wangbo/ComfyUI
1
0
Fork 0
mirror of https://github.com/comfyanonymous/ComfyUI.git synced 2026-01-15 00:30:55 +08:00
Code Issues Actions 12 Packages Projects Releases Wiki Activity

Update kandinsky5.py

Browse Source
This commit is contained in:
kijai 2025-12-02 14:36:11 +02:00
parent 4462fab1c8
commit 1a88fde6d9
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
comfy/text_encoders/kandinsky5.py
View File

@ -9,7 +9,7 @@ class Kandinsky5Tokenizer(QwenImageTokenizer):
# yes the typos "promt", "scren" were also in the original template... todo: check if it matters
self.llama_template = "<|im_start|>system\nYou are a promt engineer. Describe the video in detail.\nDescribe how the camera moves or shakes, describe the zoom and view angle, whether it follows the objects.\nDescribe the location of the video, main characters or objects and their action.\nDescribe the dynamism of the video and presented actions.\nName the visual style of the video: whether it is a professional footage, user generated content, some kind of animation, video game or scren content.\nDescribe the visual effects, postprocessing and transitions if they are presented in the video.\nPay attention to the order of key actions shown in the scene.<|im_end|>\n<|im_start|>user\n{}<|im_end|>"
self.clip_l = sd1_clip.SDTokenizer(embedding_directory=embedding_directory, tokenizer_data=tokenizer_data)
def tokenize_with_weights(self, text:str, return_word_ids=False, **kwargs):
out = super().tokenize_with_weights(text, return_word_ids, **kwargs)
out["l"] = self.clip_l.tokenize_with_weights(text, return_word_ids, **kwargs)