I just got into lora training, did a crude and quick one and holy shit it seems promising.
I want to train one to generate photorealistic stuff of my ex based on pics + videos(that I will extract snapshots of) OC that I have of her.
Any tips to optimize this and extract the best from the data?
