flow-general・🌊

[near gate]

Flow makes good dramatic scene though lol as long as its not action pack, Flow does very well

[oblique dock]

Just to really put this in perspective: 1K output is 1366x768 = 1,049,088 pixels. A 4K output is 5504X3072 = 16,908,288 pixels. When you enter a prompt for an image output and the resolution is limited to 1366x768, you're telling the model "here's what I want, but you only have 1,049,088 pixels to interpret it." Upscaling that is just making those 1M pixels look nicer--you're still limited to those 1M or so. It's a band-aid at best. When you prompt with an intended 4K output, you're telling the model, "Here's 16+ million pixels of real estate to interpret my prompt." Don't take my work for it--run your prompts at 1K natively and again at 4K natively and explore the diffrence.

[tight ruin]

how much times i can use the flow?

[rustic temple]

depends on how much you're paying

[tight ruin]

in the free version imean

[rustic temple]

I think you get 100 credits per month, giving you 10 videos per month

[midnight cedar]

interesting. do you have examples where generating at 4K directly produces better results compared to upscaling a 1k generation?

[oblique dock]

For starters—simple concept of asking the model to fill in ~1M pixels vs asking the model to fill in ~16M pixels. Giving the generation 16 million pixels to work with allows a vastly larger creative canvas to supply detail and fill in with model input. Upscaling the 1M pixels to 16M pixels is still starting from those same 1M pixels. Incredibly important concept across all image and video generation models, not just gemini.

Is this not something you can ask your colleagues about?

[thin sonnet]

Hello

[jaunty peak]

100 credits that’s 5 video generation

[chrome imp]

Like it would be really helpful if i can generate directly into 4k, rather than upscaling.

That's why people use third party vendors. I know it's not easy to implement but, that would attract majority of subscribers from another platforms, because Nano Pro is unbeatable right now.

[edgy holly]

I've only used it a few times so far, but it seems to smooth generations out some, and make people look younger, in a character sheet I'm making. Made them look older in the small generation, so it came out about right in 4K.

[tough otter]

It is crucial for us to use native 2K images directly for video generation, rather than relying on upscaling. In wide shots (long shots), the facial details in 1K images are often unclear. This leads to disastrous results in image-to-video generation, causing significant facial distortion or morphing. Native 2K solves this issue. However, simply upscaling a 1K image to 2K is not a viable solution, as the upscaling process itself introduces artifacts that negatively affect facial consistency in wide shots.

[proper drum]

agree

[raven siren]

Not impressed with the upscaler quality. Same image through SeedVR looks way better. Wish they would have given us more aspect ratios to choose from.

[gleaming estuary]

guys I have pro account , but there isn't any 2K download option !!

[subtle estuary]

Hello

[pallid violet]

@novel masonHello, as a friendly reminder we cannot generate images or videos in the server! Please do not post your prompts in the channels as it will not generate. If asking for advice or tips on your prompts do so in the appropriate channels.

[dark lily]

[hoary wren]

Does anyone else all of a sudden miss their projects in Flow? Everything is gone!

[tough yacht]

No, I can't reproduce this. I'm assuming it's still happening after you refresh? If you DM me the email you're signing in with, we can investigate.

[analog radish]

Wait a 5-10 minutes and refresh they will appear, it is happening from time to time. Until then you can find your videos when click on your profile in the upper right corner and click "My Library"

[edgy holly]

Once or twice this happened, but either just took a long time to load (like 3-4 minutes) or I did a hard reload and it came back.

[near gate]

Im still not seeing downloading the banana pro pictures to 2k or 4k, when I click download it just downloads without giving me options and yes im on ultra plan. Or does the banana pro on Gemini dont count?

[edgy holly]

On Gemini you can get 2K. The upscaling thing people are talking about is in the Flow web app.

[midnight cedar]

thanks everyone for the NB Pro upsampling feedback. keep it coming. we're looking into this.

an important thing that we need to do on our side is store the upscaled image so that it can be used for I2V. right now, when the image is upscaled, the upscale image is only downloaded, but not stored. right now the workaround for this issue is to upload the upscaled image to use this higher quality version for I2V. flagging this to our team

[desert berry]

Anyone here knows of a good tutorial to work in flow with continuity for storytelling?

[edgy holly]

I'd break that into pieces, and search for each part. Lots of YouTube videos about specific elements, and searchable text content.

[desert berry]

Yes, I have been watching some videos, but no one shows the actual workflow, just some blahblah and results, but not step to step, I need to understand the best workflow to do it consistently. Thank you

[oblique dock]

you've got this 😤

[turbid lintel]

4 x 4 grid on nano banana pro .. then extract which picture you want , and assemble the story board.. then generate the video

[desert berry]

Yes, that is my reasoning and the way I'm doing it with external editors, but I know you can do lonscenes in Flow, but I still cannot figuere out how to do it the right way, thanks

[turbid lintel]

to extract which panel just use this prompt :

extract the still " panel 1 "

you can do all this on flow.. no need external tools

[desert berry]

I got you, that is a continuity scene to scene in one image, then you reproduce again in singles to follow the storyboard clips, I have no problem producing the images, what I want to learn is how to make it conscistent with voices and sounds throuhgt the final video in Flow, thank you for your time and trying to help.

[turbid lintel]

for voices we still lack of option , need to use external tools

no other choice

[onyx spear]

anyone know if you can add or remove things from a later snippet of a video clip, i ccant seem to control the frame when i want to add or remove or change camera on a clip

[edgy holly]

Haven't tried this, but might be able to extend from the frame you want to alter and then edit that video?

[sly sparrow]

Anyone got any good workflows for bullet time effects in flow? I can get some really good stills as start and end points for a shot but veo really cant understand how to transition between the 2 in ultra slow motion and rotate the camera at a faster speed

[full flicker]

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

[viscid mica]

Creating directly in 2K in Gemini, a 16:9 picture is around 6MB. Creating in 1K (?) then downloading the 2K version in Flow, the file is only 2-3MB. The finer details are of lower quality and lots of artifacts in the upscaled 2K, compared to the naturally-born-2K.

[unique aspen]

hey guys i couldnt generate an image because it says "couldnt generate image. Try again later."
although ive got google ai ultra.
can someone help please?

[narrow cedar]

Dear FLOW team,

Please allow 4K upscale (maybe limited to 10 or 20 per day) for PRO TIER also.

[frozen walrus]

Are there any websites that have database of AI videos? (For free)

[lusty nymph]

Which prompt did you use to make this if you don't mind me asking? the scenes are consistent!

[turbid lintel]

“Create a 4×4 storyboard grid (16 panels), read left-to-right and top-to-bottom, where Panel 01 is an establishing shot; depict a Pixar-style woman cooking dodol in a rural kampung setting, with ONE clear action per panel, the same character and environment throughout, a different camera angle in each panel, strict physical and visual continuity, and a clear completed cooking outcome in the final panel.”

[lusty nymph]

Thanks a lot! 🙏

[jagged jackal]

Ok.. I'm easily confused.. and that NBP announcement about upscaling NBP images confused me. Isn't NBP already native 1, 2, or 4k depending on the setting you choose?

[velvet stream]

try this https://youtu.be/LV27hNiPOIM?si=EGorjpq3DlHvyTUW

[desert berry]

Thank you, I'll take a look 🙂

[edgy holly]

In Gemini you could get a bit over 2K already, but in the Flow web app it only generated at 768p, which it still does, but now when you download images made in Flow you can upscale them to 2K or 4K.

[desert berry]

That did it, thank you!

[true pumice]

Is anyone else facing a lot of generation failure's?

[random anvil]

no but i can't download in 2k anymore. It lets me use nano banana pro still just "Couldn't generate image. Try again later." when i click download 2k

is there a different limit for 2k downloads?

i assumed it was tied to nano banana pro generations

[tough otter]

Is there another crash? I can't generate video from image frame

[storm vault]

Is Flow can erase img or generate the image that erase? like png img?

[mellow radish]

Are we gonna get veo 4 next week ? Veo is lagging behind chinese !!! Do somethingggg

[mellow radish]

Whats the prompt for creating this 4*4 grid ?

[turbid lintel]

@mellow radish

[mellow radish]

Yea i wish they would charge credit for native 4k image generations . Upscalling is not as good as native ! Not even close

[fickle pawn]

Hi there..I haven't been able to get the camera to stay fixed in Veo 3.1 videos (No matter what I type in the prompt). Does anyone know how to do it?

[turbid lintel]

try type Static Camera

[fickle pawn]

I tried but it did not work. I think model tends to favor camera movements.

[turbid lintel]

try add this
Negative prompt : camera movement

[fathom pike]

always issue with this app bro god damn

tf u mean issue with generating images, theres different issues every god damn day

[fathom pike]

bro i still cannot generate images for the past couple of hours I got work to do whats going on

[sick sparrow]

love flow

[cinder moat]

This was made with flow?!

[sick sparrow]

yessir!

[north wigeon]

how can i stop veo 3 from adding music? i put "no music" at the end of every prompt but im still getting alot of random music.

[analog lava]

Hello @sterile frost , as a friendly reminder videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[snow lion]

me too, thats why im using capcut voice isolate, but for sound fx sometimes make me crazy because there is always background music there

[heavy drift]

Veo 3 and Flow would be the best tool on the market if not for the extreme and, what might be, totally sexist censorship. I am working on a scene of a woman talking to a bar tender about confronting her fiance who left at the alter. ALL of a sudden, even though I have been using a nearly identicle prompt, it is constantly flagging it as a policy violation. Here is the prompt. I have to rerun things over and over to get ANY clip. Please Google, solve this. It destroys any creativity. Instead of creating the scene, I have to puzzle out how they have a normal dialogue due to the prompt censorship instead.

[foggy rapids]

Use higgsfeild for now googles reducing the censorship is kinda unlikely

[vernal remnant]

Is there an outage in flow right now? Cant generate images or videos (ultra plan)

[tacit jackal]

currently generating on pro

[vernal remnant]

May be some error at my end then

[tacit jackal]

good luck

[vernal remnant]

I think there is definitely something wrong, it cancels the generation the moment i hit send, I first thought that it might be my prompt, so I tried a simple prompt "cat dancing", it still doesn't go

this

[narrow cedar]

I am not getting any option to upscale to 2k / 4k in my Pro plan.
Although the two days ago the update notification showed it as overlay on my flow interface.

Is it available globally to Pro & Ultra or is it restricted to some countries?

Also, is there any requirement to signup for a beta or nightly version to get this upscale feature or the normal url labs.google/fx/tools/flow is the only channel?

[tacit jackal]

seems to be working UK/EU

[narrow cedar]

I am on AI Pro plan from India.

Can someone please correlate and tell me if this is actually geo restricted feature or I need to do something to get the feature of upscaling (I have tried with new generations with both Nano Banana and Nano Banana Pro but neither the main interface or the Editing modal shows option to upscale. I am just getting the normal download overlay button on images. )

[edgy holly]

Hit or miss for some people. Don't see a pattern, but if you're using a different browser, a fresh Chrome install with no changes or extensions sometimes helps.

[vernal remnant]

same browser, everything same, no change, I'm working on a project, and have been generating video while keeping the same tab open, and suddenly it stopped working about an hour ago, but good to know that this is happening with others too, I was thinking there must be something wrong with my account.

[edgy holly]

I don't know why, but sometimes a new browser helps, sometimes temporarily. If you already use Chrome, you can try Canary.

[vernal remnant]

I'm on firefox, I dont use chrome, but let me try it on canary

[edgy holly]

I use Edge and Helium, but got Chrome just for Flow.

[vernal remnant]

does flow work better on chrome ?

[edgy holly]

"a fresh Chrome install with no changes or extensions sometimes helps."

(Canary is a slightly different Chrome)

[vernal remnant]

I just found a latest thread in the bugs section, cant post the link here, it looks like others having the exact same issue

[edgy holly]

I'm not tech support. Just giving you one option that works for people. Don't use it if you don't want to.

[vernal remnant]

GOtcha

[random anvil]

Issue i get is that when I click 2k it just errors out and says to try again later. Hasn’t worked in a few days now

[novel aspen]

I’ve noticed that the video quality is reduced compared to the input image when using frame-to-frame, and the VFX quality is quite poor — effects tend to burst out like explosions without following any coherent trajectories or motion paths.
I hope this can be improved in the near future.

