Quick start: your first generation

This page takes you from a fresh install to your first saved image. Everything here runs locally on your own NVIDIA GPU, so nothing you type or generate leaves your machine.

1. Install and launch

Download the installer and run it. On first launch Windows may show a SmartScreen notice; choose More info, then Run anyway (the app is code-signed). The first start downloads the base model it needs, which takes a few minutes on a normal connection.

2. Write a prompt

In the generation panel, type a short, concrete description of what you want, for example: a portrait of a woman with red curly hair, soft studio light, 85mm photo. You can add a negative prompt to push away what you do not want (blurry, extra fingers, watermark).

3. Choose a size and generate

Pick an aspect ratio (portrait, square or landscape) and press Generate. The first image on a given size is a little slower while the model warms up; the next ones are quick.

4. Save and iterate

Save the result, or send it to the built-in gallery. To get a variation, keep the same prompt and generate again (a new random seed each time), or lock the seed to refine the same image.

That is the whole loop. From here, learn to write stronger prompts, train a character with a LoRA, or turn a still into a video.