Rogue Marvel Rivals. It's a pretty versatile lora and it works with realistic checkpoints as well. The best results I had between 0.6 and 0.8, but you are free to experiment.
UPDATE!
I experimented with data training again, just like with the Lady Loki model. V1 is the closest to the game model and V3 has some modifications in the style, while V2 is in the middle.
