(Plus Size: 1

SeaArt 官方模型加载器已经支持FLUX标签和筛选，所以之前对于怎么找flux checkpoint和lora的内容就不再赘述。 目前flux是支持多lora的，但为避免应用界面过于复杂，我仅开放了一个lora位，另外默认加载了FaeTastic Details和mjV6，权重均为1。 单图的算力消耗约30PT，2-4图则会降低单图消耗，最低能到19PT每张图。建议增加出图的batch size获得最高的性价比。 SeaArt's official model loader already supports FLUX tags and filtering, so I won't go into detail about how to find flux checkpoints and lora before. At present, flux supports multiple lora, but in order to avoid the application interface being too complicated, I only open one lora bit, and the FaeTastic Details and mjV6 are loaded by default, with a weight of 1. The computing power consumption of a single image is about 30PT, and the consumption of a single image will be reduced by 2-4 images, and the minimum can be 19PT per image. It is recommended to increase the batch size of the plot to obtain the highest cost performance.

上传一张动物图片，填写生成文本，生成对应内容的动物图片 （使用说明：第一个选项categories是蒙版词汇，可以精准定位如鸭子。weight是参考图片的权重，数值越高相似度越高，但对画面的想象空间会降低，text第一个是正向提示词，第二个是负面提示词。图片尺寸默认方形1024x1024并且1.5倍超分。）

Based on Wan 2.1 Model animation video generation tool。This process can be used to combine static images with action reference videos，Achieves smooth video transition through RIFE VFI technology，Supports pose key point detection and rendering，Suitable for creative short film production and visual effects scenarios。Provides high-quality output and customizable size、Frame rate and other parameter settings。