We have trained this LORA model using real gesture photos, which supports pure natural language input. Currently, the V1.0 version is still undergoing continuous optimization. We sincerely invite you to provide valuable feedback to help us further improve and enhance the model's performance.