Frame-by-Frame Controllable Image-to-Video Workflow by PIA and AnimateDiff: Make Any Image Blink!

Using the PIA (Personalized Image Animator) model in AnimateDiff allows for controllable image-to-video creation, providing stable and high-quality results.

In this video, I’ll demonstrate how to use the PIA model in ComfyUI, combined with Prompt Travel in AnimateDiff, to make the character in your input image blink at specified frames. Additionally, I’ll show you how to use IC-Light to re-light the scene, avoiding the issue of brightness fluctuation caused by PIA.

CustomNodes involved:

model path: ComfyUI\custom_nodes\ComfyUI-AnimateDiff-Evolved\models

[denisdodo]

your videos always LIKE, tnx a lot for your work

[aivrar]

Thank you tons, this is the only comfy PIA tut I can find on YT.

[thibaudherbert]

is this the best way to create videos ? I don't know there are so many models like animatedif, animate anyone, mimic motion, liveportrait....