Installation

Create and activate a virtual environment:

python -m venv myenv

source myenv/bin/activate  # On Linux/macOS

myenv\Scripts\activate     # On Windows

Clone the repository:

git clone https://github.com/AbrahamPaulJ/diffusers-webui.git

cd diffusers-webui

Install dependencies

python install.py

Run

python app.py

Features

• Supports SD 1.5 and SDXL models.

• Image To Image tab: Allows image-to-image, inpainting and outpainting with the APIs: StableDiffusionImg2ImgPipeline and StableDiffusionInpaintPipeline

  • Implemented features:

    • ControlNet support with StableDiffusionControlNetPipeline : Canny, Depth and Openpose.

    • Support for multiple LoRA and embedding files (textual inversion).

    • Inpaint Mode:

      • Manual Brush Tool: Allows users to create a image mask (inpaint mode).

      • Gaussian Blur Slider: For applying blur to the mask edges.

      • Post-Processing Option: Retains original non-masked areas after image generation.

    • Outpaint Mode: Outpainting is done using the inpaint pipeline, where the a black background canvas is used as a mask.

      • The transform tool allows cropping of the image placed on the black canvas, which makes outpainting in any direction easy.

      • You can choose where to place the image on the canvas using the "Image Positioned at:" option.

      • You can control output image size by adjusting the "Maximum Width/Height" parameter.

• Text to Image tab: Generates images from prompts using the API: StableDiffusionPipeline.

  • Implemented features:

    • Hi-Res .fix - 2x latent

    • Support for multiple LoRA and embedding files (textual inversion).

• PNG Info tab: For previously generated images, featuring a view of the generation parameters and options to send the parameters to other tabs.

Images

Notes

• You can add new checkpoints/models to the models folder. Supports ".ckpt", ".safetensors" and diffusers formats.

• Similarly, there are folders for LoRAs and embeddings.

• Prompt format: To add prompt weighting, please use Compel Prompt Syntax

• Ensure you have the correct version of Pytorch and CUDA supported by your device, in case you want to enable GPU for inference.

• To increase generation speed with GPU, you can install xformers if it is supported. It will be detected automatically once installed.

Known Issues

• Controlnet Strength Slider - doesn't work (diffusers bug).