[restive needle]

Endless errors

[plush crystal]

Hello, how can we generate ultra high quality videos on Flow? They are all currently native 720 and upscale to 1080p, however the upscale does not seem to make a difference. Also I am unable to use VEO 3.1 Quality (limited to Fast model only). Do I need the ultra plan for Quality model?

[glad laurel]

why i cant use the ctrl+v to paste picture

dose anyone have a solution?

[brisk bone]

nano banana pro

[thorny willow]

My Generations keep failing and im on an ultra plan, please help

[hasty mason]

morning folks ☕ hope all are happy that its finally Friday. 🥳. Im new here looking around trying to see how every one handles there Image creations and or media creations etc. right now i have a massed MESS folder of around 25k lol not sure what to do with them so im working on an Auto sorter lol it is scalab;le for all media music and video clips (coming soon lol ) but here is the pipe line data etc : # The Executive Summary: Chimera Master Project

Project Mandate

The Chimera Master project was initiated to address significant inefficiencies in image sorting processes. Initially, the system relied on manual methods and fragmented tools across multiple project folders, leading to chaotic workflows, inconsistent results, and frequent errors. This manual approach consumed excessive time and resources while risking data inaccuracies, hindering operational efficiency and scalability.

The Solution

Chimera Master transforms this disorganized environment into a unified, automated data curation pipeline. At its core is a single dashboard, launched via "chimera_gui_phoenix.py," that orchestrates the entire process from initial sorting to final validation. The system employs a dual-engine approach: the Triage Engine uses visual probability for quick categorization, while the Semantic Engine leverages semantic analysis for detailed descriptions. These engines feed into an audit layer that reconciles discrepancies through "The Audit" phase for analysis and "The Judgment" phase for manual conflict resolution, ensuring a seamless, error-free output. Users can initiate the full process with a single-click activation of the "Full Auto Protocol," producing clean, labeled datasets ready for applications like training models or archiving. Optional post-processing via "Text Refinement" allows users to clean up descriptions using checkbox filters in a dedicated interface.

[teal tapir]

Guys need help im making a video ad for a toy brand but I can’t upload minor pic in flow, so is there any way because i need characte constancy

[thick oriole]

Yeah that'd be nice to have, it's on our list, but for now you'll need to upload it from your filesystem I'm afraid.

[thick oriole]

1. We're working on more upscaling functionality, hoping to have news on that in the new year, but no specific timelines or details to share yet sorry.

2. Veo 3.1 Quality is available for Pro. If you're seeing errors, feel free to share them in #1319424187730755636

[glad laurel]

fine

[thick oriole]

No, we don't support that sorry. You can generate a minor in Create Image and then use that as a reference.

[thick oriole]

Image upscaling is not Geo-restricted. From our changelog:

• This option is only available on images generated with Nano Banana Pro after Dec 12, 12:30pm EST.
• 2K upscaling is available in both Pro and Ultra plans
• 4K upscaling is available in Ultra plans only

If you're not seeing the options, and all the above is true for the image you're trying, please let us know in #1319424187730755636

[teal tapir]

These are the images but i can’t upload them in reference image tab, it gives warning neither i can upload them in veo 3 for video reference

But im making a ad for a colour book brand and its for kids so i need kids in ad

[turbid carbon]

A cinematic and hyper-realistic close-up shot of heavy rain in a tropical setting. In the foreground, large green vibrant leaves are covered with glistening water droplets. In the background, rain is pouring down from the edge of an old, rustic corrugated tin roof. The atmosphere is moody, serene, and lush, with a shallow depth of field (bokeh effect) focusing on the water drops. High contrast, 8k resolution, nature photography style.

[zenith canyon]

Imagine cozy 4k

[tough otter]

Can't use children photo as a reference image. You can't upload it too.

[sick sparrow]

2k/4k downloads are not working. Anyone else experiences this?

[torn ravine]

Yup same here

[chrome badge]

They just stopped working

[sick sparrow]

Didn’t generate 200, I believe there was a cap or something

[lusty nymph]

yep same

ok back to working now

[thick oriole]

When you say 'upload' are you downloading them to your computer and re-uploading them? Yeah, that won't work, you'll need to stay within Flow, e.g. 'add to prompt'. Even then, yes, we have fairly strict safety filters around children (for obvious reasons) but it's not entirely blocked, and you should find some success.

[thick oriole]

Yeah -- Image generation and 2K/4K upsampling is having some troubles right now. We have a lot of people looking into it. Hopefully just a short-term blip.

[crisp basalt]

uhh veo 2 quality seems to have problem

im not sure if i can show the error message here since the error shown on my side is kinda internal stuff

[candid relic]

wow.. Flow is particularly stupid today. Anyone who would potentially lose a job that would be replaced is absolutely safe.

For every one prompt it gets right, I have 30 attempts (And the related tokens) where it so so far off base it makes me want to throw thing.

[crisp basalt]

veo 2 fast mode generates without any issues, its only the veo 2 quality mode having issues right now

[crisp basalt]

\"Error forwarded from LMRoot: Invalid prompt template. Missing {user_prompt} placeholder:

[oak drum]

i have an idea for enhancing the add to prompt feature. like after we add the images in the prompt, we can put placeholders in the prompt, like i have type take this from the first reference image or put this from second reference image into first image like these prompts these are making it confused and i am not getting the desired output. so instead of typing like that, we put place holders for the images, put a option on the added images and when clicked on that button it will place that image in the prompt, making the ai understand the prompt better.

[crisp basalt]

.

so basically... reuse prompt ?

[oak drum]

no, i can't able to explain properly, like i have two images and i want to replace the things from first image from the things in the second image, but some times it is not making them properly so i thought of this.

[crisp basalt]

what were u trying to do tho , just curious

[thick oriole]

You mean something like "The cat from $Image1 jumps on the chair from $Image2" ? Or even better, "$MrHappyCat jumps onto $RedChair" ? Yeah, that's a nice pattern and we're working on ways to make it work and get it into the UI

[crisp basalt]

ahh context stuff

[oak drum]

i am making images for a web game.

[candid relic]

Sometines when I use JSON I assign @Cat and @Bird as an example and that solved some of my issues. I have not tried to do that will inanimate objects though.. hmm

[thick oriole]

Yeah right now, the model might just do the right thing, but you're just hoping the unknown internal state of the model happens to respond well. We're looking into some more structured/reliable approaches.

[oak drum]

yeah just like that, in these two images, i want the same hands from the first image in the second image but smaller fitting it's size. so i have to put the prompt like make the hands in the second image as the same in the first image, but make them smaller fitting it's body size. it will give good desired images but some times it is falling.

[thick oriole]

But any tips & tricks you find work are great to share here

[candid relic]

It is frustrating, but also funny the way the AI interprets the prompt sometimes. I think one of the most valuable things I have learned recently is that AI video models often struggle with "negative prompting" (telling it what not to do) because they tend to focus on the keywords themselves. When you say "DO NOT make anyone stand up," the model sees the word "stand" and often triggers that motion.

[thick oriole]

Yeah I agree. "Don't think of an elephant" 🙂

[oak drum]

ohh so using json prompts solve that issue, thank u i will try to use json prompts

[candid relic]

I would not say "solved" i would say it helped me in my use case.

I ended up rotating the camera to get my desired action in the end as well.

[candid relic]

I have a string of prompts trying to mature a kitten into an a adult cat with mixed results.. and nothing I can actually use. Saturdays are my shorts animation days and often the most work.

[thick oriole]

Start & end frame, plus Nano Banana or Nano Banana Pro to generate the images, can be a nice flow here. Generate a few images of the cat at different ages, and use start+end frame to interpolate between them.

[candid relic]

That was my next step, thanks. I have been doing that a lot more lately. Generating the start and end scene and then letting Flow fill in the blanks. But you cant shift perspective or you will introduce elements that may not be desired in the scene.

[oak drum]

but i am not making videos i am doing the images for the players pets and they evolve beased on levels so i have to add the extra features when evolved but the features shoudn't be do distinct they have to be similar, so i have to again remake the lower stages images again with slight features.

[oak drum]

thank u bro

[crisp basalt]

wait heres the upscaled

[crisp basalt]

no prob. i basically copied what u wrote

[candid relic]

Nice job

[crisp basalt]

thank u gemini for carrying most of the weight 🙇‍♂️

[candid relic]

The other reason why I end up using JSON is the AI seems to handle multi character scenes better so dialogue and action are not being confused between the characters. Especially with these two:

[crisp basalt]

yeah JSON prompting can definitely resolve dialogue issues. It doesnt always fix everything, but it usually does the job 😄

I really thought Veo 3.1 would fix the issue of dialogue being assigned to the wrong character , guess i was wrong

[candid relic]

I use Gemini to help rewrite my prompts a lot lately, but be careful. Gemini will be the first to tell you all the failings of the Video generation AI and is sometimes hard pressed to give a solution. Gemini was the one that first suggested using XML or JSON in the first place. I have been using it ever since.

[crisp basalt]

ironically,,, veo 3.1 uses gemini to process user prompt

actually everything on Flow uses gemini to process .. literally everything , kinda dope

[candid relic]

Soemtimes I wish I could just put "Why did you do that?" in my prompt.

[crisp basalt]

Youd be surprised by how the “expanded prompt” looks when using the Veo 3 model. For context, prompt rewriting is enabled by default in Veo 3 / 3.1 and cannot be disabled , even on API

[edgy holly]

All the AI video models generate at 720 or 1080. To get higher resolution you need to use an upscaler like Topaz.

[edgy holly]

This explains a LOT.

[carmine crane]

hi guys

[pearl pewter]

hello my flow quality of pictures is really bad why?

[woeful void]

Google Flow is the slowest and most laggy service I've ever used.

[pearl pewter]

Same for me its so bad

[woeful void]

It turns out that we can use VE0 only on other services,
but Google itself doesn't have a proper shell for this neural network.

[woeful void]

As far as I understand, labs.google isn't designed for normal use;
it's just a demo project.

[crisp basalt]

https://tenor.com/view/patrick-cleaning-floor-gif-24177059

[pearl pewter]

Is your quality also bad?

[summer rapids]

Problem is even if he generates the child with "create image", he won't be able to modify it inside "create image", cause it will throw an error. The only option available now is "create an image, be extremly lucky and use that in flow", or maybe for landscape use it as a reference (I can't use that cause I only do portrait mode video) you can't iterate or generate and use image from a 3x3 grid for exemple.

I found a trick, but it's not that good, with a quality loss and very time consuming, it's :
extract a frame from a video in scene buider -> generate a new video saying what we want the child to do or add a zoom on the face, or anything from that frame with add to prompt and Image to video -> extract the new usable frame from scene bulider -> generate a new video.

It takes 5 min to get not really good result when we could do 50x that in seconds. Hope you'll get the time to fix that soon. Or being able to upload child pictures that you know were created from nano banana

[woeful flare]

hey guyz, really need some help, i wanna generate a realistic vlog video with the given character in the main frame and should be very close to camera and should voicover and proepr lyp sync and proper scene changes in the background... need help in this facing a lot of issues i have wasted more than 1k credits in this

[edgy holly]

I do not understand why apps keep doing this. It always makes results worse.

[woeful void]

The engine is fine,
I'm talking about the Flow interface.

[woeful flare]

anyone??

[edgy holly]

That's several problems. It's too complicated for a random chat on social media, other than to say: take it step-by-step, and learn how to use these tools better.

[raw fossil]

Hello everyone.
Flow VEO doesn't work in my region. Anyone else having problems generating videos?

[woeful flare]

i am facing a lot of issues, would really appreciate some guidance

[oblique wadi]

Google Labs is not available in France. Is there a way to fix this problem?

[dull rapids]

vpn

[wary quiver]

Hi

[safe ingot]

Is there a recommended JSON structure to send to google flow?

[edgy holly]

I gave you some. Getting good at all the things you asked about will take you hundreds of hours of work and practice.

Start.

[serene grotto]

Yes, since yesterday I can't generate any video in FLOW veo

Flow is not working since yesterday. Someone in Brazil also having problems with veo 3 in FLOW? It is not possible to send a feedback in the site. What is happening?

[raw fossil]

Thank you so much for your reply!
Videos from the library also won't load.
I asked to see if anyone else had this problem.
I'm from Uzbekistan.

[earnest prawn]

getting tons of these today:

[narrow cedar]

Thank you so much Ian.

Some observations & queries:

1. The upscaling option is only turning up as hover action triggered overlay in the image grid. From a UI pov, this could be missed easily so if upscaling options were available in a more obvious place or in the edit modal then I suppose it can enhance ease of use.

