用了刘*菲，刘*诗，全*贤，宋*乔做的女性lora，希望大家喜欢tip1:注意步数再20到28 即可，过低的步数出图会扭曲，过高的步数会出现过拟合，tip2:然后是lora的权重最好再0.6~0.9之间，太低出图就不像中国人了，太高就很像原图没有随机性，可以参考我做的图的参数最后，记得点赞，点赞，点赞呀，This is a message about a female Lora created using Liu*Fei, Liu*Shi, Quan*Xian, and Song*Qiao. The author hopes that everyone will like it.Tip 1: Pay attention to the number of steps between 20 and 28. If the number of steps is too low, the image will be distorted, and if it is too high, overfitting may occur.Tip 2: The weight of Lora should ideally be between 0.6 and 0.9. If it's too low, the image won't look like a Chinese person, and if it's too high, it will look too much like the original image without randomness. You can refer to the parameters of the image I made.Finally, remember to like, like, like!补充：我不知道上传的图片为什么没有原始数据我补充一下作为参考best quality, ultra high res, (photorealistic:1.4), masterpiece, Kodak portra 400, film grain,blurry background ,indoors,1girl, sleeveless turtleneck, (busty), ((huge breasts)), large breasts, (brown long hair:1.3), (looking at viewer), (full body), (closeup) &lt;lora:Chinese doll idol 鹅蛋脸 (oval-shaped face):0.66&gt;Negative prompt: NSFW,paintings, glossy skin, wet, sketches, (worst quality:2), (low quality:2), (normal quality:2), lowres, normal quality, ((monochrome)), ((grayscale)), skin spots, acnes, skin blemishes, age spot, glansSteps: 28, Sampler: DPM++ SDE Karras, CFG scale: 7, Seed: 2090486726, Face restoration: CodeFormer, Size: 512x768, Model hash: fc2511737a, Model: chilloutmix_NiPrunedFp32Fix, Clip skip: 2, ENSD: 31337补充：对不起，各位，我不知道vae和采样方法的重要性，所以没写，现在补上。、大模型：chilloutmixvae:vae-ft-mse-840000-ema-pruned采样方法：DDIMHere's the translation:Sorry everyone, I didn't mention the importance of VAE and sampling methods earlier, so I'm adding them now.The name of the large model is "chilloutmix", and it uses the VAE model "vae-ft-mse-840000-ema-pruned" with the sampling method "DDIM".

Files with SHA256: c9a33a403208ac1d6534064ac04421f6095a732731e8832dfad150bfb2c7cd7e