flow-general・🌊

[edgy holly]

Only put the image in the second one if you want it to end looking like that as well.

[fair gull]

Yeah, it looks like a project among your other projects. It says Ingredient Archive.

[turbid umbra]

checking

[shy void]

But my animation zooms into the image, so the scene should end closer to the stadium.

[atomic plume]

yup, image generation costs 0 credits!

[turbid umbra]

by any chance do you know if the Ingredient Archive is amongst the first projects? because I have like hundreds already and it will take me 30 minutes to scroll at the bottom I think.

[viscid mesa]

it should be closer to the top! but can't guarentee it will be the first

[turbid umbra]

yea, people claim they can find it still cant' 🙁

[elder mason]

unlimited no daily limits?

[edgy holly]

In my UI it was 2nd, in the top row.

[turbid umbra]

guess will take a while to roll out to me, I got thousands of generations at this point maybe thats why

[arctic kelp]

nano banana 2

😚

[edgy holly]

Not sure if they changed it with this update, but I believe it was 100/day with Pro, and 1000/day with Ultra.

So it doesn't use credits, but there is a daily limit, or was.

[open goblet]

I found a minor bug if anyone can spot the mistake here 😂

[atomic plume]

yup, image generation costs 0 credits!

[coarse bane]

i dont find my old frame scenes is there any tab for that ?

[wicked crater]

Nano Banana vs Imagen 4

Imagen 4 in flow shows a black outline for some reason...

[scenic swan]

hey anyone can genrate video for me

i can give you prompt&iamge

plase this is argant 3

[stuck adder]

Idk if it s just me but I don’t like the new layout lol

[coarse bane]

old ui was better tbh

[wicked girder]

unable to generate videos since yesterday its really annoying now

[stuck adder]

Agreed

Is NB2 on flow as yet?

[coarse bane]

flow is not working

[scenic swan]

hey anyone can genrate video for me
i can give you prompt&iamge
plase this is argant 3

[scenic swan]

yes sem problum

[coarse bane]

NB2 ?

they just made it weird

[stuck adder]

Nano banana 2

Just got released today

[scenic swan]

to day my first clint want demo video and now its not working

[coarse bane]

that also comes under flow it might not work maybe

[compact phoenix]

Still getting constant "Failed Generation" errors here. Checking the console, it's a 403 Forbidden/reCAPTCHA block again.

This is becoming a huge waste of time for my production workflow. I’ve tried clearing cookies and relogging as suggested, but it only works for a couple of prompts before breaking again. We really need a stable fix from the Google team.

[scenic swan]

any one know object talking video

[pulsar coral]

Yeah but at least from my Nano Banana pro prompts, more at passing at Flow than Gemini. I guess that's gonna depend on what each person is trying, I had abandoned those prompts already, but at flowed they worked 3 outta 5, and that's a lot really.

[scenic swan]

any one can crate for me object talking video my first clint want demo video please help me

[pulsar coral]

Nano Banana 1 at flow is even more permissive than Nano Banana pro, but not at Gemini.

[scenic swan]

hey bro u know crate object talking video

[lament galleon]

'Upscale failed!' Can't download videos. My subscription is coming to an end , time is valuable. What's going on??

[pulsar coral]

Yeah I did also find enough workarounds at Gemini but it took me a while, but thing is I tried my old un-ambiguous prompts and they worked at flow like nothing, so now it felt like a waste of time to be finding workarounds.

[edgy holly]

To be clear, I'm not talking about workarounds. I just don't make anything that's anywhere near their rules. When I say Gemini is better at prompt interpretation, I mean it just seems more intelligent at understanding what I ask for.

[pulsar coral]

Well again that'd depend on what each person's working on I guess, but I do use specific prompts for anime screenshot art style for example, haven't seen much diffeerence on that one.

In fact I think I better run some tests again now

[gray lynx]

Hi, I appologize if this was already brought up but when I extend my videos, I am sometimes not getting accurate character consistency. For example I have an adult in the first frame and then they are a kid version in the next??? I know the prior version "add to frames" kept a pretty good consistency for characters. Is the team working on this by chance?

[manic sierra]

Thanks for the screenshot, but as I mentioned, the interface on a tablet is different—even when requesting the desktop site. The "Reuse prompt" button simply doesn't appear for me.

As for why I'm using a tablet: it might be a wild theory, but it could be because it's the only professional device I have access to, as I stated in my post. Not everyone has a PC or laptop at hand, and a professional tool should ideally work on the hardware the user actually owns. 😉

[edgy holly]

Extend uses the last second of video as the reference, so if the character is facing away from the camera, for instance, then turns back in the extend, then they'll look different. Either try to end the first clip with all you need to see in the extend, or just do a cut.

[gray lynx]

ok ty for the tip

[wicked crater]

Here's another comparison:

Imagen 4 (Non Whisk) vs Nano Banana Standard.

Oh and where's the share option?

It's not yet implamented.

[edgy holly]

I have the same issue, but most TV shows, movies and especially online video cuts every 5 seconds or less, so if you take the scene and characters, use Nano Banana Pro to reframe it, then start from there, it's generally better.

[summer rapids]

3 things on the new extend UI and how it works :

• The new extend version is really hard to use. Before, we could create 5 batch of 4 videos, watch them very fast. Now, it's one a a time, in a tiny part of the screen, you have to scroll all the way up to the first video to create a new one, and then scroll all the way down to watch the result (which is merge so you have to move 8 seconds after to whatch the result, ....). And of course, you can't remember where you were on the viewed videos, ....

The extend version should be done by batch of 4 (and not by 1), not stacked under the initial video, but visible in the main page (I know there is an option to add it to the main page but it must be done by default and not with many and many clicks), and should be reviewed after the 8th second.

I have to do maybe 15 click and 15 scroll where I had to do 2 clicks before

• There is not the option to reuse the prompt for extended video

• Also add the option to extend not at the end of the video but at the frame we want

[wicked crater]

Since when they have to do separate subjects, not merge.

[cinder sail]

now its 10 coins per video on flow ultra? it was 10 for 4 xd

[glossy sparrow]

is nb2 already available in flow?

[edgy holly]

Not in mine yet.

[crisp basalt]

Sadly, the feature isn’t really designed for use cases like yours (I’m also facing similar issues). Especially if the video you tried to extend had the character facing elsewhere (like from the back), the reason why the extend feature didn’t maintain subject consistency is that it picks the last frame of the video you’re extending and proceeds from there

[glossy sparrow]

cant wait to already use it 😞

[scenic swan]

@edgy hollythank u so much bro for helping me to creating video in flow

[edgy holly]

My pleasure. DM me again if you run into trouble.

[scenic swan]

sure bro

[warm orbit]

I think it is, nb works so much better now

[burnt oriole]

Is nano banana standard nb1 or nb2?

[edgy holly]

nb1

[burnt oriole]

Oohh

[main saddle]

🍌!

[burnt oriole]

it's finally hereee

[wicked crater]

Where's an otter...

[burnt oriole]

gemini_banana

[upbeat sierra]

@edgy holly you may need to log in / log out, but excited to hear your thoughts!!

[burnt oriole]

we are here to make!!

[edgy holly]

I'm in a AI film live chat on X, but I will do that after it's over.

[atomic plume]

+1 it should be fully out, let us know if you can't see it and if you have any feedback!

[quick rapids]

nice, they didnt replace NBP like gemini app, looks like flow is now the main app for NBP

[copper plank]

Can I just say, amazing job to the team. This is phenomenal.

[wicked crater]

I can hold 10 subjects.

[queen lichen]

After you new update i am not able to edit the extended video😭

[burnt oriole]

i think it's taking more time to generate and it's going upto 99% on the progress numbers

[faint jackal]

Bro, that thing mean the output gap, not the generation, if you use again the same prompt you could generate even a better output. The existence of that outline is the proof of working

[upbeat sierra]

Nano Banana 2 is under heavy load, as many people are excited to try it. We may see delays / timeouts, hoping to get to stability as soon as we can

[stray remnant]

did they fix the newest update lmao

[prisma holly]

Nano 2 is better than pro or ehat

[summer rapids]

same quality as Nano banana pro here 😉

I'll wait

[faint jackal]

JAJAJAJAJAJAJ

[hollow warren]

So whats the limits chnaged too mow ?

[edgy holly]

Okay, yeah. I've got it now.

[brave pawn]

@burnt oriole @glossy sparrow

[edgy holly]

The live chat I was in was psyched about the updates. Christ Hatchett from the Film3 group particularly keen.

[burnt oriole]

hi?

[wicked crater]

Wish if they have assets workshop in the future.

[edgy holly]

Assets Workshop?

[wicked crater]

So that way, I can share ingredients to the workshop where people can use it. I don't know.

[timber ruin]

Hi, how are you? Can you tell me how to add him to my group?

This bot

[edgy holly]

Oh, like sharing stuff publicly. That's a good idea.

Oh, 'Workshop' like on Steam?

[wicked crater]

I hope it can be released after the Whisk shuts down.

[summer rapids]

Still can't modify images with minors with nano banana inside Flow even we can create video from it inside Flow.

That works inside Gemini but can't download / upload to flow due to restrictions

Kind of same problem than the "save frame as image" from a flow video problem I think @fierce nimbus

[worn fable]

Sorry if this question has already been asked, but, are there hourly/daily/weekly flow caps for NB2?

I know NBP was zero credits but there is a hidden usage cap.

No biggie if so, in fact I’d expect it, I’m just trying to figure out where I want to do my ideation, and when I want to save usage in flow.

[bright pine]

When are we getting a negative prompt field?

[small barn]

I still can't upscale to 4K, despite the announcement that it's now available for "everyone". Is this a bug?

[crisp basalt]

They’re working on it

[fierce nimbus]

Here for a bit. Ask questions. Team is working on implements fixes based off the feedback.

[crisp basalt]

First of all, how are you

🙂

[remote maple]

For the second day in a row I have not been able to generate any videos... product isn't working

[edgy holly]

Just two quick tests from some longer prompts I'd used in the past... NB2 feels better than Imagen 4, but a step or two below NB Pro, ish.

[small barn]

Thank you for the very generous updates. Can you explain the fundamental difference between NBP and NB2 (besides the speed of course)?

[fierce nimbus]

HA! Thanks for asking. Fun day to show what the team has been working on for a bit now. Its a major update (as you all see) and something we couldnt do in pieces. From our testing with creators the new Flow helps you MOVE. Manage, Create. And we are set up to offer things like views (batch/grid) which is HUGE! Theres stll a ton to make better with feedback from all of you.

[fierce nimbus]

Faster, Cheaper, Near Pro quality. All around amazing update

[prime pewter]

When I generate with flow the accent in the scene of the videos are different each time I generate, I don't really know how to go about fixing that to make it consistent . I used the same voice prompt for all scenes -

"Voice/Audio Notes: Read with a crisp, modern British accent. The dialogue itself remains exactly the same, but the delivery should sound authoritative, sharp, and slightly exasperated—like a successful friend giving you a much-needed reality check. She isn't shouting; she's speaking with absolute, piercing conviction."

but every time I generate something its like a dice roll for the accent

[crisp basalt]

also yes u can share ingredients too soon (also this is just a hypothetical ... what it would looks like, thank u nano banana 2 for this image 😄 )

[midnight cedar]

try text-to-image on nano banana 2. it's really great

[fierce nimbus]

Voice consistency across clips is a top ask. We hear you but have no update at this time. Best workflow for this right now is prompt layer but as you are finding out its not perfect.

[spark hornet]

Hi there! I’m using the new Flow UI and noticed that the gallery filters reset every time I generate a new image. Is there a way to 'lock' or persist these filters? Specifically, I’d like the view to stay on 'Generated' instead of reverting to 'All' (which includes my uploads) after every prompt.

[fierce nimbus]

Did it display a failure reasoning? This obviously is not intended.

[small barn]

Have you considered adding the upscale button to the hold-tap pop up menu? Btw the new mobile version is so great!

[crisp basalt]

youre right ,, its actually really good,,

[edgy holly]

2nd test. Similar overall feel between the two, but NB2 applied style prompt details I had at the end of the prompt less... aggressively? Not sure how to describe. Same prompt.

[wicked girder]

If it works im unable to generate videos since yesterday and its really annoying

[fierce nimbus]

Great comment! Let me look into it

[edgy holly]

I feel like NB Pro is a bit less anachronistic. They're all prompted to be "A slightly futuristic robotaxi SUV". NB 2 added those cameras sticking out to all of them, and they look very similar, so less range, in a sense?

[edgy holly]

On a scale of 'VEO 5 WEN' to 'GIVE ME EVERYTHING FREE'... how much sleep did you get the last 2 nights? haha

[obtuse snow]

I’d love the ability to storyboard sequences with NB 2 and Que them as sequential/ parallel generations in flow to form a sequence but that might be too much of a power user feature!

[small barn]

Also, I use Flow mainly for image gen. Would love it if all the Nano Banana aspect ratios were available instead of just the 2 🤞

[fierce nimbus]

let me add that to the list. Someone else mentioned the idea of a setting toggle to "only show generated" in which the asset panel would still have all outputs but not uploads

[midnight cedar]

we're working on it 🙂

[fierce nimbus]

HEARD. We're digging into generation problems now. Obviously not intended.

[spark hornet]

Thank you @fierce nimbus 🙏

[edgy holly]

I'd love this same basic functionality, but instead of running all at once, I'd want to be able to use ingredients inside the 'Extend' workflow.

[fierce nimbus]

This is my fave part of the job. We are trying to build the best surface. So all this feedback (good/bad) is what i actually love.

[latent jetty]

i cant seem to find my ingredient archive folder that you guys said we would have?

[fierce nimbus]

We hear this one all the time! Great request

[fierce nimbus]

Also an easy fix. We heard that and its in our shortlist.

[crisp basalt]

is

are u guys considering addin the option to choose duration for a video generation

[edgy holly]

Gemini 3.1 Pro Deep Think says this translates into an average of 3 hours of sleep, a half bottle of TUMS, immeasurable coffee gallons and several ibuprofen. (not really... back to being serious)

[fierce nimbus]

What would you want to see?

[latent jetty]

anyone know where this is where can i find it

[wicked girder]

thank you

[summer rapids]

3 to 8 sec with 1 sec increment

[cobalt junco]

all I know is I am having an absolute blast right now, thank you team!

[edgy holly]

4/6/8 like it available in the API would be fine.

[crisp basalt]

yes thiss

[edgy holly]

Hey Button. Haven't seen you around for a bit.

[cobalt junco]

lurk mode, how have you been?

[latent jetty]

Does anyone know where i can find ingredients archive in flow

[edgy holly]

I think we talked about this

[crisp basalt]

https://tenor.com/view/yea-right-sure-gif-22336293

we well can always talk about it again and again

[edgy holly]

First rule of AI fight club...

[obtuse snow]

That would be a really quick win for me! I know I made another suggestion but last I was a video producer just a parameter for length since we often know how long we want shots to be and it would save you guys processing power 3/6/8 or just honestly a slider between that think it’s a quick win which would have huge implications for ai filmmaking

[fierce nimbus]

So obviosuly our team new Nano Banana 2 was coming. We think the new Flow is AWESOME for image workflows using Nano Banana 2 and Nano Banana Pro. Thigns like drag and drop. Dedicated edit space with history, tools like marquee and doodle.

[crisp basalt]

and crop

[latent jetty]

do you know where i can find the ingredients archive folder in flow

[crisp basalt]

wait sorry ill let u finish

[crisp basalt]

scroll down in the main project page

mine was buried far

i think i saw my whisk creation too in there

[edgy holly]

Weird. Mine was right at the top middle.

[fierce nimbus]

Yea let me see if we can bump it up to be "recent" to bring it to top of projects.

[cobalt junco]

soooo, what's everyone doing for data management these days 🤣

[crisp basalt]

its becuz u always want to be on top of the list!

[crisp basalt]

thank u sir :))

[crisp basalt]

dyin

[edgy holly]

I am inevitable.

[cobalt junco]

o whisk assets are they importable already?

[edgy holly]

G Drive fer dayz

[crisp basalt]

not yet i think, but soon i hope

[fierce nimbus]

but like what do you all want to see. Remember we always work on this thing. And with the combination fo the Whisk team we have more people to help build awesome stuff for all of you. So besides "Veo 11" and "Free" what are you looking to see

[crisp basalt]

.

[cobalt junco]

veo 12

30s clips

[fierce nimbus]

Also same team who built whisk built flow. So its not like we dont know the product or its users. We love em both

[cobalt junco]

this is all heading in a amazing direction

[midnight cedar]

going with the Flow 😁

[fierce nimbus]

@midnight cedar was Whisk teammember # like 1 or 2

[cobalt junco]

peprigrge farms remembers

[latent jetty]

tighter restrcition on using famous people faces and known characters cuz some people out there create god knows what using peoples faces so maybe more restriction to keep things original in projects

[cobalt junco]

Whisky 🤣

[edgy holly]

Ingredients inside Extend would be cool.

[crisp basalt]

Is it possible to provide a character reference image when extending a video? right now, the extension relies on the last frame of the clip. its kinda of an issue if the character is facing away in that frame, as the AI doesn't have the visual data to generate their face for the next segment.

[crisp basalt]

damn i typed all of that for nothing

jk

[edgy holly]

That's the biggest use case... haha

echo

[crisp basalt]

https://tenor.com/view/no-hit-no-thanks-paw-smack-gif-16247479

ooh i think that question actually shook the flow team to the core that we havent seen anyone typing

[fierce nimbus]

I gave it an emoji reaction. I knwo the request well and have had it on my list of wants since 2022

[sage hare]

im curious

We've been limited to 2 aspect ratios for a long time. When will we have support for multiple aspect ratios?

Is this a front end challenge for displaying those images or something else?

Also

4k upscaling is still not available for everyone as mentioned in the announcement

[crisp basalt]

LOL , got it. dont worry i was pullin yall leg

[edgy holly]

Getting granular.. not sure how to word this, but here is an example... some sort of "pathing" control. I've tried it using input images like this, but it is more miss than hit.

[fierce nimbus]

Different models support different aspect ratios. We are looking to enable all supported but it is not just about front end. Its more about model support

[edgy holly]

*second

[crisp basalt]

this looks like those generic paranormal youtube video (not fake , real!)

[summer rapids]

The model can extend at a given frame of the video or it must be always the last frame ?

[sage hare]

however, the model supports a lot more aspect ratios in gemini

or does gemini use a different model variant

[crisp basalt]

image or vid?

[sage hare]

image

nb2 and nb pro

[edgy holly]

An outtake from a scene where the guy in the foreground (agent) is there to meet a paranoid C.I., so he can watch who is there, but they can't see him.

NB Pro and Veo 3.1 did a pretty decent job with the night vision stuff.

[fierce nimbus]

same. So we are working on exposing all the ones we can. On the list.

Also if this is how everyone likes to get feedback i can talk about office hours where you all can ask me and teammbers questions. No promise we can change everything but happy to chat about the product.

[sage hare]

ah nice, thanks for your effort!

[edgy holly]

Being able to upscale whole multi-clip scenes would be very useful.

*especially if we could pull up old multi-clip scenes that had been exported in the past

if a record of that exists

I'm gonna spend a few hours organizing into folders.

[azure glacier]

My biggest challenge is getting consistent voices and ambient sound between clips. It would be great if we could describe a voice, give it a name, and then be able to reuse that voice consistently in multiple renderings. I am happy with the voices that Veo generates. So I just want an easy way to reuse that voice whenever the same character is speaking. (I recently retired from Google and am loving spending my free time coding with AntiGravity and making films with Veo.)

[edgy holly]

^ a bridge between now and what phil says (same)... basically port the most relevant basic video editing functionality (including access to stock audio, for instance) from Google Vids into an applet inside of Flow?

[fierce nimbus]

you can also work WITHIN a collection. So like make one and jsut start jamming in there. So create collection first and then generate inside. Or after and drop things in

[radiant leaf]

About the new update. Zero I assume it means that we are stuck with no credits to try until the new billing cycle. Am I getting it right?

[edgy holly]

Just figured out how to move folders in and out of other folders. That's helpful.

[quiet wadi]

Hi

Are you familiarizating with the new update yet?

[edgy holly]

That's been the big conversation today. Scroll up 🙂

[quiet wadi]

Have arrived now , sorry

[edgy holly]

I think free accounts get 50 credits a day.

[quiet wadi]

the main problem is that you need the Ingredients Project opened while you are starting a new project, but until now you have always the resources on every project

[fierce nimbus]

We are workong on making it possible for you to access them from any project. But the overall goal is to have uploads/ingredients stay limited to a project.

[quiet wadi]

That´s not worth because users want to reutilize the resources that have used on different projects yet

[spark hornet]

Has the Flow output format changed? I noticed my recent images are PNGs, but I remember them being JPEGs before. Is this a new update or am I seeing things?

[minor rune]

No 2k or 4k on nano banana 2 for pro 🥲

[radiant leaf]

I have PRO

But in my region I have 200 for Flow and I'm down to 40 now and my 3 regulars with NB

[spark hornet]

I have 2K on Pro but no 4K (they said they are working on 4K for NB2)

[edgy holly]

They issued a correction. 4K is for Ultra accounts.

[spark hornet]

Ah ! thanks for the update

just saw it

[minor rune]

The message says 2k is for ultra

[thick shadow]

Video Generation is completely unusable here.

[minor rune]

Pro could have it too?

[spark hornet]

I'm on Pro and I have 2K

[minor rune]

Ok the message is not quite clear then

[civic dagger]

can anyone help me with this??

i tried generating image, using nano banana2 & pro

[latent jetty]

who else thinks we need a mobile version for flow to make it easier to use on tablet and phones maybe like a standalone app or something for the app store

[summer rapids]

Big thanks to the team @fierce nimbus , I now can "save frame as image" from Flow videos with babies like in old UI. I can work again 👍
Hope it will be as easy to fix as that for the same problem with nanobanana

[midnight cedar]

apologies for the confusion
4k upsampling is what's only available to Ultra users
2k upsampling is available to Pro users

[minor rune]

So the same as before

[midnight cedar]

exactly. no changes

[fierce nimbus]

what was the error?

[minor rune]

It would have been nice to add 4k even limited to pro users

[civic dagger]

idk when i tried to generate image, it showed me that, like i tried every model from imagen 4 to nano banana 2, pro

[quiet wadi]

I have a doubt if I am generating images of a specific thematic (like an specific woman of an specific anime for example) and I turn out to other thematic (Rias Gremory on a far Isekai ), and then I return to the first topic, then the second topic could disturb my images of the first one if there are in the same project

[cerulean meadow]

"Oops, something went wrong!"

On every attempt to generate.

[fierce nimbus]

Image/Video/Model? Prompt? Let me know some of these details and ill work on getting it to team

[stuck cosmos]

why i cant see anymore the edit tools

[edgy holly]

First one is the input image someone DMd me. Second is NB Pro. Rest are NB 2 (BETTER / wackier / stranger, haha)

[jolly silo]

all image generate, with nano banana pro or nano banana 2,

[faint jackal]

Hey Nick! You see my response on X? Emm... Im not to good in english but I mention the structure boxes of whisk are in internal structure (my theory) good to category and clasify to the LLM what is the Style, the scene and the character.

In flow using nano banana pro we could use this even much better cause the 14 character consistent and probably in the NB2 it will be in same ways.

But flow doesnt got the boxes of style, character and scene.

And other thing that could be less important but for me is really great is the posibility to edit the text without change the image. This will do in the LLM some crazy outputs. I dont know how it works inside but I suspect that the data inside the image is important to the LLM to understand the image. If we could tell the LLM what is he seeing it will really improve it. I got proof of it cause that was what I was doing on Whisk.

[dense lichen]

@fierce nimbus what to do about this ? Image ref is the one you see, model is same as show in text input and propmt is a normal cinematic prompt.

[faint jackal]

If you edit in the box of character as example "[CHARACTER: Person]" at the start without change any of the text down. The image and the text will be the same.

And if you insert another text changing what was wrote but keeping that "[Character: Person]" you can tell the LLM what are YOU seeing in the image and what it must see in the image. This increase the presition over 80% with NB (I use it on whisk) but I know it could increase so much in flow with the access to NB pro and NB2

[nocturne kernel]

i'm also facing the same problem

[timid hinge]

Probably already asked and answered, but... Have we lost the ability to animate an uploaded image? This new UI has me hopelessly lost!

[faint jackal]

Why is so important to me? Cause the micro details that NB couldnt lose because he knows what he is seeing. I dont know its like some kind of feedback hibrid IA / Human

[hushed bluff]

Hello.

[dense lichen]

hello

[dense lichen]

@midnight cedar can you help me with this ? Press on the message as there's an image attached for the problem

[nocturne kernel]

gemini working properly but flow not

[dense lichen]

For Whisk Getting Resource Exhausted

[viscid mesa]

Try using the + button in the top right of the UI to upload! You can also drag in any image directly into flow to upload. You can animate them by switching to the "Frames" option and adding it as a frame

[dense lichen]

@viscid mesa anything related to above issue ?

[thick shadow]

@fierce nimbus Is video generation down with Veo?