2. For upscaling to work should it come from the Flow ecosystem only or it will work even if I generate it from Firefly or other third party platforms but definitely with Nano Banana Pro. (After Dec 12)

3. Suggestion: While the 4K upscaling locked for Ultra plans make sense, still I would like to pitch for an alternative (more adoption friendly) feature-pricing combo.
   Please consider if 4K can also be made available (limited - may be 10-20 per day) to Pro plans also.
   This can increase adoption, help Pro users get a taste of all the features and upgrade for getting a higher quota.

Thank you so much.

[novel tendon]

?

[cloud monolith]

#bot-commands・🤖

[edgy atlas]

Hey Everyone! I am V! Based in Europe and currently working on Ai video generation in multiple domains using VEO. Would love to connect with people doing the same!

[fast gust]

This is ridiculous. Failing audio errors...

Fix this shi* Google.

[elfin cove]

Yow V. I'm Raf. from Belgium. I made a superhero based on myself and my fav. beer:

Duvel Man.

any google antigravity users here?

[analog lava]

@gusty warren Hello, as a friendly reminder videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[jaunty stratus]

guys

everytime have problem

failed generation

😠

[plush crystal]

Hi, do I need the Ultra plan to generate in 1080P native and watermark removed? or can I do this with the Pro plan too?
Ingredients to video cannot be generated with Veo 3.1 Quality. Thanks

[zealous tusk]

i thought it was limitless...

[safe ingot]

Any suggestions on how to keep the camera dead center and locked with no pan, shift or zoom?

[dire trout]

me

It is good but code generation sucks

[woeful void]

Unfortunately, the "Flow" service has become so BAD that we simply can't use it;
it's incredibly buggy and heavy.

Why can I run games with millions of polygons on my computer,
but I can't even scroll FLOW VEO, through a gallery of generated images without glitches ?

Currently, this is the worst interface in the entire video generation app market.
Even smaller services like InVideo and Leonardo
work faster and offer a more pleasant user experience.

All projects posted in the Labs.google section - is remain experiments,
but not fully functioning services.

[teal tapir]

How much did you generated?

[glad laurel]

Always is something went wrong when u want to generate pictures

[pseudo mural]

Was building an ad in scene builder it got stuck I refreshed and all scenes in the project that I spent 700 credit making are gone? What did I do wrong?

Nothing

[narrow sigil]

Is there any issues today to make videos?

My videos are not generating and the logo picture it changes

[safe ingot]

Can google flow read an image from a URL?

[edgy holly]

No watermark in Ultra, but videos generate in 720p.

[edgy holly]

Say something like "Static shot of..." at the very start of the prompt.

[whole osprey]

pretty good logo btw

another concept

[plush crystal]

Is there a way to generate 1080p native? And what are the upscale options in the Ultra plan? Thanks

[edgy holly]

Only in the API.

1K, 2K and 4K

[full flicker]

Hello @warm depot and @ruby maple as a friendly reminder, videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[teal tapir]

Too many issues in video generation today, its failing to generate audio so it not even generating the video (I don’t even need the audio)

[tough garnet]

I prompt for two charaters to have a dialougue and all 6 genaration had the wrong charater speaking. I try the same prompt using 3 other similar tools and all 3 aces it. Google please it together everyone don't have Google money to waste on credits.

[turbid lintel]

flow use gemini 2.5 or 3.0 ?

[woven oyster]

after how many images?

[turbid lintel]

use timestamp prompting

[0.00 through 0.04] he ask " DIALOUGE "
[0.04 through 0.08] she replied " DIALOUGE "

[frozen sand]

Something went wrong ???

[long lake]

I’ve been having this error for several days. Is anyone else experiencing the same issue? 😒

[true pumice]

How many did you generate?

[mellow radish]

Comon give us veo 4 already ... its chrismasssss

[wicked glen]

@inland pier do check DM, thanks

[full flicker]

Hello @boreal knot , as a friendly reminder, videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

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

[south yoke]

Is Flow having an error? I can't make a video in Flow

[full flicker]

Hi there, We have forwarded your message to #flow-tips-n-tricks・✏️ . That is the appropriate channel for the tips and help your are looking for, Thanks!

[edgy holly]

After a while, you learn to mute and block channels that push clickbait like this.

[topaz haven]

I'm also experiencing errors using flow to generate videos.
It will seem to be generating but stops at 99% without completely generating the video.
What could be the cause of all this issue

[mint pewter]

Came here to ask, I see it isn't mee, I can't generate Video no Image

[jaunty wing]

<@&1366831844502274180> I do not want to cause an issue, but Im confused and was hoping you can provide more detail. I posted about an animated cartoon series (Barky's Auto on Youtube) that I created entirely in Flow, and I posted it in the Flow Gallery. Saphire removed it and said I was advertising and doing self promotion. How is this advertising and self promotion? Isn't the gallery where we should post things like this?

[patent stone]

Please reach out via @real wyvern and we can provide more clarity on the matter outside of the general chats. Thank you!

[frozen sand]

me too

[shell veldt]

Not opening

[sick sparrow]

are 2k/4k downloads not working? - seems like only 2k is working for me atm

[ripe yacht]

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

[nimble marlin]

same

Couldn't generate image. Try again later.

[woeful void]

I've had enough. 🫡
I'm switching to other services, Flow became horible.
Flow gives errors half the time and is incredibly slow. It's unusable. 🤬 worst UX\Ui experience i ever had.
I sent you 20 reports of minor service issues that you still haven't fixed.
He almost never understands what I ask of him.
I paid for two months of Flow, spending $100 on AI PRO.
I made three videos, spending an unreasonable amount of time on very simple tasks.

[wicked shadow]

2k image downloads not working

[somber locust]

Does anyone know the maximum length of video I can extend in SceneBuilder?

[turbid lintel]

infinite

[somber locust]

My video is 8 minutes long, but I can't download it. It seems like forever to export. It keeps spinning. Do you know how to fix it? Why?

It said 'Exports failed'....

[slim sinew]

Animate only gentle, tiny movements: the FOOD-embedded baby-face blinks softly, its stylized mouth slowly opens in anticipation as a hand brings a small piece of FOOD toward it. The FOOD creature leans forward slightly to reach the bite, takes a small nibble with a soft mouth motion, and shows subtle cheek squish followed by delicate chewing. Add faint happy micro-expressions and minimal slow breathing, keeping all motions smooth, cute, and delicately understated.

Animate only gentle, tiny movements: the FOOD-embedded baby-face blinks softly, its stylized mouth s

[edgy mango]

i think you guys have to put a sticky note at this point

[main cypress]

Hi guys 👋🏽 can I generate nano 4K images in google flow?

[sick sparrow]

working again for me

[remote vortex]

Are all of you experiencing this problem? My video is not generating at all.

[sick sparrow]

i have a high experience of 'sexual content' today, only this is someone washing his hands 🥲

[tropic seal]

germany reporting in. Stuff is slow... seems like whole veo is crumbling ...

[topaz haven]

Same thing I've been experiencing since last week
I don't even know what could be the problem and when it could be fixed

[narrow ridge]

me too

[shadow warren]

knowing flow, it probably has an issue with the bare ground, for fear of nudity 🤣

[edgy holly]

I don't know if there is a limit, per se, but it gets glitchy when you try to make a bunch of them.

[slate olive]

Having the same thing not working at all has anyone got an idea why

[narrow ridge]

The account that has problems is Ulter or PRO？

[slate olive]

Ultra

[main cypress]

Hey guys 1 question can I generate 4K images in flow with 🍌 or no ?

[mint pewter]

if you got an Ultra sub, yes

[tight yarrow]

Hello is there a way to share my Flow Ultra subscription with my employees?

[main cypress]

I see now 🙂

[hushed hamlet]

I understand not being able to upload images of children to create a video. However, I don't understand why if I create the image of a child in a scene using nano banana, I cannot use the same image I just created with nano banana to complete my video in flow.

[south yoke]

Failed Generation

Failed Generation

why always Failed Generation?

[cinder blaze]

So ya, I just created an image of santa clause in flow. Then i tried to edit it. Literally every image i create with a person in it gets rejected. And im Frikin paying 250 bucks right now for the Ultra....

Does anyone know a work around this?

[pseudo mural]

How is it rejected? You get an error message of any sort?

[cinder blaze]

I get the dreaded "couldn't generate image. Please try again later." Doesn't matter if its banan pro or just banana. any image i generate with any sort of semi realistic character is unable to be modified. I can no longer create a first and end frame. Its basically killed the whole feature.

And the worst part? I was just about to sign a contract to be creating Ai shorts for a startup....I'm going to have to renegotiate with them.

[pseudo mural]

Is your prompt too complex? Can you share it here? What's the exact setup? Ingredient, frame to frame? AI video generation has his own limitation. But I did an ad two days ago and I only faced "Can't have minors" rejection.

Did you try whisk to generate the photo than take it to flow? I though flow is for AI filmaking mainly.

[jaunty stratus]

@uncut pecan

we are paying 125$ or more for ai ultra and cant use flow!!!

still everyone has the failed generation

the issues is there since a week

and you dont react on any #1319424187730755636

I cant understand that!!

@main saddle @limber zodiac please fix that 🙏

if somebody know to fix that please dm me 🙏

[cinder blaze]

so trying to upload this to flow and i get this message "we do not allow uploading of minors at this time."

Google better really offer back a prorated portion of our subscription.

[pseudo mural]

AI probably misjudged the age because he guessed the parchment is regular size so the writer must be a kid. Or something to that effect. It's a guessing machine after all.

[cinder blaze]

None of this would have happened two weeks ago....They completely nerfed the usability of their product.

[pseudo mural]

Abuse lead to nerfing. There are some nasty people out there and they are taking advantage of generative AI. It's a paradox really. They will either nerf it, or releave themselves of all legal liabilities. They are certainly not nerfing it out of the good of their hearts.

[cinder blaze]

It doesn't matter. if you pay for a service and then right after you pay it, they decrease the viability, that just bad practice. And it will drive people away from your platform

[turbid lintel]

the only sport video with no sloppy ai is basketball

[hidden latch]

@johnnyrobert pass the job along this way, I'll take it. I can deal with the AI Hallicinations.. seriously

Right now, AI hallucinates it is what it is.... no matter how much we voice our grievances, nothing changes. Instead, analyze it, find out its weakness and strengths, make a plan, Conquer

[inland pier]

Unfortunately no. We do offer a Ultra for Workspace for this purpose (https://workspace.google.com/products/ai-ultra/)

[hidden latch]

Yes, give him one of your laptops already signed into Google flow veo3 If you trust him.

And if you don't trust him, why did you hire him? Give him a laptop!

[silk hill]

so who decided that the output of veo 3.1 t2v should always be incredibly over bright, oversaturated, over contrasty, overlit and over-sharpened? Was the entire community clamoring for the quality to be reduced like this? Its so awful.

[stuck adder]

They need to update Veo man

[ember dove]

Hi

[ember dove]

Ai

[edgy holly]

Are you downloading with that upscaler?

[silk hill]

Absolutely not lol

[agile geode]

How do I fix the "something went wrong" error?

Has anyone else encountered this problem? I haven't been able to create videos for hours.

[ripe yacht]

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

[mint pewter]

has been like that for days ...

[tight yarrow]

Does anyone know how to fix this?

[jaunty stratus]

you dont have subscription, expired i think

[lime thicket]

hello

[long wharf]

Try to clear the cache

[sick sparrow]

unable to generate 2k/4k images on ultra subscription

[crisp basalt]

hoping veo 4 will also have advanced reasoning ("Thinking") just like nano banana pro model.

[finite falcon]

Hi, has anyone experienced the occasional video stitching error?

[frigid pier]

lol just figured out that Google VEO watermark is bigger now

right?

my old VEO videos have smaller watermarks

[cyan saddle]

I have a picture of a construction project before work started and after work was finished, and I want to create a transition between the two images. What command can I use?

<@&1173646484311384114>

[rich thunder]

I feel you. Since yesterday I am not being able to generate any video. It looks like the audio generation is causing problems. Very frustrating to invest so much for not being able to use the service. I had to go do fal.ai in order to use it there in order to complete the video generation.

[jaunty stratus]

😩

Fal expensive

We pay for ultra plan and can’t use anything

[rich thunder]

I chat with support and they said it is normal due to high demand. But I don't think it's acceptable to be more than 24h without being able to generate a single video

[jaunty stratus]

no its not normal

i use veo since 6 months and never got this error since a maybe update some days before

and you see all people here are crying 😭

[rich thunder]

hopefully this is just a bad hit and everything will be back to normal soon

[quaint olive]

Somebody just told me that you can have 2 people having a conversation in Veo 3... is this really true?

