Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
ninjaneural
GitHub Repository: ninjaneural/webui
Path: blob/master/memo/comfyui_anidiff_lora.md
3275 views

영상링크

https://youtu.be/MCG_PJyazhs

워크플로우

  • 시작 워크플로우

https://raw.githubusercontent.com/ninjaneural/webui/master/memo/comfyui_anidiff_lora/workflow1.json

  • 기본 워크플로우 (페이스 디테일 없이)

https://raw.githubusercontent.com/ninjaneural/webui/master/memo/comfyui_anidiff_lora/workflow.json


  • FaceDetailer (안정성이 떨어짐)

https://raw.githubusercontent.com/ninjaneural/webui/master/memo/comfyui_anidiff_lora/workflow2.json

  • FaceDetailer대신 SimpleDetector(AnimateDiff) + Detailer (추천)

https://raw.githubusercontent.com/ninjaneural/webui/master/memo/comfyui_anidiff_lora/workflow3.json


AnimateDiff LCM 적용 beta schedule LCM이 추가되면서 순서가 조금 바꼈어요 Content Options Standard Static으로 변경했어요

  • FaceDetailer

https://raw.githubusercontent.com/ninjaneural/webui/master/memo/comfyui_anidiff_lora/workflow2-1.json

  • FaceDetailer대신 SimpleDetector + Detailer (AnimateDiff)

https://raw.githubusercontent.com/ninjaneural/webui/master/memo/comfyui_anidiff_lora/workflow3-1.json

  • denoise 1.0 / lora / openpose / ip2p (전체적으로 변경할때 예시)

https://raw.githubusercontent.com/ninjaneural/webui/master/memo/comfyui_anidiff_lora/workflow5.json

(마우스 오른쪽버튼을 누르고 링크 저장을 눌러주세요)

사용된 커스텀노드

https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite

비디오 저장

https://github.com/ltdrdata/ComfyUI-Impact-Pack

디테일러

https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved

AnimateDiff

https://github.com/Kosinkadink/ComfyUI-Advanced-ControlNet

AnimateDiff + ControlNet

https://github.com/Fannovel16/comfyui_controlnet_aux

ControlNet Preprocessors

LCM-LoRA 모델 다운로드

https://huggingface.co/latent-consistency/lcm-lora-sdv1-5/resolve/main/pytorch_lora_weights.safetensors

- lcm-lora-sdv1-5.safetensors - models/loras

AnimateDiff 모델 (선택)

https://huggingface.co/CiaraRowles/TemporalDiff/resolve/main/temporaldiff-v1-animatediff.ckpt

- 다른 모델과 큰 차이는 없는거 같아요 - custom_nodes/ComfyUI-AnimateDiff-Evolved/models 이곳에 저장

SDXL용

https://huggingface.co/latent-consistency/lcm-lora-sdxl/resolve/main/pytorch_lora_weights.safetensors

- lcm-lora-sdxl.safetensors - models/loras

프롬프트 정보

긍정 프롬프트

(masterpiece, best quality, high quality:1.2), 8k, extremely detailed, girl, cute, (smile:0.7)"

부정 프롬프트

(worst quality, low quality:1.2), (makeup, mole), (nsfw, nude)

ControlNet 모델

https://huggingface.co/lllyasviel/ControlNet-v1-1/tree/main (오피셜)

https://huggingface.co/comfyanonymous/ControlNet-v1-1_fp16_safetensors/tree/main (fp16버전:저용량)

다운받아서 models/controlnet 에 저장

ComfyUI-VideoHelperSuite 노드가 나타나지 않는경우

  • 실행할때 아래 메세지가 나오는지 확인해보세요~

no module named 'cv2'
  1. Manger 이용 방법

Manager 메뉴에서 Install PIP Packages 를 누르고 opencv-python 를 입력해서 설치해보세요 (설치 완료까지 대기후 재시작)

  1. 직접설치 방법

윈도우의 경우 아래처럼 "명령 프롬프트"에서 comfyui 설치폴더로 이동후 opencv-python을 직접 설치해주세요

python_embeded\python -m pip install opencv-python