The model obtained by training with a few simply selected pictures may have average effect.
vrc style_ani4 xl_v1-000010.safetensors