SDGO is a small project that aims to simplify working with the original SD 3.5 repository and provides a Go-based front-end. Features:
- Image generation as in the original repository
- Negative prompt support
- Minimal non-browser interface with tabs
- Size sliders have a step of 64, plus there is a megapixel counter in the UI
- Dynamic previews with a step of 5
To do:
- Add config saving button functionality
To run SDGO your machine must meet the following requirements:
Hardware: 32Gb of RAM, CUDA GPU with 12 Gb of VRAM, 20Gb of free space OS: Windows 10 or 11, x64
The recommended way of running the program is by downloading the latest release from the Releases section of this repository. There, you will find sdgo-portable-release.7z
, which you should unzip and populate the models
folder with files from huggingface: SD 3.5 Medium, Clip-L, BigG, T5-XXL and VAE.
Then, launch sdgo.exe
and wait until the console window says that the models are loaded. The image outputs will be located in the outputs
folder, but one can copy and rename the original file from within the UI via the Save button.
As of now, the original files of the project do not have a defined license, but the backend files have the original MIT License stated in SD 3.5 repository. However, if you wish to copy and/or modify the files for distribution, you are free to do so, but must provide a link to the SDGO repository.