The overall operation structure
Stable Diffusion model construction first downloads diffusers, and then installs, the command is as follows:
git clone https://github.com/huggingface/diffusers.git
pip install diffusers
cd diffusers
pip install .
Both ubuntu and win systems are available
Wenshengtu, the code of Tushengtu and the trained model can be found on Baidu Netdisk (the trained model is very large, more than a dozen g)
Modify the model_id in txt2jpg_inference.ipynb to be the local model address, prompt to be the prompt word, and add specific descriptions to negative_prompt, separated by commas, and the proportion of influence on the result will decrease in turn:
:
result:
Modify the model_id in jpg2jpg_inference.ipynb to be the local model address, and prompt to be the prompt word:
Run the video demo :
https://www.bilibili.com/video/BV1DM4y1a7Mb/?vd_source=ffc65fd2c117392dd6599cf1d3bbaf20