Merge versions of the Isegye Idol.

Each version may contain two, three, or more members of the group, and it's still a work in progress.

I realized my limitations when I tried to merge six members into a single LoRA. I don't have enough knowledge about merging, and each dataset contained many of the same tags. Therefore, a LoRA that can represent all six members will not be released anytime soon.

Purpose: To generate images of interactions between members based on the ANIMA model's natural language understanding capabilities without style bias caused by using multiple LoRAs.

Limitations: Costume details are less accurate than in the individual LoRAs. If interactions are not needed, the individual LoRAs are recommended instead.

Goal: Creating merged LoRAs for diverse member combinations. My first priority is Unniz and Dongsaengz, each containing three members. I thought separating the members into these groups would minimize overlapping tags between the datasets. Additional versions are also planned, including Gaenariz, Hentaiz, and costume-themed sets such as SFY_Unniz, SFY_Dongsaengz, and more.

Tips: You can write natural language prompts based on trigger words, like "A girl with grey hair and purple eyes…"