A work in progress app to create content using generative AI.
It uses the stable-diffusion.cpp library, which implements popular text-to-image model architectures.
The user needs to have downloaded an SD 1.5 .safetensors model, at the moment the model path can be changed from ./src/dl/sd_instance.cpp.
Clone the repo with the submodules:
git clone --recursive https://github.com/stefanasandei/diffusion-studio.gitOr if you forgot --recursive, to fetch the git submodules after clone:
git submodule update --initAfterwards it's the classic CMake setup:
mkdir build && cd build
cmake ..Now open the project with Visual Studio, CLion or your editor of choice.
Relevant files are in ./src/gfx/ (vulkan code for the imgui window), ./src/app (ui logic for the panels) and ./src/dl (where the magic happens, coroutines are used with a generator to stream the image renders to the ui).
MIT © Asandei Stefan-Alexandru. All rights reserved.