Back, trying out wan 2.2 loras. All my SD 1.5 loras are trained on and generated with EveryDayArt v5. They don't transfer well, so use that base model if you wish to have similar output