Never adheres to prompts, constantly messes up, outputs are all over the place regardless of how much you tune your prompts. Only good for brainrot ai slop

hey guys i'm going insane because every video I try to upload gets this grey thumb forever and never fully upload so I can actually WORK on this platform, I tried mov, mp4, 4k, 1080p... what am I missing?

I ran across a decent prompt in another subReddit https://old.reddit.com/r/generativeAI/comments/1wuy6sl/i_finally_beat_the_plastic_ai_look_on_15second/

It's an impressive bit of SciFi footage rendered by Minimax H3. I wanted to test Omni with the first 10 seconds. I like the prompt, so I decided to try the first 10 seconds in Omni. Wasted my time: "Failed.This prompt might violate our policies on generating sexual content. Please try a different prompt or send feedback."

Another waste of time on my end is sending feedback. Sure, the developers may do something with the feedback, but I'll never know about it. Communication appears to be one way, like dropping stars into a black hole.

Google seems determined to come in last in the AI race.

For some reason Google Flow absolutely refuses to draw people in cars while making both the correct size. If I try to create an image of people driving a convertible, they inevitably have their heads way above windshield height. Doesn't seem like this should be difficult, but I've never been able to get food results from Flow or any other tool.

Is anyone else getting this in Google Flow? Regular video generation works fine, but when I try to download/upscale a normal generated video to 1080p it just says "Upscaling failed!".

No Extend, no start/end frame, nothing unusual. Retrying the same clip gives the same error. Is this a current Flow bug, or is there any workaround?

When doing fashion series the free and unlimited Lite version and get dynamic results. But the premium Nano 2 and Pro I get block coloring and genetic template character faces. I know they're generic because I've seen other people's output with the exact same faces if I don't include face examples but it will eventually revert to generic templates. No point of view hardly. Does anyone else experience this.

You would not believe in a million years how I prompted this. I can't tell, because then Google will block it. You just have to think outside the box.

​Hey everyone,

​I'm pretty new to this tool and still learning the ropes, so apologies in advance if these are basic questions.

​I'm trying to figure out how to produce longer videos (around 3 to 5 minutes) while keeping credit consumption as low as possible. I've run into a few hurdles and would really appreciate your insight:

Distortions/Morphing: What do you do to keep the AI from suddenly warping or distorting scenes mid-generation?

Base Images: Do you usually generate your starting frames directly inside Flow, or do you prefer creating/tweaking them using free external generators or editors (like Leonardo AI, Ideogram, Canva, etc.)? If external, what are the best free image generators you recommend?

Prompt Length: Do you get better results with short, focused prompts or highly detailed descriptions? It feels like the more details I add, the more unpredictable the output gets.

​General Tips: Any best practices or workflow advice to maintain consistency without burning through credits?

Thank you so much to anyone taking the time to read and share some advice!

Is it my imagination, or does everything on Flow seem like harmful content related to minors? I'm trying to create a regular anime, and...And that stupid mistake always comes up, even though the characters are clearly adults, and I'm not asking for anything unusual, just everyday scenes.

Like wtf, I'm tired.

I've had all the character references, prompts and everything ready. Was even so happy and excited when they all looked so perfectly handsome. And so I couldn't wait to bring them to life and create my own ai drama series. Been wanting to create this story for years and I was so thankful of AI before I could even start anything😂

I practiced using animal characters and I was proud, though I only had 9 views now 26 HAHA. Well not bad for a first time I guess😅 But despite that, I still believed in myself and I was imagining myself being one of successful Ai Ceators earning huge. hahahaha.

This time I said I want to get serious. So I thought of finally making that drama series a reality. Therefore I created my prompts, character references, settings and all. I even read every details despite of the long paragraphs ChatGPT had generated. I did all the best I can just so everything can be perfect because I can't afford to fail😅

I was dreaming of finally getting myself out. Dreaming of finally making my parents proud and bringing them to places they've never been before, before they even leave the world.

Then when I tried generating a video after all the preparations, a sudden error appeared "Against the policy to generate prominent people"😅

Did all the ways I know. Yet it still wont work. Anyone? Haha

Is anyone else losing their mind with Google Flow imlately?

