update consisid

Author: SHYuanBest

src/diffusers/pipelines/consisid/pipeline_consisid.py

@@ -45,16 +45,27 @@
         ```py
         >>> import torch
         >>> from diffusers import ConsisIDPipeline
+        >>> from diffusers.pipelines.consisid.util_consisid import prepare_face_models, process_face_embeddings_infer
         >>> from diffusers.utils import export_to_video, load_image
 
+        >>> face_helper_1, face_helper_2, face_clip_model, face_main_model, eva_transform_mean, eva_transform_std = prepare_face_models("https://huggingface.co/BestWishYsh/ConsisID-preview", device="cuda", torch_dtype=torch.bfloat16)
+        >>> face_helper_1.face_det.to(device)
+        >>> face_helper_1.face_parse.to(device)
+        >>> face_clip_model.to(device, dtype=dtype)
+
         >>> pipe = ConsisIDPipeline.from_pretrained("https://huggingface.co/BestWishYsh/ConsisID-preview", torch_dtype=torch.bfloat16)
         >>> pipe.to("cuda")
 
         >>> prompt = "A woman adorned with a delicate flower crown, is standing amidst a field of gently swaying wildflowers. Her eyes sparkle with a serene gaze, and a faint smile graces her lips, suggesting a moment of peaceful contentment. The shot is framed from the waist up, highlighting the gentle breeze lightly tousling her hair. The background reveals an expansive meadow under a bright blue sky, capturing the tranquility of a sunny afternoon."
         >>> image = load_image(
         ...     "https://github.com/PKU-YuanGroup/ConsisID/blob/main/asserts/example_images/1.png?raw=true"
         ... )
-        >>> video = pipe(image, prompt, use_dynamic_cfg=True)
+
+        >>> id_cond, id_vit_hidden, image, face_kps = process_face_embeddings_infer(face_helper_1, face_clip_model, face_helper_2, eva_transform_mean, eva_transform_std, face_main_model, device, dtype, img_file_path, is_align_face=True)
+        >>> is_kps = getattr(pipe.transformer.config, 'is_kps', False)
+        >>> kps_cond = face_kps if is_kps else None
+
+        >>> video = pipe(image=image, prompt=prompt, use_dynamic_cfg=False, id_vit_hidden=id_vit_hidden, id_cond=id_cond, kps_cond=kps_cond)
         >>> export_to_video(video.frames[0], "output.mp4", fps=8)
         ```
 """

src/diffusers/pipelines/consisid/util_clip/__init__.py

@@ -0,0 +1,11 @@
+from .constants import OPENAI_DATASET_MEAN, OPENAI_DATASET_STD
+from .factory import create_model, create_model_and_transforms, create_model_from_pretrained, get_tokenizer, create_transforms
+from .factory import list_models, add_model_config, get_model_config, load_checkpoint
+from .loss import ClipLoss
+from .model import CLIP, CustomCLIP, CLIPTextCfg, CLIPVisionCfg,\
+    convert_weights_to_lp, convert_weights_to_fp16, trace_model, get_cast_dtype
+from .openai import load_openai_model, list_openai_models
+from .pretrained import list_pretrained, list_pretrained_models_by_tag, list_pretrained_tags_by_model,\
+    get_pretrained_url, download_pretrained_from_url, is_pretrained_cfg, get_pretrained_cfg, download_pretrained
+from .tokenizer import SimpleTokenizer, tokenize
+from .transform import image_transform

src/diffusers/pipelines/consisid/util_clip/bpe_simple_vocab_16e6.txt.gz

@@ -0,0 +1,2 @@
+OPENAI_DATASET_MEAN = (0.48145466, 0.4578275, 0.40821073)
+OPENAI_DATASET_STD = (0.26862954, 0.26130258, 0.27577711)