Marimo
Another LoRA for the masses, trained on sprites and several works, refer to gallery images to see best practices, hope you enjoy!UPDATE 12/03/2025 ILLUSTRIOUSIllustrious support added, along with the experimentation of higher accuracy FP training using Float 32. This appears to have massively improved stability, accuracy and consistency. However it takes much longer to train. This model is MUCH better compatible with other base models and seems to have no visible instability issues. However outfit comprehension is overall lower. This mode of training may be more suitable for outfits specifically and will experiment with upcoming outfit models.Outfit supportUN Combat Uniform: Combat BDU, blue combat shirt, blue combat pants, beltUN Uniform: UN Uniform, formal uniformNegs: white uniformGreen uniform: Green Formal UniformFortified: Fortified Suit, green bodysuitOriginal PonyAnother rundown of the capabilities of this model...Well trained for creating memorable outfits from her VNs.Her Civilian Sweater, UN Military Outfit and Fortified SuitUnfortunately the model was trained equally on everything, so the UN BDUs are difficult to make, and making her green JSDF uniform is difficult as it tends to lean towards the UN uniforms. However as always the main spirit of the character is preserved!Let me know who you want me to make next!I will retrain this model as I'm still new to making SDXL LoRAs if there is demand for it.
