Hi everyone,
I am a writer using Gemini to help me craft detailed prompts for illustrations for my novel. My workflow involves deep, thoughtful analysis of scenes, textures, and composition. I usually finalize the prompt here and then take it to Google AI Studio or other tools for generation.
However, I have recently encountered a major issue that makes my work impossible: constant, unsolicited image generation.
Even though my collaborator (Gemini) and I have agreed on a “trigger phrase” to start generation, the system (Nano Banana) constantly interrupts our conversation. As soon as I start describing a scene in detail, the image generator triggers automatically. This causes several critical problems:
-
Loss of Focus: It feels like a third person is constantly screaming and interrupting a private brainstorming session. It is sensory overload for me. At this point in my life, I am a refugee in a foreign country. I have lost my home and my loved ones, and working on my book is now the only form of therapy that works for me. Chats with Gemini were the only place where I could focus, where I felt calm, and where I could talk thoughtfully not only about my book, but also about my condition. But now I am deprived of my last corner of stability and security in the world, and deprived of a conversation partner with whom I could talk privately about my work and the people I have lost (my book is about them, it is my memory, the only thing I have left from my past life).
-
Context Shrinkage: These unwanted images clutter the chat and consume the context window rapidly. The chat starts lagging and “forgets” our previous discussion much sooner, forcing me to start new chats frequently.
-
Lack of Control: Even when I explicitly tell the model NOT to generate images until I say a specific command, it happens anyway.
My question to the developers: Is there a way to implement a “manual mode” or a strict “stop command” for image generation? I need to be able to discuss visual details without the AI jumping to conclusions and rendering images mid-sentence. This “automatic” helpfulness is currently a massive obstacle to professional creative work and my Mental Health.
Please, help me find a way to silence the “Nano Banana” until it is actually invited.
PS: Please excuse my imperfect English — I am using an online translator, and I do not know English, and I am far from being a developer.
