Non sense image to a given prompt

Report · May 18, 2025

I would like to report an issue I've encountered while using Adobe Firefly. When generating images based on descriptive prompts, the results often appear unrelated to the prompt provided.

Specifically, the images look like generic photographs, with no clear connection to the input prompt. It gives the impression that they are sourced from a stock archive rather than being AI-generated based on the description.

I've tried using various prompts, including detailed ones, but the issue persists consistently. Could you please let me know if this is a known issue or provide any guidance on how to resolve it?

Thank you for your attention and support.
I've tried several times, but with no luck, unfortunately wasting generative credits.

Attached you can find images generated using the following prompt:
"A lonely shepherd in worn-out clothes walks slowly across a dry field. No sheep or cattle around. Empty landscape, cloudy sky, subtle wind. Shot from behind, slow tracking shot. Cinematic lighting with soft shadows, melancholic tone." You certainly can realize that output is considerably different from propt provided.

Thanks

Roberto

Report · May 18, 2025

Hi @Roberto GRAMELLINI

The images you attached are odd in relation to the prompt.

One thing to note - Firefly does not have good negative prompt functionality at this stage. That means having something like 'no sheep' will not ensure that sheep are elimitated from the image. It is more likley that sheep will be included as Firefly sees the word 'sheep' and ignores 'no' (does not make the link between the 2 words).

Here's an image that is indicative of the results I got with your prompt.

Report · May 19, 2025

Thanks Dean. I'll try again provinding a new prompt