven content rather than generic visuals.

I’m learning how to combine screenwriting, direction editing and AI tools to build complete visual stories. This project is something I’m taking seriously experimenting, improving, and understanding how far AI can be pushed creatively
If anyone here is exploring AI video storytelling or experimental filmmaking I’d love to connect and exchange ideas

[turbid umbra]

It started working again bois, go go go

[marble path]

Which model do you prefer when u want to edit or create a video from the image?

[turbid umbra]

the veo3 fast with ingredients to video in my opinion works best so far

for consistency

They recently updated that feature, it's clearly improved in my opinion.

[marble path]

Newest one right?

[turbid umbra]

However, for the TikTok account I'm growing I find frames to video more comedic, I generate with nano banana the initial comedic frame, then go Frame to Video.

I officially depleted all the Ultra credits 25k, 11 days left until they replenish hahahah, time to see how the low prio goes until then.

[meager edge]

Is this happening with some else also , still the generation fails hope to get the Google service back ASAP .

[pearl basin]

anyone having this issue? cant generate any video, since yesterday

[junior hazel]

DOWN AGAIN TODAY

[minor badger]

If you mean 60s, okay 😄

[fluid abyss]

entire teams or individual creators as well?

[coarse oyster]

hey the episode duration will be 60-90 seconds. Total duration of the show will be around 60 mins. Happy to share more details in the dm 🙂

[coarse oyster]

currently looking for teams (typically 5-7 folks). please dm if you want more details.

[regal raptor]

Flow nano banano pro reverting to the old crappy model generations while still saying PRO, seems like a capacity issue when their over capacity they go back to the old model bad generations without actually telling you

[turbid umbra]

Nano banana not working again.

[chrome yarrow]

Why is flow not working over and over again. after generating a few video and images

Here's a tip that is working well with me. Try to change the settings of video like if you are generating on [BETA FAST] try the second setting below it if you encounter problem. same with the nano banana. if after several tries and still not working. login on incognito mode.

[tepid geyser]

march

**May

[kind bison]

I compare them with 1) real human acting and 2) every other model that now generates dialog.

[signal maple]

I need Help with veo prompt, I need to make a static scenario, the camera needs to be static and do not move. But aways move anyways the prompt is.... Anyone can help me with that??

[edgy holly]

They usually don't annouce releases far ahead of time.

[edgy holly]

If you want help, that's for the #flow-tips-n-tricks・✏️ room, and share your inputs (prompts and any images you use), and outputs that show what you're getting that don't match what you want.

[frosty comet]

Huh

Prompt: create a social media page

[frozen oak]

I always get Failed Generation.. dont understand why

[livid gust]

hi guys i need a help with my flow someone can help me?

[surreal ruin]

Doing similar thing. Making documentary on Psychology, Self-Help type of Niches.

[dusky wraith]

Mine doing the same thing💀

[opaque wagon]

OK, the world was saved.👍

[stiff ledge]

Is there a way to get rid of these failed generation errors? I was deleting them as they failed but I logged back in and there are a ton of them still in there.

[spiral pollen]

Is the Flow server working properly right now? Nothing is generating at all.

[stiff ledge]

Same here.

[edgy holly]

Getting a lot of Nano Banana returning with images that have no alterations. Just giving me the same image I uploaded.

[robust cipher]

hey does anyone know why I keep on getting this error whenever I try to generate a video on flow? "Audio generation failed. Please try a different prompt or send feedback. You have not been charged for this generation"

[trail maple]

same

[rose vigil]

looks like thry limit the number of image/video we can generate per IP. I changed my IP and video/image generation works again.

[stray vortex]

wonder where are you guys usling flow not workig in Japan and usa how can we fix this

[opaque wagon]

new bug, my history is missing.

[frozen oak]

i really wish there's auto caption feature in Google Flow

[gray sluice]

Guy's Need help with AI character consistency
Working on a series and struggling to keep characters consistent across scene's Any proven workflows or tips would help

[turbid lintel]

veo 3.2 is coming

[turbid lintel]

why nano banana pro quality drop?

look like i use nano banana

[karmic current]

any Idea why I can't export from Flow's Scenebuilder? 🙁

[sick sparrow]

NBP down again ...

