Player not loading? Watch on YouTube
Kevin Stratvert walks through ComfyUI desktop setup and a first image generation workflow. He describes ComfyUI as free and open source, then selects local execution during installation. He also discusses a paid cloud option for people who prefer not to manage hardware.
The text-to-image starter template reports a missing model. Stratvert uses Show Missing Models and Download All to install it, then enters a chocolate chip cookie prompt and a negative prompt. He explains image dimensions, batch size, and how changing the seed produces variations with the same prompt. He recommends that beginners keep the sampling defaults. Generated images save to the computer and appear in the Assets view.
A second example changes the cookie prompt and increases the width from 512 to 768 pixels while retaining the 512-pixel height. The tutorial then explains how connected nodes pass data through the workflow. Stratvert adds a 2x upscaling node between decoding and saving, runs the workflow again, and saves it as Upscale Image Generator for later reuse.
The closing section introduces templates for tasks such as image-to-video, image editing, image-to-3D, and text-to-song. These are presented as starting points that viewers can customize after downloading the required models.
