ip-adapter_sd15.safetensors:基本模型, 平均强度 ip-adapter_sd15_light_v11.bin:与ip-adapter_sd15相同,但更兼容文本提示 ip-adapter-plus_sd15.safetensors:比 ip-adapter_sd15 更接近参考图像 ip-adapter-plus-face_sd15.safetensors:人像模型 ip-adapter-full-face_sd15.safetensors:更强的脸模型,...
"Node name for S&R": "CLIPVisionLoader" }, "widgets_values": [ "ip-adapter-plus_sd15.bin" "IPAdapter_image_encoder_sd15.safetensors" ] }, { "id": 5, "type": "IPAdapterApply", "id": 3, "type": "IPAdapterModelLoader", "pos": [ 650, 290, 60 ], "size": { "0": 21...
"Node name for S&R": "ControlNetLoader" }, "widgets_values": [ "sai_xl_depth_256lora.safetensors" ] }, { "id": 6, "type": "ControlNetApplyAdvanced", "pos": [ 850, 900 ], "size": { "0": 315, "1": 166 }, "flags": {}, "order": 33, ...
"sd15/realisticVisionV51_v51VAE.safetensors" ] }, { "id": 2, "type": "VAELoader", "pos": [ 890, 820 ], "size": { "0": 300, "1": 60 }, "flags": {}, "order": 1, "mode": 0, "outputs": [ { "name": "VAE", "type": "VAE", "links": [ 12 ], "shape":...
"Node name for S&R": "CLIPVisionLoader" }, "widgets_values": [ "IPAdapter_image_encoder_sd15.safetensors" ] }, { "id": 1, "type": "CheckpointLoaderSimple", "pos": [ 290, 280 ], "size": { "0": 300, "1": 100 }, "flags": {}, "order": ...
The results are often not as good as SDXL. Optimal weight seems to be from 0.8 to 2.0. The Style+Composition node doesn't work for SD1.5 at the moment, you can only alter either the Style or the Composition, I need more time for testing. Old workflows will still work but you may ...
https://huggingface.co/h94/IP-Adapter/resolve/main/sdxl_models/image_encoder/model.safetensors 一共大概需要下载30多G，耗时在概是18小时。下载好，放置在正确的目录，重启，就可以了，如果你的网络不好，不能科学上网，你也可以关注我，在我的作品，文章下留言，我可以把我下载好的百度网盘连接给到您，...
下载clip_h.pth:https://huggingface.co/h94/IP-Adapter/resolve/main/models/image_encoder/model.safetensors,放到/ComfyUI/models/clip_vision,将文件名改为CLIP-ViT-H-14-laion2B-s32B-b79K.safetensors。 下载buffalo_l.zip:https://github.com/deepinsight/insightface/releases/download/v0.7/buffalo_l...
https://huggingface.co/h94/IP-Adapter/resolve/main/sdxl_models/image_encoder/model.safetensors 一共大概需要下载30多G,耗时在概是18小时。 下载好,放置在正确的目录,重启,就可以了。 如果你的网络不好,不能科学上网,你也可以关注我,在我的作品,文章下留言,我可以把我下载好的百度网盘连接给到您,直接下...
faceid_sd15.bin" lora_ckpt = "ip-adapter-faceid_sd15_lora.safetensors" device = "cuda" noise_scheduler = DDIMScheduler( num_train_timesteps=1000, beta_start=0.00085, beta_end=0.012, beta_schedule="scaled_linear", clip_sample=False, set_alpha_to_one=False, steps_offset=1, ) vae =...