Sign In

ZhRealisticV1

54
414
4
Updated: Oct 20, 2023
stylerealistic
Verified:
SafeTensor
Type
Checkpoint Trained
Stats
414
1,725
Reviews
Published
Oct 20, 2023
Base Model
SD 1.5
Training
Steps: 60,000
Epochs: 30
Usage Tips
Clip Skip: 1
Hash
AutoV2
3ED265EAA7

ZhRealistic 致力于写实模型真实人像的还原,很大程度上优化了AI的磨皮感,欢迎大家尝试。weibo/bilibili:铜锣烧栗子,全平台同名。

模型没有经过融合,非常适合用于继续训练,如果你使用该模型用于训练,请给我一个大大的赞!推荐配合 bad-picture-chill-75v, badhandv4, EasyNegative, ng_deepnegative_v1_75t使用。训练使用的素材均为1024px及以上,请尽量不要使用低于768px的分辨率出图。

对于realistic, photorealistic进行了大量的训练,nikon-d8x0,canon-eos-5d等相机型号都为触发词使用,使用者可以多多发掘其中的玩法!更多的风格以及二次元模型将在sdxl以及未来的版本中更新,敬请期待。

ZhRealistic is dedicated to the true restoration of realistic model portraits, greatly reducing the AI's 'airbrushed' look. Everyone is welcome to try it out. Bad-picture-chill-75v, badhandv4, EasyNegative, ng_deepnegative_v1_75t are recommended. Training materials used are all 1024px and above. Please try not to use images with a resolution lower than 768px.

Find me on Weibo/Bilibili at '铜锣烧栗子', with the same name across all platforms.

The model hasn't undergone fusion and is very suitable for further training. If you use this model for training, please give me a big thumbs up!

We have conducted extensive training on 'realistic' and 'photorealistic'. Camera models such as the Nikon-D8X0 and Canon-EOS-5D have been used as trigger words. Users can explore various ways to play with it! More styles and 2D models will be updated in the SDXL version and future releases. Stay tuned!