Who writes the prompts
Every prompt in this library was written for Photo AI Prompts. The wording, the example images and the notes on each prompt page are produced by this site; nothing is copied from another prompt library.
How a prompt is written
Each prompt follows the same structure: a subject sentence, a lighting sentence, a lens and framing sentence, then a realism guard that fixes skin texture and rules out text, logos and watermarks. Prompts that start from a person's photo open by locking the face, so the tool changes light and styling rather than identity.
How prompts are checked for duplicates
Two prompts are treated as duplicates when they share the same subject, location and lighting and only swap a wardrobe colour or an adjective. When that happens the pair is rewritten or one is retired. Each prompt page links to a comparison prompt and explains what makes it different, so you can judge the difference yourself.
What is tested, and what is not
Prompts are tagged only with the tools they were written for: Gemini, ChatGPT, Dreamina and CapCut. A tag is not a guarantee that a prompt produces an identical result — models change, and the same prompt behaves differently in different versions. Midjourney and Flux pages republish prompts written for the tools above; they are explicitly marked as not tested in Midjourney or Flux.
Example images
Example images show what a prompt produced in a session. They are not stock photography standing in for a result, and they are not presented as the output of a tool that was never run. Where an image is unavailable or its provenance cannot be confirmed, the prompt is marked as blocked instead of being illustrated with an unrelated picture.
How user feedback is handled
Every prompt page links to a feedback address with the prompt name pre-filled. Reports about a prompt producing a bad result, a broken image, or a description that does not match the text are checked against the page and corrected. Corrections that change what a prompt does are dated in the site changelog.
Corrections and updates
Prompts are reviewed when a model they are tagged for changes behaviour, when a reader reports a mismatch, or when an image no longer loads. The library is not rewritten on a fixed schedule: it changes when there is a reason to change it, and the count on the homepage reflects the prompts that exist at that moment.
Questions about this policy: contact us.