flow-general・🌊

[glad isle]

Yeah its a mess. And it’s google, so you know how they don’t handle things.

[unreal viper]

hi all, please use #1238621511401734254 to submit any bugs! these threads are actively monitored by the team and so may be seen faster!

[paper ermine]

Lol.

true

[glad isle]

Ronnie from google also nonchalantly said they won’t reveal daily quotas, just because.

Seriously, their attitude is getting worse every year.

[cedar fractal]

The "Compute" system meanwhile is still borked. I already canceled my Pro sub and I'll keep an eye on the discord but this is not the way to treat your customers.

[fading dirge]

i will say this, omni is far better at explination video than veo 3.1

[flat quiver]

Is anyone having an issue upscaling specific images? I have about five images that refuse to upscale. All fashion poses, the character's legs are showing (no nudity). I think it's a safety filter issue but the only error is "Upscaling failed", no other info. Other characters with the same pose and exact clothing upscale just fine so I'm stumped.

[calm spoke]

i did so many times,,, useless

[calm spoke]

same here, but i refresh before every upscale, it works

[quick oasis]

from the pricing session in google flow. It'll show you of what's free without subscription and what is paid subscription

[flat quiver]

I'll give that a try, thank you

[neon oasis]

Hey there, I have a question: how do I use the Flow Agent to make sure the image generator follow my instructions? Nano Banana Pro can't generate my character with 5 fingers and toes.

[inland parrot]

Is Flow working again?

[open plover]

I tried to feed flow with audio as it is promoted but it doesn't accept mp3 or wav...how can i feed it with audio ?

[echo spade]

Feel great to pay for something I can't use.

[brittle canopy]

Is there a limit on how much the agent can be used ?

[rigid nest]

Hi @main saddle

Thanks for the update!
Where can we find documentation about no. of photos and videos allowed for each tier?

We are seeing a lot more "Generating too quickly" error this week than before.
The reset duration seems to be random with no clear instruction on when an user should try next.

Can you please share the documentation, so we are aware of what to expect and be more precise with our allowance?

[gilded pewter]

Irony Mode: ON... Such lovely videos (not). This prompt—which contained absolutely no sexual or violence-glorifying content, and simply asked for a woman to move like a model in front of the camera—worked perfectly just a few days ago. Now? I get this garbage.
I’m going to cancel my Pro account, because I have zero desire to pay for an "AI Bug Generator." Sorry, Google—but you need to fix your system first and properly fine-tune your bizarre content filters so that the tool is actually usable. Or, at the very least, give users a list of "forbidden words" they aren't allowed to use. It’s just become utterly ridiculous—sorry—and yes, I’m starting to get genuinely angry now, because the situation with image generation is even worse and has simply become a massive annoyance!😠

[sterile fulcrum]

Hi! I'm really confused about how the character maker works...like I don't know how to reuse a prompt, and sometimes I generate a new picture for the character, and sometimes it just replaces one I already had for her! I have 2 right now and I can't figure out how to add more...and can I make different outfits for her, or just one?
So is there a guide somewhere to explain everything?

I'm sorry to keep asking so many questions, but I can't find any detailed tutorials..

[rigid nest]

Sorry need to update it, Multivoting is not working.

[grim nest]

Boss getting watermark logo add that in vote

[rigid nest]

Can you please add a little more detail, I am not aware of this issue.

[grim nest]

I am getting watermark of veo at my generated videos even i have ultra subscription

[rigid nest]

Updated the Poll. I hope it can be used to visualise which errors need to be fixed on priority

[chrome imp]

I hope they get to fix or improve the prompt adherence, there are a lot of times that it generates something i don't want

[last lichen]

It's bugging me now on uploading some photos or screenshots, most work, but the others don't work?

[proud edge]

in the new whisk, how can you change the aspect ratio?

i remember being to make the aspect ratio

[rancid timber]

does google flow creddit reset dayily or monthly

[rigid nest]

I have seen this too. Happens for 5% of cases for me

[rigid nest]

In Flow (New Whisk) when you click on Nano-banana button, it allows you to change aspect ratio.

[glad yacht]

Seems right now video are left with each ai plan tier credits to decide how much video you can create.

[rigid nest]

My credit has not reset for months, but looks like each generation is costing 10 credits.

But I am trying to preserve it for emergency, incase it does not reset.

[glad yacht]

Each tier unused credits won't roll over each month, even if you think about saving them as an emergency.

[proud edge]

ok ok sick

[rigid nest]

Ah Ok. Sorry, I do not use video too often.
Where can we find more information about tier limits? Do you know?

[glad yacht]

Here is the latest tier information

[rigid nest]

Thank you!

