SD.Next
SD.Next is an open-source web interface under Apache-2.0 for AI image and video generation, built on Stable Diffusion and descended from AUTOMATIC1111's WebUI. It covers text to image, image to image, text to video, inpainting and outpainting, with LoRA, ControlNet and IP-Adapters, and on-the-fly quantisation that cuts video memory use up to four times. It runs on NVIDIA, AMD, Intel, Apple and DirectX GPUs, or a CPU.
What does SD.Next do?
SD.Next is an "all-in-one WebUI", a "server application and web interface (WebUI) for AI image and video generation, built on Stable Diffusion and supporting dozens of advanced models." Its workflows cover text to image, image to image, text to video, image to video, inpainting, outpainting, a detailer and high-resolution refinement, with LoRA, ControlNet, IP-Adapters and prompt enhancement. It has desktop and mobile interfaces and is translated into about 15 languages.
Its SDNQ quantisation lets you "use pre-quantized or run with quantization on-the-fly for up to 4x VRAM reduction with no or minimal quality and performance impact", the project's own figure, and balanced offload splits memory between CPU and GPU. Captioning uses more than 25 built-in language and vision models. Reference models download automatically. The README credits AUTOMATIC1111's WebUI as the original codebase and links out for the list of supported models rather than naming them.
Key facts
- Licence: Apache-2.0, with no added conditions.
- Text to image, image to image, text to video, image to video, inpainting, outpainting, detailer and high-resolution refinement.
- LoRA, ControlNet, IP-Adapters and prompt enhancement.
- SDNQ quantisation for up to 4x less video memory, by the README's figure; balanced offload between CPU and GPU.
- Hardware: NVIDIA CUDA, AMD ROCm and ZLUDA on Windows, Intel Arc, OpenVINO, any DirectX GPU on Windows, Apple M1 and M2, ONNX, and CPU only.
- Captioning with 25+ built-in language and vision models; about 15 interface languages.
- Install: git clone and the webui script, with a built-in installer that handles updates and dependencies; Docker recipes for CUDA, ROCm, Intel and OpenVINO.
- No GitHub releases, only date-named tags (newest 2026-07-14); last push 2026-09-16; 7,342 stars on 2026-09-17.
What does it replace, and where does it fall short?
SD.Next replaces Midjourney for someone with less common hardware, an AMD or Intel card, a DirectX GPU on Windows, or a machine with little video memory, who wants image and video generation from a familiar form-based web interface. Its quantisation and hardware list make it the pick for limited or unusual hardware on the open-source alternatives to Midjourney page.
Where it falls short: the README does not list which models it supports, pointing to its documentation instead, and it has no GitHub releases, only dated tags. The memory saving is the project's own measurement. For a node graph and the newest models listed explicitly, ComfyUI.
How does SD.Next run?
Clone the repository and run webui.sh on Linux or macOS, or webui.bat or webui.ps1 on Windows; the built-in installer handles dependencies and updates. Docker recipes cover CUDA, ROCm, Intel IPEX and OpenVINO. No hosted version exists; the project takes GitHub Sponsors.
Who is SD.Next for?
Someone on AMD, Intel or a low-memory GPU who wants a full-featured image and video web interface. Someone who wants to build node workflows should use ComfyUI or SwarmUI's workflow tab.
What limits does the README state?
None stated in the README, which points to troubleshooting and debugging guides.
Questions people ask
Is SD.Next open source?
Yes. SD.Next is Apache-2.0 with no added conditions. It credits AUTOMATIC1111's WebUI, which is AGPL-3.0, as its original codebase. The image and video models you run in it are separate downloads, each with its own licence to check before commercial use.
Does SD.Next work on AMD or Intel GPUs?
Yes. The README lists AMD through ROCm, and ZLUDA on Windows, Intel Arc through OneAPI, OpenVINO devices, any DirectX GPU on Windows through DirectML, Apple M1 and M2, NVIDIA CUDA, ONNX, and CPU only. That is the widest hardware list on this page.
How does SD.Next save video memory?
Through SDNQ quantisation, run on the fly or with pre-quantised models, which the README says cuts video memory use up to four times with little or no loss, and balanced offload, which splits memory between CPU and GPU. Both figures are the project's own.
How does SD.Next compare with Midjourney?
SD.Next runs image and video generation on your own hardware, including modest or non-NVIDIA cards, through a web interface, for free. Midjourney is a hosted prompt service. The open-source alternatives to Midjourney page compares it with ComfyUI, InvokeAI and SwarmUI.
Is SD.Next related to AUTOMATIC1111?
Yes. Its README says main credit goes to AUTOMATIC1111's WebUI for the original codebase. SD.Next has since added video generation, quantisation to save memory, much broader hardware support and its own installer with automatic updates.
Sources
- SD.Next README and Apache-2.0 LICENSE: github.com/vladmandic/sdnext, read 2026-09-17.
Compared with the others
On the open-source alternatives to Midjourney page, SD.Next is the pick for modest or unusual hardware. Apache-2.0 web interface with LoRA, ControlNet and quantisation its README says cuts video memory up to four times, on AMD, Intel, Apple or DirectX GPUs and CPUs.
Also on that page: ComfyUI for control, and the newest models, InvokeAI for a canvas for inpainting and editing, SwarmUI for a simple tab now, a graph later.
More image and video generation
The GPL-3.0 node-graph engine for image, video, audio and 3D generation that supports the newest open models, from Stable Diffusion and Flux to Wan and LTX video, on NVIDIA, AMD, Intel and Apple hardware, offline.
An Apache-2.0 creative image tool with a unified canvas for inpainting and outpainting, node workflows, and support for 28 model families from Stable Diffusion to Flux and Qwen Image, run as a local web app.
An MIT image and video generation web interface, formerly StableSwarmUI, with a simple Generate tab for beginners and a raw ComfyUI workflow tab for experts, that can spread work across several GPUs.
Added September 17, 2026. Every claim above comes from the project's README, LICENSE or model card, read on September 17, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/vladmandic/sdnext.
Running one of these inside your own environment, with your own data and your own security rules, is the kind of work Reveneau does. Read how a forward deployed engagement works.