(This template is used for Workflow Contest)

What this workflow does

👉 Output stable video based on pictures

How to use this workflow

👉 [Please add here]

Video to video or picture to video. Select one to hide the other. 
Filter the samplers you need and hide the samplers you don't need. 
Adjust parameters and run!

Tips about this workflow

👉 [Please add here]

Global seed:

Control seeds for all samplers

Switch method: Select the node and press Ctrl M




Diff sampling:
Reduce size, redraw using AnimateDiff

HD sampling full:
Original size, redraw the entire image

HD sampling person:
Original size, only characters redrawn

Refinement sampling face:
Original size, refined face




Image Source:

1-Picture directory
2-Video frame breaking








Latent type:

1-EmptyLatent
2-Image to Latent









HD sampling model:

1-Use initial model
2-Use additional models








Tagger reverses:


Right click and select WD14 Tagger

🎥 Video demo link (optional)

👉 [Please add here]