The easiest way to generate killer AI images isn’t prompt engineering.
It’s reference engineering.
Master it and you'll stop asking people for prompts.
For this image, I gave AI two pictures:
• Image 1 = pose, composition, background, camera angle, outfit Inspo
• Image 2 = face, hair, accessories
Then I wrote:
"Replace girl in Image 1 with girl in Image 2. Maintain the exact pose, camera angle, expression, and background from Image 1. Use the facial features, hair, accessories, and overall appearance from Image 2. Dress her in a pink version of Image 1's outfit. Change the phone pouch to pink. Preserve the original art style and lighting."
That's it.
The trick is separating what each reference image is responsible for not writing better prompts.
Reference A = structure.
Reference B = identity.
Prompt = instructions for merging them.
Once you understand this, image generation becomes 10x easier.
RT if you gained value.
W🦋