comfyui-singlepose-clone

VNCCS Simple Clone — Single-Pose (QWEN)

Standalone bundle for the single-pose character clone workflow built on Qwen-Image-Edit. Upload this folder to a git URL and point the compose's SIMPLECLONE_REPO at it.

Contents

__init__.py                      # custom node: VNCCS_SimplePose (+ web editor)
web/simple_pose_editor.js        # 2D/3D pose editor + OpenPose JSON import + PNG passthrough
workflows/VN_Clone_SinglePose.json

VNCCS_SimplePose renders one OpenPose control image (512×1536) from an editable pose. It can also pass through a ready-made OpenPose PNG (optional openpose_image input). The 3D editor reuses ComfyUI_VNCCS's Pose3DEditor when present.

Dependencies (handled by dc-singlepose-qwen.yml)

Custom nodes:

• ComfyUI-GGUF (UnetLoaderGGUF)
• ComfyUI_VNCCS (VNCCS_QWEN_Encoder + 3D pose editor)
• this bundle (VNCCS_SimplePose)

Models:

• unet/qwen-image-edit-2511-Q3_K_M.gguf
• text_encoders/qwen_2.5_vl_7b_fp8_scaled.safetensors
• vae/qwen_image_vae.safetensors
• loras/qwen/Qwen-Image-Edit-2511-Lightning-4steps-V1.0-bf16.safetensors
• loras/qwen/VNCCS/poser_helper_v2_000004200.safetensors

Use

1. docker compose -f dc-singlepose-qwen.yml up -d
2. Open http://localhost:8188 → load VN_Clone_SinglePose.
3. LoadImage = your character → VNCCS Simple Pose (Open Pose Editor) → edit the positive prompt for custom instructions → Queue.

Tuned for an 8GB GPU. First run after boot may OOM at the sampler (cold start) — just re-queue.