(Position to be spanked :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.

更新日志： V3: 输入一张单个明确主体的图像，即可给出该主体旋转不同方位后的图像（九宫格图*1张） V2: 输入一张单个明确主体的图像（工作流自动扣除背景），不需其它参数，即可给出该主体旋转不同方位后的图像（*6张） V1: 输入一张单个明确主体的简单背景图像，和仰角/方位角（0.1±180.0），即可给出该主体旋转指定方位后的图像（1张） Update log: V3: Input an image of a single clear subject, and you can get the image of the subject after rotating to different positions (nine-grid image * 1 picture) V2: Input an image of a single clear subject (the workflow automatically deducts the background), and you can get the image of the subject after rotating to different positions without other parameters (*6 pictures) V1: Input a simple background image of a single clear subject, and the elevation/azimuth angle (0.1±180.0), and you can get the image of the subject after rotating to a specified position (1 picture)