s1x7e/krea2-pose-controlnet

Krea 2 Turbo — OpenPose ControlNet LoRA Demo

Generate characters that follow a body pose while your prompt controls appearance, clothing, style, and scene.

Base model: krea/Krea-2-Turbo Pose LoRA: thedeoxen/Krea-2-pose-controlnet

Status: Full Gradio + ZeroGPU app is in this repo (app.py). The Space is temporarily Static while a community GPU grant is requested (free accounts need PRO or a grant to host Gradio/ZeroGPU). Once approved, frontmatter switches to sdk: gradio and the interactive demo goes live.

Two input modes (when Gradio is live)

1. 1.Pose Image — upload a ready-made OpenPose / DWPose skeleton (black background works best).
2. 2.Regular Image — upload a photo; DWPose extracts the skeleton, then you can drag joints before generating.

Tips

• —Recommended LoRA weight: 0.8–1.0
• —Typical settings: ~10 steps, guidance 0.0 (Turbo distilled)
• —DWPose skeletons on a black background give the strongest pose adherence
• —Lower LoRA scale (~0.6–0.8) if the pose feels too rigid

Example prompt

A high-resolution render of a futuristic cyber-ninja in a low-profile stealth pose. The suit is matte black with subtle blue and red LED accents. It holds two glowing plasma katanas low to the ground. The character is balanced on a high-rise ledge overlooking a rain-slicked cyberpunk megacity at night. Style: Cyberpunk aesthetics, realistic rendering, dramatic lighting.

No special trigger phrase is required — the pose comes from the control image.