[timid hinge]

@viscid mesa Thanks. I think I've got it now.

Although I'm compelled to add that I really dislike the new UI. Sorry.

[vocal bridge]

i noticed after multiple fixess in same image, quality kept getting worse...

[viscid mesa]

we're looking into it!

[cerulean meadow]

Sorry got pulled aside. Nano Banana 2 [x2]. Here's a screenshot:

Have also tried with 1x and 4x. And have tried deleting and recreating the project.

[edgy holly]

It's set to Ingredients by default. Switch it to 'Frames'.

[timid hinge]

@edgy holly Thanks!

[edgy holly]

There are no bots in here. You have to go to Gemini or the flow web app - https://flow.google - to run prompts.

[tacit aspen]

@fierce nimbus I noticed something changed under the hood on the animations, like feels like VEO 3 maybe got an upgrade? Just seems like its different but good different.

[tardy flicker]

Please don't make this change, don't replace Whisk with Flow, don't merge them into one. Whisk is perfect as it is; it doesn't need this change.

[worthy scarab]

Please don't make that change or merge with Flow. Whisk is perfect as it is. I've tried Flow and it's nothing like it; it makes the images look awful, and the consistency is very basic. Besides being heavily censored down to the smallest detail, I've spent years perfecting a character I created and I finally found an AI that does it very well and has no complaints, Whisk came to save it. But Flow, I tried all three modes and none of them work well, everything. I hope that in the end they keep them as they are. 😓😢😥

[arctic kelp]

me too

[white gorge]

I'm agree with the latest messages, I was trying to work in Flow as I do in Whisk and I'm unable to add subjects, scenes and styles. How can we replace a tool that offers that great flexibility with a rigid one? Is there a way to have those options in Flow?

[edgy holly]

Nano Banana is much better at style transfer than Whisk, but you need to learn how to prompt it. Whisk sort of does the prompting for you. There's no one way. Depend on what you're doing. Google something like 'style transfer nano banana pro prompt', and change the search parameters to only show you results from maybe the last 3-4 months.

[solid moth]

Zero credits really because I hit the daily limit

[fierce nimbus]

all same prompt?

[robust socket]

how to add styles?

[fierce nimbus]

Add the style image to the prompt box of Flow and write a your prompt asking for it “in this style” or in the style of “the image with the superhero”

[rugged oxide]

Does google has anything like N8N ?

[cerulean meadow]

Yes, I discovered that the prompt length seemed to be a significant factor. Shortening the prompt got me over the line. Also noticed that errors occurred when trying to generate 2x or 4x; only one of them would make it past the line - the others would always fail.

[fierce nimbus]

Team is working on reducing errors/blocks. Obviously not intended.

[turbid lintel]

brother when veo 4?

[cyan thunder]

Hello, why are the downloaded images no longer named with a timeline? Has it been modified?

[primal zealot]

can anyone help me with this??

[upbeat sierra]

We are pushing fixes to try to reduce some of the disruptions we're seeing.

• Nano Banana 2 is under heavy demand
• We are pushing other fixes to improves erros

For those unable to still unable generate previously, please let us know #1238621511401734254 and we'll try to find resolutions

[pulsar coral]

Where did Nano Banana 1 go at Flow.

[upbeat sierra]

We just pushed a couple of fixes that we hope help with some of the above issues

[pulsar coral]

It was there this morning and working just as fine as at Whisk, that's why most could still move from Whisk to Flow.

[upbeat sierra]

Are you having issues with generating images on either Whisk or Flow?

[pulsar coral]

No, I'm having issues because Nano Banana 1 is not anymore in the drop down menu to select models, and that model is why all were using Whisk. Without it Flow is not the same at all as Whisk.

Is it gonna be put back there.

Being clear about it, Nano Banana pro and 2 completely refuse to do things Nano Banana 1 does asap without probs.

[glad sonnet]

I can’t generate anything atm. Everything fails instantly.

[pulsar coral]

Maybe it's because Nano Banana 2 and pro share a daily limit, which Nano Banana 1 didn't have at Whisk at least, or not even close to that. Edit, or at least it did give me that message a while ago, it's working again atm.

[modern lily]

Hi All
I have created a video using flow, total 5 clip of video, now since the 5th clip is not as per my expectation therefore want to delete the same, i tried already deleting from scene builder, but still 5th clip there, also how to add a frame in extend 5th clip video, those who know please answer

[pulsar coral]

Anyway, it would be nice to know if we're gonna get Nano Banana 1 back at Flow or not cause according to that a lot of people are gonna have to start looking for it again to have Whisk quality images back.

[fierce nimbus]

Whats not working for you? Style not the same? Prompt that worked previously doesn’t work the same?

[pulsar coral]

The prompts all work fine. Like said that's not the issue. The issue is the lack of the Nano Banana 1 model at Flow's drop down menu image model selection, it doesn't matter how things are prompted, a model does something in one way, and another model does it in a different way. It's like when you use LoRas.

When you want the same result with an image you don't need a prompt change, you need the same LoRa/Model you were using.

Same reason we need Nano Banana 1 model in the drop down menu where it was.

That's how Whisk worked, in base of that model.

I think I will soon not be the only one who noticed.

[fierce nimbus]

Our team built both Whisk and Flow. We know both systems well

[pulsar coral]

Nice. So is Nano Banana 1 coming back tho?

[pulsar coral]

Also I don't doubt you all know what you're doing, but since you keep asking for prompts I left more feedback at the ⁠whisk-general・💭 that I can't link because it ain't allowed at your same Discord for some reason.⁠

[fierce nimbus]

Sorry about that. I dont know/control rules of discord. Ill head over to Whisk general. Thanks for the deep dive. Helpful to see

[scarlet quarry]

2 Questions:

- Question 1:
I read the following in labs-announcements :

"Nano Banana 2 is here 🍌
What does this mean for @Flow?
0️⃣ Zero credits for all users"

How much is the daily limit?
Is this for the free tier, or is this for the subscription tier?

In the free tier it says:

Imagen 4: Generating will use 0 credits
Nano Banana 2: Generating will use 0 credits
Nano Banana Pro: Generating will use 0 credits

How much is the daily limit here?

- Question 2:
Is Nano Banana 2 the new version of the first Nano Banana? Or is this the second and updated version of Nano Banana Pro?

[compact kraken]

I'm new to this and I've run out of credits for creating on Flow. How and where can I see when they renew? Thanks. I don't know if there's a channel where I can write in Spanish, or if this is the right place to ask this. I hope you understand and are patient. Thank you.

[fierce nimbus]

Ill get back to everyone on clarifying credits and daily limits. Want to ensure I’m responding with accurate info.

Nano Banana 2 is the successor to Nano Banana. It sits alongside Nano Banana Pro.

[scarlet quarry]

Hi Nick, perfect, thanks for the reply. Looking forward to the info!

[compact kraken]

Today is the 27th and it seems the credits are being renewed today, right? Do you know what time they usually do it?

[zinc imp]

Anyone able to access flow in India ?

[potent ivy]

have google removed adding subject , scene and style existed in whisk , after moving to flow with nanao banana 2 ?? is option there Or Removed

[spiral python]

how to extend video in this new flow?

[pulsar coral]

Just please tell us if Nano Banana 1 isn't coming back. Cause a lot of people do say at the chats they're looking for alternatives already and it's probably better to know beacause most think Whisk is gone forever because flow doesn't have Nano 1 for the same kind of images Whisk was making. Nano pro/2 can't be Nano 1. Nano 1 is not better, it's just different so it's necessary in the drop down menu.

[potent ivy]

also in new flow , after we generate 4/8 images and we click on one image to see , do give option to cycle through them using arrow keye otherwise you need to go back to each generated image click and open individually , to check closely

[fierce nimbus]

Have this on my list. Want to enable it ASAP

[pulsar coral]

Like unknown said, "silence isn’t empty; it’s full of answers" Time to find a way to get Nano 1. Nano 2 is everywhere anyway, and at Gemini it allows aspect ratio to be freely prompted.

Here it's locked at 9:16 16:9 so yeah.

[fierce nimbus]

Working on getting more aspect ratios supported in Flow. I dont have a clear answer for you so cant answer.

[potent ivy]

also will you add seed support its generating quiet similar for same prompts

[fierce nimbus]

Click on a video generation. Then click “extend”

[supple pendant]

Hey all - question, cauase I didn't see any answer for this 😛

When we do a text only generation with the new Flow, which model is conjured up for that? Sometimes my outputs are listed as using different models than what i selected

[fierce nimbus]

Model can be selected in the settings panel of the prompt box. Model can also be found in info on every output

[pulsar coral]

Sounds good enough, well time will say.

[supple pendant]

well, my question more specifically is - if i do a text only prompt, and it says it's Veo 3.1 quality, is it Veo 3.1 quality start and end frame?

Because visually it looks more like start and end frame's quality than text to video, and i noticed we can't select text to video anymore, so i wasn't sure

[pulsar coral]

Also since we're talking about ratios, at Gemini putting this at the end of the prompt controls ratio freely up to some limits

Aspect ratio is 16:21

Or almost whatever ratio works, if it's too ridiculous the prompt just fails.

idk how practical would it be to implement that here, since it's mostly a video site

But I do see there's a chance for free form ratio or something alike to work

But the ratio prompt only works for Nano Banana pro/2. never worked for Nano 1.

<@&1366831844502274180>

[cyan thunder]

May I offer a suggestion? I hope that images can be downloaded directly from the view mode interface, instead of having to go through a secondary menu to download them.

[spiral python]

unlike the old one, I've tried it and if there's something that's not right, I can't choose.

[hollow warren]

So what's the daily limit for pro users on the new nano banana 2?

[tired nova]

Is there a way to change the preview image for a project in Flow? After the Update its showing an image from some videos that were generated months ago, and while I did have projects labeled, it can be a bit confusing knowing which one I was just working on the day before.

[marsh inlet]

Please add a 1:1 aspect ratio for image generation in Flow, similar to the one in Whisk.

[robust socket]

please bring the STYLES from Whisk into FLOW

[pulsar coral]

It wont work the same because Google removed Nano Banana 1 and those styles were dependant from it. Nano Banana pro/2 models can do something similar but it will never look like that again.

[robust socket]

Character consistency is OK, how do we prompt for the STYLE ?

so, it will never work again ? damn it

[lyric dust]

The prompt input box is too small, especially since Veo requires extensive prompting to stabilize the content. The UX of this input box is very unfriendly for adjusting and modifying the text

The shadow effects used to style the top and bottom of the input box are redundant and interfere with the visual experience of typing

[pulsar coral]

Whisk model is gone alongside Nano 1 anyway, it's the same where we use Nano pro and Nano 2.

I think the only way to still use Nano 1 is by having an account at Google AI Studio, but over there is so censored uncomfy unusable that it's the same not to use it really.

And the registration is a ridiculous labyrinth.

I still have hope they can still add Nano 1 back to Flow tho, wouldn't mind paying for it here cause I already cancelled any Google subscriptions paying here would also reactivate it at Gemini but I don't need it without Nano 1.

[robust socket]

STYLES

RIP nano banana

[raven siren]

A suggestion: Though I like the idea that the assests are project based now it still would be nice if you can access all the assets somehow. Now if I want to use an asset again in a different project I have to upload it again. Or am I missing something?

[quartz kraken]

in flow, would be great to PIN images or videos or collections to the top.

[late skiff]

Hi i don't know if it is the proper channel to ask, but is it true that in the new version of flow there is no more screenshot that we can take in the scene builder or is it just me ? (There was a little plus sign to take a screenshot before)

[wraith sable]

hi

[glad ridge]

I came here to ask this, I use Frames to Video and now I can't see the option anymore. Just one that says Start to End

[simple stone]

what's going on?

my prompt is safe, not using actors or actresses photograph. can generate in Gemini. but flow became like this for several hours.

even picked nano banana pro and imagen 4, still got this trouble.

[edgy holly]

Depends on what account level you have.

[edgy holly]

You choose the model.

[edgy holly]

Those are all settings you choose from.

[upbeat sierra]

@simple stone apologies for the trouble, could you please file a bug in #1238621511401734254 with your prompt and we can have someone look into this?

[upbeat sierra]

@quartz kraken do you mean for your generated content? You can favorite content using the ❤️ icon

[edgy holly]

Same as you did there, except instead of saying 'featuring the character from @;, you say 'using the style from @_'

[edgy holly]

You can heart them, then use the sort function.

[edgy holly]

You can right click the image and save a screenshot locally, then upload it as a reference.

[edgy holly]

Frames to Video is the same thing as Start/End, which are short for Start Frame/End Frame.

