Type | |
Stats | 503 |
Reviews | (79) |
Published | Nov 28, 2023 |
Base Model | |
Usage Tips | Clip Skip: 2 |
Trigger Words | meidusha,red dress |
Hash | AutoV2 A5377B254D |
perfectWorld_v6为底模训练，推荐用这个生成图片
全身照用ADetailer脸部修复
禁止用于商业用途！！
禁止出售此模型生成的图像！！
禁止出售此模型或合并！！
模型：perfectWorld_v6Baked
采样：dpm++ 2M Karras
步数：28
分辨率：512*768
高清：4xUltrasharp_4xUltrasharpV10 0步 重绘幅度0.4
CFG：7
此lora可生成2.5D和3D风格，取决于正反画质提示词
如想生成3D风格，推荐使用Lora:more_details，C站就有
本人图像生成所用正反提示词lora，本站都可搜到
Lora权重建议0.5-0.9，如服装不像可拉高权重，但同时背景会过拟合，还需自行斟酌
最后欢迎返图和建议
perfectWorld_v6 is used for base model training. It is recommended to use this to generate images.
Full body photo with ADetailer facial restoration
Commercial use prohibited! !
The sale of images generated by this model is prohibited! !
No sale or merging of this model is allowed! !
Model: perfectWorld_v6Baked
Sampling: dpm++ 2M Karras
Number of steps: 28
Resolution: 512*768
HD: 4xUltrasharp_4xUltrasharpV10 0 steps redraw range 0.4
CFG: 7
This lora can generate 2.5D and 3D styles, depending on the front and rear image quality prompt words
If you want to generate a 3D style, it is recommended to use Lora:more_details, which is available at Station C
The positive and negative prompt words lora used in my image generation can be found on this website.
The Lora weight is recommended to be 0.5-0.9. If the clothing does not look like it, the weight can be increased, but at the same time the background will be over-fitted, so you need to consider it yourself.
Finally, feedback and suggestions are welcome