Image2Image Lora Dataset Generator Workflow

[Quantum]

Lora Dataset Generator Workflow

This workflow uses Qwen Image Edit Nunchaku + Control Net to create multiple poses from one input image.

Model:
Qwen Image Edit Nunchaku - https://huggingface.co/nunchaku-ai/nunch...afetensors
Text Encoder - https://huggingface.co/Comfy-Org/Qwen-Im...t_encoders
VAE - https://huggingface.co/Comfy-Org/Qwen-Im..._files/vae
Control Net - https://huggingface.co/Comfy-Org/Qwen-Im...afetensors

Download Workflow: Click Here

[khalil mahbouli]

thanks for sharing

[review]

Total Disk Space Required for above models = ~ 33 GB

[Dwayson_Rockham]

WHAT . DOES . THIS . THING ?!?!?!

And what path is needed here?!
"Load Images For Loop"-NODE

if its blank the wf will not start.

[Quantum]

WHAT . DOES . THIS . THING ?!?!?!

And what path is needed here?!
"Load Images For Loop"-NODE

if its blank the wf will not start.

You need pose images to use this workflow. You can use this workflow to create pose images: https://diffused.to/Thread-Image2Image-C...r-Workflow
In this workflow put the folder path where you have the normal images to create pose images.

In the Qwen edit workflow:
"Load Images For Loop"- In this node you need to put the folder path where you have all the pose images. The above workflow saves the pose images in output/pose folder so put C://your/comfyui/path/output/pose in this node.

[comfycoffee03]

ksampler getting a dtype error.

[xocalbee]

thanks for sharing

[spencer8]

Thank you very much for sharing that workflow!