The training were based on game assets from swf file and used svg to png conversion to train 768x768 images.
msa2.safetensors