[rancid timber]

is that 1000 credits per day

[glad yacht]

Per month, 🤔. We also get daily credits 🤔.

[rancid timber]

so what happen if we use up al our ai creddits

does it reset the next day

[neon raven]

why does it state 10000 or 25000, which is which

[rancid timber]

i have pro

to gemini

on the google plan i have pro

[rigid nest]

Updated the Poll. I hope it can be used

[glad yacht]

2 ultra plan, 1 cost $100, the other $200

[neon raven]

ah yeah

but both have low priority mode right

[glad yacht]

Paid tier can purchase extra credit before tier reset kicks in.

[reef mural]

But still free gens of veo 3.1 lite ?

Low priority?

[glad yacht]

No more free unlimited Veo fast Gen for ultra

[reef mural]

That’s really bad

[neon raven]

what? where does it say that

[reef mural]

It the reason many uses this

[neon raven]

yeah, i havent seen any update about that online

[glad yacht]

Google has changed the unlimited generation tier for Ultra users from Veo 3.1 Fast to the lower-tier Veo 3.1 Lite model.

[neon raven]

yeah but its still unlimited lol

[reef mural]

Yes still free

[neon raven]

they did that 1 month ago

[reef mural]

Works perfectly for my Lego animations no need for veo 3.1

[glad yacht]

If you want to use the higher-quality Veo 3.1 Fast model, it will now drain your monthly pool of credits.

[reef mural]

It has always cost

Only but images are it still free?

For my Lego images I use ChatGPT because it’s looks nice and can batch render 10 at a time

And animations in veo 3.1 lite low priority

[calm spoke]

what about that "unusual behavior" flag??????????

[reef mural]

Looks gorgeous

I hope it’s gone, haven’t had it for three weeks now

[rancid timber]

so im pro plan on google

so can i buy it with google play money creddits

[glad yacht]

If your credits run out before renew refresh kicks in, using any form of payment methods Google accept will work.

[rigid nest]

Sorry I am not aware of it.
You may add it to the thread I created earlier.

Server guidelines do not allow me to reference it in my response.

[rancid timber]

is there any cheaper credit to buy it

i can use my google play money i think

i think i can use my google play balance

but im saving it for clude and gemini pro plan

@glad yacht

[glad yacht]

Several digital platforms offer Google Play codes at prices below their face value.

Here is the alternative cheaper deal to refill your Google Play balance.

[rancid timber]

what do you mean

[glad yacht]

Google it

[rancid timber]

i tryed to google it on gemini but it did not give me anything

[elder canopy]

https://youtube.com/playlist?list=OLAK5uy_mTveuK1nvYaaQwEUVgAbR1vmT2PqiDaic&si=9Ho0zp7Ctd3oQcCr can someone help me make music videos I'm in Atlanta

[pulsar coral]

From all the cool useful requested features from Google AI studio they could bring, they had to bring the most annoying useless one that no one asked for which is the chat assistant agent. I can't even continue a project because it keeps confusing images with videos even if I turn it off.

Who hates this project so much to be adding this stuff all the time.

[fallen spear]

How does the AI credits in flow work? Are they fixed or do they reset after everyday/week/month?

[zenith current]

I really can't find any reason to subscribe to a Google AI Premium.

[rotund elm]

Is nanobanana down on Flow?

[pulsar coral]

For the love of god consciously beta test stuff before just pushing a button and adding things

Flow can still do image generation but one has to start a new project all over again

it's not possible to continue an old one, the agent needs to be trash'd asap.

On actual good news tho, it seems plus plans and above can generate 25 images at once again, idk how free plans are doing atm.

And even better seems the censorship at least for NB2 has gone down just a bit I think, nothing like 2 months ago when it was just fine either way.

[proper path]

Good morning

[pulsar coral]

Also they removed the "Reuse prompt" line from the generated images, but I imagine it's fine since the return icon from them does the same... it's just not as noticeable tho

Nvm now it's just 14 image generations at once... tho it did give me 25 at once an hour ago. Well that's still something.

[signal roost]

Been messing with flows tools. Quite good the scope creep is insane though. Minor changes and huge parts get simplified. Is that just me or others notice that too? Also yesterday in my tool i could select a character via the flow-input but now it soley shows images. Was that changed as it now complains the sdk does not support it.

[pulsar coral]

If you're continuing a project like something you were doing yesterday things are messed up unusually, I'm still trying to figure what's really going on, right after they added this "agent" option tool. But if you start a new project all seems like it's everyday lately

Be that it's a good thing or not

[signal roost]

All my old ones seem ok. Its the tools area. I made a tool click add character it used to show charactes in the flow ui selector in the tool. I made no changes but the tool now only shows images and uploads.
Not a huge deal breaker though i was just using it as a creator aligned to a character but instead i can just make a new chat and @ the character to speak to it and work that way instead. Untill they add that back to the tool il not continue it further. Just be nice to not have to leave the tool.