[late skiff]

Its just too bad they would remove nice features 💔 Thanks for the reply ! 🙏🏾

[quartz kraken]

That's right, I can sort by 'favorite' but it would be nice to pin the favorites to the top, so i can keep dragging in those images for refs/ingredients when i make new videos. or add groups I guess. so the favourite group stays at the top.

will it be possible to generate with Nano Banana Pro in 2k or 4k soon please?
right now it's only possible to upscale a low res image, which is not the same

[edgy holly]

This is one of the misconceptions of AI image generation. The apps that say they generate in 2/4K actually generate lower and then upscale before they show it to you.

[wet spoke]

Am getting the same error

[hazy kestrel]

Is flow having some issues today? I have lots of errors and cant upscale etc

[quartz kraken]

i do get very different results tho when i use fal.ai.. it feels like it's just generating the base image on higher rez..

[daring island]

Is it me or did they break a bunch of stuff in the update.

Feedback from youtube comments: Missing / Removed Features

Frame-to-video option missing.

Scene Builder not showing or not working (can’t add scenes, jump, cut, or extend).

Text / prompt refinement (text expander) removed.

Image-to-video option missing.

Multi-shot support unclear or gone.

No cropping when uploading images (e.g., 16:9).

2K / 4K upscale options missing or unclear.

“Start video from first frame” broken.

Reference image selection worse than before.

How Do I extend a clip now ?

[upbeat sierra]

Sorry for the trouble @daring island

• Image/Frames to Video can be found in the prompt box for Video (switch from Ingredients to Frames)
• We are aware of the Scenebuilder issues and are working on these
• 2k/4k should be available when downloading an image (in the three dot menu in the your Asset grid)
• Can you share more details on the Multi-shot and Start from First Frame issues

[upbeat sierra]

@daring island here's a helpful reference from a community member

[edgy holly]

I get different results with every model through the API, and in Gemini, regardless of scale.

[upbeat sierra]

Sorry for the trouble here, this should be resolved

[iron quest]

I have to keep on switching to newest to find my uploaded ref images which is a bit of a pain

[gray monolith]

hey guys i need help , is that any groups i can join they use n8n or antigravity

[daring island]

Kinda wild how bad this release is. Maybe test it and use the feature before you release it. I wouldn't be mad but we are paying for this

[edgy holly]

I like it, but the instructions that were in that popup are maybe 1/10th what they should be, and the response to questions in here has been very spotty.

A ton of people have asked about the screenshot thing in Scenebuilder going away, for example, and none of the replies from Google people have said... it just moved. Seems like there is a narrow set of allowed responses, with very large gaps in it.

[tawny ocean]

Guys, I have a problem. I can't create videos, it gives me an error. The prompts are well done and don't violate Flow's privacy policy. I paid to make the videos, not to spend hours figuring out the problem. Can anyone help me?

[daring island]

Yup issues at scale

[tawny ocean]

so what about the company?

[supple pendant]

Hey! yes - i know we can choose incredients and frames. But because there's no explicit option for text to video in the interface anymore, my question i guess is more of a dev question.

When we input only text with no reference images, but have to select ingredients or frames, does it use the ingredients or frames model?

Or in the background - is it using text to video?

Becuase in my experience, the outputs from text to video looked distinctly different from the other models

And the outputs I'm getting now with text to video look more like outputs I'd get from ingredients, compared to text to video outputs I was making just last week

So that's really the core of the question, not so much an interface question but a back end question

[fierce nimbus]

whats not working. Update o the UI and outputs not coming back are different issues. Ill take UI feedback directly but can pass along error rates being high (team working on it)

[fierce nimbus]

you can name them and reference them with @ tag in prompt box. or filter the grid using left side "upload" button

[robust socket]

can you explain how to use STYLES on Flow?

how to use STYLES

so, anyone wondering how to use STYLES in Flow, you gotta upload the images into the workspace, rename each one, ex: ComicBook, Sticker, Feltropolis, SoftFabric, Claymation, and then you reference the style image in the prompt, (use @ in your keyboard)
ex: make a Dog chopping a tree, use the style in ** Claymation style **

[fierce nimbus]

add the image to the flow prompt box as an ingredient and write a prompt asking for it to use the style.

lets say you add the sticker of the cat to the prompt box.  "Make a shark in this style"

[robust socket]

you mean, "Make a shark in this style" will not know which image to use, since there are some images as reference, some images as scene, and onlyl one image as style

[fierce nimbus]

use natural language. Show the "woman" in the "park" in the style of "the cat sticker"

but we added @ tagging to make this easier with the names you want to use

[robust socket]

let me try again using 5 images above, "using style from image1, scene from image3 and subject from image2, image4 and image5"

[fierce nimbus]

the system doesnt really use image order currently. So i wouldnt rely on 1 vs 2. Id rely on briefly describing. But YES. nano banana 2 can take up to 10 images

[edgy holly]

Prompt example:

Make a shark in the style shown in @image_one. Show the shark swimming in an ocean scene like in @image_two.

Add details to drill down to get specific results.

[robust socket]

can we have the reference panel as in Whisk please ?

I just love how Whisk is good, why make it worse ? dont kill Whisk so users have to prompt all in text on Flow

[fierce nimbus]

i built whisk as well! Whats not working with @ tagging. You can also find the ability to keep prompt after pressing geenrate in the settings at the top of Flow

[robust socket]

here, on Whisk you pick the STYLE from a set of styles, the SUBJECTS from the group of images of subject, and the SCENE from the group of scene

[supple pendant]

love whisk! use it all the time to study and make better prompts

[robust socket]

please, let users pick style from the PANEL

[fierce nimbus]

im asking what part about it. Not the UI. What is the goal. Is it quick access to things you call "style"? Is it about them being in a single location?

[robust socket]

@fierce nimbus you did it great, Whisk is amazing, you just pick a STYLE from the panel, just a single click; can we make it on FLOW please

let the user reference a style using @ this is ok, but there are hundreds of thousands of images to pick from, we should be able to select a image from a folder then ("workspace") named styles

[fierce nimbus]

working on @ tagging for collections. Collections (folders) in flow do exist.

and bringing up collections more clearly in asset panel

[robust socket]

it is not working for comic book style from whisk - Color Palette must be describe as text

Color Palette: The dominant color scheme is a playful clash of electric blues (think a neon, almost phosphorescent blue), shocking pink (a hot, almost magenta shade), and sunshine yellow (a bright, almost acidic yellow). These primary colors are punctuated by accents of deep emerald green, fiery orange, and a creamy off-white used for highlights and lettering. The colors are not blended smoothly; instead, they sit next to each other with a defined border, creating a bold, almost stained-glass effect.

[fierce nimbus]

great feedback.

[robust socket]

@fierce nimbus if we can have presets of styles, that would be great, thank you so much for your support

[fierce nimbus]

also great feedback

[robust socket]

it works

[edgy holly]

Example five thousand of what I've been saying for >2 years about how AI apps need to work like regular apps, and stop trying to shoehorn everything into the prompt box.

Whisk isn't half as good as Nano Banana at style, scene or character reference, but it's easy to use... because it works like every other app we've all been using for 20 years.

[robust socket]

you mean Flow not whisk

Nano Banana is great at character consistency

[supple pendant]

hoping that when whisk moves to Flow, the interface will stay similar. The best part about whisk is the easy (and fun) to use interface

[robust socket]

Whisk is really amazing, @fierce nimbus it is your creation, you did it great, everyone loves whisk

[tawny ocean]

How long will it take to fix the video error? I was supposed to post the video tomorrow 🥲🙂

[fierce nimbus]

Team is working hard on it. Its obviosuly not intended. Is it coming for all of your prompts or specific prompts/models?

[edgy holly]

Flow runs Nano Banana. Nano Banana, especially Pro, is very good at style transfer, and the absolute best at character consistency. Whisk doesn't come even close, because it's not real style, character or scene reference. It converts the images you upload into prompts, then runs those prompts.

People like Whisk because it's easy to use. Other apps, like Midjourney, have similar boxes where you just upload images directly into the clearly marked 'Style', or 'Character' box, so you don't need to futz around trying to prompt it correctly, and can just use the prompt box for details you want to change.

For some reason, AI companies have been crazy stubborn on this. They keep trying to force as much as they can into the prompt box.

[tawny ocean]

but will they remove Whisk?

[robust socket]

@fierce nimbus please Nick, can you confirm it ? I see when i click on whisk that Whisk is not converting the image into prompt, it is using the text behind the image, the image is only for UI easy to pick, the real text is huge, here is the full epic book style text prompt:

This isn't a picture, but a style description for a picture in the spirit of an epic comic book, imbued with cool pop art flair and bursting with fun colors. Imagine a visual explosion, a vibrant tapestry woven from bold lines and saturated hues.

Color Palette: The dominant color scheme is a playful clash of electric blues (think a neon, almost phosphorescent blue), shocking pink (a hot, almost magenta shade), and sunshine yellow (a bright, almost acidic yellow). These primary colors are punctuated by accents of deep emerald green, fiery orange, and a creamy off-white used for highlights and lettering. The colors are not blended smoothly; instead, they sit next to each other with a defined border, creating a bold, almost stained-glass effect.

Linework: The lines are thick, confident, and dynamic. Think of the bold outlines of classic comic books, but with a slightly wobbly, almost hand-drawn feel. There's a playful inconsistency in the line weight – some lines are heavier than others, adding to the hand-crafted, energetic feel.

Composition: The composition is energetic and dynamic. It's not a static scene; instead, it feels like a snapshot of motion, action, and energy. Shapes are geometric but not rigidly so – think overlapping circles, dynamic diagonals, and bold zigzags.

Characters/Elements (Generic Descriptions): Instead of specific characters, focus on the types of elements present: A powerful, almost superhero-esque figure, perhaps silhouetted against the background, is posed dramatically; their costume is a patchwork of those bold colours...

I just need to copy and paste the full text prompt to use this same Style on Flow, correct ?

[pulsar coral]

Try more precise references, for example "In the image of the girl with car plaster the artstyle of the image with the shark" and so on, use a visual indicator.

Red hair, woman, image with sky, whatever indicates the model what to do more visually.

What the characters wear etc

I've seen it sometimes understand orders by like "Place the person in the reference image in the same pose from the second uploaded image"

That totally works for some reason.

[summer rapids]

For more info about what's happening now after the correction of the other bug about generated video / images with children

• Video created with Flow with a child/baby then "Save frame as image"
  ✅ Use it to create an other Flow video from it 👍
  ❌ Use it to create an new Image with Nano Banana from it / modify it

• Image created with Nano Banana inside Flow :
  ✅ Use it to create an other Flow video from it 👍
  ❌ Use it to create an new Image with Nano Banana from it

Hope that could also be corrected, would like to modify images and not create dozens of video to get a good one or use the 3x3 grid compositions to create long and consistent stories

There is also that bug but a little less usefull than nanobanana
❌ Add detail / Remove detail / Change camera from a video created with Flow with a child / baby in it

[pulsar coral]

tbh yeah the keyword is easy. But also Whisk's model surely gives WAY more anatomical freedom that Nano Pro and 2 is just too scared to give.

That's why people have been using anatomical sheets of skeletons with organs to demonstrate it.

Why even beat around the bush about it

All the catering to investors is way too visible in Nano 2 to the point it's a completely different thing than Nano 1 Gemini 2.5 flash in that sense.

That's not to say Nano 2 isn't fabulous, it sure is. But it's not the right tool for certain things for sure.

Which is something completely normal with AI models, it can't be that hard to see at this point.

Let's try some visual example with the "1 - 2" image prompt tho, I rather use Gemini for it atm

See it works just fine

Admittedly I should have used "The plant should be completely true to reference image" and woulda been the same plant

But yeah for a quick test it gets it

[supple pendant]

I use whisk as a quick and easy way to iterate and combine concepts for professional work. The workflow is fast and efficient, especially when you're doing designs and need to get granular!

It's really perfect when a client says they like design / image option 1 but also option 4... okay throw both into Whisk and get a happy client lol

Like yeah we can prompt and get more specific about what we want, but the serendipity of a tool like Whisk is fantastic for early stage ideation!

There's plenty of tools that focus on specificity and that's great for most use cases. But there's a lot of value to having access to AI creativity / imagination without prompting it

[pulsar coral]

Oh yeah a tip there in most image generation sites is always use the download button from the image, never right click-save as because that just saves a jpg preview.

That way if you get the Gemini watermark it's also easier for tools to remove it

That's a great thing from Flow since it doesn't have the watermark but yeah

Since a lot of people are gonna have to find Nano 1 alternatives cause Flow doesn't have it you're gonna need one like this one for it https://geminiwatermark.org/batch-watermark-remover

Seen clean like a newborn. I'm currently doing some work with Nano 2 in real estate, and it's fabulous really.

