Request: ComfyUI workflow + settings for “cowboy singing with guitar” MultiTalk demo

[zacdagostino]

Hi,

First of all, thanks for the ComfyUI integration for Wan/MultiTalk – it’s awesome and super useful.

I’m trying to reproduce the specific “cowboy singing with a guitar” example shown in the MeiGen-MultiTalk demo page using your ComfyUI workflow.

Would you be willing to share the exact ComfyUI workflow JSON and settings you used for that clip? Specifically:

The ComfyUI JSON workflow file for that cowboy-with-guitar example (either attached directly or via a Gist / link).

The core generation settings, including:

Model / checkpoint(s) used

Sampler + steps

CFG / guidance scales (text + audio)

Resolution, FPS, and clip length (frames/seconds)

Seed / any fixed seeds you used

Any LoRA / VAE / extra models or special nodes you used (e.g. pre-processing nodes, motion / face refinement, post-processing, etc.).

My goal is to understand your exact configuration and replicate that specific result, not just get something “similar”. I’m happy to credit you and link back to the repo if I ever share results or a write-up.

If it’s easier, a single exported .json ComfyUI workflow file with a short note like “this is the cowboy singing workflow from the demo” would be perfect.

Thanks a lot for your time and for maintaining this wrapper!