(Intimate details: 1 - เครื่องมือ AI ด่วน

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.

Xi_rel_260420