NMKD Stable Diffusion GUI

NMKD Stable Diffusion GUI is a downloadable Windows application developed by N00MKRAD that packages the Stable Diffusion image-generation model and all of its Python dependencies into a single installer with a graphical user interface, so users can run AI image generation locally on their own GPU without manually setting up Python, Anaconda, or Docker. The application is distributed on itch.io under a name-your-own-price model and the source code is published on GitHub under the GPL-3.0 license.

Key Features

• Text-to-image and image-to-image generation, plus InstructPix2Pix instruction-based image editing for natural-language edits such as adding an object to an existing photo.
• Advanced prompting tools: multiple prompts at once, negative prompts, attention and emphasis weighting, wildcards filled in from inline lists or text files, a prompt queue, and a searchable prompt history.
• Support for custom Stable Diffusion checkpoints, custom VAE models, Textual Inversion embeddings, and LoRA concepts, characters, and styles including LoRA training.
• Built-in post-processing with RealESRGAN upscaling and CodeFormer or GFPGAN face restoration, exposed from a dedicated toolbar button.
• A built-in image viewer that shows generation metadata, an option to create seamless or tileable images useful for game textures, an option to generate symmetric images, and a built-in updater that can re-install or repair the application.
• A built-in safety check that scans downloaded models for malware before they are used, plus a developer tools menu that includes Stable Diffusion CLI access, model merge and prune, and CKPT/Diffusers/SafeTensors conversion.

System Requirements and Performance

• Operating system: Windows 10 or Windows 11, 64-bit.
• Minimum GPU: an Nvidia card with 4 GB of VRAM using the Maxwell architecture (2014) or newer, or any DirectML-capable GPU with 8 GB of VRAM for limited feature support; AMD GPUs are supported in an experimental mode through the ONNX implementation.
• Memory: 8 GB of RAM minimum with the Windows pagefile enabled, and 16 GB is recommended.
• Disk: roughly 10 GB minimum and 12 GB on an SSD recommended, plus about 5 GB of free space for temporary files.
• Performance scales with the GPU: the developer reports under one second per 512x512 image on an RTX 4090 and under two seconds per image on an RTX 3090.

Distribution and Licensing

• Downloaded from the official itch.io page at nmkd.itch.io/t2i-gui as a name-your-own-price tool. Version 1.11.0 is offered as a 3.3 GB package that includes the SD 1.5 model and a 1.5 GB package without model files for users who supply their own checkpoints.
• Source code, the main guide, and additional documentation including an inpainting guide and an AMD GPU guide are published on the official GitHub repository n00mkrad/text2image-gui.
• Licensed under the GNU General Public License v3.0. The developer states the application is uncensored and unfiltered, no data is shared with or collected by the developer or any third party, and the developer is not responsible for content generated with the tool.

Intended Users and Support

• Targeted at hobbyists, digital artists, game texture creators, and experimenters who want to run Stable Diffusion on a Windows desktop with a single installer and GUI rather than configure a Python environment manually.
• User support is provided through a stable-diffusion-gui channel on the developer's Discord server, and Patreon supporters gain access to newer builds of the developer's related video-interpolation tool Flowframes.