[worldly willow]

NBP down again ...

[upbeat meteor]

...

[wraith furnace]

is flow down?

[prisma lotus]

Why image generation not working

[upbeat meteor]

you are not alone

[meager edge]

IS NBP working for anyone as Mine is down again

[analog radish]

Looks like flow is down both image and video generations

[tepid geyser]

sigh!

[sand nymph]

hi! Just started using this service recently. Can anyone tell me how long it will take to fix the issue with generating images?

[clear glacier]

NBP still down ?

[sand nymph]

yep

[serene cradle]

Still here, it keep failing at 99% loading

[sand nymph]

same

[rapid idol]

same

[sand nymph]

Does anyone know of a service that is currently working? Nano Banana Pro works in Gemini, but its functionality is, of course, very limited

[noble carbon]

99% as well

[tawny jolt]

still :))

[tepid geyser]

))

[civic dagger]

guys anyone here?

I need helpp

[worldly willow]

NBP down again ...

[civic dagger]

when will that get fixed, I got the same problem

[modest lotus]

man i just realised how hard life with out nano banana pro ..

pleas fix your stuff google

[versed garnet]

NBP generally down and we're paying for it..

[viral folio]

SAME ISSUE GUYS

DOES ANYONE KNOW ANY THIRD PARTY SOFTWRE FOR ACCESSING NANO BANANA PRO?

[toxic shale]

higgsfield is failing too or taking too long

[modest lotus]

try artlist

[iron sleet]

Why is nothing working today?

[viral folio]

Alright, thanks for the input

[opaque sand]

What's up with Nanabanana Pro on Flow? It hasn't been working for two hours now...

[viral folio]

Fun fact, yesterday I was speaking with one of my friend about what if the system of AI collapse, and boom - Today it happened!😭

[opaque sand]

The deadline is approaching - I'm “smoking” pictures through Gemini chat.

[viral folio]

Same man! I run a marketing agency, and I need to create some video and upload for my client's business💁

[sand nymph]

oh god can anyone tell when flow will be fixed.. 🥲 i guess we all need to work and have deadlines

[toxic shale]

Is gemini working consistently?

[sand nymph]

yes, working perfectly

[toxic shale]

Got it thanks

[opaque sand]

yep, But the workflow clearly suffers in terms of speed.
Although, if you remember how it used to be done in minutes in сomfy)))...

[fallow dove]

google Flow creating images and videos keeps failing at 99%... why?? 😢

[civic dagger]

<@&1366831844502274180> when will this get fixed

[copper mica]

The mods can only assist with Discord related issues.

Any bugs please submit to #1319424187730755636

[civic dagger]

whom do i contact for this issue?

[copper mica]

#1319424187730755636

[little stone]

Perhaps it’s a global issue caused by system instability, or problems that arose during the upgrade of the server programs.

[meager edge]

did server are back for GOOGLE models NANO BANANA PRO and VEO 3.1

[copper osprey]

yes

[junior hazel]

ANOTHER DAY DOWN! THIS IS A SCAM!!!

[unreal viper]

please can all be reminded of our first rule: Be compassionate and treat others with kindness and respect. we have #1238621511401734254 for feedback specific questions/ queries

[little stone]

not yet

[tawny sandal]

already hasnt working anyone can fix ?

[gloomy verge]

Please, can someone help me?

[tepid geyser]

😅 Relax, No need to scream.

[gloomy verge]

Can I ask somthing?

[little stone]

maybe next few hours

[junior hazel]

I pay 200€, dont tell me relax

[little stone]

ultra user is painful

[warped wind]

yep

Dont charge 250$ for a unusable service

[little stone]

7€ per day😤

[modest lotus]

his right

[worldly willow]

still not working

[deft summit]

Is Flow always this buggy? Everything I generate seems to get stuck on 99%. I thought it was a moderation issue, but even doing something like “a car” doesn’t work. I haven’t been subscribed in a while so I didn’t know if this is normal.

[modest lotus]

its been like that since yesterday

[deft summit]

Well I picked the poor time to subscribe lol

[digital spire]

For those with the nano banana issue, a kind of workaround is to refresh the screen after it is stuck at 99, after reloading it, it sometimes have the images that it was creating.

[modest lotus]

didnt work