Go to OpenArt main site
Upload workflow

【Update】Comfy Daily | Harry Potter Style Live Newspaper + Single-person poster

5.0

11 reviews
197
23.5K
4.9K
40
Description

ComfyDaily by Hyacinth & Simon Lee


ComfyDaily is a workflow to generate live portraits in Harry Potter style.


Update:

Add a Single-person poster version.


NOTE:

1. This workflow is super memory hugry in 1 run. If you run out of CUDA memory, you could split it into 2 steps. 1st generate the photo, 2nd generate the live portrait.

  • 2. Ollama is not necessary. If you haven't installed it, you can disable that node and write the prompt yourself.
  • 3. You can record any facial expression video to replace face.mp4

  • ————————————————


    INSTRUCTION:


    Base Model - RealVisXL V4.0:

    https://civitai.com/models/139562/realvisxl-v40


    LoRA:

    Harry Potter Style / Uniforms XL Lora:

    https://civitai.com/models/301881/harry-potter-styleuniforms-xl-lora


    American Newspaper front paper:https://civitai.com/models/555597/american-newspaper-front-page


    ————————————————


    Ollama:

    https://ollama.com/

    https://github.com/stavsap/comfyui-ollama


    InstantID:

    https://github.com/cubiq/ComfyUI_InstantID


    PuLID:

    https://github.com/cubiq/PuLID_ComfyUI


    Live Portrait:

    https://github.com/kijai/ComfyUI-LivePortrait


    ——————————————————


    License


    © Since this workflow was created with InstantID and Pulid, which used Insight Face, this workflow is avaiable for non-commercial research purposes only.



    ❤ Have Fun!


    by Hyacinth & Simon Lee

    Discussion

    (No comments yet)

    Loading...

    Author

    3
    18.2K
    658
    77.0K
      Single-person poster.zip (9.6 MB)
      BG_Canvas.jpg (2.4 MB)
      Openart_LiveportraitOnly.json (33.3 kB)
      MASK.png (14.5 kB)

    No reviews yet

    • - latest (4 months ago)

    Primitive Nodes (15)

    Anything Everywhere3 (2)

    DownloadAndLoadLivePortraitModels (1)

    ImageConcatMulti (3)

    LivePortraitProcess (4)

    Note (1)

    PrimitiveNode (3)

    Simple String (1)

    Custom Nodes (59)

    Allor Plugin

    • - ImageCompositeAbsolute (1)

    • - CR Image Grid Panel (2)

    ComfyUI

    • - InvertMask (2)

    • - JoinImageWithAlpha (2)

    • - ImageScaleBy (1)

    • - LoadImage (3)

    • - CheckpointLoaderSimple (1)

    • - ImageScale (1)

    • - ControlNetLoader (1)

    • - DifferentialDiffusion (1)

    • - LatentUpscaleBy (1)

    • - EmptyLatentImage (1)

    • - LoraLoader (2)

    • - VAELoader (1)

    • - CLIPTextEncode (1)

    • - MaskToImage (1)

    • - SaveImage (1)

    • - CLIPSetLastLayer (1)

    • - easy imageScaleDown (3)

    • - easy imageRemBg (1)

    • - ImageCrop+ (4)

    • - InstantIDFaceAnalysis (1)

    • - InstantIDModelLoader (1)

    • - ApplyInstantID (1)

    • - LayerStyle: Stroke V2 (1)

    • - OllamaVision (1)

    • - PrepareImageAndMaskForInpaint (1)

    • - VHS_DuplicateMasks (1)

    • - VHS_VideoCombine (1)

    • - VHS_LoadVideoPath (4)

    • - BatchPromptSchedule (1)

    • - GrowMaskWithBlur (1)

    • - Image To Mask (2)

    • - PulidEvaClipLoader (1)

    • - PulidModelLoader (1)

    • - PulidInsightFaceLoader (1)

    • - ApplyPulid (1)

    • - ImageRGBA2RGB (1)

    • - StyleAlignedBatchAlign (1)

    • - Seed Everywhere (1)

    • - Masks Add (1)

    Checkpoints (1)

    realvisxlV40_v40LightningBakedvae.safetensors

    LoRAs (2)

    aninewspaper-sdxl.safetensors

    harry_potter_v1.safetensors