Why Your Midjourney Prompts Keep Producing Generic Results

There's a recognizable "AI look" that a lot of generated images fall into: evenly lit, center-composed, vaguely photogenic, and completely forgettable. It's not a limitation of the model — it's a symptom of underspecified prompts.

Lighting is the first lever. "Soft studio lighting" is doing almost no work compared to "low-key single-source light from camera left, hard shadow falloff." Named lighting setups push the model toward less average output.

Lens language is the second. Real photographers think in focal length and aperture, and so should your prompt: "85mm f/1.4, shallow depth of field" reads completely differently to the model than no lens information at all.

Mood is the third, and the easiest to skip. A single evocative adjective — cinematic, clinical, nostalgic, brutalist — reorders everything else in the prompt around it.

Put those three together and the generic-stock-photo effect mostly disappears, without changing the subject of the prompt at all.