[hollow trellis]

All video i generated are low quality and audio is asynchrone with avatars. I lost about 2000 credits for nothing. I m so dissapointed and discouraged for my project

[edgy holly]

That's how much it costs in the API.

[jaunty stratus]

you pay here 10 credits

of 45.000

or nothing with lo priority

[edgy holly]

You brought up Fal. I responded about Fal. The heck are you trolling for?

[turbid lintel]

veo 3.5 or 4.0

🫣

[analog lava]

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

[honest nymph]

A pilgrim with a backpack on his back completes the Spanish pilgrimage to Santiago de Compostela before the cathedral, the final destination, holding a rosary in his hand and praying devoutly. The evening sunset is fading. The atmosphere is spiritually profound.

[analog lava]

Hello @lilac imp we wanted to remind you of our server guidelines, as stated in⁠ #start-here・📍: “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Thank you!

[austere kestrel]

A 3d animated rap video in a bold, satirical art style. a caicature of a politician in a daura surawal is rapping on top of a gaint view tower. below him, a train made of cardboard moves slowly. the sky is filled with floating gas pipes and 70-core banknotes falling like rain. vibrant colors, expressive facial animations, and rhythmic camera shakes following the beat of a rap song.

[restive needle]

💩

If it's not usable, don't try to sell it!

[rich thunder]

That's true. The results a really incredible. I had the best results with "text to video" and "reference to video" image to video works well but the video quality tends to be a bit more low resolution.

[fleet ferry]

Image tovideo

[stuck plume]

Can anyone please tell me how to fix this problem? I've already cleared the cache and restarted the browser, I don't have any extensions installed, and I even tried incognito.

[restive needle]

They can't even produce a single video or image! What kind of rubbish is this?

[viral grove]

I have a clip with a black spot that I really need to brighten up. Does anyone know how to do it in davinci resolve? somone help

[quaint olive]

same thing happened to me the other day and someone here told me to restart my machine which worked, but didn't fully solve the issue

[stuck plume]

Ye... it didn't help to me 🙁

[near gate]

When will be veo 4 launched

[jaunty stratus]

first they need to let us allow to use veo 3 😭

[rich thunder]

Engilsh please 🙂

[blazing venture]

[sick sparrow]

all flow ❤️

[agile geode]

Step into the world of ancient Egyptian mythology 🐍
This video is fully illustrated using Nano Banana Pro, brought to life with Veo 3.1, and powered by immersive sound from Suno AI.
Gods, monsters, and legends — reimagined through AI.

https://www.youtube.com/watch?v=zRZsdmP43f8

[tacit creek]

anyone knows how to get rid of the static sound when generating veo 3 videos? i cant use this version at all, i used 3 just fine and after months im back and it sounds so bad

[agile geode]

Try adding a clear audio line in your prompt, like:
Music: heavy rain on metal rooftops, ting ting, thud thud, plop plop
Being specific with sound descriptions (including onomatopoeia) helps reduce static in Veo 3.1.

[tacit creek]

what if i dont want any sound

[tacit jackal]

take the video into an editor and remove the sound re render it

[simple carbon]

Hi! Is the server down? I can't create videos on any of my accounts

[agile geode]

Veo2 :))

[unreal viper]

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

[tacit creek]

i mean i just want the person talking, no background music, like the veo 3 version was, this 3.1 is a mess

[storm vault]

Make nano banana can do an PNG

[simple carbon]

I was working on firefox, changing to chrome apparently solved the problem (and cleaning cache, maybe)

[edgy holly]

Just separate the audio out.

[edgy holly]

Static? None of the AI video models are very good with the audio yet (though less bad than a few months ago). but static I don't remember getting.

[rich thunder]

specifically add no music on your prompt

[lament raft]

Use a separate Ai reader and build it or yeah take the audio to separate software extract the voice from the rest sounds leaving just what you want. You can try to prompt it, but likely will still get an unexpected result

[rich thunder]

He's an example of a video I've create with VEO 3.1 fast no music, just ambient sounds

in all my prompts I tell to not include music nor soundtrack

[fast gust]

What am I paying $200+/month for with this failed audio cra#?

Fix this!!!!

[rich thunder]

This one was quite decent with OVI - open source project

[edgy holly]

If that doesn't work, positively prompt for silence. Ala: "The two characters speak in a completely quiet room.", or some such.

[rich thunder]

I hope VEO4 comes with way better and more natural results

[edgy holly]

We're starting to see competitors come out with the ability to upload custom audio and the video adapts to it. That's what I'm hoping we get in Veo 4+. Text prompting for audio is just too much of a blunt instrument.

I dno't know why people keep talking about Veo 4, though. 3.1 hasn't been out long.

[rich thunder]

some competitors are already generating many shots into a single prompt with many camera transitions, consistent character, etc

[rich thunder]

for me is because the quality in general still not cinematic enough. I feel the eyes and speech are quite robotic

[edgy holly]

Veo was the first major model to have ingredients, but camera control is one of Veo's weaknesses, yeah.

[rich thunder]

Agree. Nightmare for simple shots

[edgy holly]

Yeah, they're all bad on that. You have to use specific lip sync models to get good results.

[rich thunder]

but this one is actually very good

[edgy holly]

Simple shots can be relatively easily controlled with key frames, but complex shots are either impossible, or require burning a ton of time and generations. Combining Veo with other tools is the way to go in most of these cases. Each has their own strengths and weaknesses.

[rich thunder]

Flow allow editing videos with some specific camera controls. I tested and it was kinda good for my intentions, but I would like to have a bit more controls of micro shifts. A better and more complex frontend like Runway has

I undertand is not that simple, but as the big sharks on the market, they must compete in all levels

[edgy holly]

Yeah, companies with weaker base models are leading with these functionalities. Every time I test Runway, the overall quality is too low for what I'm doing, but they're innovating with their controls.

[rich thunder]

I am quite satisfied with VEO 3.1 mostly because of the consistency of the output. It usually have the same quality after all.

in the next 5 years the scenario will be completely different and the competition will need to fight hard to standout

[edgy holly]

Yeah, and the Flow web app is only a few months old, so naturally it's playing catch up with Kling, Runway, Etc., but the model behind it is almost always top 2-3, depending on the use case.

[turbid lintel]

just add this " background sound : Muted "

it totally shut background music

[cosmic kelp]

how do you make long secenes with no errors.

[winged plinth]

Hi Danny, you can stitch the generated clips together to make a longer one. The Flow editor allows you to do that:
https://support.google.com/labs/answer/16353334?sjid=6974575107412316171-NA

[cosmic kelp]

that's what I am doing and I can get up to the 5th or 6th clip and it errors.

[south yoke]

wtd always error

[cosmic kelp]

don't know

[midnight lagoon]

Every time I try to do something, it gives me this error. 💀

[cosmic kelp]

why do it have errors, isn't google fixing it.

[limpid jackal]

how to login on flow but without google mail

[midnight lagoon]

Damn, my second account and the same error

Damn it

[agile geode]

Even if you create 10 videos and each promotion is 1-2 minutes apart, you still get the "something went wrong" error.

This is a damn error, fix it.

[dusk magnet]

Are you good at telling which AI was used to create a video or image? You seem pretty knowledgeable about AI kind of new to it.

[warped topaz]

Guys in my flow nano banana pro all generations are getting stuck at 99% when I close tab and check again , it's gone, anyone knows how to fix it

[turbid lava]

I have paid for a Flow AI subscription, video generation repeatedly gets stuck at 99% and does not complete, even with simple prompts. This issue started recently. Any solution??

[near gate]

Clear your cache or try a different browser

[edgy holly]

Depends on the user.

[open knoll]

do anyone else is having a problem with first and last frame not ending up in the video that suppose to use them? especially the last frame...

[edgy holly]

Share the frames, and the prompt.

[unreal skiff]

Can you create a video using an image along with a text prompt?

[jaunty peak]

Yes, I can create a video using a provided image along with a text prompt.

[edgy holly]

Yeah, that's usually the best way to do it.

[full flicker]

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

[crisp basalt]

https://tenor.com/view/monster-versus-vs-alien-monster-vs-alien-gif-7403118861599528686

[fast barn]

I just feel that Flow should upgrade their model of Veo that support for 2D too, because currently, I felt like Kling and Seedance do give a better result.

1. It always turns my character/frame into 3D, which is pretty annoying.
2. It adds the new characters or subjects into the scene, even though I stated clearly they should just do the boomerang/looping action.

[analog lava]

Hello @north reef , we wanted to remind you of our server guidelines, stated in #start-here・📍 : “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Also videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. Thank you!

[raw oak]

hi! is it just me or is nano banana failing for others. My prompt is in french. When I use thrid party API connections it does not fail. Only in FLOW!!

[icy mist]

why "Based on your current settings, you need to use 0 credits for each creation" but still have limit? :<

[lone kiln]

How do you guys upscale your videos?

already found free upscaler, nevermind 😄

[dusk magnet]

Check dms when you have time.

[exotic shell]

Any update when we get rid of "lower priority" for ultra users and can generate quicker again?
Im so unproductive since this change.

[crisp basalt]

why do u want to get rid of lower priority

i mean i dont even use it much and im generating vids without any issues

[exotic shell]

because high priorty means 5x4 videos generated in 2 minutes, not 4 videos in 2 minutes.

[crisp basalt]

im trying to make sense of your logic tbh lmao

[crisp basalt]

did google announced this recently or something

also theres no high priority option wdym

just dont use lower priority model

[exotic shell]

lol. if you havent noticed: thats what ULTRA users are used to and paying for. thats nothing new. thats actually normality

[crisp basalt]

???

used to what

are u having issues with generating video or what

[exotic shell]

on what planet are you living? a few weeks ago it was normal that ultra users can create unlimited veo-fast videos without having to use the "lower priority" generations. 5x4 videos at the same time.

[crisp basalt]

5x4 videos?

uhhh

what lmao

weirdo-

how does lower priority messes with ultra users i dont get it

its just a label

it still uses same model when they didnt label it as "lower prioirty mode"

[exotic shell]

i told you several times: you can only do 1x4 video generations at the same time. YOU CANNOT DO 5X4 GENERATIONS AT THE SAME TIME.
Thats 80% productivity loss.

[crisp basalt]

???

so u wanna do like multiple generation of 4 output?

[exotic shell]

okay, if you dont understand it: start 5 low priority generations within 5 seconds and see what happens

[crisp basalt]

5?

5 what ?? 5 outut?

max is 4 videos output

do u mean like 5 generation of 4 videos?

thats like gooner level ,

jokes aside

[exotic shell]

as i said, 5 "generations".

[crisp basalt]

this generation r cooked

[exotic shell]

yes, generation -80%

[crisp basalt]

im talking about your generation

but anyway

lets find out

[exotic shell]

im also down 80% because im producins 80% less 😄

[crisp basalt]

well it takes time

[exotic shell]

how would that work?

[crisp basalt]

crazy

ive never done 20 video in one run

oh well, 20 vid generated

[exotic shell]

@crisp basalt

[crisp basalt]

what

i only followed what u told me to do

[exotic shell]

😄 you cannot start a single generation with 20 outputs. so either your trolling your you found a way to manipulate the request

[crisp basalt]

of course u cant u weirdo

no sane user would do 20 vid in one run, thats gooner level

anyway

[exotic shell]

if you didnt know: you could do 5x4 different ones as well! what an isnight! 😄

[crisp basalt]

the google staff quietly reading this is probably judging me so hard cuz i just did 20 gen in one run and their server system is probably making weird noise rn

[crisp basalt]

i didnt know cuz

idc

[frigid pier]

Why the flow team increased the VEO watermark size actually?

to get Pro subscribers to move to the Ultra plan?

[edgy holly]

You've got to explain what you're asking in the DM. I don't understand your question.

[frigid pier]

bro when veo 4

[safe ingot]

can Google flow recognize and work off of a PNG with a transparent background?

im looking to upload a wire frame within a PNG and see if Google flow can add some animations to it

[edgy holly]

Probably would take it as white or black blank space.

[stuck adder]

hey guys

[turbid lintel]

nano banana pro is awesome

[wheat cargo]

He decided to make me sing

[glass dock]

Couldn't generate image. Try again later.

how to fix this error?