[rigid nest]

I have created a poll to track the most common errors for flow. If anyone can, please pin it.

There is also a thread to discuss additional errors, which have not been added to poll.
The idea is to collect all major issues in one place and pass it to DEV team for fix based on popularity.

[pulsar coral]

I think a couple of those errors stopped happening already, and others just radicate on reaching limits, like the upscale one is related to that too.

But yeah putting it together that's one huge scary list.

Might just end at Trust Pilot or something at some point really... so it's better to just keep record already.

[sage otter]

Day 95 of not being able to search all my projects

[errant gull]

My flow still can't generate images 🥲

[spiral pollen]

nanobanana is broken?

[kind dirge]

Everyone getting this issues on flow last few days now. Keep getting " Your generating image too quickly" I KEEP GETTING THIS ANNOYING MESSAGES EVERY 2 OR 5 Generating image NO MATTER IF I USE NANO BANANA or nano banana pro. It slowing down the image i want to make and i have to move to other ai site while waiting for this either annoying bug or this stupid feature that they're not telling us.

[pearl lava]

Both Nano Banana Pro and Nano Banana 2 models appear to be down or unavailable on my end. Only Imagen is running normally during this period.

I've been experiencing for approximately 10 hours now.. Is anyone else experiencing the same issue right now? Any workarounds you've found while waiting for a fix?

[plush gate]

Could you bring back the “Recent activity of family group members” section in Flow credits?

A few friends and I purchased and are using the $200 Ultra plan together, and right now some members in our family group are having a very hard time dividing and tracking how many credits the other members use.

Please bring back the feature that shows the credits used by other members in the same family group.

[tame falcon]

I actually think that such radical censorship to prevent deepfakes is unnecessary. Most people use it normally, and the model itself prohibits explicit or indecent content.

[pulsar coral]

NB2/Pro was way more than censored enough as it was two months ago, you couldn't really do any explicit content at all with any of the Nano models, not even with Gemini 2.5 flash image.

And if aiming for that one could just use Stable Diffusion, LoRAs etc, they are much better for that content anyway.

No one was using Nano for that kind of stuff, so all that the censorship did was ruin consistency of other prompts... tho admittedly I think it's working like 5% better today, nothing like it was before tho.

Google must've had some confusion on how the Nanos were being used.

[pearl lava]

any tips for this problem guys? please help me 😭

[pulsar coral]

Or maybe the decisions were taken based on images that weren't made with them.

[pulsar coral]

Did you try clearing cookies and going slower? Do you have a paid plan or free?

[pearl lava]

allready clearing my cookies, and i had a paid plan

[pulsar coral]

Was it working fine some hours ago? like how long ago?

[pearl lava]

so it will be 11 hours right now

so this is come out when i ask it to the agent on the module

[pulsar coral]

The "oops something went wrong" message is indeed tied to connection probs, it could be a server issue on Google's side. But sometimes it's also due to connection probs on user's side.

Whatever the issue, Flow needs a lot of ram and banwidth to display images and videos.

If one side fails it wont be able to do it

But you could still try restarting your wifi and freeing all the resources possible from your device so Flow gets as much power as it can.

Like if it's a comp restart it, if not it's on Google side and not much to do about it atm, you'd need to report the isssue at this Discord's labs-bugs-feedback area, they'll get on it.

[pearl lava]

i dont think its my network, i running fine with another account, on same device too

ok, thank you.
will hit the labs support

[pulsar coral]

Well the other solution is trying from a different account, sometimes the "server based" issue won't affect a different one. If you see it works you could get another $5 plan right there just to work the night if it's too urgent.

[pulsar coral]

Oh also your same Flow Google account also links to Gemini, you can make an image at once there and it has watermark but it can be removed. And if you got Pro plan and above it also links to Google AI studio, it's the same deal there.

This thingie removes the spark watermarks, but you need to have downloaded the image with it's download button not right click save as or it won't work https://geminiwatermark.org/batch-watermark-remover

[civic kraken]

Does anyone know how I can remove the voice from my already created character so I can use more than one character in the video generation with ingredients?

[pearl lava]

okey, thank youu.
I could just generate using another account and send it to my main account, but why should it have to be that way when everything can be done centrally from my main account? After checking, I found several other users experiencing the same issue in the labs-bug channel — let's hope the problem gets resolved soon."

[pulsar coral]

Well you gotta inform them about it anyway, bugs here are very inconsistent and don't always affect everyone the same way. I think it's about randomly assigning available resources.

