Basic Image Generation

Getting Started with Basic Image Generation

Learn how to create your first image using ComfyUI's node-based interface. This tutorial will walk you through the essential steps of setting up a basic generation workflow.

First, we need to load our base model and the required VAE.

• Click the "Add Node" button or right-click in the workspace
• Search for "Load Checkpoint" and add it to your workspace
• Select your desired model from the dropdown (e.g., "sd_xl_base_1.0.safetensors")
• Add a "VAE Decode" node - this will be connected later

Create positive and negative prompt nodes to guide the image generation.

• Add a "CLIPTextEncode" node for your positive prompt
• Add another "CLIPTextEncode" node for your negative prompt
• Connect both to the Load Checkpoint node's CLIP output
• Enter your desired prompts in each node

Set up the parameters that control how your image will be generated.

• Add a "KSampler" node to control the generation process
• Set your desired steps (20-30 is a good starting point)
• Choose your preferred sampler (Euler a is commonly used)
• Set an appropriate CFG Scale (7-8 works well for most cases)
• Connect your checkpoint, positive prompt, and negative prompt nodes

Configure the final steps to generate and save your image.

• Connect the KSampler output to your VAE Decode node
• Add a "Save Image" node and connect it to the VAE Decode output
• Configure your desired save location and file format
• Your workflow is now ready to generate images!