v2

Audio Reactive Group

This took some work. What I did was use the audio reactive nodes and the prompt scheduling with Lora word triggers. Using Loras and letting the audio influence the trigger word strength through the pw a and pw b inputs.

This isn't permanent, still working on a better solution, but it works for now.

To use regular prompts set the conditioning switches to true and bypass the audio reactive group.

To use Audio Reactive Lora prompts, un-bypass the Audio Reactive Group, set the conditioning switches to false, set your lora strength between 1 and .80 and set denoising strength to around .20 to .40. This will give you the best results from my testing so far.

Updated video upload, converted IP adapter to tile IP adapter, downgraded ksampler, added a means of influencing audio for now. Some slight improvements.

v1

Corrected controlnet connections

AnimateDiff

SVD

Controlnets

Loras

IP Adapter

Cascade

Text2Img

Text2Vid

Vid2Vid

Img2Vid

Img2Img

Canvas (Paint Node)

Layers

Remove Background

Upscale

Film

Group Bypass

Input Switches

Labels and Notes with links