Image not the same as description

Hi laurentiax!

Thank you for sharing your feedback and the screenshots. We understand how frustrating it can be when the generated image doesn’t match your description.

From what you've described, it seems the tone and visual focus of the image were interpreted differently by the system.  Firefly uses AI to interpret text prompts, and while it aims to follow the description closely, certain creative elements like emotion, styling, or character roles may sometimes come through differently than intended.

We’d be happy to help you refine your prompt to better capture the mood and characters you envisioned. For example, you could emphasize that the woman should appear scared and that the spectral men, covered in black lace, should be the focal point and appear frightened as well.

You can use Generative Fill and get faces together.

Let us know how it goes.