Multiview-Avatar

repository for multiview character generation based on SDv1.5

Requirements

• torch
• torchvision
• torchaudio
• diffusers
• accelerate
• xformers
• triton
• transformers
• 16GB GPU

Installation

git clone https://github.com/ArcherFMY/Multiview-Avatar.git
cd Multiview-Avatar
pip install -r requirements.txt

Getting Started

Download Models

Download models from Baidu Disk or Google Drive. Unzip models.zip into the root directory of the project.

${ROOT}  
|-- data  
|   |-- test.jpg
|-- models  
|   |-- base-model
|   |-- control-imgs
|   |-- controlnet-lineart
|   |-- controlnet-pose
|   |-- wildcards
|-- mvavatar
|...

Or download the models from hugging face (link1 or link2).

Inference

from mvavatar import MVAvatar

model = MVAvatar.from_pretrained('models', device='cuda')
image = model.inference('1 girl')

image.save('data/test.jpg')

see more in demo.py

License

This code is released under the Apache License 2.0 (refer to the LICENSE file for details).