kimVixen

Model Overview

The Vixen series Flux model is specifically designed for female subjects. Due to the nature of Flux, I didn’t use an extensive number of images to create it. Using LoKR, I trained both the UNet and text encoder to preserve as much diversity as possible.

To achieve the ideal poses and realistic style, I used a finely curated dataset for training.

This model is intended for generating realistic, photography-style images, and I will be using it as the base model for all future Flux LoRA generation.

---

Recommended Parameters:

- Sampler Method: Euler simple

- Sampling Steps: 20-30

- Recommended Resolution: 1024x1536 or 1280x2048

- CFG: 3.0

======================================================

模型简介

Vixen系列的Flux模型是女性类专用模型，由于flux的特性我并没有用特别多的图片去制作它。使用了lokr同时练了unet和text尽可能的保留了更多的多样性。

为了理想中的人物姿态和写实类风格，使用精筛集训练。

它可以用来制作真人摄影写真方向的图片生成，我将用它来配对之后的所有flux的lora生图底模。

---

推荐参数：

采样方法(Sampler)： Euler simple

采样迭代步数(Steps)：20-30

推荐尺寸：1024*1536 或 1280*2048

CFG：3.0

Q交流群： 852450267