ComfyUI Stable Diffusion Interface

ComfyUI is a powerful and modular graphical user interface and backend for AI image generation models, designed around a node-based workflow system that provides fine-grained control over the entire diffusion process. Created by comfyanonymous in early 2023, ComfyUI has become one of the most popular interfaces for Stable Diffusion and other generative AI models, favored by power users for its flexibility and extensibility. Key features: node-based workflow editor where users construct image generation pipelines by connecting graph nodes (checkpoints, VAE, CLIP text encoders, samplers, schedulers, latent operations, conditioning) into a directed acyclic graph (DAG), enabling unlimited customization of the generation process. Model support: Stable Diffusion 1.4/1.5/2.0/2.1, Stable Diffusion XL (SDXL), SD3, Stable Cascade, FLUX, Kolors, Hunyuan DiT, PixArt, ControlNet, LoRA, embeddings (textual inversion), hypernetworks, and custom checkpoints. Sampling: support for all major samplers (Euler, Euler a, DPM++ 2M, DPM++ SDE, DPM Fast, LMS, Heun, DDIM, UniPC) and schedulers (normal, karras, exponential, simple, ddim_uniform). ControlNet integration: depth maps, Canny edge detection, OpenPose pose estimation, lineart, scribble, segmentation masks, and tile-based upscaling. Image-to-image and inpainting with mask support. Upscaling: Latent upscaling and pixel-space upscaling with models like RealESRGAN, ESRGAN, SwinIR, and 4x-UltraSharp. LoRA loading: dynamic LoRA strength adjustment without re-encoding. Queue management: batch generation with customizable prompt variations using wildcard syntax and prompt weighting. API: built-in REST API for programmatic workflow execution. Custom nodes: extensible architecture with community-contributed node packs (ComfyUI Manager, Impact Pack, WAS Suite, rgthree) adding hundreds of custom nodes. Model optimization: xformers, PyTorch 2.0 SDPA, and low-vRAM mode for running on GPUs with 4GB VRAM. Open source under GPL-3.0.