Try size: 512 x 768 (SD 1.5) / 768 x 1024 or above (Pony and ILXL)
Weights: 0.8~1.0 (Adjust it according to the basic model you use)
Clip skip: 2
Training material from bing
Pictures from CalicoMix 8.5 and CalicomixPonyXL_v10 and kantanmix_v16LCMTurbo
Note: It’s easy to make mistakes with the girl's arms and body. (SD 1.5)
Note: The "solo" will increase the chance of success, but will affect the effect of nsfw, so recommended to use "male pov". (Pony)
Note: Sometimes the "convex mirror" will be too big and even float in the air, but the probability is not very high, so it has little impact. Because ILXL has specially trained the "traffic mirror" (danbooru tags), so I specially added it this time. (ILXL)