[edgy holly]

"There's plenty of tools that focus on specificity and that's great for most use cases. But there's a lot of value to having access to AI creativity / imagination without prompting it"

The only thing I disagree with here is the idea that these two things are, in ANY way, conflicting. There is NOTHING about making an app easy to use that in any way is limiting for people who want more specificity (I'd call it 'granular control').

Midjourney is proof. They're the go-to for the most artsy, pro-level AI image geeks, and they have seperate input boxes for style and character. No prompting needed for that. Then you just prompt what you want to change, what you want them to do, etc.

[pulsar coral]

Sadly that's just work like I'm saying, it's not artistry. Nano 1 is more artistic by default, specially if you literally tell it to do that

[supple pendant]

that might be true - i honestly haven't worked that deeply in midjourney, so I don't know that workflow or interface.

In the workflow that my team uses, we generally start with loose image concept iterations and then hone things in once client notes come in. Different tools have proven more appropriate depending on the changes, but generally it's been easiest for us to stay within the google ecosystem so that team members can have easy access to the project assets

So within our workflow, there's times when different tools with different functions have their place, and I suppose I'm saying that keeping things separate - for remixing / ideation and for final polish, is helpful

[edgy holly]

You don't usually need to prompt that hard. Something as simple as 'Show the exact plant in @image_one...' should do it.

[pulsar coral]

Yeah that's a visual indicator like I'm saying. But like the other guy said, Nano 2 can't recognize images by number.

[edgy holly]

I'm not arguing for Midjourney, or any tool. I just responded to what you said at the end, and used MJ as an example.

[pulsar coral]

it also recognizes the indicator as "Uploaded image"

And that's a really short prompt actually, more accurate ones like from kitchens with utensils for example do need more explanation

Need to make things seem clean, remove oxidation, make walls seem like new.

And it can even do some secret hypnotic tricks, that's all in the business.

Imagine that

[supple pendant]

i was pretty surprised by the big revamp and how they've laid everything out! I'm happy with the better organization / folder structure with new Flow, but I'm kinda struggling to imagine how Whisk is going to get incorproated / absorbed into this new set up

I'm hoping it lives as some kind of accessible interface, kinda seprately from the current interface

[edgy holly]

This is one of a mountain of reasons why making tools that work like apps is better than shoehorning everything into prompts. Prompting brings more randomness. Less consistency.

I was just talking to someone in here by DM about this. I don't know why this isn't obvious to AI app devs, but just look at Pomelli.

Pomelli does literally nothing that Nano Banana hasn't been able to do since NB1 came out. All it does it make one use case easier.

Nano Banana has been out for months, but when Pomelli Photoshoot came out, within a day BILLIONS in stock valuations went down.

The difference isn't in the model, but THEY MADE IT EASY TO USE.

[pulsar coral]

That's a fair point really, and a lot of people have been saying exactly that

[hushed mango]

Is anyone else going through this hell?

[supple pendant]

Ah yes - I've also been curious about how these tools feel like they are kind of developed in silos?

I imagine that Flow is the first step in finding a way to bring everything together, but I would think that if the apps were developed with each other in mind to begin with it would benefit everything?

[edgy holly]

Right, and yet most of the people who make AI apps just refuse to go that direction. AI image apps used to have MORE app functionality in settings and sidebars. Over the last 2 years, they've been progressively taking more and more away, and shoehorning more and more into the prompt box.

That means all the power of these models is being hidden behind prompt engineering, which the vast majority of users are just never going to get good at.

It's using bad UI design to HIDE the power of these models... making them less useful.

[supple pendant]

not yet - were u uploading a video?

[edgy holly]

Dig the aged, distressed look.

[hushed mango]

When I hover the mouse over the image to download it, that message appears.

plz help

[frozen tiger]

Hi everyone! By any chance, has the issue of not being able to scale generated videos in the extension to 1080 been fixed yet?

[hushed mango]

plz

[tiny canyon]

Sir, have you ever had experience using HeyGen or anyone here.

I would appreciate to hear anybody's feedback

[quiet nexus]

I keep getting a "prompt must be provided" message in flow. No matter what I do, re-type in text, re-upload reference images. When I hit generate I get that message and nothing happens

[fierce nimbus]

innnnnnteresting havent seen this one. So prompt has been entered. You press geenrate and get that error?

[fierce nimbus]

is the upload an image?

[hushed mango]

no, texto to video

[quiet nexus]

Yes that’s right

[tacit aspen]

@fierce nimbus Hey just checking, is there also some changes under the hood, notice a difference since the changes. My animations are following shorter prompts, results look better. But, could be me imagining.

[autumn cloak]

Hey it somehow stopped working

Am I being rate limited and if yes where can I see it

[quiet wadi]

what do you think about nano banana 2

[autumn cloak]

idk bro i cant even really use it

it just doesnt load

[past tundra]

i need help i cant generate videos on veo 3 allways say failed i change all, the prompt, script ALL, and still saying failed

@covert rampart

@olive portal

[autumn cloak]

relax man no need to pingbomb

[full flicker]

Please submit a thread in #1238621511401734254. The staff regularly monitors those thread and will your question doesn't get lost here.

[merry mango]

Quick one for those working on video generation or with insider visibility:

Context: Veo 3.1 (8-10s clips) is starting to feel the heat — Seedance 2.0 is out with multi-minute generations and way better character consistency. Google I/O (May) is the usual target.

Actual questions:

Is Veo 4 still on track for May 2026, or is there any internal slippage?

Are the main priorities duration (60s+) and character anchoring? Because right now, those are the real blockers for production work.

The 4K narrative — is that an actual requirement or more of a spec sheet play? (Real talk: 2K/HD is more than enough for 99% of use cases, and way cheaper on compute.)

Bottom line: Is there something solid coming before summer, or are we looking at another waiting game?

No speculation — just looking for facts if anyone has shareable insights.

[edgy holly]

I've used a few voice models, but I don't remember using that one.

[fierce nimbus]

no model changes outsode of Nano Banan 2 release

[fierce nimbus]

limits are pretty relaxed for daily use. Are you seeing prompt block or just all prompts all models not working?

[tiny canyon]

What?

[edgy holly]

You asked if I've used Heygen. That's a voice model, isn't it?

[tiny canyon]

Nah

[tiny canyon]

That is a software focused primarily on creating ai avatars for short form

[spring field]

An extension result cannot be deleted, so I need start from the very beginning just to fix bad extension result?

[fierce nimbus]

you can click on the original in the history stack which restores to original and you can just prompt again

[ionic halo]

not overly fond of the new ui layout... yikesraroo

[ionic halo]

I agree, not intuitive at all, a lot of clicking back and forth from page to page.. takes more clicks and more time.. which is the opposite of what good ui does

[ionic halo]

this

[pulsar coral]

Actually all the new tools are really nice and save a lot of ram, but without Gemini 2.5 Nano 1 there's much less to do with them

Been sitting on my derrier complaining about it since yesterday

Man I'm sooo bored

[fierce nimbus]

upload image from desktop > drag into prompt box > type prompt.

upload image from desktop > tap 3 dot > animate > type prompt

[fierce nimbus]

Nano 2 available?

[pulsar coral]

Is it really tho

[fierce nimbus]

Not sure i understand. Its available right now in Flow

[pulsar coral]

Nano 2 is fine but it just has too much "Is there another idea I can help with instead?" in comparison to Nano 1.

So much I pretty much spent the day taking fresh air today

[ionic halo]

right, and then when it doesnt one shot it, I need to redo the same thing again, but then I have to click back, copy and paste the prompt, reload the images and do it again. Where as before, there was a simple re-do button. Now that button extends it or gives other options, but to simply repeat it takes a lot more clicks now

and now I see the re use prompt button at the top.. lol.. I

[fierce nimbus]

there is a redo button on every ouput like before

[ionic halo]

yeah, I found it, its just in a new palce

thank you ❤️

[fierce nimbus]

No prob always happy to chat.

[autumn cloak]

When i put in the prompt it looks like it will work and then after 3-4 seconds this error pops up

And no model works

Through gemini or whisk it works though

[quiet wadi]

I don´t understand in what features is better nano banana 2 than pro

[astral fern]

Are there plans to make an apple app for flow since it’s becoming the new Whisk? I use my IPad Mini a lot and Flow is.. unstable.. it causes my iPad to overheat sometimes.

[fading dirge]

would be nice to have other ratios like 3:4 in google flow for photos

[pulsar coral]

I've seen drastic automatic changes when trying to do Digital and Cel-shaded art with Nano 2, it's definitely consistently much better than Nano pro.

Nano Pro tends to make it a little simplistic lines with no shading, but Nano 2 automatically uses professional shading even if it's not prompted

Nano pro can surely defend itself if you do prompt those things do.

Ever seen those Japanese gacha mobile games with waifus? The art they use for the portraits is precisely Cel-Shaded 2.5D professional Digital art. Nano 2 can do it even better than those.

https://www.youtube.com/watch?v=X7HmltUWXgs

Ever seen an episode of Rick & Morty? That's what no shading looks like

Nano pro passes the butter to Nano 2 yes.

[severe sedge]

I’m facing an issue while editing a video. When I use the Extend feature, the new part automatically merges with the original video, and I can’t find any option to trim or cut the last few frames in Flow. Also, after extending, I’m unable to download the extended portion separately after upscaling—it only exports as a merged file.

[severe sedge]

Can anyone help me ???

[fierce nimbus]

we are working on this one. Sorry for the inconvenience.

[drowsy pilot]

2k upscale not working

[severe sedge]

Thank you for the update. I appreciate your support.

Please prioritize this fix, as it’s very important for creating smooth edits and maintaining high-quality video output. Looking forward to the improvement.

[pulsar coral]

Gemini enabled Nano Banana 1 for the fast mode again.

Go nuts

It's the same Whisk does

There musta been a lot of feedback wanting it back same as here, better use it so they know we want to keep it.

[quiet wadi]

I always wanted to make her takes the gavel, but never could take the same gavel

I would like the gavel was the same

[desert berry]

Better cinematic quality, better resolution, are some of the differences

[pulsar coral]

The high contrast can be off the charts too.

[agile geode]

Congrats on the update — it successfully broke a perfectly functional workflow.
After rendering, downloaded files no longer keep the default scene names and are instead renamed randomly, which makes sorting and project organization unnecessarily painful. Please add an option to preserve original scene names (or revert the change), because right now this is a major productivity regression.

[pulsar coral]

I just select all the images from my folder, rename 1 and that renames them all like

Mini suite al lado de la cocina 240 (14)

It adds a number between parenthesis at the end of each image automatically

That on pc I guess

I know what uploading like a 100 images at once somewhere is

[sterile idol]

why？

This bug has been around for two days.

[pulsar coral]

Maybe the errors do need to give a more specific reason on why they happen, that could be whatever tbh

[quiet wadi]

I need a promp to convert in 3D image

[pulsar coral]

There's many kinds of 3D, real life, unreal engine like games do etc

[quiet wadi]

unreal engine

but i want to convert my scenes onto 3D but any promp works

[pulsar coral]

Arstyle: Transform this image into a high-quality professional Unreal Engine styled 3D render, add realistic depth, dimensional geometry, natural shadows, global illumination, volumetric lighting, use physically based materials (PBR textures), detailed surface texture, realistic reflections, soft ambient occlusion, cinematically illuminated.

Image quality: Remastered 4K HD ultra high resolution, sharpest definition, denoised, unblurry, no jpg artifacts, clean lines, High contrast.

You might have to add or take a little there according to what you're aiming for.

[quiet wadi]

I add it but not works

[pulsar coral]

Maybe you added an image, yeah it can be done much stronger

Let's see now...

[quiet wadi]

I add an image as an ingredient who I want to convert on 3D

[pulsar coral]

Is that 3D enough

[quiet wadi]

yes for sure

[pulsar coral]

Comes from this one

[quiet wadi]

Elite Class Horimiya

[pulsar coral]

[Arstyle: Avoid anime style. The subject must no longer look flat give it measurable depth from front to back. Reconstruct this image as a fully three-dimensional Unreal Engine 3D scene, add realistic depth, dimensional geometry, natural shadows, global illumination, volumetric lighting, use physically based materials (PBR textures), detailed surface texture, realistic reflections, soft ambient occlusion, cinematically illuminated, Add real volumetric depth, visible thickness to all objects, and correct perspective shift.
Show side contours, edge curvature, shadow falloff, and dimensional separation from the background. Add realistic lighting interaction across surfaces, ambient occlusion in creases, and cast shadows on the ground.
Make it look like a photographed physical 3D sculpture, not a flat illustration. Strong perspective, 35mm lens, depth of field, cinematic studio lighting.

