📷PhotoMaker Adapter Loader locally

The local twin of the pack's hub-based adapter loader, doing the exact same job with a file that's already on your disk. PhotoMakerAdapter_Loader_local takes a path to your downloaded PhotoMaker adapter, loads it into the pipe from your base model loader, and - same as the fromhub version - registers the img trigger word and swaps the scheduler to EulerDiscreteScheduler. If you've got photomaker-v1.bin sitting in a models folder and don't want the hub node to re-download it, this is the one to use.

Use it exactly where you'd use the fromhub version: right after BaseModel_Loader_fromhub or BaseModel_Loader_local, before the generation node. The graph is identical, only the source of the file differs. There's no real advantage to one over the other once the file is cached locally - this one just skips the download and gives you a visible path you can point at a different folder if you keep models on a separate drive.

The inputs:

• pm_model_path - the folder containing the adapter, not the file itself. The default is a placeholder that says "enter your photomaker model path" - you have to replace it or the load will fail. This is the field people most often get wrong.
• filename - photomaker-v1.bin by default. The node joins folder + filename to find the file, so if yours is renamed, change this too.
• pipe - the MODEL out of your base loader.

Output is a MODEL (the adapter-loaded pipe) that flows into LoRALoader or NEW_PhotoMaker_Generation. You can get photomaker-v1.bin from the official TencentARC/PhotoMaker repo on Hugging Face if you don't have it.

Pack install is the standard one, once per machine - ComfyUI Manager (search "PhotoMaker") or:

cd ComfyUI/custom_nodes
git clone https://github.com/ZHO-ZHO-ZHO/ComfyUI-PhotoMaker-ZHO
cd ComfyUI-PhotoMaker-ZHO
pip install -r requirements.txt

Restart ComfyUI and point the path at wherever your adapter lives. Worth stating plainly: this whole pack is an unmaintained diffusers-wrapper port (quiet since April 2024) and it eats around 10GB VRAM at generation time. It works fine on SDXL, but if you're setting up fresh today, the native PhotoMaker nodes in ComfyUI core are the lighter, actively-maintained alternative - this pack is for the older workflows or the stubborn among us.