SD2.1 (15 GB) uses pivotal tunining, slow editing	SD1.5 (8GB) uses latent alignment, fast editing

Out of AI presents a flexible tool in Gradio to manipulate your images. This is our first version of Image modification tool through prompt manipulation by reconstruction through diffusion inversion process.

https://github.com/user-attachments/assets/de8a4f9d-cdad-4b73-962e-2cc42bbc2e0d

Download the project

git clone https://github.com/OutofAi/OutofFocus.git
cd  OutofFocus

🚀 Run Gradio Demo locally

After downloading the project and navigating to the correct folder

Install dependencies:

pip install -r requirements.txt

Run the app

python app.py

If you running on a GPU with less than 15GB VRAM, try running the turbo variation as it only requires 8 GB VRAM, this model uses SD 1.5 and can handle any aspect ratios as well Run the app

python app_turbo.py

🚀 Run Gradio Demo locally in Virtual Environment

If you face any dependency, token or cuda issues we higly recommend running it in a virtual environment.

To run it in a virtual environment use these commands instead, after downloading the project and navigating to the correct folder

Create a virtual environment:

python -m venv venv

Activate the environment:

Windows:

.\venv\Scripts\activate

macOS/Linux:

source venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Run the app in virtual environment:

python app.py

🚀 Run Gradio Demo on Google Colab (GPU based)

If you want to run it on Colab either click on the badge below

or create your own notebook, make sure you got a GPU notebook enabled (Views->Notebook Info) a A100 GPU or a L4 GPU and run the following commands

!git clone https://github.com/OutofAi/OutofFocus
%cd OutofFocus
!pip install -r requirements.txt
!python app.py --share

https://github.com/user-attachments/assets/1f71553d-d41d-45d6-a256-df3e06f93182

More Examples: