Works fine with all sd 1.5 models. Trained on this
strength 0.6-0.8, hiresfix or img2img for best results(it's works for every LoRa).
Train models isn't easy, crypto donations on some coffee are welcome:
Boosty, subscribtion with additional content: