TencentARC
Howto: Windows Installation
For windows users out there, I have created a fork with an easy set of instructions and updates to facilitate the use on windows. You can find the 1st release at:
https://github.com/bmaltais/PhotoMaker
This fork also support non square output format so you can produre results like:
You will find the width and height setting under the Advanced Accordion in the GUI:
Once you have followed the installation instructions you simply start the Gradio GUI with: .\gui.bat
按说明安装出现以下信息
另外如果使用python_embeded的可以安装吗？
My solution require a local installation of python 3.10.9... I don't think it will run properly with another type of python installation... sorry ;-(
I've forced the install
torch==2.0.1
torchvision==0.15.2
Now that I'm stuck in the place where I downloaded the model, where do I store the model I downloaded from huggingface?
Very cool! Thank you @bmaltais. We have listed your contributions in README file.
Very cool! Thank you @bmaltais. We have listed your contributions in README file.很
Do you consider the situation of download interruptions?
@xhoxye Look like you still have some network download issue... causing the timeout...
The download seems to be complete, and it shows a timeout when it loads
Humm... this is something happening with the app.py code at this point. I wish I could help you out more but this is outside facilitating the installation and calling the app.py
Keep in mind... running gradio python app is not a walk in the park... there are many things that can go wrong as config on individual system can have a significant impact.
I modified the model path and finally got into the UI, but it was very slow and the GPU didn't seem to be being used?
Still no, come back tomorrow
The download seems to be complete, and it shows a timeout when it loads
same problem. First, it loads the models for about 10 minutes, after which it starts. But photo generation doesn’t work, it loads endlessly. The video card is not loaded at all.
Any ideas about this problem?
This is odd... maybe I need to add some other pre-requirements like Visual Studio... I do have it installed on my computer... it is possible that it is required to build photomaker... strange... I will have to do some investigation
I found out that you cant install binsandbytes with cuda support on windows ..so it will not use GPU but only CPU... so what now ?
Edit: What is this ? https://github.com/Keith-Hon/bitsandbytes-windows
Any ideas about this problem?
This is odd... maybe I need to add some other pre-requirements like Visual Studio... I do have it installed on my computer... it is possible that it is required to build photomaker... strange... I will have to do some investigation
I deleted the last dot in the requirements-windows.txt file, and now everything works
Any ideas about this problem?
This is odd... maybe I need to add some other pre-requirements like Visual Studio... I do have it installed on my computer... it is possible that it is required to build photomaker... strange... I will have to do some investigation
I deleted the last dot in the requirements-windows.txt file, and now everything works
Really strange. I deleted the venv, removed the wheel part, re-installed and it work fine...
Can you try to do a git pull in the terminal to check if you have the latest code?
I found out that you cant install binsandbytes with cuda support on windows ..so it will not use GPU but only CPU... so what now ?
Edit: What is this ? https://github.com/Keith-Hon/bitsandbytes-windows
I managed to get PhotoMaker to work yesterday, i used https://github.com/jllllll/bitsandbytes-windows-webui/releases pip uninstall bitsandbytes pip uninstall bitsandbytes-windows pip install https://github.com/jllllll/bitsandbytes-windows-webui/releases/download/wheels/bitsandbytes-0.41.1-py3-none-win_amd64.whl
Good question. Installing python 3.10.9 directly from python is the supported method for the underlying so-script from Koby’s. Might work with anaconda but I have not tested it.