Does the tool generate the image itself?
No. The tool is a structured composer: you pick your ingredients from the dropdowns, and it assembles the output as optimal English tags you then paste into the image service of your choice. The free-form subject field is left as-is, so read our full prompt-engineering guide if you want to fine-tune the wording yourself.
Why does the negative prompt contain so many anti-minor tokens?
Because some image models have training biases that can drift toward problematic outputs if you don't explicitly push against them. Those 9 tokens (underage, minor, child, teen, young, youthful, childlike, loli, shota) cover the main variations found in English and in training datasets. They're non-negotiable, ethically, legally and practically.
Why is the minimum age 25 and not 18?
Because image models tend to visually age down their outputs relative to the requested age. Explicitly asking for "25 years old" typically produces a result that reads as 20–23 to a human eye. A prompt saying "18 years old" therefore produces a visually ambiguous result that can be read as underage. The 25 floor is a technical safety margin as much as an ethical one.
Does this tool work with every service?
Most of them. The generated tags are standard and compatible with SDXL, FLUX, Pony, Illustrious, NoobAI, Wan 2.1 and derivative pipelines. Services that internally rewrite your prompt (Candy.ai, partially Promptchan) can produce slightly different results, but the structure still works as a skeleton. For services that expose a separate negative prompt field (Promptchan, Seduced.AI, most self-hosted ComfyUI interfaces), copy both fields.
Do you collect the prompts I type?
No. The whole tool runs in JavaScript inside your browser. No request is sent to our servers, no prompt is stored, no cookie is set. You can cut your internet connection after the page loads and the tool keeps working.