Are there any resources based on our PhotoMaker? Please leave a link here and we will acknowledge and list it in the README.

[Paper99]

Any form of resource is acceptable, including but not limited to Windows version, MacOS version, comfyui and WebUI versions. The only connection between them is PhotoMaker 😄

[yorickvP]

Hi! We've been working on https://replicate.com/jd7h/photomaker. We're currently working on some bugs and features and we'd like to transfer it be under the TencentARC org when it's done, but feel free to cherry-pick datakami-models@db6f38f as a link in the meantime.

[Paper99]

Cool! We will update this in the README.

[Vkasdubey]

🔥🔥

[bmaltais]

I have updated some of your code to make the installation and running on Windows easier. Here is the current release in my fork. You might consider including some of the updates in your project to make it easy for Windows usage:

https://github.com/bmaltais/PhotoMaker/tree/v1.0.1

It also add support for non square output:

I added a . at the end of the requirements-windows.txt to make it built the module during pip install.

[Paper99]

Cool! Thank you @bmaltais

[Paper99]

We'll update it.

[choigawoon]

cool this is what i want

[githubUser01946]

Why is it downloading things several times?

[mtangminghaiguge]

cool ！i am learnning...

[bmaltais]

@Paper99 Would it be possible to update the README.md link to my repo to:

- [Windows version of PhotoMaker](https://github.com/bmaltais/PhotoMaker) by [@bmaltais](https://github.com/bmaltais), easy to deploy PhotoMaker on Windows. The description can be found in [this link](https://github.com/TencentARC/PhotoMaker/discussions/36#discussioncomment-8156199).

[Paper99]

Of course!

[bmaltais]

Here are the two ComfyUI node project I know about:

https://github.com/ZHO-ZHO-ZHO/ComfyUI-PhotoMaker

https://github.com/StartHua/Comfyui-Mine-PhotoMaker

[Paper99]

Thank you @bmaltais. We'll update them.

[ri131211]

Hello

[machineminded]

I'd like to integrate this into Fooocus, however their pipeline is customized. Was hoping to get some guidance on this since Fooocus doesn't use diffusers out of the box. Can the model be converted to a format expected by Fooocus?

Edit: It has been integrated. More features coming soon:

https://www.github.com/machineminded/Fooocus-inswapper

[bmaltais]

You may want to watch out. I have done testing on that ComfyUI implementation and the results are not as good as the diffuser implementation. Specially on longer prompt where that likeness of face fade away…

probably related to a difference on how prompts and ksampler is applied?

But it is a good starting point…

[machineminded]

@Paper99 @bmaltais Very cool - thank you for your work. I'm not too deep into the AI architecture but I should be able to integrate this into Fooocus. Might take me some time so if someone else wants to take a stab at it feel free.

Originally I was going to just use it as diffusers in Fooocus, and it would just have its own special pipeline. That still may be the best option given your experience @bmaltais

[Paper99]

Please let us know if you get the job done, thank you.

[machineminded]

I have a basic implementation for this here: https://github.com/machineminded/Fooocus-inswapper/tree/feature/photomaker

I need to do some more testing, mostly to make sure pip package compatibility. Otherwise I think the basic implementation is ready for use. It has been implemented with 🤗 diffusers and still hooks into the Fooocus async tasks. Preview images don't look as nice as ksampler but I think it is a limitation of diffusers.

[Paper99]

Hi @machineminded,
Thanks for your efforts.
We've updated your project in README file.
Commit message 78a0794

[woai18ytgjiang]

how to control two different face image mixing percentage in prompt?

[ToniDuque]

Does anyone know how to fix this after running pip install -r requirements-windows.txt?:
Building wheels for collected packages: photomaker
Building wheel for photomaker (pyproject.toml) ... error
error: subprocess-exited-with-error

× Building wheel for photomaker (pyproject.toml) did not run successfully.
│ exit code: 1
╰─> [28 lines of output]
Traceback (most recent call last):
File "C:\Users\Toni\Desktop\photomaker\venv\lib\site-packages\pip_vendor\pyproject_hooks_in_process_in_process.py", line 353, in
main()
File "C:\Users\Toni\Desktop\photomaker\venv\lib\site-packages\pip_vendor\pyproject_hooks_in_process_in_process.py", line 335, in main
json_out['return_val'] = hook(**hook_input['kwargs'])
File "C:\Users\Toni\Desktop\photomaker\venv\lib\site-packages\pip_vendor\pyproject_hooks_in_process_in_process.py", line 251, in build_wheel
return _build_backend().build_wheel(wheel_directory, config_settings,
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\api.py", line 58, in build_wheel
return WheelBuilder.make_in(
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\builders\wheel.py", line 88, in make_in
wb.build(target_dir=directory)
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\builders\wheel.py", line 124, in build
self._copy_module(zip_file)
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\builders\wheel.py", line 269, in _copy_module
to_add = self.find_files_to_add()
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\builders\builder.py", line 166, in find_files_to_add
for include in self._module.includes:
File "C:\Users\Toni\AppData\Local\Programs\Python\Python310\lib\functools.py", line 981, in get
val = self.func(instance)
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\builders\builder.py", line 92, in _module
return Module(
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\utils\module.py", line 79, in init
PackageInclude(
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\utils\package_include.py", line 29, in init
self.check_elements()
File "C:\Users\Toni\AppData\Local\Temp\pip-build-env-n7f5zrnf\overlay\Lib\site-packages\poetry\core\masonry\utils\package_include.py", line 85, in check_elements
raise ValueError(f"{root.name} is not a package.")
ValueError: photomaker is not a package.
[end of output]

note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for photomaker
Failed to build photomaker
ERROR: Could not build wheels for photomaker, which is required to install pyproject.toml-based projects

[mhamzasani]

Is there an easy way to Install it and use it on Linux. I've tried hard on my server but run into errors. Want create a Rest API to use this?

Need help!

[mhamzasani]

Also, does it only run on Large GPU? Can it run on Oracle Cloud 4 Ocpu?

[bmaltais]

If you can run ComfyUI then I suggest you install this custom node to make use of photomaker. Might work better with lower VRAM GPUs: https://github.com/shiimizu/ComfyUI-PhotoMaker

It is the best node I could find out there as it integrate mostly using native ComfyUI nodes to load the LoRA part and create the image using the native ComfyUI KSampler.

[defisynner]

How can i input imgs of 2 diff people and have it in the prompt?

An example: uploading an image of a man and another image of a lady. And create a photo of them kissing

[Paper99]

Please refer to #112

[defisynner]

i tried having img twice in the prompt and it will reject my prompt

[Paper99]

Hello, you need to make modifications to pipeline.py. We currently cannot guarantee performance.

[bssrdf]

Hi, I have implemented PhotoMaker in stable-diffusion.cpp using purely C/C++/CUDA. The results are not as good as the official version though, maybe because SD is not using the diffuser pipeline and I may have missed something. It does generate some interesting ID faces.

[Paper99]

Thanks for your awesome work.
We've added your work in Related Resources.