看上去像是人脸迁移，那相比目前的换脸工作流原理有何优势呢？而且缺乏明确的代码和工作流的使用讲解
我倒是挺想尝试一下的，但是搞不懂，看上去似乎需要两个lora协同工作，并且走图生图流程？
I made the directories as indicated, but none of the Hyperlora nodes find the corresponding files (Not found!)
Same here, I have trouble getting model files recognized in ComfyUI.
I double checked I have the required files in folders, and HyperLoRA configuration complains that each of the files are missing. Latest Comfy UI, all custom nodes updated.
I first thought maybe it is the fact that my models are loaded from another drive, (using extra model paths), then I tried symlinking folder, but that didn't help, and finally I copypasted the folders into comfyUI\models\hyperlora, then restarted ComfyUI completely from command line, but that made no difference.
I also found one YouTube video by AiorBust (HyperLoRA: The best model to create AI portraits Right Now! (Better than Pulid and Ip-Adapter)) but that video didn't do anything differently.
* HyperLoRAConfig 2:
- Value not in list: image_encoder: 'clip_vit_large_14' not in ['Not found!']
- Value not in list: image_processor: 'clip_vit_large_14_processor' not in ['Not found!']
* HyperLoRALoader 3:
- Value not in list: model: 'sdxl_hyper_id_lora_v1_fidelity' not in ['Not found!']
我倒是挺想尝试一下的，但是搞不懂，看上去似乎需要两个lora协同工作，并且走图生图流程？
@jiuluowkl 是文生图，只需要id lora就可以，github repo里有示例工作流 https://github.com/bytedance/ComfyUI-HyperLoRA
@titocrovetto
@eses
clip_vit_large_14 , clip_vit_large_14_processor , and sdxl_hyper_id_lora_v1_fidelity are three subfolders. Please confirm that they are respectively located in models/hyper_lora/clip_vit/, models/hyper_lora/clip_processor/, and models/hyper_lora/hyper_lora/ directories.
Thanks, I got it working now.