<aside> ⚡
EVOLVING AI · PROMPT LIBRARY · IMAGE EDITING
</aside>
Ten prompts that actually hold up. Field-tested image-editing prompts for AI image models — refined, reliable, and formatted for copy-paste use. Every prompt below is written out in full, with the example output underneath it and a breakdown of what each constraint is doing.
INDEX — click any prompt to jump straight to it
HOW TO USE THIS LIBRARY
01 · Copy it verbatim
Select the whole prompt and copy it. Every constraint is doing work. Trimming lines is the single most common reason these stop working.
02 · Swap the variables
Anything in square brackets or CAPITALS is yours to change. Only two prompts here have them; the rest run untouched. Each one says so at the top.
03 · Attach and run
Attach your image, paste, send. These are written to work across the major AI image models rather than being tuned to one.
REPAIR · NOTHING TO EDIT
Rescues group photos where someone blinked. Opens closed eyes without touching anything else.
The prompt
In the uploaded group photo, detect every person whose eyes are fully or partially closed and open only those eyes. Preserve each person's original eye shape, eyelid crease, iris color, pupil size, gaze direction, and the lighting on their face. Keep all facial structure, expression, skin texture, pores, hairstyle, head tilt, and camera angle exactly as they are. Do not remove eyeglasses, reflections, or shadows. Do not add, remove, or reposition any person. Do not enlarge, brighten, or sharpen the eyes beyond what matches the rest of the face. The final image must look like a single real photograph taken at the same moment, with everyone's eyes naturally open and consistent with the original scene.
<aside> 🖼️
Drop in → 01-eyes-open-fix.png
</aside>