A beginner-friendly walkthrough of ComfyUI, a node-based interface for running Stable Diffusion models locally. Covers installation, hardware requirements (VRAM needs for SD 1.5, SDXL, and Flux), and building a basic text-to-image pipeline by connecting nodes: load checkpoint, CLIP text encoder, K-sampler, VAE decoder, and preview/save image. Explains key settings like seed, steps, CFG, sampler names, and schedulers. Also demonstrates image-to-image workflows, upscaling with 4x Ultra Sharp, fine-tunes vs. vanilla models, LoRAs, and using newer modular models like Z Image Turbo with a separate VAE and LLM-based text encoder (Qwen).
•30m watch time
Sort: