user/pass: demo / demo
The demo relies on community resources and may be overloaded at times. Please DO NOT use this for automated or production workloads; it is subject to change.
Home Diffusion is an open source project that uses Stable Diffusion and ControlNet to revolutionize the way home design is done. With Home Diffusion, you can run your own design studio and craft beautiful, unique designs with ease.
Home Diffusion is easy to use and requires only a GPU to run. Here's what you need to get started:
- A GPU with suggested minimum 12GB of VRAM
- An internet connection
- Home Diffusion
Once you have the necessary hardware and software, you can start designing your dream home with Home Diffusion.
Home Diffusion is a revolutionary open source project that utilizes Stable Diffusion and ControlNet to create highly accurate and customizable home designs. Home Diffusion is a powerful tool that allows users to design their own homes with unprecedented accuracy and precision.
Users can customize every aspect of their home design, from the overall layout and structure to the smallest details. Home Diffusion also provides users with access to a library of pre-made designs and templates, so they can quickly find the perfect home design for their needs.
Home Diffusion is an innovative open source project that enables users to create design concepts for their homes. Through a fine-tuned implementation of Stable Diffusion and ControlNet, Home Diffusion provides users with an intuitive and powerful tool to create dream homes that fit their exact specifications.
You can run your own design studio with Home Diffusion if you have your own GPU.
Either way, the first step is to download the file and place it in ./models.
git clone [email protected]:HomeDiffusion/HomeDiffusion.git
cd HomeDiffusion
# Download the model, this is ~5.4 GB, so grab a coffee. Thanks to lllyasviel for ControlNet!
wget https://huggingface.co/lllyasviel/ControlNet/resolve/main/models/control_sd15_mlsd.pth -P models/
conda env create -f environment.yml
conda activate homeDiffusion
# Get another coffee. This be slow the first time, but faster in subsequent uses. Varies based on your internet and machine
docker-compose up --build
# Get another coffee. This be slow the first time, but faster in subsequent uses. Varies based on your internet and machine
Once the build is done, with either conda
or docker-compose
, you should see a message like this at the end of the output:
homediffusion | Loaded model config from [./models/cldm_v15.yaml]
homediffusion | Loaded state_dict from [./models/control_sd15_mlsd.pth]
homediffusion | Running on local URL: https://0.0.0.0:7860
homediffusion |
homediffusion | To create a public link, set `share=True` in `launch()`.
Open a browser and load https://0.0.0.0:7860
. Your local instance of Home Diffusion should be running and look like this:
After running thousands of iterations locally, we have set some sane defaults for you. An advanced user guide is coming soon!
Enjoy! Please give us a star if you like Home Diffusion :)
Thank you to many open source contributors who have made this possible, but especially: