another badly trained lora

v3 latest version

v1 has more fanart of him in dataset

v0.8 is same as above but at 8th epoch

v2 has more game screenshots in dataset and lower learning rate and a failed experiment

all bad xd

trained on NAI

prompts like teamred/teamblu, red jacket/blue jacket can help change the team color of soldier

helmet, red/blue jacket, bandolier, two grenades, covered eyes, on prompts can help