Version 2.0 UPDATE
Now the workflow has been rejigged and no longer uses the repose lora
The model has been updated to the Qwen Edit 2509 model
The workflow now uses the new model's Control Net features using DW Pose
New Image concat outputs an info image with inputs 1, 2 and the output.
Version 1.0
Using Qwen Edit and a repose lora to repose a reference character to a new pose.
Brie's uploading her Qwen Edot Lazy Repose workflow. Using the power of Qwen Edit and an amazing repose lora by @kingroka , we're able to repose characters to your heart's content.
Inputs:
Character Image - An image of your character, preferably A-pose or T-pose.
Pose Image - An image of a target pose, preferably forward facing without any extreme hairstyles, clothing overhangs and tattoos, which may carry over to the target image.
Outputs:
Output Concatenated Image - An image of both the target pose on the left and the reposed character on the right.
Note: The reason I can not crop it automatically is likely because Qwen Edit can not produce a pixel perfect composition of the concatenated input images. This may be remedies in a future Qwen Model.
The cropping nodes are included if you want to play around with it. If both input images are the exact same size, there's a high likelihood of a successful crop.
Require Resources:
📂 ComfyUI/
├── 📂 models/
│ ├── 📂 diffusion_models/
│ │ └── qwen_image_edit_fp8_e4m3fn.safetensors
│ ├── 📂 loras/
│ │ └── Qwen-Image-Lightning-4steps-V1.0.safetensors
│ │ └── pose_transfer_v2_qwen_edit.safetensors
│ ├── 📂 vae/
│ │ └── qwen_image_vae.safetensors
│ └── 📂 text_encoders/
│ └── qwen_2.5_vl_7b_fp8_scaled.safetensors
Cheese and have a good one!
Brie Wensleydale ~