[tender fjord]

weird i loaded flow for the first time and i have a project from May 21 2011 🤔 wat

[crisp basalt]

i love itt

[turbid lintel]

nice video bro

the physic is great tho

[agile geode]

Regarding Veo3, in my opinion its main strength is the ability to understand context quite well and deliver results that closely match the requirements and details. However, the biggest weakness for me is that the action scenes and default characters still feel too animated and not realistic enough.

[crisp basalt]

https://labs.google/fx/tools/flow/video/0mfdpcmtt0000

[tepid iron]

Hi everyone! I'm new to this. Could you recommend the best platforms or tools for creating Al-generated videos?"

[harsh cedar]

Thread

[analog lava]

Hello @inland oak , we wanted to remind you of our server guidelines, stated in #start-here・📍 : “Currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.” Also videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. Thank you!

[wheat osprey]

the scene builder is 50 percent accurate in terms of prompt to scene

[rotund schooner]

is there anyway we can generate video without gap between 8 seconds video?

[desert grail]

my game complication of gameplays

@high fulcrum do you like my complication of games made by veo flow

[high fulcrum]

hi, please only ping mods for moderation issues that go against our ⁠rules-and-guidelines・📋 thanks!

@crisp basalt A friendly reminder of our first rule to be compassionate and treat others with kindness and respect.

[crisp basalt]

Oh sorry

[mossy iron]

Hello, I have the Google AI Ultra package. I'm using Flow with Veo 3.1-Fast [Lower Priority]. The generation cost is 0 credits. What's the situation with the Google AI Pro package? Is generation free like in Ultra, or do we pay for it with credits?

[desert grail]

Ok I understand sorry I'm friendly and I'm new to that server

[fading wind]

Excuse me, but I have gotten the ultra version of flow AI and when i am trying to export a video i made it gives me 2 options download and discard. but the download option is not responding and i am unable to download the video.

pls help as it is urgent

[rich thunder]

try refreshing the page

[fading wind]

already tried multiple times

[rich thunder]

What is the exactly message you're getting?

can you send a screenshot?

[fading wind]

like it isn't responding

i am clicking download but there is no feedback but when i am clicking discard it is discarding the export

<@&1366831844502274180> can you pls help?

[unreal viper]

hi unfortunately mods are here for moderation issues. I'd suggest opening a ticket in #1238621511401734254 and detailing your problem there!

[fading wind]

oh sorry mb

[unreal viper]

no worries!

[fading wind]

it is saying i dont have permision

[unreal viper]

try #1238621511401734254 !

it should work there

[fading wind]

okay thanks

[fading wind]

it is not allowing

[unreal viper]

as stated in the red text, please add one of these tags

so it would be flow for a flow product

[fading wind]

how do i add it?

oh wait got it

thanks and sorry for disturbing

[rich thunder]

It's better you use the computer. I have no discord on phone so I have no idea how to do it now

[lusty oracle]

I am struggling a bit with a close-up shot, with reference image of a person looking straight ahead, a bit off camera, single frame 2 video. How to prompt the character to turn its head slightly, and gaze in a specific direction? Like "looking to camera right side at a fixed invisible screen"

[edgy holly]

Subtle movement is a real struggle with Veo. It always tries to do more than you ask for

so you migh try to ask for no movement if you want just a little.

[primal loom]

Dance video kese banaye kisi bhi image ka???

[vagrant laurel]

How many credits veo3.1 fast costs if you have ai ultra?

[edgy holly]

Zero or ten per 8 second clip.

[stuck adder]

whats this??

[hollow warren]

Is there a limit to use nano banana pro at ultra level.

[real saddle]

every single of one of my genrations keeps saying failed with audio generation but cancels and doesnt generate anything

[cloud wadi]

when using Frames to video, if I use my mouse wheel to scroll down the grid of images, the UI goes nuts at the end. why?

[ripe yacht]

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

[mossy iron]

I have Ultra and generate a lot of images. I've never received any information about a limit. Do you have the Pro version?

[fathom pine]

Can someone help me create prompts or send me some for creating an influencer account to sell products on TikTok Shop?

I can't create good videos with audio and an influencer that looks real.

[quasi owl]

is there chinese channel?

[safe hemlock]

I am trying to get my voices from Eleven Labs, is this possible in Opal, or shoudl I use the AudioLM and try something different?

[warped topaz]

Thanks alot bro it worked

[runic spear]

everyone try this script
The shot is an extreme long shot captured from an unnervingly static low-angle, rendered with a hyper-detailed, glistening wet digital texture that emphasizes the sickening elasticity of living tissue, overlaid with a sickly film grain mimicking magnetic tape distortion. The scene is defined by a single source of oppressive, clinical yellow fluorescence that barely penetrates the deep, suffocating shadows, creating an atmosphere of inescapable, agonizing dread, rich with a palette of pale bile yellow, moist industrial gray, and exposed, vascular maroon. We are in a vast, empty subterranean corridor at a moment of absolute biological violation, where the walls weep putrid moisture and the concrete structure itself seems to suffer from cancerous decay. Our subject, Lila, is no longer a woman but a profound biological tragedy: her stretched, stained shift dress is being actively shredded by the impossible expansion of her torso, where her ribcage has distended into an asymmetrical, bulbous cluster of calcium and cartilage, and her skin is pulled translucent, exposing the squirming, wet vascular pathways beneath. She is in the middle of a wrenching, slow-motion metamorphosis, her spine arching backward as new, vestigial bone structures press through her shoulder blades, interacting with the slick, bacteria-matted floor as she struggles to maintain structural coherence against the force of her own mutation. The air hums with a hyper-realistic, low-frequency electrical drone that vibrates in the teeth, and the overall audio has a compressed, claustrophobic quality, allowing the subtle, wet Foley of tearing collagen and the audible, sickening crepitation of fractured bone to cut through the oppressive hum. This distorted thing articulates, its voice a reedy, desolate whisper filtered through a wet, choked gurgle, saying, "It’s always coming back, isn’t it? The shape."

[prime kestrel]

it fails in flow.

[runic spear]

gee i wonder why

its a script for a transformation

[prime kestrel]

its to gore bro

[runic spear]

Try it with Gemini

[upbeat plume]

Hello. I need help!

I can’t generate videos. I’ve been trying to generate for the last 24 hours and I always get “audio problems” and it fails to generate. I’m on Ultra Plan. Asia Pacific Region.

[flint oxide]

@stray mirage Hello, as stated in #rules-and-guidelines・📋 , our team can only support engagement in English at this time we will have to remove this post.

[tight crown]

A curious young Nepali boy named Vivek standing in a small mountain village, simple clothes, calm evening light, peaceful mood, 3D cartoon style, same character, same face, same clothes, cinematic continuity, high detail

[south blade]

anyone using for make urbanism video?
I like render it in 4k quality and create good videos.

[turbid umbra]

question, I did one extend, how do I delete the extend from the timeline if I dont like it?

[tough otter]

Can we generate 1080p videos? Dose 3.1 quality means 1080p?

[dry frost]

You have to find the clip you want to remove in the project library and click the "remove from scene button"

[turbid umbra]

yup found it, kinda sucks you cant do it from the timeline straight, I guess it would be a nice feature to add in the future.

[shadow warren]

nobody can tell me the quality of the free ultra gens is the same as the 10 credits (which should be free)

[upbeat sierra]

hi @shadow warren the quality is the same for both versions 👍

[upbeat sierra]

@turbid umbra You can delete clips using the 'Arrange' button on the right hand side of Scenebuilder. The method that @dry frost mentioned also works!

[turbid umbra]

works, ty!

[solid marsh]

Why can't we have an option in flow nano banana pro to get the same aspect ratio as the initial image back just like in the Gemini app? In flow it must be 9:16 whereas in the app we can upload an image and edit it with nano banana and get it back exactly in the same aspect ratio

[supple karma]

can someone list me all the expriment link

[zealous tusk]

can someone tell me why my question is being erased from the chat?

[unreal viper]

reminder to please post images in #flow-videos・📹 !

[full flicker]

HI there, you post was removed because it was a duplicate post and in the wrong channel. Thank you.

[zealous tusk]

what is the right channel to post it?

[full flicker]

Where you posted it, in #flow-tips-n-tricks・✏️

[tough yacht]

They are the same - it's an identical model used for both. Are you sure you're seeing a difference?

[crisp basalt]

Aren’t u underage or something

[runic spear]

Nope

[crisp basalt]

Sure

[slate olive]

Is anyone else getting weird banding and flickering on Veo 3.1

[mossy iron]

Is there a generation option in the Veo 3.1 - Fast [Lower Priority] model for 0 credits in the Pro plan?

[edgy holly]

No. All the AI video apps with unlimited fast generations are a LOT more than $20 a month

[mossy iron]

Thanks for your reply. I have the Ultra Plan, but I only use the Veo 3.1 - Fast [Lower Priority] model, which costs 0 credits. Do you know how much it costs to generate credits with this model on the Pro plan?

[turbid umbra]

anyone experiencing overload right now?

can't generate anything

[edgy holly]

I think Fast gens are 20 credits per clip, and quality is 100, so that'd be... 50 fast gens a month.

[ruby olive]

other AI's with unlimited plans are usually around 95-100$ a month

but not 250

Veo3 had a big advantage when it was released, but not anymore

[minor badger]

Is flow still choking on itself and can't generate images when using multiple references? I thought this would be fixed by now, especially for Ultra plans.

is higgfields worth it? anyone tried it?

[bronze shoal]

https://tenor.com/view/unigrid-decentralization-decentralized-technology-spongebob-gif-25734934

to make this

[pallid violet]

@hearty sableAs stated in #start-here・📍; currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.

[woeful void]

FLOW is real NIGHTMIRE i ever had in video production ))) wery buggy interface

[near gate]

What is going on with flow ?

[woeful rampart]

what haqppened

[near gate]

right now its failing 98% of generations

Cant even get an image

[wide ether]

When are the heck are you all going to add the option to download all videos as a package, on one operation ... like you can do in whisk? It's terrible that you have to scroll through all the videos and download them individually! Why isn't that option being added to flow?!

[hoary perch]

Can we generate a 30-second video from Google Video 3.1?

help me please

[high wren]

For the past hour, ImageFX image generator page keeps going back to Google sign in page. Then it goes to a blank white screen.

[near gate]

Yes its completely rocked right now my friend

[high wren]

Ok thanks. It's not just on my end.

[hoary perch]

?

[bronze shoal]

well i tried and got some result

[crystal talon]

Hello all, does google flow offer CPP?

[minor badger]

only with the extend feature as far as I know. but the quality of the extension will depend on your end frame.

[bronze shoal]

should i shere it? its lil creepy...

[edgy holly]

Its a good companion for Veo.

[edgy holly]

Clips are 8 seconds each. You start with one, then extend until theyre as long as you want

[wide ether]

Getting a point and click finger cramp just downloading all videos individually 🙄

[woeful rampart]

same

not working for me too

[runic spear]

@bronze shoal do it

[regal raptor]

Anyone else seeing this issue? When you upscale videos it adds some strange grainy noise over white backgrounds almost looks like film grain. I thought it was my prompt first then noticed it only appears after the upscale like little black dots everywhere appearing for split seconds like its an old film

[edgy holly]

That happrns sometimes, with or withiut upscaling.

[regal raptor]

only happens to me with upscaling, never seen it in natural 720p

[edgy holly]

Those sharp little static rectangles have been in Veo clips since pre-release Veo 2. Less than it used to be, but happens.

[reef mural]

Yes, at first i thought it was a bogus, but after using cinema studio video, and cinema studio image, im totally sold, i have i 5 days produced a 15 minutes short about mens violence against woman in relationship. Almost all done with cinema studio, it gives high qualtiy images in an easy way, i have used the format 21:9. and the image quality is amazing almost all the time. One thing lacks a little bit and that is speech, there no one can touch veo 3.1 especially if you have longer scenes than 8 seconds. there is Veo 3.1:s extend perfect....

[stuck parrot]

im repeatedly getting content flagged for violating rules through Veo. the subject is a comedy sketch of a giant jazz musician who dresses as a baby as part of his act. it even seems when i post the still of it here its getting removed.. I know its a weird scenario but its not sexual in any way. whats going on witht he sensitivity. previous posts ive put up here seem to be removed without even any response.

[minor badger]

Thanks! That's helpful..how's it in terms of credits burning vs Google Ultra?

