Stable-Diffusion Openpose

Stable Diffusion 목차

 

https://github.com/fkunn1326/openpose-editor

Openpose를 설치합니다.

 

 

 

 

 

 

 

https://huggingface.co/ckpt/ControlNet-v1-1/tree/main

control_v11e_sd15_ip2p_fp16.safetensors 을 다운받아 extensions/sd-webui-controlnet/models에 넣어줍니다.

 

 

 

 

 

 

 

 

 

 

 

https://huggingface.co/lllyasviel/ControlNet/tree/main/models

control_sd15_openpose.pth을 다운받아 extensions/sd-webui-controlnet/models에 넣어줍니다.

 

 

 

 

서버를 재구동해서 OpenPose Editor를 확인 합니다.

 

 

 

Positive:
wearing jeans in the concert dome, with white cardigan,full body, (wide hips:1.3), (thigh gap:1.2), ( (intricate details:1.2), (face, mature female:1.2),Depth of field, Cinematic Light, Lens Flare, Ray tracing, perspective,RAW, (1girl), detailed skin, sharp focus, delicate, korean, instagram, pop idol, korean beauty, korean mixed, best quality, extremely fine and beautiful, best illumination, semi realistic, aegyo sal, illustration, adult, pale white skin,  cinematic lightings, soft volumetric lighting, subsurface scattering,(night photography:1.3), (solo:1.2), solo focus, (crowd:1.2),blurry background, (walking:1.3),(bangs:1.3), (looking away, long hair,:1.3)

Negative:
(low quality:1.5), (worst quality:1.5), text, nsfw, nude, naked, underwear, innerwear, many people

 

이미지를 생성합니다.

 

 

 

 

포즈를 수정해서 Send to txt2img 합니다.

 

 

 

 

 

기대치와 다르지만..-_-

 

 

 

 

이미지 사이즈를 맞춰줍니다.

 

 

 

 

 

 

 

모델을 변경하니 기대치에 조금더 가깝게 나오네요.

 

 

 

 

 

 

 

 

 

 

 

 

Stable Diffusion 목차