Back to Model
K

Krea-2 Pose Control-LoRA

Model Nov 3, 2026

Adhit's community Krea-2 Turbo adaptation for image generation conditioned on project-format body-pose skeletons, with a ComfyUI custom-node workflow.

Visit primary link 3 sources attached
Author-published generated samples for Krea-2 Pose Control-LoRA show fantasy characters, seated figures and paired travelers in several visual styles.
Generated samples published by Adhit alongside the English model card and technical post. These are the author’s selected outputs, not an AITrending reproduction or proof that arbitrary poses will match. Source: Adhit / Krea-2 Pose Control-LoRA

What happened lately

Krea-2 Pose Control-LoRA timeline

The project supplies a mix-025 control checkpoint and custom ComfyUI nodes for converting a reference image into a PoseBridge skeleton and generating with Krea-2 Turbo. It uses 17 COCO-style body keypoints, with no direct control over fingers or facial expressions.

Its Hugging Face repository dates to August 26, 2026. A November 2 author write-up explains the training and evaluation, while November 3 repository updates refine ComfyUI workflow selectors and documentation. Reported adherence results are the author’s own evaluation on selected training images, and do not establish general performance across arbitrary poses.