🏢 University of Bologna

LLaNA: A novel Multimodal Large Language Model directly processes NeRF weights to enable NeRF captioning and Q&A, outperforming traditional 2D/3D-based methods.