Regarding everyone’s feedback on the inconsistent proportions of the model’s body: The sample picture is 1024*2048, which is a fashion display photo, mainly to highlight the specific cut size of the clothes on the model.If you don't like this size, you can set it to the default size, such as -512*768, 768*1024. It is recommended to use controlnet to fix the finger problem.Recommended number of steps: 30-150 (30-50 is basically enough, unless you need to add more details on the clothing, you can open the large tag to adjust)Sampling recommendation: spm++sde karras (I usually use this, you can test the others by yourself)Recommended magnification algorithm: 4x-ultrasharp (redraw amplitude is about 0.2, magnification factor is generally 2)Recommended size: 512x1024*2 (if you mind if your person is too long, you can use: 512*768/768*1024) cfg recommendation: 2-11 (if you are looking for some distinctive clothing or effects, 2-3.5 is recommended)There isn’t much to share about controlnet.Maybe it would be easier to understand by watching a video tutorial.Version 1.0 is still very good, but version 2.0 is equally excellent, so choose yourself! E-commerce or other commercial graphics can be produced using post-processing software, so I won’t introduce them here.

JAY-Photographic Realism

jayPhotographic_v10.safetensors

v1.0

步数推荐：30-150（30-50基本够用，除非需要增加更多服饰上的细节可以开大tag调整）采样推荐：spm++sdekarras（我一般用这个，其它的可以自行测试一下）放大算法推荐：4x-ultrasharp（重绘幅度0.2左右，放大倍数一般为2）尺寸推荐：512x1024*2（介意人太长的可以用：512*768/768*1024）cfg推荐：2-11（如果是抽一些比较有特色的服装或者效果推荐2-3.5即可）

模特

步数推荐：30-150（30-50基本够用，除非需要增加更多服饰上的细节可以开大tag调整）
采样推荐：spm++sde karras（我一般用这个，其它的可以自行测试一下）
放大算法推荐：4x-ultrasharp（重绘幅度0.2左右，放大倍数一般为2）
尺寸推荐：512x1024*2（介意人太长的可以用：512*768/768*1024）
cfg推荐：2-11（如果是抽一些比较有特色的服装或者效果推荐2-3.5即可）

Files with SHA256: a57d93930acc8e7af4e91f6a356673027ef5c585cdc1af1f120e2164104b8830