v5





Kyedae (Twitch Streamer) LoRa
Trained on 768x768 Images, Realistic Vision 2.0
Used on Vlad Diffusion
optimal settings to get best results :
LoRa weight : 0.9 to 1
Samples : 25-30
CFG : 6-7
512x768 + high res fix with 1.5x upscale. with either 8x NMKD Superscale 150000 or 4x Ultra Sharp
It works well with realistic models, and comic/cartoon ones.
Prompt examples
Positive - photo of Kyedae1 woman, wearing a blue denim jacket, and black pants, extremely detailed, intricate detail, fine detail, careful detail, up close portrait, masterpiece, best quality, ultra-detailed <lora:Kyedaev5Lora:1.0>
Negative - Deformed, extra limbs, extra eyes, extra noses, extra ears, ugly, childlike, blurry face, nose ring, artifacts, bad quality, bad ???????, jewelry,, worst quality, jpeg artifacts, signature, watermark, username, blurry, artist name, bad feet, big head, fused body, twisted,
v5
1. 转载模型仅供学习与交流分享,其版权及最终解释权归原作者。
2. 模型原作者如需认领模型,请通过官方渠道联系海艺AI工作人员进行认证。我们致力于保护每一位创作者的权益。 点击去认领
