SD Chat - AI Creation Platform

English | 简体中文

An AI creation platform based on Stable Diffusion, supporting both image and video generation. This project integrates multiple advanced AI models and provides a user-friendly web interface for creating high-quality AI-generated images and videos.

Demo

Roadmap

Completed Features ✅

1. Text-to-Image

• Stable Diffusion XL Base 1.0
• Stable Diffusion 1.5
• Realistic Vision V5.1
• Dreamshaper V8
• Custom prompt support
• Parameter adjustment
• Negative prompt support

2. Text-to-Video

• Stable Video Diffusion
• Custom prompt support
• Video parameter adjustment
• Negative prompt support

3. Image-to-Video

• Stable Video Diffusion
• Custom initial image support
• Video parameter adjustment
• Motion intensity control

Under Development 🚧

4. Image Editing

• ControlNet support
• Image inpainting
• Image outpainting
• Prompt-based editing

5. Video Editing

• Video restoration
• Video style transfer
• Frame interpolation
• Video upscaling

Features

1. Image Generation

• Multiple high-quality models:
  • Stable Diffusion XL
  • Stable Diffusion 1.5
  • Realistic Vision
  • Dreamshaper
• Custom generation parameters
• High-quality output
• Batch generation support

2. Video Generation

• Based on Stable Video Diffusion
• Image-to-video conversion
• Text-to-video generation
• Adjustable video parameters

3. Interface Features

• Intuitive web interface
• Real-time preview
• Parameter control
• Example prompts

Quick Start

Requirements

• MacOS (Apple Silicon)
• Python 3.10+
• 32GB RAM (recommended)
• Disk space: 20GB minimum (for model storage)

Installation

1. Install Conda (skip if already installed):

# Download Miniforge3 (for Apple Silicon)
curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-MacOSX-arm64.sh"
bash Miniforge3-MacOSX-arm64.sh

2. Create and activate environment:

# Create Python 3.10 environment
conda create -n diffusers python=3.10
conda activate diffusers

3. Install PyTorch (with MPS acceleration):

# Install PyTorch with MPS support
conda install pytorch torchvision torchaudio -c pytorch-nightly

4. Clone the project:

git clone https://github.com/samzong/sd-chat.git
cd sd-chat

5. Install dependencies:

pip install -r requirements.txt

First Run

1. Start the service:

python run.py

Initial run will automatically download required models, which may take some time depending on your network connection.

2. Access the interface: Open http://localhost:7860 in your browser

Project Structure

sd-chat/
├── run.py          # Unified startup script (recommended)
├── api.py          # Image generation service
├── video_api.py    # Video generation service
├── web_ui.py       # Web interface
└── requirements.txt # Project dependencies

Contributing

Issues and Pull Requests are welcome to help improve the project.

中文文档