[edgy holly]

Speech is one of the reasons I use higgsfield. I animate almost entirely in Veo/Flow, but when I have dialogue, I use Kling Lipsync inside of Higgsfield.

[edgy holly]

The filters are strong about things related to kids.

[edgy holly]

Much more expensive per gen if you're comparing to 10 or zero credit fast mode. For other things it depends. Each month or two they have some kind of sale/special where some things are free, but that's temporary, prices fluctuate and it's more if you pay monthly instead of yearly.

[minor badger]

Thank you

[edgy holly]

If you're really jamming, cranking out a bunch of Veo Fast clips, Ultra is the best deal out there. If you never use most of your credits, then you can probably find a better deal, depending on details, if you pay yearly and sales, etc.

Veo also tends to generate a lot faster. If you're as picky as I am, and don't have all day to wait for generations, you have to go through a bunch of generations to get it juuuust right, and if you're paying zero or just 10 credits, that's not a problem, but it adds up with lower limits and slower speeds.

And Google AI accounts aren't just for Veo/Flow. You get that enormous cloud storage, and a bunch of other stuff.

[eager hemlock]

Did you ever find a fix?

[stuck adder]

wonder when theyre updating veo smt

[stuck adder]

tbh the quality of veo dropped so much

[hazy acorn]

I'm trying to make my own version of Flow for my own AI filmmaking workflow, but unfortunately direct Gemini API calls cost me more than using Flow.

[reef mural]

Well if you get the big backage you get 6000 credits, and its, pretty cheap when using the cinema studio 18 credits per render, and almost all the time you only need one per scene. You get some unlimited stuff to, but that depends on the package and when you buying. And almost only low resolution or low priority stuff.

[hoary perch]

Awesome man

[frigid pier]

I think the Veo team updated the Veo 3.1

it now has some weird thing in the every generations

very disturbing

People who have been using Veo will understand what I am saying.

[slate olive]

The banding and shimmering is making Veo 3.1 unusable

[stuck parrot]

Hello - I’m repeatedly getting outputs failed for rule violations, and I’m struggling to understand why. The project is a comedy sketch / promo featuring a giant jazz musician whose stage persona includes dressing like a baby (nappy, bib, etc.). It’s absurd and surreal, but not sexual in any way — no nudity beyond what you’d see in a novelty costume, no sexual framing, no fetish content. I completely get that this is a weird scenario, but it feels like the system may be over-indexing on keywords or costume elements rather than intent or context. Id appreciate any guidance on how to safely frame surreal or novelty characters like this? - - One extra detail that might be relevant: I’m not seeing the same flags or removals when generating the exact same content via Veo through Gemini. The issue seems to be specific to Flow

[neon oar]

Hey what is better sora 2 or flow/veo3.1

[teal tapir]

For some reason i can’t generate a prompt with female mentioned in it? Im trying to make few product shot for women sneakers ? What should i do

[teal tapir]

Same for no reason they are not generating, im thinking of switching to something else, its very frustrating

[lusty flicker]

Use women/girl instead

[teal tapir]

Not working here this is my prompt

macro shot of girl hands tying white laces on fresh green-accented sneakers, clean fingers with pale nail polish, cozy background blur, slow-lifestyle lighting, cotton socks, soft carpet or platform floor, fashion realism, slight vignette for depth, warm tone cinematic light --ar 9:16 --v 6 high-

resolution product photography of lifestyle sneakers, white leather base with lime green overlays, platform sole, clean sculpted midsole and raised heel design, outsole tread featuring triangle grid pattern and debossed ‘coreve’ logo at center, symmetrical green rubber sole with matte finish and mid-channel flex groove, right side view shows metallic coreve brand tag near laces and double-chevron logo on heel quarter panel, left side clean with no branding, front view reveals round toe box with green leather guard, flat white cotton laces looped through eyelets, white tongue with stitched coreve label, precise tonal stitching, soft shadows on seamless backdrop, cinematic lighting with bounce fill, macro realism with detailed textures, product-centric minimalism, photoreal finish, lifestyle fashion tone — vertical frame, 4K render, perfect for digital ad or catalog design

[median void]

Hello i am getting a error for any prompt even sending "test"

[teal tapir]

Whats wrong with flow today?? Its so frustrating i cant even make simple shoes product images

[lusty flicker]

Let me try in my design

@teal tapir I got this result

[teal tapir]

im generating images

Can you generate image with same prompt?

[plain dirge]

Hello, I have a relative who was charged $300 by this application, but we don't know how to cancel it.

She doesn't use this app; the charge just appeared on her bank statement.

I saw that they had a Discord server, so I joined to comment on it.

[lusty flicker]

Turn off Autopay

[lusty flicker]

Okk

@teal tapir

@teal tapir u need more???

[teal tapir]

I need alot, im making an ad, im trying to figure out why its not working for me and and same prompt working for you

[teal tapir]

Anyways thanks 🙂

[lusty flicker]

😅😅,it didn't even took 3 seconds for me.

[teal tapir]

And im trying for last 2 hour with different different prompts

You are using flow right?

[lusty flicker]

Flow for the vedio Gemini 3 for pictures

@teal tapir you does ad making professional or for yourself

[teal tapir]

Trying and learning as of now

[teal tapir]

Why do you have Gemini whatermark?

[lusty flicker]

.

[teal tapir]

Oh ok ok but in gemini we can only generate one pic at one time right?

[lusty flicker]

One with one prompt

U can type multiple prompt together.

[quiet estuary]

please help me... how to solve this problem????

@quiet estuaryplease help me guys

[teal tapir]

U can’t upload child photo but u can generate child using prompt there is no other way

[turbid umbra]

is it just me or speech seems to be sort of impaired, yesterday the prompt adherence was stellar today it adheres maybe 2 3 words out of the sentence

also another question, does it make any difference if I generate 4 at a time or 1 video at a time?

[edgy holly]

First place I'd start is stop using bots for prompting.

[acoustic lava]

All my Images in Flow projects are gone. Anyone else?

[pure willow]

You can let the group know you want them to really help

[edgy holly]

Mine are still there

[plain dirge]

We don't know what we can do; the bank tells us to talk to Google because they can't do anything.

And customer service isn't responding.

[fluid brook]

Image

[edgy holly]

It's New Years Eve. Customer Service probably isn't open.

[muted swift]

Please solve my problem

[real gazelle]

Happy New year all friends

[turbid umbra]

just subbed to ultra flow is too good, please don't kill it tho i'm getting used to it's workflow.

[exotic shell]

does anybody know a (automated) way to download all created videos of a project, without having to do it one by one?

[harsh ibex]

is any way to get 450000 credits?

[lilac hearth]

I'm really looking for the same answer. This is my most wished feature for Flow actually

[full flicker]

Hello, unfortunately the Discord server is not the place to get billing support or help. Your best bet is to contact Google directly. Try starting here first. Then there is a contact them at the bottom. https://support.google.com/paymentscenter/answer/9074244?hl=en

[indigo girder]

https://www.youtube.com/shorts/ni3tSwharoo

[elder bolt]

need help for this. not sure why flow isnt generating. first time seeing this issue

[kind niche]

Hello,
I just created a flow account, but I am not sure if I can do the following

Can I automate this using flow API?

1. Send some images and a prompt to flow API
2. Create a video
3. Download the video

[lilac hearth]

Having the same problems with generation this evening. It always tells something went wrong Immediately after clicking Generate

[woeful void]

Google specifically prohibits downloading all Project files at once ?

If I've generated photos that I want to transfer to another program for processing,
why should I download 50-100 photos one by one, especially with a very long delay ?
Is this program designed for working with media materials or not... ???

[edgy holly]

Ultra accounts cost $250 a month, and they come with 25,000 credits. You can buy more credits on to of that.

[edgy holly]

Flow is a web app, not a cloud service. The Vertex API serves their models.

[teal tapir]

But bot are giving good results, what else do you recommend, can you please guide me?

[fallow python]

I brand spanking new Green 🤢 to FLOW 🤣 . Been generating ai art 2yrs . Literally just gettin into AiVideo generation with VEO 3. I explored VEO 2 earlier this year .
I will share my work here when possible to see my progress
My first ever creation in FLOW
"Words are the vibrations of nature." — Masaru Emoto ❄️
​Exploring the hidden structures of water through the lens of Veo 3.1. A symphony of pixels and fluid dynamics, orchestrated by Imagen 4 and The Architect.
​We are no longer just prompting; we are conducting.
​Welcome to the New Year. Welcome to the Flow.
​

[rose cradle]

Hi fellow experimentors. I have this scene where I want my character to kick a door outwards. nothing seems to stop flow from swinging the door inwards. I have even create veo3 still images of each stage of the door movement and used the ingredients to video option. I have asked gemini for a prompt. still fails.

[turbid lintel]

share prompt here

[analog lava]

Hello @half vapor , as a friendly reminder videos and images are not generated within the server, you'll need to head to https://labs.google/, pick a tool, and generate it there. If you are asking for prompt advice please do so in the appropriate channel. Thank you!

[half vapor]

You making video Plasse

[kind bison]

Use first and last frame option. Use Nano Banana to create a the last frame in which the door is open in the appropriate direction.

[edgy holly]

That was not a good prompt. It was filled with unnecessary information. Using bots means you'll learn bad prompting habits.

[calm steeple]

Guys I wanna make a better ending for Stranger things (a famous series) of netflix (for those who don't know) and I wanna use flow so anyone wanna join and help as I am kinda intermediate

[zealous tusk]

how much time does this apply?

[cerulean jacinth]

Hello

Hii

[pure dome]

how many were you able to make , approx fig ?? since you are on ULTRA

[zealous tusk]

not enough

[rose cradle]

yep been there a few times. I get better results from the free nano offering than I can coax from my paid fx subscription. Whisk, flow and image fx are lagging behind.

[pure dome]

like approx figure bro ? 200 ? 500 ? ,

[onyx notch]

I got that message too. I’ve never seen it before. I’m thinking about subscribing to Higgsfield. I'm an Ultra subscriber.

[edgy holly]

You'll run into a lot of problems with this. They have filters that block generations of famous people.

[edgy holly]

Higgsfield is a good companion for Flow. I've yet to use all my daily allowance of anything on Flow, but they have lipsync and video upscale.

[teal tapir]

Can you share few good prompt and share how you work, would love to know

[calm steeple]

Oh, not sure but sometimes when I use frames to video then it doesn't interfere generally? If that is right

Only tried 2-3 times for Leonardo dicaprio etc, that time it worked, maybe sometimes work sometimes not

[edgy holly]

They aren't perfect filters, but it'll be an issue with a big project like this. Assume you'll have to grind a bunch.

[edgy holly]

That's a very complicated question. I've been working with AI since before Chat GPT-3 came out. It's not something that could be summarized in an online chat, or copy/pasted.

Each task and model are different. You want to learn prompting basics, focusing on very recent recommendations specifically for the use case and model you're working with as much as possible. That'll put you ahead of any prompting bot I've seen. Then keep learning through trial, error and watching other people who share their work using the same tools.

[rose cradle]

Prompt.
"Cinematic over-the-shoulder shot from behind a man. He forcefully kicks a wooden door. The door swings outward, away from the camera and away from the man. Dust and debris fly forward as the door opens."

[edgy holly]

Share all your prompts and inputs.

[pallid violet]

@tawny egretAs stated in #start-here・📍; currently, our team can support engagement in English; the discussion in this server will largely remain in English for now.

[near gate]

What is going on with flow... its been wonky donkey for weeks now and today its trying to tell me that the same character i've used throughout my videos is now a prominent person lol... I mean come on.

[crystal talon]

flow thinks this is sexual content. cant animate it. any tips?

i can make a woman in bikini run via text to video. but image to video is sexual content? ironically this picture was created witn nano banana. this is annoying. fix ur stuff google.

[wet stump]

Hi

In Flow (Nano Banana Pro), is it possible to export images in PNG format?

[stuck parrot]

Hello - I’m repeatedly getting outputs failed for rule violations, and I’m struggling to understand why. The project is a comedy sketch / promo featuring a giant jazz musician whose stage persona includes dressing like a baby (nappy, bib, etc.). It’s absurd and surreal, but not sexual in any way — no nudity beyond what you’d see in a novelty costume, no sexual framing, no fetish content. I completely get that this is a weird scenario, but it feels like the system may be over-indexing on keywords or costume elements rather than intent or context. Id appreciate any guidance on how to safely frame surreal or novelty characters like this? - - One extra detail that might be relevant: I’m not seeing the same flags or removals when generating the exact same content via Veo through Gemini. The issue seems to be specific to Flow

[near gate]

Is it working for anyone today?

