In earlier versions of generative image models, users relied on massive lists of repetitive keywords like 'hyperrealistic, 8k, octane render, trending on artstation'. With Midjourney v6, the language understanding engine has been overhauled to reward clear, photographic descriptions.
Step-by-Step Instructions
Construct the Core Subject and Action
Start with a concrete physical subject and environment rather than abstract concepts. Describe materials and natural lighting.
Set Your Aspect Ratio (--ar)
Always define output proportions at the end of the prompt: '--ar 16:9' for horizontal headers, '--ar 4:5' for mobile, or '--ar 1:1' for square.
Apply Style Reference (--sref)
Paste the image URL of an existing aesthetic you admire after the '--sref' flag to clone its color grading and texture without copying the subject.
Remove words like 'stunning' or 'masterpiece' from your prompts. Replace them with specific technical descriptions of light direction, shutter speed, and focal length.