[pulsar coral]

And the other possibility is your account just ran out of image generation attempts for the day, but it's rare it would display that error message for that, cause that one's usually for connection probs whatever side they may come from. Guess you'd know 24 hours later after you last generated an image successfully.

[wide moon]

It’s best not to do that especially if you’re going to get the same results as your main account

[signal roost]

tried to make 4 images and got this. is this what happened to you @pulsar coral

i have hit the quota for the agent mode again might be related no way to check the agent mode quota yet as its still credit based on flow

[pulsar coral]

That's the one that usually gets fixed by clearing cookies, right now you're getting it if you try to generate more than 8 images at once, you also have to wait a couple minutes before requesting more.

But it could be allowing to generate like 14 images at once today.

[signal roost]

ah so its likely tripping some in browser anti spam mechanic.. noted il not do 4 at a time no more

[pulsar coral]

I was in a plus plan account just now and managed to generate 14 images at once per 2 mins, but yeah going faster than that could trigger that error even in paid plans.

Free accounts could have shorter limits idk, like said it works different for everyone specially according to region, I'm in south america atm.

...wouldn't pay much more than $5 with Flow current's state atm, it's risky af

[rotund helm]

Hi, every personal video I use with Google omni gets flagged as a violation of policy. Every single generation. Tried with different videos. Does omni work only with Veo generated videos? Is it a known bug?

[signal roost]

im on pro.
kinda happy with it at the mo. except for i ran outta credits fast.
just found out you can add a character to a veo fast for a voice. didnt try adding the actual character though i should have. i thought they were just omni things.

[ivory storm]

@pulsar coral Does it apply in generating videos too? Its been like this since morning and I cant generate a single video even though im using ultra

[pulsar coral]

Yeah but successful videos will still consume your credits, images don't as usual.

[pulsar coral]

However even Ultra plan accounts have a daily video/image generation attempt limit, if you keep getting an error that says "You're requesting generations too fast" after like 5-10 mins trying even after clearing cookies. then you just need to wait till your account's limit resets in about 25 hours later, even if you still have credits.

Like free accounts get about 100/200 daily image/video generation attempts according to daily resources available, and it's double that amount for each paid plan tier, Ultra does get much more tho.

For Nano Banana Pro can be as low as 10 image generation attempts for free accounts tho, and so on.

So yeah if you were using Nano Banana Pro and you hit the error limits you can try swapping for Nano Banana 2 for some more images.

... until that one is dry too so yeah.

I'm not even gonna say next day/24 hours for the reset, it's more surely 25.

So If you were having a workflow it's more often better to consume all your daily generation attempts until you get the error message so you can be sure you will have a full reset the next day/25 hours later, if not you might hit your daily limit alike 12 hours later and you'd again have to wait till it resets again.

Can't just wake up like 10 hours later and wait for it to have resetted already if you didn't consume it, doesn't work on the chant from the rooster.

[pulsar coral]

Admittedly all this stuff is completely invisible, you won't know until you have like a couple days of practice on it.

[pulsar coral]

Nvm all the stuff about projects from hours ago, I can continue projects now, for the life of me idk what was going on with that earlier, but works again now

[gray nest]

Hi, quick bug report from a Traditional Chinese user on macOS.
When I’m typing in Traditional Chinese in Google Flow, the IME requires me to press Enter to confirm the selected characters. But in the Flow chat box, that same Enter instantly sends the message, so my prompt gets submitted mid-sentence before I finish typing.
This causes major friction in the workflow, especially when writing longer or more detailed prompts. When I’m still trying to build and refine the meaning of a prompt, this issue interrupts my train of thought, which can be really frustrating.
I actually reported this issue last year, but after several updates, big and small, it still hasn’t been fixed. It feels like a real shame, because I genuinely love this platform and even pay for it.
What makes it more frustrating is that Google Search and most other Google services don’t seem to have this issue. Flow also didn’t have this problem in the earlier stage. It seems to have started after one of the updates, and the issue has continued ever since.
I really, really, really hope this small issue can be improved in a future update. It would make the experience much smoother for Traditional Chinese users.
Thanks again to the team. I truly appreciate your work.

[tame falcon]

You should learn Pinyin input method. It's very simple; there's a table for the phonetic symbols. I'm from Taiwan, but I know how to use Pinyin input method.

[tame falcon]

Taiwanese people use Zhuyin Fuhao (a phonetic alphabet), which is actually outdated compared to global trends. I didn't know how to use simplified Chinese input methods when I played Chinese games before, so I have to learn how to use simplified Chinese input.

[verbal thicket]

Agent not working

@real wyvern there is problem coming in mac omni ai agent feature is not working

@bitter flare

[open goblet]

Anyone else finding the new Characters option useless as every single character is flagged as a "prominent figure" - even though they are generated in flow?

