Diffusion-pipe Wizard GUI (WIP)

A Flet-based GUI application designed to run on WSL/Linux for constructing and managing configuration files for diffusion-pipe.

Key Features:

• Interactive user interface that automatically matches and displays default model configurations, making it easy to customize and fine-tune training parameters for supported models without manual editing of configuration files
• Built-in video/image editing and captioning manager for dataset preparation
• Model download manager using aria2c for efficient parallel downloading of large model files

Note: This application was tested on WSL (Windows Subsystem for Linux) and may require additional testing on native Linux distributions. Note: Not all functions are implemented to properly work with dataset files on server yet.

Quick Start

📁 Navigate to project folder (optional)

cd Dpipe

🚀 Run flet_app

source ~/miniconda3/bin/activate && source $(conda info --base)/etc/profile.d/conda.sh && conda activate ./dp_env && python3 -m flet_app.flet_app

1. Installation

1.1 🔧 Install FFMPEG ,aria2 and Miniconda

sudo apt update && sudo apt install ffmpeg && sudo apt install -y aria2

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
chmod +x Miniconda3-latest-Linux-x86_64.sh
./Miniconda3-latest-Linux-x86_64.sh -b -p ~/miniconda3
source ~/miniconda3/bin/activate

1.2 📥 Clone GUI and diffusion-pipe

git clone https://github.com/siraxe/Dpipe-Wizard-GUI.git Dpipe
cd Dpipe
git clone --recurse-submodules https://github.com/tdrussell/diffusion-pipe

1.3 🐍 Setup venv

conda env create -f dp_env.yml -p ./dp_env

Apply qwen_edit_plus patch (optional/wip)

chmod +x apply_qwen_patch.sh
./apply_qwen_patch.sh

Apply LongCat-Video patch (optional)

Thanks to tdrussell/diffusion-pipe#453 you can add LongCat-Video model support to diffusion-pipe by running :

chmod +x apply_longcat_patch.sh
./apply_longcat_patch.sh

This script will:

• Apply the LongCat-Video model patches to diffusion-pipe
• Automatically download and initialize the LongCat-Video submodule
• Update the necessary configuration files

That's it , you can run it now.

conda activate ./dp_env && python3 -m flet_app.flet_app

2. 🖥️ Setup WSL to be able to trigger photoshop (optional)

Edit /etc/wsl.conf in wsl

sudo nano /etc/wsl.conf

Ensure:

[interop]
enabled=true
appendWindowsPath=true

[automount]
enabled=true
options = "metadata,umask=22,fmask=11,case=off"

In Windows PowerShell : wsl --shutdown

Reopen WSL and re-test

⚙️ Edit start_dpipe.bat

# Change this to match your setup ,to use it from windows
/home/username/Dpipe