Getting Started

This guide covers the basics of generating images with SD.Next, assuming the application is already installed.

Open the web interface by running webui.bat (Windows) or ./webui.sh (Linux/Mac). The main page loads with the Text tab active by default.

The following settings are required before generating an image:

Base Model

The base model (also called a checkpoint) determines the visual style and capabilities of image generation. Models are available on CivitAI and Hugging Face, and typically range from 2 GB to 30 GB. Place downloaded models in the models/stable-diffusion/ folder inside your SD.Next directory.

Positive Prompt

Describe what you want in the image. Be specific about subjects, style, lighting, and composition.

Negative Prompt

Describe what you do not want in the image. Common entries include anatomy issues (e.g., extra limbs, distorted faces) and style elements to avoid. Negative embeddings such as bad-hands-5 can also be placed here.

Width and Height

Set the output image dimensions. For SD 1.5 models, start with 512×512 or 512×768. Generating at sizes significantly larger than a model's training resolution often degrades quality.

Sampler

The sampler controls the algorithm and number of steps used during generation. More steps do not always improve quality and always increase generation time. Euler A typically produces good results in around 24 steps. See the Samplers page for details.

CFG Scale

Controls how closely the output follows your prompt. Values between 6 and 9 are typical. Too low gives the model more creative freedom; too high can produce over-saturated or distorted results.

Seed

A unique number assigned to each generated image. Set to -1 for a random seed on every run. To reproduce settings from an image found online, use the Process tab to import its metadata — the seed will be included.