[silent juniper]

hi! can anyone help me understand why my Ultra subscription has watermark on all veo generations,regardless if in gemini or flow?? this is a USA based account

[tired nova]

This was happening to me too yesterday, after awhile it worked but its been doing it again for me. I wonder if using the same character on VEO makes it believe its a real person or if the "Created" people are close enough to someone who is real for it to be flagged.

[open goblet]

i figured the last must have been the case so i kept trying different characters but in the end i just gave up. seemed pointless

[jade crest]

Does the $99 Ultra subscription offer unlimited video generation?

[low imp]

Hi everyone!
I have a small problem/request 🙂
Due to the current operating principles of Google Analytics and Tag Manager, it is not possible to directly track the scans of physical QR codes without having a separate tracking code running on the landing page.
Therefore, we would like to request the development of a feature that would allow the scans of dedicated dynamic QR codes to be measured directly, without the need for a separate external service provider. Thank you in advance 🙂 Of course, if you can ask for it!

[void ledge]

Whisk really lacks functionality

You write a prompt on a specific topic. The AI would automatically add random elements and generate a new prompt from them. But you didn’t have to use the prompt; you could just feed the result (the images) into the pipeline. Sometimes it came up with interesting solutions. How I miss that. Now it’s just a default on-demand generator.

[fallen lily]

Nah

[fallen lily]

Agreed native dynamic QR scan tracking inside GA/GTM would be extremely useful for offline attribution. Definitely a feature worth requesting.

[plain robin]

I am not able to create avatar it shows me this

[trail spoke]

"Inusual activity", "policy violation", "unappropiate", "children", "celebrities", "sexual", "guess why"... It is being a headache. I am just getting 10%-20% of the results. I work on advertising and the new censorship in flow is clearly overreacting since the last update. I am wasting money and time. Does anyone else feel this is not working?

[glad isle]

Everyone. I canceled my plans. Its better to just use 10-15 free accounts.

They are acting like a startup “in development “.

[low imp]

Hello. Have a nice day!! Topic: In longer conversations, the CONTEXT unfortunately gets ruined/misplaced due to the many side questions. I tried to summarize at intervals but unfortunately it was not good, it did not lead to the best result. The context got messed up and then I had to go back and do everything again. For example, if the conversation and questions started in the wrong direction and I realized that this was a wrong direction, then if I wanted to return to another good direction, the wrong direction sometimes/often interfered with the conversation. The context caused problems. No matter how much I tried to summarize or do anything, the wrong direction kept interfering and this was a problem. Sometimes chaos developed in the context and in the text as well. The whole communication was ruined because of these disturbances. Then I thought about what if…. I am writing here and I ask you to judge what the idea is. This is just a suggestion!!! There are people much smarter than me… you know what I mean… 🙂

​Suggestion: a split-screen, two-track chat interface. The essence is that the main train of thought (Context) remains continuous on the (left side), while other questions can be handled in a separate lane (right side). [These are sometimes needed for a clearer understanding of the content.] The right side takes into account the main context on the left side with its answers, but does not overwrite, disrupt, or do anything with it. /To sum up: it would leave the context completely intact, but in its answers to the questions I ask it, it would take them into account in their entirety./ The “main clean context” would be taken into account during the answers, for more efficient collaboration. The right side could integrate the call to the “temporary chat activation” function, which is already ready, which “stores information for 72 hours”. Complete separation, no interference with the context, ugly!!! I apologize if I did something unprofessionally, I was just pressed for time.I hope the translation went well.

Például nekem nagyon tetszik ez a felosztás a fényképen:

[forest raptor]

The video model is very good on the video..... but really bad on the audio.

[silent dagger]

guys any solutions when keep getting "FAILD we notice unusual activity etc...." ?

@pallid violet please report this guy @vapid briar he is redirecting to fake support discord group

[pallid violet]

Hey, already handled, thank you!

[glad isle]

Cancel sub…

[devout ledge]

No. Both AI subscriptions use monthly credits. If you run out, you can purchase more.

[silent dagger]

Better alternatives ? Preferably with high gen limits

[glad isle]

Just use free accounts. You got more flow credits anyway compared to a pro plan. Pro gets 1000, free gets 100 and 50 daily (1600 combined).

[flint cloak]

Anyone looking for a real UGC creator? I have over 5-10 UGC creators and who can make a realistic testimony video

[tawny hollow]

Hello

[surreal hare]

I don't know why agent mode working for Pro plan 🤔

[burnt oriole]

I was wondering, if Omni is a native video model,
can it have context of previous videos and then generate the next one? I mean right now flow's approach for continuing a scene is by taking the last frame of the previous video and continuing from there, so i was wondering if Omni can just have the previous clip as context and continue from there.