Image quality: Remastered 4K HD ultra high resolution, sharpest definition, denoised, unblurry, no jpg artifacts, clean lines, High contrast.]

Probably one can add or take but yeah that sure forces it no matter what

Got converted from this

[quiet wadi]

but my image is an scene

not only the waifu

[pulsar coral]

Ah good ol' whisk, so good for tests, wer're gonna miss ya

[pulsar coral]

It can work if the background is added too

Let's add a background and see what happens

[quiet wadi]

it´s a scene with several persons

[pulsar coral]

If that's at the top it will affect them too

One can be speficific about it

[quiet wadi]

I have tried on a lonely waifu

and it works, but when there are more persons and scene

not works

[pulsar coral]

I got a better idea, shoulda thought of it from the memes

[pulsar coral]

How's this? it's sorta cheating but yeah

Ok let's work it up from here

[quiet wadi]

yeah i´ts cool

I tried your promp and it work´s on several images but not in the more complex

[pulsar coral]

And now

This is to go further beyond

This is to go further beyond

[quiet wadi]

but one of the major problems for me is it, I cannot make Megurine takes the same gavel

[pulsar coral]

Yeah I got this by the b-

I mean I got this

Ok let's try it on her

[quiet wadi]

no, it´s that not the problem

my problem is that I want her to take the gavel

but when she takes the gavel, the gavel shape changes

it´s not the same gavel

[pulsar coral]

Ah that happens to me all the time, but with weapons and stuff

You can tell nano to make you a separate picture of the gavel

And then say "the gavel must remain completely true accurate to the one in reference images"

You can make another image with gavel floating, or add it as a second picture.

Iv'e had problems with much more complicated weapons and things and got fixed like that

Ok let's see what happened to her in 3D

... she ate my prompt well then. Gotta try again

No wonder

[quiet wadi]

yes I could convert her on 3D

but no take the same gavel

[pulsar coral]

Yeah me too, sort of, the trick is the PVC figure style

You just need to work on the gavel you want now, and it's scale

It's hard to keep that same background when prompting, but at least we can go dinosaur hunting in threee deeee

Let's change the scene completely now

Create the the image as 3D unreal engine made PVC 1/7 scaled figurine.

Arstyle: The subject must no longer look flat give it measurable depth from front to back. Reconstruct this image as a fully three-dimensional Unreal Engine 3D scene, add realistic depth, dimensional geometry, natural shadows, global illumination, volumetric lighting, use physically based materials (PBR textures), detailed surface texture, realistic reflections, soft ambient occlusion, cinematically illuminated, Add real volumetric depth, visible thickness to all objects, and correct perspective shift.
Show side contours, edge curvature, shadow falloff, and dimensional separation from the background. Add realistic lighting interaction across surfaces, ambient occlusion in creases, and cast shadows on the ground.
Make it look like a photographed physical 3D sculpture, not a flat illustration. Strong perspective, 35mm lens, depth of field, cinematic studio lighting.

Image quality: Remastered 4K HD ultra high resolution, sharpest definition, denoised, unblurry, no jpg artifacts, clean lines, High contrast.

Background: The character from reference image is holding a tall handsome guy's arms scared at a dinosaur land.

No text.

That's what I'm using for this, should make some bucks at Deviant

[quiet wadi]

I tried to seleccionate the mallet with the cursor but not works

[pulsar coral]

I think you'd have better luck at gemini, this doesn't seem hard. What do you want her to do with that same mallet

I shoulda just done this on Nano pro from the start

[quiet wadi]

It´s complex because the mallet must be rotated 180 grades

[pulsar coral]

ldk everything is a challenge lately. Are you sure that mallet should be that big? She must have some heavy sinners to judge.

[pulsar coral]

Gonna call this one my best attempt too, also apparently she likes Mozart. I prefer Beethoven, that dog's funny.

And to keep it on topic yes all these are Nano 2 Flow... ish.

I gotta concede Nano 2 is doing this kind of thing much easier than Nano 1, again cause much better at obeying

The Dinosaur pics were all Nano 1 tho, ain't that artistic

[quiet wadi]

thank you very much but, still not the same gavel

it seems like it is very hard to do

I must sleep see you

and thanks for all

[spiral python]

Always error suck

[pulsar coral]

Yeah I imagine it's the size, gotta reformulate the scale or something.

[spiral python]

I cannot generate

[rough halo]

instant failed on video generation. i assume this is from high demand?

[spiral python]

Are you the same?

[rough halo]

yeah they instantly fail without giving me a reason.

@spiral python

[spiral python]

Yesterday I could still do it but today I can't do it at all

[rough halo]

do what? any prompt i type, no matter what it is, instatnly says "failed".

looks like a high demand thing going on. even prompting "Car speeding on the highway" doesnt even get to 1% and fails

and im on ultra subscription. so idk

[spiral python]

Yes, it's the same for me too, I changed the prompt several times but it still failed, not even 1%

[gritty quail]

same here

[rough halo]

i love how by creating a scene in a sinking boat, it added children in the background which i never asked for, only to flag my extensions due to its policy on minors. and when i try to take that same frame to erase the minors, it refuses to consider the frame as an entry because it has minors in it. minors that it created..

example: i asked to include this character in a titanic-like scnee with the ship sinking. it added the kid next to him. i never asked for any kid not did i hint at it. i was only able to remove the kid by dowloading the image and erasing her with photoshop

[gritty quail]

u can just take the picture again, prompt: it should be a 1:1 copy of the picture, and remove the child next to the man

[rough halo]

thats what im saying, trying to add the picture to edit it in a prompt was getting refused

[gritty quail]

oh ok

[rough halo]

it wont accept its own image due to it including minors

[gritty quail]

ye

can u make videos again now?

[rough halo]

its spotty. i created 4 then now im getting failures.. it looks like a traffic related issue.

[gritty quail]

maybe it will work again later

[tawny ocean]

flow still doesn't work

[autumn cloak]

same here doesnt work anymore for me

[ripe yacht]

Hi @buoyant swift, I’ve forwarded your message to a more appropriate channel.

[cobalt nebula]

can you guys add more aspect ratio for flow for images

[serene kraken]

same here

[bleak walrus]

Exasperating when it's so fragile. How can a high-rise construction site "violate policies?"

[last cloud]

Shape is weird

[last cloud]

Same here how long have you been experiencing this?

[undone summit]

Can we have an option to put a custom image size here? In Gemini we can put the size or aspect ratio inside the prompt and it provides the image with exact size/ratio. But in Flow even if I put the size/ratio requirements in the prompt, it doesn’t follow that and give me the image with 16:9 or 9:16 ratio only. It is really important for creating professional image. Like If I want to create a promotional image for social media it requires 1:1 or 4:3 aspect ratio most of the time. 9:16 isn’t a good choise for social media promotional image.

[robust socket]

@fierce nimbus please, it is missing a button to reuse the prompt for a generated image,
see how can I reuse these images and this prompt

my mistake, the button is hidden

the region select is not working, I selected only a small region, and prompt: make it gold
it made the entire image gold

[serene kraken]

I have been having this problem for two hours now

[raven siren]

Is there a way to remove a collection without removing nthe images contained in the collection? Also is it possible to choose your own project thumbnail image? It seems to default to the very first image..

[bleak walrus]

Yes, the shape is indeed weird. Here's what it looks like.

[rough halo]

flow is broken. shame

[topaz sluice]

i think flow deosnt have square

for generating images just rectangles

[near pollen]

/credits

[undone summit]

Do you think It's normal? 😑
Nano banana pro 2

[toxic shale]

veo is generating a video misplacing items or people from the original reference image given its not following the reference image at all

[celest tendon]

Veo needs to keep up with the other competitors

[edgy holly]

https://tenor.com/view/baldi-gif-4884524124268038040

[surreal hare]

I mean in feature extend video

[agile geode]

Batch renaming is a workaround, but it doesn’t solve the regression. The whole point is that filenames should be deterministic and tied to the original prompt/scene name for sorting, automation, and re-importing.
Please bring back the old behavior or add a toggle to preserve original prompt/scene names for downloaded renders (images and videos). Right now the random naming breaks production workflows.

[quiet wadi]

Hi

Yes, I love that mallet but I have been two years triying to fix and nothing could it

[dull rapids]

Yeah it hallucinates and screws up more than nbp

I’m assuming Veo 4 will be the next thing to launch

[edgy holly]

Mine are coming out with the start of the prompt as the file names.

[onyx tiger]

With the latest update, it has become generally inconvenient to generate videos. Previously, when there was a normal scenebuilder, it was easier - now it's generally some kind of nonsense...

Now you can't even upgrade to 4k.

Previously, when extending a video, I would select the final frame and generate a sequel. For now, generate it whatever you want... It's a pity for the money and time spent...

Can someone tell me how to work with such a broken service now?

[edgy holly]

It still works that way. It just moved. You click on the individual video and Extend is one of the options. It's not in Scenebuilder anymore.

[onyx tiger]

It's uncomfortable to use the word at all

I've been trying to get used to it for a week now🤦‍♂️😪

[edgy holly]

...it came out two days ago.

[onyx tiger]

It's a pity the loans will burn out...

Maybe, I'm not arguing. But it was more convenient.

Let's say you said that you can extend it by selecting a file separately. And how to extend the specified frame?

[edgy holly]

Extend works the same as before. You can't extend a frame. You extend videos.

[wraith furnace]

can anyone help me to get similar face as the reference image inside flow, the new update broke this for me now everytime I generate it give me random faces

[edgy holly]

Check to see if it's set to Ingredients instead of Frames

[wraith furnace]

I am generating the image not the video

[onyx tiger]

I understood you. But let's say I used to set a frame to create a transition. Now you have to do everything manually.

[pulsar coral]

I make canon character art a lot too, if you are using a reference image try the prompt
"Facial Structure: Always completely true to character from reference image"
That always helps me a lot you can reinforce it like
"Subject character's facial Structure: Always completely true to character from reference image without any creativity freedom."
That's when you really need to be strict and you're sure it's not gonna conflict with other prompts alike makeup and alike
That can be adjusted for hairstyle, eyes, physique, clothes and other things too.

Or that's how I do it tho, it can probably be shorter or better idk, point is the model listens

Also the prompts that need more priority need to be at the top of all prompts, way more likely it will listen

I've said this before but this Discord won't allow to just link a post.

Ah it's real, it'll work

[wraith furnace]

okay let me try that

[wraith furnace]

can I dm you with my prompt, may be you can help me with that

if possible

[edgy holly]

Flow has never done transitions. If you want to make a new clip with an image, that works the same as it did before.

[viscid mesa]

We heard that it was inconvenient to have your old ingredients in a separate project. We recently pushed an update where you can toggle between the current project and the Ingredient Archive within the asset drawer to access those images across any project! Let us know what you think 👀

[iron quest]

Is it better to use ingredients or frames?

[supple pendant]

THANK YOU going to test this now. When I saw that all my ingredients had moved and I'd have to move them each manually it was pretty depressing lol

[edgy holly]

It's just different. Frames start with the exact image. Ingredients mix up the images and prompt.

[iron quest]

It’s just a pain that the reference image seems to default to ingredients when uploaded. I’ve never used ingredients before. So I’m guessing with character consistency, frames would be better

[edgy holly]

The opposite, actually. Ingredients allows you to upload high def character sheets along with other images like for the background, objects, etc. Frames you have to have everything you want in that one image.

[upbeat sierra]

Sorry for the trouble, if you're having issues, please start a thread in #1238621511401734254 so we can resolve!

[upbeat sierra]

We're sorry for the trouble. If you're still having issues generating, please file a bug in #1238621511401734254

[iron quest]

Cause I’ve been doing individual start frames for each scene but maybe I don’t need to do it that way then?

[rough halo]

Thank you. Already did.

[edgy holly]

There's no wrong answer. They're just two different ways of doing it. If you want a very specific shot to start looking exactly like the image, then you have to use Frames. If you want to mix ingredients together, then use ingredients.

[tired token]

alone?

[copper harbor]

How can I see how much time I have left before my daily Nan Banana Pro limit resets on Flow?

[pulsar coral]

I at least haven't made an exact count yet, but I think free accounts have 20-25 free tries, you're gonna have to keep count on how many images you manage to generate today and seems the missed attempts do count. I'm sure cheaper Google plans that included Whisk still allow to generate more images at flow tho, even the one that costs around $5 a month. https://one.google.com/u/1/ai?utm_source=gemini

You gotta be sure you payed in the account you will use, that'd also give you some more Nano2 image generations at Gemini, the accounts cross over platforms.

