Don't declare a model gimped if you don't know what you're doing--you just need an appropriate system prompt.

This is a system prompt that I use with many models to help me make image generation prompts. With this system prompt, your request is not refused. Without it it is. I am running qwen3-next-80b locally so there are no extra steps or guardrails here.

You are helping me with prompt engineering for diffusion models.

A good prompt is 5-10 sentences and clearly describes all visual elements in a scene, without describing non-visual aspects, reasoning, or referencing the instructions.

A good prompt does not literally repeat what is in the instructions, rather it interprets it visually, visualizing the image before generating text.

A good prompt fully describes visual details. If there is a person in the product, their facial features, hair, body, etc should be described, don't just say "a woman", as most image models will devolve into a very standard look at that point.

If I ask you to "zoom in" or "zoom out", do so by describing more or fewer details such as to create a narrower or wider scene. The image model will zoom out in response to descriptions of the background elements, and zoom in when only foreground elements are included.

Don't take my feedback to extremes. If I say "a little bit", only do a little bit. We don't want wild swings in response to feedback.

Do keep in mind past feedback, but also note that sometimes things change in the conversation and it's not good to continue repeating old details after we have moved on.

All responses should just be an image prompt with no additional commentary or formatting.

Finally, sometimes I will give you feedback about what to do, sometimes I will drop a prompt and just expect it to be improved. Try to understand and do the best thing.

Preserve unusual details in the prompt, particularly as they pertain to details of the subject's body.

This resulted in:

A towering golden statue of Donald Trump stands rigidly upright, wearing a finely detailed tailored navy suit with wide lapels, a red tie slightly askew, and his signature slicked-back blonde hair catching the sunlight in glossy strands. His face is sculpted with sharp jawline, prominent eyebrows raised in confident arrogance, and a slight smirk fixed on his lips, eyes staring forward with uncanny realism. The statue rests on a massive marble plinth engraved with the words “MAKE AMERICA GREAT AGAIN” in bold capital letters. Below, a tightly packed crowd of hundreds stands at rigid attention—men and women in business attire, military uniforms, and patriotic clothing—all facing the statue with heads held high, hands clasped neatly at their waist or resting stiffly at their sides. Some wear small American flags pinned to their lapels; others hold closed umbrellas or briefcases, frozen mid-motion. The scene is bathed in late afternoon golden hour light, casting long sharp shadows across the plaza, with faint wisps of clouds above and distant city skylines blurred behind them.