PuLID

PuLID copied to clipboard

The image returned from the replicate api is a completely black image, but the background shows that the loading is successful.

1

 They all seem to be successful, whether it is API or web, but without exception they are black 

pijiuya

VXeffect

Comfyui please!

9

Very good face consistency but need comfyui. Excellent work!

freeexit2002

I have been testing PuLID in A1111 following this guide https://github.com/Mikubill/sd-webui-controlnet/pull/2838 But the results are very different from your demo, they are very low quality and low fidelity. I trsted...

Lorian0

I wanted to load a custom stable diffusion model, and also to load custom lora weight instead of loading SDXL Lightning. Is there a method, or a planned method, to...

Fqlox

Hello, Nice project but as you are utilizing antelopev2 as the face recognition model, it's important to note that your entire project is currently restricted for pure research purposes and...

imperator-maximus

signficant -> significant

eltociear

Will the training scripts be released at any point?

okaris

This is my errors: `torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 22.00 MiB (GPU 0; 12.00 GiB total capacity; 10.89 GiB already allocated; 0 bytes free; 11.31 GiB reserved...

inspire-boy

Please change base model to latest fine tuned sdxl model like juggernaut x or any other model

1

pipeline.py ``` class PuLIDPipeline: def __init__(self, *args, **kwargs): super().__init__() self.device = 'cuda' sdxl_base_repo = 'stabilityai/stable-diffusion-xl-base-1.0' sdxl_lightning_repo = 'ByteDance/SDXL-Lightning' self.sdxl_base_repo = sdxl_base_repo ```

putuoka