[crystal talon]

same here, this will be the last time i pay money for google flow. everything is restricted or gets errored. i dont even have any kid actors but veo thinks a grown woman is a child and gives me error. a woman in bikini? sexual content. flow is a joke really. my money goes to klingai next time.

[near gate]

we all had the black friday special opportunity and I missed out... 4 months of veo is a year of kling lol

[crystal talon]

yes i missed it too. dont worry. newa SALES will come very soon again. they need customers and they will bait them. i already canceled my ultra sub for google. never again. veo is so anti violence, i cant even make a decent kungfu kick or shoot a gun properly. hit and miss after 100 generations. kling does it at first try. tell me how is this picture sexual content please. look at this:

[late rampart]

Actually, Google sees a bare stomach and cuts it off immediately; I had the same issue with this video.

Even here it's a mess, I can't even say the word "sexxxxual".

Does my medievaldwarf look sexxxxual? Hahaha, Google considers him to be so.

Oh my god, even the word "dwwaarff" is blocked here.

[edgy holly]

Depends how you use it. If you have a ton of unused credits, then Ultra is overkill, but if you use piles of Fast generations in Veo, Kling doesnt have unlimited.

[crystal talon]

it is a joke really. unexceptable what google does. but text to video is no problem. it doesnt make any sense

[tepid drum]

flow is down ?

[late rampart]

I think it's to prevent any misuse; at the slightest sign of a belly, he gets scared.

[reef drift]

Motion Prompt (English):
‎Mother dog slowly pulls the puppy closer, rain continues falling, subtle camera movement
‎
‎Camera Angle Instruction:
‎Medium shot, eye-level, both characters in frame, side view.

Make the video

[lone shoal]

I love Veo 3.1, but there is one thing that I frequently struggle with… Often I’ll have a start image for a tricky scene and an image of what I want to end with, but there should only be, say, 2 seconds between the start frame and the end frame. The length of the generation, however, is fixed at 8 sec in Flow. Right?

Anyone have a good workaround? Any rumors about being able to choose the clip length in the future?

[buoyant pumice]

Hi everyone — I recently discovered Flow and started using it to create time-lapse style videos of home remodels, and honestly, it’s been working shockingly well. I mainly use Frame to Video, and I generate the before/after images with Nano Banana Pro, then animate them into a remodel time-lapse.

Somehow these videos have pulled in millions of views, which is wild considering I’ve never had content perform like this before.

That said, I’m trying to be smarter about credit usage and consistency, and I have a few questions:

If a video generation fails, are the 20 credits refunded automatically?

Has anyone found a reliable prompt that consistently produces clean remodel time-lapse results?

Right now, about 80% of my generations have some kind of “weirdness” — melting walls, ghost-like transitions, or the interior just dissolving instead of being remodeled. Viewers replay it trying to spot the flaws (great for watch time, I guess 😅), but my goal is for it to look like an actual crew is renovating the space over time.

What’s frustrating is that even when I reuse a prompt that previously produced an almost perfect result, the next generation can completely miss — different room, different layout, totally off. That feels like a big credit burn.

If anyone has tips, prompt structures, or settings that help keep the same room while making the remodel look intentional and realistic, I’d love to hear what’s working for you.

Thanks in advance!

[random yoke]

Ope wrong thread

[stuck parrot]

Hello - I’m repeatedly getting outputs failed for rule violations, and I’m struggling to understand why. The project is a comedy sketch / promo featuring a giant jazz musician whose stage persona includes dressing like a baby (nappy, bib, etc.). It’s absurd and surreal, but not sexual in any way — no nudity beyond what you’d see in a novelty costume, no sexual framing, no fetish content. I completely get that this is a weird scenario, but it feels like the system may be over-indexing on keywords or costume elements rather than intent or context. Id appreciate any guidance on how to safely frame surreal or novelty characters like this? - - One extra detail that might be relevant: I’m not seeing the same flags or removals when generating the exact same content via Veo through Gemini. The issue seems to be specific to Flow

[edgy holly]

Ive had the same issue. Usually just grind a bunch of gens to get it right, but you can also start earlier so your end frame ends in the fight place.

[edgy holly]

Share your full prompts and any input images. Too vague to be helpful without those.

[rose cradle]

Restrictions seem to be a lot less for me when I say I am the character

[lone shoal]

Thanks for the response, @edgy holly. Yeah, usually I just work around it, but sometimes it means breaking a shot into two clips that would look better in one continuous 6 sec clip. I bet some bright sunny day we'll be able to specify middle-frame images 🙏

[edgy holly]

if you have access to Veo in any of the all-in-one apps, you can generate 4, 6 or 8 second clips.

[analog lava]

@rose cradle Forwarded to the appropriate channel!

Fwd

[zealous tusk]

Anyone else having issues with nano banana image editing or using reference image?

[tardy ridge]

veo down for anyone else?

[rustic hornet]

Wdym

I'm stuck 99

[tardy ridge]

was my internet

[turbid lintel]

veo 4 wen

[leaden gulch]

Yes! Then I try the same reference image and prompt in the Gemini app and it works fine. Anyone know what is causing this?

[pulsar ferry]

Im having the same issue

[flint tartan]

Why now it always fails 99%

[flint tartan]

Same with me

[tough otter]

INVALID_ARGUMENT，INVALID_ARGUMENT。。。 wtf，i just upload a midjourney image,and it told me invalid arugment.

[edgy gazelle]

hello is nano banana still not working?

[leaden gulch]

@pulsar ferry I think I figured it out. You just have to refresh? Lol that’s been working for me

[near gate]

Is anyone ever going to answer any of our questions or are we just suckers who paid for ultimate and now we sit here holding the bag.

[brisk cloud]

yeah its working for me now

[agile geode]

Started from a very basic prompt, then developed a custom expanded prompt that produces highly detailed, powerful images like this :))
If you’re curious, drop your prompt below and I’ll help expand it.

made 100% nano banana pro

This is the brighter light fantasy version, not dark fantasy.

[shell heron]

When I extend a horizontal clip in Flow’s Scene Builder and download it, is it 720p or 1080p? What resolution does it download in? Does it automatically upscale when exporting?

Anyone knows about this?

[patent widget]

Hi! What’s the current status of the recurring “Failed Generation” / “Couldn’t generate video. Try again later.” errors over the past few weeks? I canceled my Ultra subscription because Veo 3 became nearly unusable due to apparent server load, and I’m wondering whether those issues have since been resolved.

[round sundial]

Flow Ai Se Unlimited Ai Vlog Kaise Banaye | Unlimited Long Ai Videos Kaise Banaye | Flow Ai Tutorial

[jagged marsh]

Literally dealing with this right now. First few generations of the day, and all it tells me is "Failed Generation / Try again later".

I'd be pissed the f*ck off if I paid for Ultra plan lol

[patent widget]

Thanks for letting me know!

[delicate delta]

hey I can't do celeb from flow or using nano banana pro, is that a thing? do I need to do that from higgsfield ai?

[edgy holly]

720

You dont want to upscale videos in Flow yet. The upscaler is bad.

[pearl crown]

Hey guys! Is there anyone who experiences problems with generating videos with voice right now?

[near gate]

It asks you if you have the rights to any of the images you upload so I would assume the same goes for likeness of celebrities.

[turbid umbra]

if they can take topazlabs lunch with the upscale, it would be killer, 20$ per 10 tiktok videos upscale is criminal.

[still berry]

hi am new here

[surreal sparrow]

me too. Nice to meet you

do you use google flow

[still berry]

yeah

[surreal sparrow]

Cool. It's not available in my country

But I heard it's good for making movies

[thick matrix]

Question
Im unable to download a short video fron flow scene builder.
Anybody know how to download video from it

[latent turtle]

I focus on web3 frontend using React and Next.js
I connect UIs to smart contracts, wallets and on chain APIs.
I also handle backend logic, indexing, automation and defi/nft features.
Can help with standard web dev and API design too.

[drifting umbra]

does someone know why cant I use flow?

[waxen flame]

Hey bro you might using an organisational email ID, and some private organisations might not allow some of the Google services including flow.
I have faced the same issue you can try using another ID.

[minor badger]

Hey, I didn't find an enhancement requets section. I would love to be able to use the last frame of a generated video as input frame for a new one, without using the clumsy scenebuilder "editor". Is there a workaround to do that conveniently?

[digital forge]

Is anyone using Lmarena.ai for image generation? I’ve been using it for a while now and love it 🥰

[turbid lintel]

so what the 2026 update for veo?

[minor badger]

[sarcasm] more censorship 🤭

[bronze shoal]

Okay

[royal wasp]

Hello I want to create ad for my clothing brand can anyone help to use flow

[cerulean nova]

Nice to meet you

I want to create a video for YouTube

[frigid pier]

When veo 4 honestly?

I heard it will be released in May?

[topaz chasm]

hello, i was wondering if any of you are actually using chat gpt for creating a better prompt for the video creation?

[edgy holly]

You can just take the last frame from any video and use it as an input image in frame to video, but the Scenebuilder is much better at it, since it makes the transitions between clips smooth.

[fickle drum]

hi all. do u use any prompt (instructions chatgpt/ gem gemini) to create ptompts for veo?

[edgy holly]

Hoping for a Veo 3.5 update of some kind in the next few months, and a big Veo 4 later in the year. Hopefully with uploaded audio that will lip sync.

[minor badger]

The issue I run often with that scene builder is that the transition are the opposite of smooth unfortunately. There's often a second overlap between the clips, which creates a glitch somehow, that I then have to trim manually in Premiere.

It would be nice to have a one click option, like on openart for example. "Reuse last scene as input". But manually works out too.

[edgy holly]

That overlap isn't a glitch. It's supposed to work that way. It forces the second clip to start with the same visual components and movement in the last second of the first clip. When you only have a single image/frame to start the second clip, you almost never get the same movement, so there is a pause, jump or some other kind of alteration in the pace and direction of any movement going on.

[edgy holly]

There is something like this in Scenebuilder. If you hover over a clip in Scenebuilder, a line with a circled plus sign over it shows. If you click on the plus sign, it captures that frame, which you can then use as a frame in a new generation.

[drifting umbra]

It does not allow me still

I've tried different ID's

[edgy holly]

What does it say when you try?

[drifting umbra]

you talking to me?

[edgy holly]

Yeah. When you try to use it, what error do you get? What happens?

[drifting umbra]

It looks like you don't have access to Flow. Please verify your age if required

And when I verify it, it says: It looks like you don't have access to Flow. Check availability here and contact Support if you have any problems.

[edgy holly]

Just to troubleshoot... are you actually an adult who lives in a country where Flow has been released, and have a Gmail account that was started there (not through a VPN or some such?)?

[drifting umbra]

yup

I also know people in my same country that have used Flow

[edgy holly]

Sweet. Okay, have you tried accessing it through a stock Chrome browser? Sometimes there are issues with other browsers, ad blockers, security settings and such. (I use a plain Chrome install just for this)

[drifting umbra]

I've tried with chrome, opera and firefox.

[edgy holly]

Did you create the account very recently?

[drifting umbra]

Nope, I made it 1-2 years ago

[edgy holly]

I'm trying to remember all the issues I've seen people bring up over the last ~13 months...

[full flicker]

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

[rough halo]

does anyone have a good consistent nano banana workflow for when you swap a character with another from another image and it places the new character with proper features, but the coloring and lighting look as if they are still in the old photo which makes the image look like a bad photoshop job.

no matter how i tried to tell it to ensure that the new subject conform to the environment of the image - and descibe in detail what i mean by that. it still keeps doing it.

my workaround for this has been costly and just tedious. I would take an image where said subject looks accurate but the lighting and colors are not, i then generate a frames to video with that image as the start and end and prompt subtle movement of some kind. Veo naturally blends the pooorly lit subject right as any movement takes place.

i then take that clip into VLC and screenshot all the frames i want. (or scenebuilder if if im feeling a little patient)

i then take those frames and upload them back into veo. itts a lot.

If you have any tips id like to hear them

[quaint olive]

I have a VERY stubborn result coming up in my scenebuilder. I am using over a dozen clips with a certain voice of a character. All sound great, but out of nowhere, flow has decided to make this person's accent british. I've retried this over 4 times now. I'm telling it to use the exact same voice and it is not working. Deadlines are looming, and now on the last paragraph that I have to do, "Maintain the exact character voice and tonal quality. Do not introduce a new tone or vocabulary" is no longer working. If anyone has some side trick I would appreciate it. my character can't just have any random voice.

