AI Image Generator

An AI image generator creates a picture from a written description. You type a prompt such as "a lighthouse on a cliff at sunset", pick a style, and a model draws a new image that matches it. This one makes a 512 x 512 PNG and offers eight style presets: Photo, Illustration, Watercolor, Anime, 3D render, Pixel art, Oil painting and Line art, plus None to leave the prompt as you wrote it.

The model is SDXS-512 DreamShaper, an open distilled version of Stable Diffusion released under the OpenRAIL++ licence. It is built to draw in a single step instead of the twenty or more most image models need, which is what makes it fast enough to run on an ordinary server processor. It runs on the Toolz server, not on a cloud AI company, so nothing is downloaded to your device and it works on phones too.

Your prompt is sent to the Toolz server to make the picture and is not stored or logged. The image comes back to your browser and is kept only there, until you download it or close the page. The last eight images you make stay on the page so you can compare them.

Be realistic about quality. A small one-step model is well below large cloud generators such as DALL-E or Midjourney. It is good at moods, landscapes, animals, objects and illustration styles; it is weak at hands, faces in crowds, readable text inside the image and exact counts ("three cups" may come out as two or four). 512 x 512 suits thumbnails, blog headers, mood boards and concept sketches rather than print.

It is free, with no signup and no credits. There is a fair-use limit of six images per ten minutes per visitor. A picture usually takes a few seconds; when several people are using the AI tools at once, yours waits its turn, because the image model and the writing model share one server and take turns in its memory.