不知不觉，玩AI也一年了，认识了很多朋友，得到了很多大佬的帮助，特别要感谢DukeG大佬的技术指导和帮助，跟着学到了很多，但是我比较懒，懒得投入学习，学的不精。 本意想练个龙年祝福的模型，sd1.5和sdxl都练了，可惜自己原因，加上临到春节才练，模型实际效果并不好，绷脸、手残，斗鸡眼都出来了，发出来纯属是纪念一下，介意者勿下。推荐权重0.6-0.64，垃圾模型采样方法: 2m karras 或 sde karras， 高清设置 放大算法: 4x-ultrasharp 重绘幅度: &lt;0.4 cfg scale: 7 全身图绷脸，手崩的概率大，你可以通过使用upperbody来控制出图，避免脸崩====================================================== 版权说明：我的模型训练使用sd生图进行训练，所上传的模型成品在本人未允许的情况下不得用于数字模特、转载及搬运等商业用途。如果在商业用途中使用了该模型，那么一切风险与责任追究与本人无关，本人不承担任何法律 感谢DukeG大佬的技术支持和指导!Without realizing it, I've been playing with AI for a year now, getting to know many friends and receiving a lot of help from some big names. I especially want to thank DukeG for his technical guidance and help, from which I've learned a lot. However, I'm quite lazy and reluctant to devote myself to learning, so my knowledge isn't very deep.My original intention was to train a model for Lunar New Year blessings using both SD1.5 and SDXL, but unfortunately, due to my own reasons and starting just before the Spring Festival, the actual performance of the model wasn't good, resulting in distorted faces, awkward hands, and cross-eyed effects. Sharing these results is purely for commemoration. Those who mind, please do not download.Recommended weights are 0.6-0.64, for a model that's not great.Sampling method: 2m karras or sde karras,High-definition settings Upscaling algorithm: 4x-ultrasharpRedraw magnitude: &lt;0.4 CFG scale: 7There's a high probability of distorted faces and ruined hands in full-body images, but you can control the output by using "upperbody" to avoid facial distortions.======================================================Copyright notice: My model training uses raw images from SD, and the finished models uploaded are not to be used for digital models, reproduction, or any commercial purposes without my permission. If the model is used for commercial purposes, all risks and responsibilities are disassociated from me, and I will not bear any legal responsibilities.Special thanks to DukeG for his technical support and guidance!

Files with SHA256: 4750442254a50fe6f7bb5c5bd34b0e2dce5b695ff6c7a11e712da9c2f59904ce