A fine-tuned LoRA model of camera movement task