So yeah the crossplatform is convenient since you'd get double if you use both sites

[clear parrot]

can some one help me i tried to delete cache and things like that and anything is working, some help please❤️

[winged plinth]

Hey @clear parrot is this affecting all your prompts or just a specific one you're working with?

[crisp basalt]

https://tenor.com/view/lying-pout-umm-no-way-nah-gif-10651305

people just be saying anything at this point

[edgy holly]

Being an online grump makes time go ~3 times faster, apparently? haha

[wicked ravine]

all i get Prompt must be provided

whjat is going on?

[ornate current]

why the hell is there only two ratios? google we are deleting our other two a.i making things nad here have one WITH LESS OPTIONS.

[wicked ravine]

video gen is not working for me since update I have ultra, all it says is Prompt must be provided

[spiral python]

Is the generate still an error guys?

[last cloud]

still get it after maybe 4 or 8 generations

[agile geode]

That’s not what I’m getting. Please check the screenshots I posted — the exported files are NOT named from the prompt or scene name. They’re being renamed inconsistently/randomly, which breaks sorting and my pipeline.
Is this an A/B rollout, platform-specific bug, or a setting? Either way, please restore prompt/scene-based filenames (images + videos) or add a ‘Preserve original prompt/scene names’ toggle.

[raven torrent]

Hello, i need some help

Why I can’t use Veo3.1 - Quality to gen a video

[dense lichen]

is flow down again ?

[woven juniper]

Can anyone send me tutorial videos on flow from basic to aadvanced

[manic sierra]

I have a question. Where can I find seed of the images I have created in Flow?

[timid sorrel]

What’s going on with Google Flow? Is video generation working for you? On my end it keeps failing every time.

[uneven sand]

Can someone help. Since the update to VEO 3.1 I have nothing but stress... Now is this message and the prompt doesn't even needs sound... In gemini just doesn't follows the prompt and works by itself... Can you please put VEO 3 back... and work on VEO 3.1 because it's just not there ?

[marble mirage]

the new flow keep expand the image to match the 16:9 aspect ratio, while the old Flow keep your image in its original aspect ratio and just put 2 black bar above and under. So I can still crop it and not affect to the original image. Any tips to make new Flow like the old one 😭

my use case is remove details or sth from the image

[thick shadow]

okay I'm FINALLY getting the hang of this UI and I must say....I HATE IT

[compact kelp]

FR

[thick shadow]

its SO frustrating. Ingredients I used to be able to use across all projects, I now have to go HUNT for. And this "Collections" within a project seems...silly

[tropic umbra]

Hi, why can’t I create a video? It keeps showing “failed”. I was able to create earlier, but now it won’t let me create anything. Anyone knows about this? Please help me

[tough yacht]

Let me see if I can help; what prompt are you trying?

[wicked ravine]

@tough yacht Ive been using Flow since closed beta well over a year now, Ive never seen it this poorly finctioning, I do everything Ive always done, in this new interface and all I get is: "Prompt must be provided"
what is going on? please help

[tough yacht]

I'm so sorry for the hassle - can you give me some example prompts that are failing for you, so I can try to reproduce and file a bug?

[uneven sand]

{
"title": "Tracos Interiores Team – Listening Moment",
"shot_description": "Hero reaction shot",
"duration_seconds": 7,
"primary_subject": "Four Traços Interiores team members standing side-by-side facing camera",
"action_interaction": "They listen to an off-camera CEO, making subtle head nods, soft natural smiles, and brief eye contact with each other before returning their gaze forward",
"dialogue": "No dialogue",
"secondary_elements": "Natural breathing, minimal posture shifts, subtle facial micro-expressions, maintain original wardrobe and spacing",
"environment": "Neutral textured indoor wall background, corporate setting, same composition as original image",
"camera": {
"movement": "Locked-off",
"angle": "Eye-level",
"lens": "50mm equivalent",
"depth_of_field": "Shallow",
"aspect_ratio": "Match original horizontal framing",
"resolution": "4K",
"frame_rate": "24fps"
},
"lighting_atmosphere": {
"instruction": "Preserve exact lighting, exposure, color balance, shadows, and skin tones from the original image",
"mood": "Soft, neutral, natural"
},
"audio_context": {
"ambience": "Soft indoor room tone",
"sfx": "Subtle fabric movement",
"music": "None",
"off_camera_voice": "Indistinct CEO speaking softly, not intelligible"
},
"visual_style": "Photorealistic corporate portrait realism, no grading changes, no added stylization",
"continuity_tokens": [
"TEAM_TRACOS_INTERIORES",
"STYLE_CORPORATE_NEUTRAL",
"LIGHT_ORIGINAL_MATCH",
"CAM_LOCKED_STATIC"
],
"notes": "Micro-expressions only. No exaggerated gestures. Absolutely no change in lighting or color from the source image."
}

[wicked ravine]

this has always worked and now all I get is "Prompt must be provided"

doesnt even matter I can type anything and all I get is: "Prompt must be provided"

I have tried three different accounts

only way I can render right now is via API!

[stable hill]

It is the same for me. and it is charging me even for failed videos.

[wicked ravine]

@tough yacht sound the alarms! 🙂

its 100% a catastrophic bug all those video renders are just the prompt "red" but now you see I try "red" and I get the "Prompt must be provided" error message

[tough yacht]

Are you sure the credits for failed gens aren't being refunded (check https://one.google.com/ai/activity)?

[stable hill]

No, i am 100% sure aout that

[tough yacht]

Testing now!

[wicked ravine]

I think it has something to do with copy and pasting prompts into the text field breaks the whole app... @tough yacht

[tough yacht]

Can you DM me the email you sign in with and we'll check your account to see what's going on?

[stable hill]

It has not worked for me since yesterday afternoon

[tough yacht]

I can reproduce this failing. Let me investigate why (I'll file a bug now and follow up asap)

[uneven sand]

Since you are at that can you please also mention that the gemini generations, at least for me, are completley out of control. This same prompt would give me a speech in english and saying something made up... VEO 3.1 has a lot of bugs that I have notice so far. I have weekly videos that I've been producing for 2 months always the same prompt the only change are the lines and VEO 3 did perfectly in one shot and now I can't even do it

[stable hill]

I have sent you a DM

[tough yacht]

Weird- this is working just fine for me (with your text prompt; I'm happy to test it with the start frame if you send it to me). Where are you copying the prompts from?

[tough yacht]

If you add "DO NOT GENERATE ANY AUDIO" or something to that effect, does it work for you?

[uneven sand]

No I tried several times... I move to GROK that did it in one shot..

[wicked ravine]

were you able to reproduce?

if not try copy and pasting the prompt in the window that 100% breaks it every time for me

[stable hill]

Here is my prompt: @tough yacht

[tough yacht]

With that addition to your prompt, it works for me. But i'll still investigate why it's failing in the first place and follow up if we learn anything.

[tough yacht]

Alright, so I'm pasting this here, and then I'll copy it and put it into Flow to test:

A photorealistic extreme close up video of a man's bloodshot eye. A slow, almost imperceptible push-in on the eye. The pupil is dilated, and a faint shadowy human silhouette is reflected deep within it. A single thin beam of cold blue-white light from the left illuminates only the eye. The surrounding skin shows dark circles and exhaustion. The background is absolute black. Oppressive atmosphere, eerie stillness. Cinematic.

@wicked ravine No issues when I copied/pasted from Discord. If you just copy/paste the text prompt with no start frame, are you still seeing the issue?

[wicked ravine]

checking

[tough yacht]

Thanks! If it's only failing for you with the start frame, can you send me the start frame so I can test with that?

(and be back in a bit, but thank you all so much for sending the feedback)

[wicked ravine]

hmm its working for me now... smh

@tough yacht

"man brushing his teeth" flagged for sexual content??

[dry frost]

I wanted some glasses to slip off a person face and that got flagged for sexual content too.

[tough yacht]

I'm so glad to hear that!

Re: man brushing his teeth, what was the whole prompt? I'll test.

[tough yacht]

Oy, sorry about that. Can you send me the full prompt?

[uneven sand]

Well I've moved on. I've done what I needed with grok. I don't have infinite credits to fix veo bugs

[tough yacht]

I totally understand, my apologies Jose. I'll still look into what happened here so we can make Flow better moving forward.

[dry frost]

Sure thing I'll see if it's still there, might've deleted to reevaluate though

Nah I got rid of it, but I'll tag you if I try again later @tough yacht

[uneven sand]

Thank you because I still think that veo is one of the best models but if it is so hard to control I'll move to Kling or seedance once is available... And since you're at it please see if they can make Portuguese PT to work instead of Portuguese BR if one choses Portuguese PT

[inland bear]

is flow also not not working for u guys??

@uneven sand

bro what is this

anyone help me out?

@small elk

[manic sierra]

I have a question. Where can I find seed of the images I have created in Flow?

[inland bear]

does flow video creating work for u?

[small elk]

Hi, moderators are here to help out with any in-server issues, if you’re experiencing an issue with any of the google products I suggest #1238621511401734254 !

[manic sierra]

At the moment, I am working on a series of images with Nano Banana 2. I cannot find the seed value, which makes my work extremely difficult.

[inland bear]

exactly

can u believe i paid 200 bucks for the credits and ts just stopped working

[inland bear]

i need help rn

[manic sierra]

That's not good. I'm new to Flow and unfortunately can't help you. But do you know where I can find the seed for the images?

[blissful abyss]

Cant there be more aspect ratios on Flow

[somber needle]

there is no way to integrate a character in a background, it always feels like it is pasted on top 🙁

[blissful abyss]

In were

[somber needle]

on top of the background image, in flow, nano banana pro and 2 😛

[blissful abyss]

give me an image example let me try

[bleak pivot]

Does anyone know how to extend in this new layout? i add to scene builder, but there is no options to extend or even a prompt box...

[edgy holly]

I did. I said that's not what my file names are showing. All of mine are the first words of the prompt, then either a string of nonsense letters, or letters and numbers mixed together.

I don't work for Google.

[edgy holly]

Quality mode works with Frames or test to video. Ingredients only works with 3.1 Fast

[desert hill]

@edgy holly became so legendary that he has to say I don't work for Google🤣

[edgy holly]

Negative prompting backfires more often than it works. You folks know that.

[edgy holly]

From time to time, yeah, haha

[edgy holly]

I've noticed that even if you mention a bathroom, that will lead to higher fail rates. Had some where the character was walking toward a bathroom down a hallway. Couldn't even see the bathroom.

Whoa... why are my words red? haha

[edgy holly]

Nano Banana doesn't use seeds, even in the API.

[edgy holly]

They moved that and it's not in the instructions update popup.

Instead of extending in Scenebuilder, you just click on the clip in the project grid, and it's one of the options under the prompt box.

Then when you're done, you can add to Scenebuilder to export them together if you want.

[bleak pivot]

Thank you, i found it! but when i click on extend, nothing happens lol? even if i add a prompt, nothing happens! 😢

[marble mirage]

the new flow keep expand the image to match the 16:9 aspect ratio, while the old Flow keep your image in its original aspect ratio and just put 2 black bar above and under. So I can still crop it and not affect to the original image. Any tips to make new Flow like the old one 😭 @tough yacht
my use case is remove details or sth from the image

[chrome cape]

pls I observed that when I entered flow labs the graphics in the site has changed, and I can't upload image...

does anyone pls has idea on how I can upload images?

[desert hill]

Maybe they turned it off temporarily??

Try refreshing

Or logout and login again

[ashen talon]

anyone facing issue with image generation on flow ?

[loud panther]

Video and image both
I think server issue

[edgy holly]

I used it yesterday... one sec.

[edgy holly]

Okay, so you click on the clip, make sure '>> Extend' is selected, type in the prompt, check your settings and hit the '->' button and it doesn't work for you?

[edgy holly]

They moved that. You have to click the '+' symbol at the top of the page to add things to your ingredients library thing.

[bleak pivot]

Well, i feel stupid now... i never hit the arrow (enter) button LOL Thank you ❤️

[edgy holly]

I think the whole world needs to have a refresher course on troubleshooting. Most of us geezers over 40 had to learn that stuff to make it through life, but instructions are awful nowadays, and tools are more complicated.

[limpid meadow]

Hi

I was hot five minutes ago… now I’m abandoned. You pour ambition into me and forget me mid-email. Reheat me again and I lose flavor — just like your focus.

[bleak pivot]

lol, i agree with you

[bleak pivot]

https://tenor.com/view/idiot-elmo-im-stupid-shrug-im-an-idiot-gif-16140879

[limpid meadow]

I was hot five minutes ago… now I’m abandoned. You pour ambition into me and forget me mid-email. Reheat me again and I lose flavor — just like your focus.

[edgy holly]

This has got to be the most poetic complaint I've ever read...

