sdnext

Why we included this project

SD.Next is a self-hosted web interface for Stable Diffusion that handles the full image and video generation workflow in one place, from text-to-image and inpainting to upscaling and captioning. It also reaches into video generation, with both text-to-video and image-to-video support, and includes tools for model quantization and memory balancing that let larger models run on modest hardware. The built-in captioning and tagging utilities use LLM, VLM, and CLIP-based taggers, and there is a color-grading toolkit for post-processing. It runs on NVIDIA, AMD, Intel Arc, Apple silicon, and CPU-only systems, so teams and hobbyists can deploy it on hardware they already own.