Also, if it can handle multiple videos as input and generate one video as output. like if I attach a video with a moving background, and another video with the character moving, and maybe another image for style and then ask it to combine them all together into one video clip

[modest terrace]

@main saddle The user experience on Flow has honestly become very frustrating. Image generation works sometimes, but basic features keep failing.

Downloads randomly fail with the error: “Failed to download media.”

Upscaling frequently fails or gets stuck for no reason.

Sometimes the website/app barely works properly and feels unstable.

Features are inconsistent, making it hard to actually use the app seriously.

This is ruining the overall experience because users spend time generating content and then can’t even download or upscale it properly. These issues need urgent fixing and stability improvements.

[elder cosmos]

Hi, I need help with Google Flow Ultra. I paid around 100 CHF, but for several days I haven’t been able to generate a single usable video. Even with my own uploaded footage and harmless prompts, the system keeps blocking generations with policy/filter errors. I tried simplifying the prompt many times, but it still fails.

Can someone please help me fix this or tell me what I should do next?

[tranquil fog]

I too upgraded to Gemini Ultra and I can't generate anything in Flow. I have cleared all cookies and all browsing data in Google Chrome, and still no luck.

[lean hollow]

I have a crazy one. I have two Google Ultra plans. One the new Agent works perfectly, the other always times out saying SOmething went wrong. Same computer, same browser, cleared history, and cached, etc. Why does one account work, and the other doesn't? All generations tools are fine, just the Agent won't talk.

[full flicker]

Hi everyone if you are having issues/bugs or have feedback to share, please head on over to #1238621511401734254. Be as specific as possible with your details, while remaining respectful. The team regularly monitors those threads. Your messages can get lost here in the chat. Thanks!

[lean hollow]

Thanks!

[bold falcon]

🍻 Can I just say how happy I am for the Flow Team and the fact that we get to retain our 25,000 credits on Ultra, separate from thew Gemini quota's? Thanks for all that y'all do. 👍 🔥

[void crater]

[burnt pilot]

So i cant add media from here no more??

Its so annoying adding it from the top left then adding it to the prompt

[dense mortar]

Hello everyone, can anyone tell me why Omni flash doesn't see my attached references and only generates videos based on the text prompt, even though there is a start and end frame?

[broken fox]

I am facing a problem with google omni video to video, violation come everytime

[burnt oriole]

which new feature of Flow can be used as a multi scene story board generator?

I was thinking about the agent one. Though I haven't tried yet

[torn obsidian]

Does anyone else feel overwhelmed by the updates made to flow, especially with the tools?

[fair gull]

You could probably build one in the Tools.

[burnt oriole]

really? I was trying the Agent feature

and it looks really cool too!

much easier than copy pasting the prompts from Gemini to Flow

[ivory radish]

use the storyboard studio, this is a great one to start

[sinful lotus]

Does anybody know if the "top up" (get more credits) is gone now from pro plans? It just forces you to upgrade to $100 per month and doesn't show any options to purchase 2500 credits anymore (used to be $25).... really? That's a disaster if so....

[ivory radish]

use the storyboard studio its great

[burnt oriole]

oww really? is it a built in tool?

[ivory radish]

yes it is and it matches assets, characters and scenes etc.

[burnt oriole]

oo nicee! ill surely try it

[jagged imp]

Anyone else not understand the new Flow layout? Agent pops up now to confirm settings after I hit render now not before? How do you pick image or vid? When you reuse prompt it just ads it to the settings que?

oh, nevermind, I accidentally turned on Agent. doh!

[sinful lotus]

Sorry for the churn/alarm. My bad, didn't understand the new interface. You have to click on "Google Flow credits"... Then you get to this screen where there is this little button you can push to get more. See the red box I added. (not exactly obvious, but it worked)

[proud edge]

bruh is it saying that your requeting too quickly as well

[near garden]

same. i havent even touched them yet other than trying to make a character once. i'm already annoyed at the video generation. before it would do as i stated, now it randomly makes my character speaking.

[torn tulip]

I have a quick question maybe someone knows how to use Scenes tab/workplace. I created characters and generated locations. So logically you got to scene tab to build the scene and work around it, but whatever I generate in Scenes tab goes to all media and my scenes workplace is empty. Is it a glitch or can anyone explain how to use that tool?

[upbeat sierra]

You can toggle the Agent on and off in the prompt box on the bottom left hand side (only supported on Desktop)

[upbeat sierra]

this seems like a glitch, can you please share as many details in #1238621511401734254 and we'll take a look as soon as we can?