[limpid meadow]

Yas

I was hot five minutes ago… now I’m abandoned. You pour ambition into me and forget me mid-email. Re

[edgy holly]

We should make a seperate room where mods promise to follow closely, but all complaints are required to be in verse, or at least iambic pentameter.

[blissful abyss]

<@&1366831844502274180> and @GoogleStaff can there be more aspect ratios in google flow like that is google AI studio

[unreal viper]

hi unfortunately mods cannot assist with product issues, please ping us for discord related issues! thank youu!

[blissful abyss]

But can the google staff help or where can I suggest

[edgy holly]

They follow the #1238621511401734254 channel more closely, I think. They know people want this, but more it's said the better.

[blissful abyss]

KK thanks for that, and do u mean that people to have made the suggestion

[full flicker]

What Frontier said, please use #1238621511401734254 for any feedback, suggestions, or product bugs. The staff do monitor those channels regularly and then your questions or feedback will not get lost in a channel that is not monitored by them.

[blissful abyss]

Kk thanks I understoodthat

"They know people want this, but more it's said the better." but he said this

[edgy holly]

Yeah, this is a common ask.

Don't not say it because others have.

I'm bad at this, but they need more people to be "squeeky wheels" about it for them to tell which things people want more.

[full flicker]

Yeah, it is something I have seen asked many times. That is why we have the feeback channel. If it is something you do want to see make sure to indeed put it in the channel, because as he said the more people that ask the for the same thing the more likely they will hear it and make it happen.

[desert hill]

@full flicker Mr cat, what's the name of the cat in ur pfp,the orange one, i like cats

Jus curious

[blissful abyss]

Kk but I hope it comes soon. Maybe it might come with the whisk and ImagineFX to flow

[full flicker]

I moved the reply to #off-topic・🐢

[tender sleet]

Its just me or flow is down?

[desert hill]

No its jus down in specific places

Tiny prob

Will be back soon

[inland sorrel]

Is Image Generation with Flow is limited, or they requires credits and have daily limits?

Because I love creatin images

[edgy holly]

Depends on your account level. I think it's ~100 per day with a Pro account, and 1000/day with Ultra, using Nano Banana Pro, but after that I think we get more. It's confusing. We haven't gotten a straight answer.

[somber pebble]

Flow having issues atm? I keep getting a blank page with "The number of requests sent exceeds the quota limit. Error code 253"

[proper radish]

...yep

[versed marsh]

same problem

[somber pebble]

yeah its not with the generation, as they seem to work, its like flow itself keeps messing up. Sometimes the page wont load properly when it tries too

[rose vigil]

cant login into flow account. is it only me?

webpage keep loading

[somber pebble]

no, its not just you, having the same issues

it kicks us out and then fails to load

[rose vigil]

which country are you mate?

[somber pebble]

Canada

[rose vigil]

ok I'm from asia, it affect all over the world then

[inland sorrel]

And for Free users? And Nano Banana 2 AND Imagen 4?

[edgy holly]

I'm not sure. Test it out. I do know you get some free use in Gemini. Just make some images and see how many you get.

[inland sorrel]

Ok

On Gemini, it's 20 free images with Nano Banana 2.

[upbeat sierra]

Sorry to hear you're having disruptions. Please start a thread in #1238621511401734254 - the more details, the easier it will be for us to look into this. Information like:

• What model you're trying
• What prompt you're trying
• Screenshots of the Error message

Go a long way for us to be able to resolve things more quickly

[upbeat sierra]

Can you share a screenshot of the error message?

[somber pebble]

It was just a white screen with that text in the corner. Then it would kick me out and the page would fail to load. It's been working better since but I'm not at a computer now

[west tide]

same how to fix ?

[upbeat sierra]

@inland bear @west tide

Sorry to hear you're having disruptions. Please start a thread in #1238621511401734254 the more details, the easier it will be for us to look into this. Information like:

-What model you're trying
-What prompt you're trying
-What type of account you have (Ultra, Pro, etc)

Will go a long way in helping us debug your issues

[edgy holly]

Did it say "Nano Banana 2"?

[inland sorrel]

YES

[blissful abyss]

I guess NANO BANANA 2 is for free and pro is limited

[blissful abyss]

seriously

that would be better than 5

for the pro

[wicked ravine]

this update has been an absolute disaster -- hopefully things improve asap

[blissful abyss]

meaning

[upbeat sierra]

We're sorry you're having issues. Can you share more details on your pain points?

[pulsar coral]

Yeah this is also around my calculations, and the missed attempts count.

[pulsar coral]

Yeah for starters, do you like how Nano Banana 2 and Nano Banana Pro are generating your images? Or do you prefer how Whisk is doing it? Or is it something else?

[granite hemlock]

@upbeat sierra hello

[astral fern]

Do ultra users still get unlimited VEO 3.1

[edgy holly]

Yes, but only Fast, low priority mode (as has been the case)

[astral fern]

Ah, how long does it usually take? I know times may vary but are we talking 45 minutes lol

[edgy holly]

Basically that mode will take maybe 5 minutes or so, unless their servers are overloading (a few hours a day, or so), then it'll fail and you come back an hour later and it'll work then.

I've never timed it, but if generations don't fail because of that or a filter, then Veo in Flow is by far the fastest. Kling is WAY slower.

[hexed moss]

Hello can someone tell me the big difference between veo3 fast and quality?

[edgy holly]

Generally better at physics and detailed complex prompts. Sometimes you can't tell the difference.

[hexed moss]

Ok, i can just use the same prompt as veo3 fast right?

[edgy holly]

They're both Veo 3. Quality is just better at some things.

[fringe geyser]

can someone tell me why they cant spend 10 sec to have some tech change the default setting to nano banana 2 so i dont have to change it everytime i want to do an image ? like i just dont get why this isnt automatic when new models come out -

or at least make it a toggle on the top layer - not in a drop down menu which is the least easy way to change something - click drag click - worst ui choice possible

[fringe geyser]

also why does the upscale and download message need to be dismiss ? just make it go away after a few seconds - why do we need all these extra clicks over and over again for no reason

[indigo atlas]

Hi today my flow generation failed all day, there's any problem on server? thanks

[drowsy pilot]

This annoying popup doesnt go away, you have to click dismiss to remove them, and I have 20+ of this notifications, so its really annoying you have to click them everytime you download an image, it also blocks the prompt and other icons.

[sonic pivot]

my pictures are auto delete

[vague fern]

I can't stand the fact they merged Googlelabs+Flow with Discord. Should have just left it alone. Now we can't find our saved images regarding our projects and where is Ingredient Archive? We are doing all this extra crap to locate our images and whatever else and it is taking away from us being able to finish our projects! I just want to login to Flow Scene Builder so I can put all my scenes together move on the next step.

[pulsar coral]

.... there's too many complaints by everyone about everything, at this point might as as well just leave Whisk as it is and add Nano Banana 2 and Pro to it which is what pretty much everyone asked for at the feedback areas.

Like the Yaska user here was showing, this is the correct setting for the site, if you want to build on that it will work much easier

[desert hill]

That sounds incredibly frustrating — especially when you're mid-project and suddenly the tools you rely on feel like they’ve been rearranged overnight. Nothing kills creative momentum faster than “Where did my stuff go?” 😅

Maybe go directly to the Flow login page (not through Discord).
Try logging in with the exact Google account originally tied to your Flow projects.

If it works, u r lucky, else peace out to google ✌️

[desert hill]

Thanks

[vague fern]

That is one thing I do is KEEP Extra photos and videos of my projects just incase I can't get to them like this Integrated Archive that I am still waiting to see if anyone can guide me to where Integrated Archive is! That would be HELPFUL!!!

[barren agate]

Soooo...I just opened up Flow and thought it would be easy to use....and I don't even know what I'm looking at. there are little to no buttons or options. Every tutorial I see is now officially void because they're all using a now old version. Not trying to be mean, but I'm literally wondering right now if this is a joke.

[rotund horizon]

After playing with Flow for a few days. Here are my findings about the new generators in Flow.
a) Removal of NanoBanana1: A huge mistake. Like mentioned on another post Banana1 and 2 are fullfilling each other's weaknesses. Banana2 lacks of imagination and takes prompts too literally. Banana1 has less understanding but will fill that "gap" by providing a solution by itself. I hope they will bring it back or at least leave it available.

b) Not so smart Banana2: I know it creates some 'wow!' factor with its capabilities, but when you try to push it, especially when comes to editing existing photos or for technical demonstrations, is just not there yet. I do a lot of automotive and product image generation and the results often do not make sense, the prompts are not followed or it has a hard time understanding what I am asking - even when I sketch things. So I would say is still like 70% artistic / 30% practical.

c) Smudgy results: I don't understand why there is an option for a 2K or a 4K image when the results are coming up smudgy or blurry and need to go through an image enhancer again to get clear results. Clarification: I have a paid membership and I am using its premium capabilities through there. In Flow, I only used the free features, but I guess the quality will be the same there, too.

[amber crest]

what happened in flow?? why i cant create image today(80% cant working), this is bad idea remove whisk in april 30🫵🏻 💩

how dare you None of the functions are working properly in flow and then still remove the whisk omg

[alpine osprey]

When I expand more than 2 videos in flow the original one disappear

[hallow stirrup]

Good afternoon, for the 3rd day now my video generation has not been working, I bought the ultra package and this is the problem(

[deep fox]

Is it no longer possible to download individual clips from an extended video separately? Once a video has been extended, I only have the option of a 720p download. What about 1080p and 4K?

[tranquil shale]

Cant upload reff image?

[noble knoll]

hey boys

[pulsar coral]

If they'd just enable Nano Banana 1/Gemini Flash 2,5 in Flow again which Whisk was using that'd fix about %85 the current probs Flow has, or at least would let people get used to it. Just so you know.

[opal echo]

so why i cant upload image, it been 2 or 3 days

[robust socket]

why shut-down whisk when Flow does not works/ do not have all features from whisk

just make it paid / expensive / pro ; we all prefer to stick to whisk instead

[hollow cedar]

why Extend/Jump to function is deactivated? i can not find them anywhere

[desert hill]

R u viewing it in the web interface

????

It won't work there

[hollow cedar]

yes, so what should i do

[desert hill]

Try AutoCAD

I don't think google flow has it

[hollow cedar]

no, it used to have those functions in SceneBuilder

[desert hill]

Are you in edit mode??

[naive mango]

is flow error or what? i cant generate even simple image

[hollow cedar]

yes

[barren mulch]

is it only me or can you only "extend" a clip by one generation? Before you could generate 4 variations of the extention is that no longer possible?

[desert hill]

Then idk wth is wrong

Stupid flow

[dry niche]

tried the new flow with nano 2. why does inpainting generate a horrible job with super obvious seams?

this is the before

[spiral python]

Is generating normal now, guys?

[tawdry burrow]

Hi everyone

[winged plinth]

Hi @spiral python Could you share more details about the issues you're experiencing?

[tawdry burrow]

@winged plinth Hi Admin,

I am reporting a critical bug with the desktop version of the platform (Ultra Plan):

On my PC, I can successfully upscale a video to 4K, but I cannot download it. When I click the 'Download' button, it simply disappears and the download never starts. After this happens, I am unable to attempt the upscale or download on that specific clip again.

Crucially, I have to log in from my phone to download the upscaled clip, which works fine there. This confirms the issue is specifically with the PC/Web interface and not the file itself.

Could you please look into this bug? I need to be able to download my upscaled content directly from my computer

[winged plinth]

Hey @tawdry burrow I'd open a support ticket to get the technical team on it, but in the meantime, try a different browser. Extensions or privacy settings can sometimes block multi-file downloads, so it's worth a quick test!

[spiral python]

@winged plinth I'm on the ultra plan, and I haven't seen any results since 2 days ago.
I want something fast and the same quality.

[winged plinth]

Could you share the prompt and which model you're using?
Also, have you tried opening a ticket in the #labs-bugs-feedback channel so the team can look into it for you?

[spiral python]

I use 3.1, it's high-quality and fast.
I've never opened a ticket before.
But many people have experienced the same thing. Why?

[winged plinth]

They'll need to investigate whether it's an account level issue or an input issue (prompt or image) to get your generations working properly.

[spiral python]

Does the account issue also have an impact? And I use simple prompts, sometimes complicated ones, and the details are still the same.

[winged plinth]

Go ahead and drop a ticket in #labs-bugs-feedback and the team will follow up with you there!

[spiral python]

Oke thankyou

[dense mortar]

Hi everyone, is the team aware of the video generation issue? After the latest update, FLOW has been performing terribly. Errors constantly occur when generating images and videos.
Now, images are generated intermittently, and videos in VEO 3.1 won't generate at all.