Translated data: Researchers from ETH Zurich in Switzerland and Google Zurich have jointly proposed a new technology called InseRF, which successfully generates consistent objects in 3D scenes through text prompts and 2D bounding boxes, without the need for explicit 3D information. Experiments have shown that InseRF outperforms existing methods, capable of inserting consistent objects into NeRF. Despite limitations imposed by the underlying model's capabilities, the team plans to further optimize InseRF's performance through future model improvements.