Gemini 2.0 Flash (Image Generation) Experimental stopped generating images

it just doesnt want to generate an image as it says it’s not capable of doing so… :upside_down_face: Am I doing something wrong?

Welcome to the forum. The model will generate an image from scratch if you describe what the image you want is expected to show in your prompt. Alternatively, you provide an image and describe what you want changed in the image you provided. So, I think you just need to change your prompt to be successful.

Hello! This is exactly what I did and it generated just text description, then I asked to generate image, not text, and it generated text again (as per attached screenshot).

Got it. Once a model starts “misbehaving”, not following instructions, there’s not much progress you can make by more prompts. You delete that chat and start fresh. The chat always maintains the full history (unless you specifically edit some of it away using AI Studio), so whatever caused it to “misbehave” in the first place remains in the context.

Anyway, I just wanted to reassure you that the model basically works as advertised, it both generates images from a description and makes edits to images based on a request to make modifications. Not always exactly the modifications you expect, but I guess that can be improved by using better prompts (still early in the testing cycle to decide for sure).

Hope that helps.

1 Like

that solution didn’t work!

Welcome to the forum.
What prompt did you try and what didn’t work as you expected?