Version 1.1 I noticed that when generating images with low resolution, the eyes may look a bit off depending on the anime model being used, especially when mixing it with another LORA. Therefore, I created this new version by reducing the number of epochs and tweaking the data. It should work better when mixing with other Lora or generating lower resolution image. To trigger Alisa with custom clothing, is the same tag and strengh from version 1 To trigger Alisa with school uniform you can also used the tag and strenght from version 1 but you get better result by adding white jacket, (((schoolUniform))) If you can't decide which Lora to pick, just try them both. They are both good in their own way 🤷‍♂️ Version 1: There are two custom tags: "Alisa Kujou" (Yes without Mikhailovana) which represents Alisa in general, and "schoolUniform" which represents Alisa wearing a school uniform. To trigger Alisa with custom clothing, set the strength to around 0.5-0.7 and use the following tags: Alisa Kujou, white hair, blue eyes, and optionally "red hair ribbon". To trigger Alisa with a school uniform, set the strength to 1 and use the following tags: Alisa Kujou, schoolUniform, blue eyes, shirt, bow, ribbon, jacket, hair ribbon, white hair, bowtie, red bow, red ribbon, red bowtie, and white thigh-highs. The models used for the showcase are: "Meinahentai_21", "meinamix_meinaV8", and "anythingV4.5".

Alisa Mikhailovna Kujou -Roshidere

Better eyes performance and better with lora

Files with SHA256: 1cb1842e9c456987f87211aa546d120a0196583703e398d877dd59a9fe5877ea