[upbeat sierra]

Hi @dense mortar the Omni Flash model does not support Frames-to-Video at this time, so you are likely using Ingredients mode. We are doing our best to resolve this as soon as we can! In the interim, Frames still works great with the Veo model family

[upbeat sierra]

Glad to hear about the video, and sorry for the poor results with Audio! If you feel up for sharing feedback at #1238621511401734254 we'd love to take a look into this

[plucky vector]

I am LOVING the tools section, such a great addition to Flow! Can we get a discord channel for sharing these? I'd love to see what other people are making!
https://labs.google/fx/tools/flow/shared/tool/954938cd-3184-4ec1-88cd-6a2cd2497a80

https://labs.google/fx/tools/flow/shared/video/bfbe0e8e-5304-4b70-a215-5473b779bc5d

[pulsar coral]

Yeah and the reuse prompt option is only at this arrow icon from the image now, there used a be an option for it at the drop down menu, but clicking this one when one's continuing a project works just like before with agent or not. The agent did get on my way too before turning it off. Funny this is like the 5th option they add in the past months that people just want to turn off... why bother.

[frosty niche]

Food For Thought(Sharing Link Feature):
It'd be nice to start a new project when you click another flow link instead of adding to existing projects via remixes button.

[pulsar coral]

Then again I wish it was that easy to turn it off at Google AI studio, at least Flow did give the option. AI assistants are annoying.

Wan for example has this option to improve and clean the prompt, Flow's agent/assistant would be much better if it'd do that instead.

[frail wave]

It's almost the same situation for me—the only difference being that everything works except for image generation using Nano Banana 2 and Pro. This has been happening specifically since upgrading to the Ultra plan on May 20th.

[proud edge]

what?? it doesnt work for me lol

i did the exact same thing with clearing cookies/cache

[pulsar coral]

That depends on what the error message says, clearing cookies is only for fixing the one that says "We've noticed some suspicious activity"

And sometimes you gotta refresh too after clearing cookies, it can be stubborn.

The "oops something went wrong" one is associated with connection problems either from Google's servers or on the user's side. Sometimes freeing your device's resources and restarting your wifi/router would improve it cause Flow needs a lot of bandwidth and ram.

But yeah if that doesn't work you need to report it at labs-bugs-feedback, more often it depends on region too.

In worst case just using a different account also helps sometimes, it's also tied to accounts for some reason, seems to happen way less to accounts in Plus plan, but sometimes not.

[burnt oriole]

Yesss haha! It's working really well and automating my entire workflow

Would submit some feedback on it after testing it even more!

[somber pebble]

Just started running into this error. Any ideas why it would say this? Havent seen it before.

[pulsar coral]

Clear cookies, if you're on a free plan don't generate more than 8 images/videos at once and wait until all of them generate/fail to generate before requesting more, or about two minutes, or you'll have to clear them again.

Plus plans can generate 12 at once, I think Pro can generate 16 at once and Ultra 20 or 25, going faster than that could cause the error again.

Oh yeah and aftrer clearing cookies sometimes you might have to refresh if it didn't work.

[fading dirge]

i know i had my say about omni and how i thought it would be better but if anyone is wondering if its an upgrade over veo 3.1, it is for sure. you can see here, and im sure you can guess which one is omni vs which one is veo 3.1

[small compass]

wow that's a fun one!

[lavish field]

flow is in a bit of a mess now..wish they would stop changing things..just get5 used to one set upo then it changes now im find it hard to see how to add and download and delete and find the orginal one i did ewtc..aghh

[main saddle]

#flow-tools・🧰 🙂

[lavish field]

Ok bought more credits i have the recipt but no credits have been added to my account? anyone have this happen?

[sudden condor]

W

[burnt oriole]

charlieeee

[somber pebble]

Refreshing the page seemed to work. Was on Ultra and hadnt gonna over that many at a time. I had agent mode create a set of 10 images a couple times, maybe that triggered it? Was an easy fix atelast

[upbeat flare]

Are we losing our 1000 credits a month for flow? I got this letter. I read the announcement that flow was staying credit based but if what I am reading they are downgrading the service to longer receive credits..
•
Usage limits in the Gemini app: For the Gemini app, we’re introducing compute-based usage limits that factor in the complexity of your prompt, the features that you use and the length of your chat. Your limit will refresh every five hours until you reach your weekly limit. As an AI Pro subscriber, you’ll enjoy a usage limit four times higher than non-subscribers.
AI credits: The product-based usage limit model is also rolling out to other products, starting with Flow and Antigravity. You can extend your limits by purchasing AI credits. While 1,000 AI credits will no longer be included as a benefit in your base plan each month, the new usage limit model that we are introducing should allow you to maintain the same experience as you are used to. To learn more about how to use AI credits, please visit our Help Centre.

