FLUX 1.0

• New dataset captured from the PC version of the game.

• Using full caption training.

• Weight: 0.5-0.8

SDXL V1.0

Base model: SDXL

This is my first experimental Lora training with SDXL, the dataset is almost the same as the 1.0 version, due to the low-resolution dataset, it may have some weird body ratio issue when directly generating the image above the resolution of 1280, but the hires.fx upscale method is not affected.

V1.0 (SD1.5)

Jote from Final Fantasy XVI

Base model: Chilloutmix-Ni

Weight: 0.6-0.9, highly recommend using it with Adetailer and hires.fx

The dataset of this character is very small compared to my other FF16 model, so the face will have some problems at some viewing angles.

Any comments or suggestions are welcome.