Pose Selector Flux Kontext Workflow — v1.2 Pose Lora Update

Pose Selector Flux Kontext Workflow is for selecting a pose for a character and producing an output image with Flux.1 Kontext; depending on the variant, it takes a DW Pose image and character image, a Reference Image with a built-in pose, or a full-body character image with a mai

It offers three variants: 1_recomended_pose_lora, classic_pose_lora, and classic_without_pose_lora.

1_recomended_pose_lora uses a new pose LoRA by thedeoxen; classic_pose_lora limits you to baked-in poses; classic_without_pose_lora uses descriptive predefined prompts and baked-in pose images without a LoRA.

For 1_recomended_pose_lora, paste a DW Pose image from the poses folder into the DW Pose Image node and your character image into the Reference Image node.

For classic_pose_lora, use a Reference Image and choose from the built-in pose list; all pose images are baked in.

For classic_without_pose_lora, choose a pose from the list, load a full-body image of your character, and choose an output image size.

For classic_without_pose_lora, choose a pose in the “Choose your pose” node, load a full-body character image, and click run.

The setup names these model files: ae.safetensors, clip_l.safetensors, flux1-kontext-dev-Q5_K_S.gguf, FluxKontext/refcontrol_pose.safetensors, and t5xxl_fp8_e4m3fn_scaled.safetensors.

The setup names these node packs: comfyui-custom-scripts, ComfyUI-GGUF, rgthree-comfy, and teacache.

Consider keeping the reference-image height above 1080 px; changing it may make pose following better or worse.

Suggestion · non vérifié

You can add a pose or output image size by duplicating an existing group whose title contains “pos” or “size”; new poses or sizes then appear in the “Choose your” list.

Review the Civitai page before use because no license is specified for this workflow.

Suggestion · non vérifié

Before running, verify that your environment includes every named model file and node pack.

Suggestion · non vérifié