Nodes/ComfyUI-Tween/Load SGM-VFI Model
ComfyUI Node

Load SGM-VFI Model

A ComfyUI node in video/SGM-VFI with 3 inputs and 1 output.

By ethanfel·Created 6 months ago·Updated 10 days ago· 14
Load SGM-VFI Model
    • model
    model_pathours-1-2-points.pkl
    ttafalse
    num_key_points0.50
    Categoryvideo/SGM-VFI

    Inputs (3)

    NameTypeDefaultDescription
    model_pathCOMBOours-1-2-points.pklCheckpoint file from models/sgm-vfi/. Auto-downloads on first use if missing. Variant (base/small) is auto-detected from filename.
    ttaBOOLEANfalseTest-time augmentation: flip input and average with unflipped result. ~2x slower but slightly better quality.
    num_key_pointsFLOAT0.500–1Sparsity of global matching. 0.0 = global matching everywhere (slower, better for large motion). Higher = sparser keypoints (faster). Default 0.5 is a good balance.

    Outputs (1)

    NameTypeDescription
    modelSGM_VFI_MODEL