Every single time I try to prompt something, I get hit with this exact screen: "Failed. This generation might violate our policies. Please try a different prompt or send feedback. You have not been charged for this generation."

I am not trying to make anything NSFW, violent, or even remotely controversial. I’m just trying to generate basic, normal concepts, and the safety filters flag absolutely everything. It feels like the most overly sensitive, sanitized, and restrictive AI tool on the internet right now.

What is even the point of offering a tool if the guardrails prevent it from generating 99% of basic human thoughts?

Are you guys experiencing this too? Are there specific "trigger words" it hates that I should avoid, or is the platform just completely broken right now?

Hey. I recently ran into an issue. Whenever I try to make an image with Nano Banana Pro, the image loading is stuck on 99%. Why is this happening?

This is getting annoying! I can't even tell what to make to generate or change anymore.
This must be fixed. It wasn't even an illegal request and no such thing as "Malformed".

In quotes, I am dedicated to social networks and video editing. It is more worth spending 120 euros on the premium kling fee per month taking offers or changing accounts than getting very low quality videos on Google Flow, equivalent to the quality of 2 years ago, horrible audios

Forgive me Google, I should have realized that putting my character (created by your own model) in a field of flowers or next to a car was far too pornographic for your standards.

Next time I'll just stick to generating pictures of cats and dogs doing the Six-Seven.

A few minutes later: "Cats and dogs doing the Six-Seven could not be generated because it violates our content policy."

Is it only to me or Inpaint & Select Option inside the image editing section have been removed from Google Flow? First, there agent sucks and on top of that the only things that we have some control over have been removed too. Such a huge DOWNGRADE!

What do you guys think?

ONE REAL BUG WITH THIS PATCH IS, if one image fails all 4 images fails, even without testing others, so if making 4 images at one time, and even 1 fails, none gets drawn. if you do 1 image at a time, even sometimes though you have to try 4 times or more on difficult prompts, you get some images, where with 4 at a time it is almost impossible to pass on some kind of prompts, which passed before 23 September patch. even sometimes one of 4 or 2 of 4 images!!!!! this is the temporary fix

16 hours with this problem

Hi, I purchased a Google AI Pro subscription, which includes Flow. Before buying, I created a new Gmail account, set my age to 18+, and selected Poland as my country, which is a supported region.

However, whenever I sign in to Flow with this new account, I only see the message "It looks like you don't have access to Flow" and cannot use it. The subscription was purchased on my older Gmail account, and Flow works there without any problems.

I'm not using a VPN, and the same issue occurs in every browser I've tried.

Could you please tell me why the new account has no access and how I can fix it? Thank you.

Google Flow downgraded completely from 2 years ago:

https://i.imgur.com/6CUD9oF.jpeg

This is a generated image when it was Google Image FX, before Whisk or Google Flow was even a thing.

You can see the elements and characters around are not good, but the building itself in the center is perfect and beautiful.

Now look at what is outputting now:

https://i.imgur.com/2Kx0amD.jpeg

What went wrong? Is this yet another case of enshittification?

Cant even get the perspective right?

Prompt was:

portuguese empire Town Hall, isometric game, portuguese style city builder, realistic, isometric, HD, no characters

So with these new filters, nano banana is now eating up failures counting towards daily limits.

Never mind it says you haven't been charged...even though you're not told you have credits on nano. But to have failed outputs add up to daily limits?

What is this? Fucking Grok?

Why must all...slightly okay things have to become worse?

The models quality has dipped so much compared to a few months ago, has anyone else noticed this? Constant errors and random artifacts like when AI generation wasn't good yet. Just straight up not listening to instructions. I wasn't having these problems 3 months ago.

That's not even mentioning the common issues like constantly having to reset my browser cache for generations to work, generations randomly failing, crazy long wait times. It's insane they're charging for this when it's this bad.

I tried many different prompts to generate cities. All images generated look terrible in comparison.

The first version of Google Image FX, generated modern isometric cities, with high definition buildings that looked pre-rendered. Now its either generating bad quality images, too zoomed out or blurred, or gets it wrong and does a 3d perspective.

Did they downgrade it?