>>105738232
it's not a perfect model by any means but SD 3.5 (referring to Medium here, never played with Large much, not sure how much it compares) is still the only base model that actually looks like it was trained on real photos, as opposed to Flux gens that were captioned as photos kek

this one was just `a candid amateur selfie photo of a gorgeous young white woman in a park`. No other post-SDXL base model will ever produce anything that looks remotely like this, under any circumstances, no matter what you do