[near garden]

😒

[signal roost]

curious the voices we are making for characters will there be a tts added to flow or the ability to use them else where?

[rigid nest]

As a daily reminder, we have a poll to report the most common Flow issues. Please cast your vote.

[errant gull]

My flow still can't generate images after 5 prompts (i am paid plan btw) 🥲

[rotund elm]

Either you hit a censor filter, or the images for n2 and npro don't work. 🙁

[strange forum]

Brahhhhh what is going on yall costing me money

[rotund helm]

Can you officially be just clear about whether you've rolled it completely globally or not? Have been getting "policy violation" errors trying to use omni in flow repeatedly only to hear this on gemini.

[flat quiver]

I remixed the Google Mockup Tool tonight to add some things I personally find helpful. My updates are far from perfect and still WIP, but try it out and let me know what you would find useful: https://labs.google/fx/tools/flow/shared/tool/da378d07-8cec-479b-83ba-91035e7d6d2f

I added a couple print mockups where you can specify your own title and subtitle, like these

[onyx pulsar]

Can't generate images on Flow on Ultra Plan.

[fallow dove]

Same here — I can't use tools either, not even simple image generation.

[meager lion]

i upgraded to ultra recently and everything stopped working. should i go back to pro and buy credits.

[surreal drum]

That's not fighting. You have 8 seconds for a fight. You don't have a winner. Here is Pixverse 6 for fighting with my new way of prompting. Veo 3.1 was the reason I have new way of prompting to speed up the fight process. Google has not yet created a model for fighting.

[pulsar coral]

Chinese models do show proper fighting videos, and that was months ago

Admittedly it wasn't perfect yet either, but I haven't seen a VEO video reaching close to that yet

Anyway there's a goal to reach

[ivory radish]

in flow if a prompt is invalid or needs to be filtered for keywords, why does it wait for a minute till the end of generation and say its not as per policy. why cant the engine check the prompt immediately and throw the error before generation.

[surreal drum]

Can you show a fight video without cartoon fighting? Veo 3.1 had themost graphic fight video I have created and seen. I prompted for a liver kick and got one, but the opponent's guts are hanging out.

[pulsar coral]

Nah I mostly save anime stuff, but the render is irrelevant. If you want to see more real people fight you can look for Seedance stuff in youtube.

Also the trick is mostly to add this around the last prompts of your video.

[Fixes: No unnatural clipping/intersecting/overlapping. No occlussion errors.]

That way all limbs and weapons collide instead of ghosting through each other, I've seen VEO also understands it

[surreal drum]

It looks like Veo 3.1 was trained on Mr. Rogers.

[surreal drum]

Here is my Pixverse fight video from yesterday using my new method for prompting. I wish I could try my new method with Seedance, but I can' afford a subscription. most of the time

[pulsar coral]

Your prompting seems fine, it's VEO's habit of adding jump cuts what's visibly a little off there

[surreal drum]

But my new method of prompting bypasses Veo's filters.

[pulsar coral]

Yeah having to go through those could also cause small imperfections, but it's a big problem

Facing that prblem with image generation too, and it often messes my ocean water for some reason

[surreal drum]

My new method bypasses all model filters I have tried. Here is Seedance 1.5 pro using my new method of prompting.

[pulsar coral]

I saw it already and yeah it's good, maybe you used ketchup lol. Man you might wanna delete that video asap, mods are very strict around here.

Been losing friends here left and right. All good now tho, but yeah you did make a point.

[reef mural]

Hm after getting the new cheaper ultra i cant generate any images in flow, any one else has this problem?

[echo spade]

Everyone is having this issue. Google has done nothing to resolve it.

[brave wren]

Me too. I tried generating an image about 8 hours ago and again just now, but it hasn't worked even once.

[reef mural]

Thats really bad, i have it as an income making videos so its really important that this works

[brave wren]

I also want to earn money through video production. How can I get clients?

(As a side note, when I encountered the "Antigravity" issue, my access was unilaterally restricted for a week. I contacted support, but I didn't receive a meaningful response. I encountered the same problem twice, so my project was left with a two-week gap.)

[pulsar coral]

Seems Plus tier and above can request 25 generations at once again today

[native barn]

Is the 50% off the Ultra plan for 3 months intro promo gone now? I only see the $249.99 monthly price.

Google...

Is the 50% off the Ultra plan for 3 months intro promo gone now? I only see the $249.99 monthly price.

[pulsar coral]

https://one.google.com/ai?utm_source=flow&utm_medium=web&utm_campaign=flow_account_menu_upgrade&g1_landing_page=75

Aren't you going there? Ultra should be costing like $99 now.

Whatever the cause

[open plover]

Did the omni model just jump from 30 credits to 120 ???

And Quality now 400 credits ???