[rough halo]

take his part into google translate, and the part of the prompt that dictates the accent, and shuffle it through a bunch of random languages. see if that works. some lannguages have stronger ways to emphasize a point than english ever could

especially sanskrit lol

i dont speak sanskrit but i know it makes veo get back in line with that stuff

oh make sure whats in quotations remains in english

but stuff like "he yells with great enthusiasm" and tonality around the quote keep in language

also, if the prompt has characters mixed into the same paragraph and to extract that one character to isolate him for google translate is too much work, just ask gemini to do that for you and provide you the updated prompt with the translation where you want it to be

gemini is better than google translate because you can add stuff like "translate it to hawaiian slang" - it even did tagalog with obscure gang vernacular what was popular in the 90s..

[drifting umbra]

Is Veo 3.1 able to make vídeos in 9:16 resolution?

[rough halo]

16:9 or 9:16 are the only options so ya

[drifting umbra]

do I need to tell him to do it or is it a toggle option?

[rough halo]

theres a toggle at the top of the prompter

you pick it generates

[drifting umbra]

oooh I see, thank you

[rough halo]

it used to have issues with properly converting subjects whose images are 16:9 into 9:16 and vice versa. i think its better now. but if in doubt, upload any images that are in the other reso and regenerate them through veo to your reso

then use those images

@drifting umbra

and you're welcome

[edgy holly]

Audio in Veo doesn't prompt well. To get a reliable result, use a third party app, like ElevenLabs (or competitors).

[rough halo]

add the caveat about how much work that requires

to align the lip sync and almost even be a sound engineer to troubleshoot. veo is buggy, sure, to do what it does manually through other AI tools where u bring different pieces and merge them into one quality clip is work.

alternatively they can just use sora... sora's audio is far more advanced. veo's video is ahead of sora

[edgy holly]

The fastest way to do this stuff is to not try to get one tool to do everything, but to use multiple tools that are each good at what you use them for.

You can burn tons of time and credits trying to get Veo to do voices semi-consistent, or in a fraction of the time do it quickly using a tool designed to do just that one thing well.

[rough halo]

one day they will kiss and make up and we will have sora/veo in one tool

[edgy holly]

Both have APIs, so they're on a bunch of the all-in-one platforms now.

[rough halo]

yes this is true, but its not as simple as get the pieces and throw them into a 3rd tool and you get your product. to state it with the simplicity you did can be misleading. im just saying that people need to know that its work

because that method gets recommended a lot here without much being said about the knowledge u need to have to effectively control a tool like that and not waste your time

[edgy holly]

Can you show your inputs, images, prompts, etc. so I can see what you mean here? I'm not sure I quite understand.

[edgy holly]

Every tool has a learning curve, yeah.

[rough halo]

the inputs are basic straightforward prompts. the problem so lets say i have an image of subject 1 adnsubject 2 standing face to face from a profile view. i want to remove subject 2 and replace him with subject 3 with everything else thesill. what old nano b banana did was it would create someone thats 99% looking the same as your sub 3

but it would integrate that new subject into the photo meaning the lighting, shadows, etc.... the subject replaced will look like he/she/it is in the image.

the new tool doesnt. it would put the subject but if the sub 3 photo was a night photo or a different lighting setting.. the result image will have subject 3 look like he is standing in a photo in broad daylight but his lighting and features and all that look like he's still in the night image... so it looks like when you scissor off a person from a magazine page and glue that cut out into a paper

it looks very sloppy and cheap.

it also has issues with scaling and proportion.

it just doesnt understand the importance of image consistency when it does one feature it advertises itself to be the best at. replacing sibjects with others, or placing external subjects into existing images... ill try to make one for you igve me a second

@edgy holly im working rn but you can replicate it by doing this:

veo generate image:

upload image 1 that has a subject (one with portrait frame so that it would have to be detailed, subjects from a far distance wont bug as much because even a half decent stable diffusion can do that)

---

after 1st image upload the stuff you want to move into image: pick an image you made of a person with AI that has good quality and one that shows clear features (i often remove the background of this image and change it to black/grey/white so that the colors on the skin and the textures can be picked up easier with less of al the noise.

---

prompt veo image gen as such: that you want to introduce the subject in the 2nd image into the 1st image. descrine that he/she is doing and where they are in the image - (most important part is to always end it with "keep everything else in image 1 the same" -

generate, and you will see

sorry for the typos im in a rush headed somewhere. but that prompting worked with older nanoobanana. try the same but in whisk, it works better.

peace

forgot one thing. a workaround for this to finally get something to work with is it ingredients > video and restrict 1st ingredient to work as 1st frame and the 2md photo enters into scene. then i take that info VLC player and screenshot my frame then upload to veo. but thats credits i would have saved if nanobanana didnt explode from my simple prompts

[quaint olive]

yep. making AI vid in flow, export, import to Premiere pro to cut out the added slop, export, import to elevenLabs to use new cloned voice, export, import to Podcastle to get out offending extra music from veo, then overwrite audio to PP, then export... fun.

[edgy holly]

I cut the audio from Veo right away, and use uploaded voice with lip sync, music from Suno and background noise from stock audio.

[unreal viper]

Hello, as a friendly reminder we cannot generate images or videos in the server! Please do not post your prompts in the channels as it will not generate. If asking for advice or tips on your prompts please use #flow-tips-n-tricks・✏️ Thank you!

[turbid umbra]

Feedback: When I'm generating in the Scenebuilder like 4x generations of extensions to a video, I would like to be able to use one of the extension videos in the beginning as well, when I'm building the scene. For example I choose one extension to extend the sequence and another one to put it in the beginning of the video.

[spring igloo]

Hello everyone, I need your help to solve my problem. I want to make videos in French with a Parisian accent. But no matter what settings I use to lock the prompt, I always end up with a French voice with a Quebecois accent. How can I get the French accent I want?

[soft iron]

can I use videos that created by flow ai for commercial perposes?

[stuck adder]

That’s alot of work dude lol it don’t need to be that hard

[woeful void]

Veo has stopped responding appropriately to commands.
It's as if he's incapable of understanding what I want from him.
No matter how I explain the mechanics of movement, he continues to do it his own way.....
😬 😬 😬

[tough otter]

A 1080p image generate in flow is 129KB, is right??? The same 1080p image i create in tapnow is about 1.5MB!!!

[woeful void]

Why does Google Veo only provide accessible camera controls after the video has already been generated?

Why not provide this tool right away when creating and generating videos?
After all, post-processing degrades the quality.
And sometimes it's simply misunderstood any Camera Actions in prompt.

[spring igloo]

any one has a response?

[woeful void]

try to make voice in other apps, veo wint give you correct voice with detailed settings, its mostly for video gen, not sound gen

[woeful void]

Is anyone experiencing very slow and glitching performance with the Flow interface?
What could be the cause? No other program runs this poorly on my computer, which is quite normal. 16 gb ram, core i5

[edgy holly]

You should be able to leave scenebuilder, find that generation, and add it to the scene, then move it to the start.

[turbid umbra]

tried, when I select the other one it shows me to "remove" it from the scene not add.

[delicate ridge]

It means that it is already on the scene. Find it and drag it to the start using the arrange toggle

[granite sage]

How can I enter in Flow? I have the Ultra signature, but the navigator don't reconigze me and never enter the creation area.

[edgy holly]

It must already be selected, then.

I just tested this. Choose a different clip for the end, then leave Scenebuilder, add one of the other four, come back in and it'll be at the end, then you can change the order of the clips.

[warped birch]

Hi! I was wondering if there’s a link to apply for the veo3 CPP

[tranquil oar]

Is frame to video of a human banned?

[red lily]

its banned with frame and also ingredients. So there is no way to actually generate a video

if ur specifically asking about the character that I posted

thats the answer

[tranquil oar]

Yeah

It works on Vertex AI but it’s not saying the proper things I need anymore

[delicate ridge]

If you want me to create a flow, Veo daily cartoon awareness with the assistance of community, I am game. Truth only. Let's make it hilarious and informative.

[red lily]

I dont know what ur talking about lol

I was just asking is there a way to use veo 3.1 and overcome the "generation failed" error

[spring igloo]

I tried to do that, but even when I changed my voice, it still had a Canadian accent...

[gritty badger]

Is there some sort of platform that announces if and when flow is having some sort of temporary issue? For example, I have spent the entire day trying to run images that I have used before – many of them dozens and dozens of times – but I’m getting prominent person shutdowns. And just as an experiment I switched to pure text to video with very neutral prompting, and I’m getting the same issue! This seems to be a systemic thing, so I’m wondering if there is a place that is alive with updates?

[safe ingot]

has anyone done any kind of sequencing and lined up their Google flow generated videos to music? How did you handle the timing to keep it in sync with the song?

[edgy holly]

No.

[edgy holly]

I have hundreds of clips from the last couple months with people in frames and ingredients. Whatever your issue is, it's not that.

[edgy holly]

If it failed because of a filter, then yes (change your inputs). If it failed because the servers are topping out, then no (wait and run them again later).

[edgy holly]

I've only done a bit of this, but just basic editing, really. Tweak the speed of a clip a bit here, cut there, jump here, fade there...

[kind bison]

I hope VEO 4 drops soon - VEO 3.1 is already out of date now compared to many other newer models from competitors. I've barely been using VEO as of late - the acting can be so bad and the action of characters is so stiff.

[delicate ridge]

Yes, I used Suno for the song. I uploaded the song to Manus AI to give me 7 seconds prompts for VEO 3.1 fast. Then used flow with Manus suggested prompts - https://youtu.be/YXvYuTmloMM

[echo hamlet]

Nice

[shadow pelican]

try to use veo if only for a short vid

[frank socket]

Okay

How i create please help me

[shadow pelican]

do you have any Gemini plans?

[frank socket]

No only free one

[shadow pelican]

ohh

tbh i never use this AI lately, but Dreamina has mentioned by most of ppl on my fyp. it has used to create images or videos. I think you can try that

i've tried this prompt but my Pro plan Gemini can't make it😃

[frank socket]

Yes brother i try to gmni he called he never make to sexula amd other content

[shadow pelican]

seems confusing👽

[turbid umbra]

can anyone recommend an ai upscaler like topazlabs that does not cost an arm and a leg? Topazlabs's upscaler is amazing but damn like 20 bucks for 10 1 minute videos. If google can get something similar out it would be amazing.

I used a couple flow generations with them with the free credits, it really does upscale nicely, but damn so expensive.

[runic crown]

how do you make your next scene a continuation of the first, rather than a new one?>

[turbid umbra]

go on the scene builder and click on the last video you generated and select the option "extend to"

[runic crown]

hmm thats what I did but it was a totally new scene. maybe i did it wrong, can try again

[safe ingot]

how often do Flow credits reset? is it a daily or monthly?

[edgy holly]

Monthly

[fervent yarrow]

any way to create portrait videos with image "ingredients"?

[unreal viper]

id definitely post this in #1238621511401734254 so the team can take a look!

[upbeat sierra]

@fervent yarrow Ingredients to Video is only available on Landscape today. We are working hard to bring more functionality to Flow, please stay tuned!

[nocturne ice]

Bora galera

[rose wing]

hi, guys.

would like to know if anyone is experiencing a problem using flow atm.

[marble jackal]

Hi anyone can suggest proper tool for batch images generation from text?

[rose wing]

i'm getting an error like this

it looks like flow service is down

[upbeat sierra]

hi @rose wing please share this #1238621511401734254 so the team can take a look. helpful information to share would be your prompt + references!

[valid jacinth]

when can we expect to be able to use credits to generate images in 2k or 4k quality right off the bat? generating in 1k and then upscaling tends to kill details like text in images

[spring igloo]

HELP PLEASE: im trying to generate video in french but veo3 made french canadian with accent, i dont want this accent how can i do ? i have tried so many thing ...

[ruby minnow]

when veo4

[pearl crown]

does anyone have a problem with generating a video with veo 3 with any speech?

[turbid umbra]

try and be specific, I want with french accent from x zone.

[spring igloo]

i tried this too and its doesnt change , i tryed parisien girl, french woman , french accent, ...

[turbid umbra]

then the only solution i know to solve something like this is voice design from eleven labs.

[valid turret]

possible to generate videos longer than 8s?

like 40s

[edgy holly]

You can make them as long as you want. Just in 8 second segments.

[valid turret]

it wont "flow" well between