I’m beyond frustrated with how restrictive and inconsistent the image generation system has become.
I’m working on a personal creative project with my own characters, my own settings, and zero copyright issues.
And yet, the system constantly blocks perfectly safe prompts.
Example of a completely harmless request that gets rejected every time:
“Cinematic 16:9 scene of a classic wooden ship sailing on the ocean at night. The moon softly illuminates the deck, detailed textures, realistic lighting and shadows.”
That’s it. A boat. At night. Nothing violent, nothing unsafe, nothing copyrighted.
And yet, it’s blocked.
But here’s where it gets even more absurd:
If you even dare to mention a certain high-contrast noir style inspired by a famous graphic novelist (let’s just say it: Frank Miller), the system completely loses its mind.
It either blocks instantly or gets “stuck,” like it has no idea how to handle the request — even if you don’t mention the artist’s name and only describe the style technically.
This isn’t “safety.”
This is broken design.
Right now, it feels like the centinels (your automated filters) are:
-
Flagging night scenes and ships as if they’re dangerous.
-
Treating high-contrast noir aesthetics as if they’re forbidden.
-
Punishing users for doing perfectly legal, safe, creative work.
Meanwhile, we get zero feedback explaining why prompts are rejected. No reason. No transparency. Just “blocked.”
I’m at the point where I spend more time fighting the filters than actually creating. And for a tool that’s supposed to enable creativity, that’s unacceptable.
We need:
-
Clear explanations when content is blocked.
-
A way to whitelist original IP so creators can work on their own projects.
-
Less overreach from automated moderation — especially when dealing with fictional content.
Right now, OpenAI’s image generation is crippled by nonsensical restrictions.
I love the potential of this tool, but unless this changes, it’s becoming useless for serious creative work.




