What this is
Promptsaxis is a directory of AI image and video prompts that have been run on the model they name, by a person, before being published.
Why it exists
Most prompt collections are lists of text nobody has tested. They read well and produce something else. The failure is usually the same: the prompt names a look but not the optics, the light or the render, so the model fills those in differently every time and you re-roll until something lands.
Every prompt here is written with those decisions made explicit, and the image on its card is the output of the prompt underneath it — not a similar picture, and not a render from a different model.
How a prompt gets published
- It is written against a specific model version, with the lens, lighting and render directives stated rather than implied.
- The parts meant to change are pulled out as typed variables with sensible defaults, so the prompt is a template rather than a one-off.
- It is generated. The output that ships on the card is that run. If it does not reproduce, the prompt is rewritten, not published with a nicer picture.
What is in the library today
10 published prompts — 9 image, 1 video. It grows as new model checkpoints land, and existing prompts are re-tested against them. A prompt that stops reproducing on a newer version gets re-authored rather than left to rot.
What it costs
Nothing. There is no paywall, no account requirement and no gated content. Prompts carry a free or pro label to mark how involved they are, not to lock them — every prompt in the library opens, customizes and copies for everyone.
Some links out to generation tools carry a partner parameter. Those cost you nothing extra and never change which tools are listed or in what order. The full statement is in the terms.