LLaNA: Large Language and NeRF Assistant
This work addresses the challenge of multimodal AI understanding 3D scenes for researchers and developers, though it is incremental as it builds on existing NeRF and MLLM technologies.
The paper tackled the problem of integrating Neural Radiance Fields (NeRFs) into multimodal large language models (MLLMs) to capture object appearance and geometry holistically, resulting in LLaNA, a general-purpose assistant that achieves favorable performance in tasks like NeRF captioning and Q&A by directly processing NeRF weights without rendering images.
Multimodal Large Language Models (MLLMs) have demonstrated an excellent understanding of images and 3D data. However, both modalities have shortcomings in holistically capturing the appearance and geometry of objects. Meanwhile, Neural Radiance Fields (NeRFs), which encode information within the weights of a simple Multi-Layer Perceptron (MLP), have emerged as an increasingly widespread modality that simultaneously encodes the geometry and photorealistic appearance of objects. This paper investigates the feasibility and effectiveness of ingesting NeRF into MLLM. We create LLaNA, the first general-purpose NeRF-language assistant capable of performing new tasks such as NeRF captioning and Q\&A. Notably, our method directly processes the weights of the NeRF's MLP to extract information about the represented objects without the need to render images or materialize 3D data structures. Moreover, we build a dataset of NeRFs with text annotations for various NeRF-language tasks with no human intervention. Based on this dataset, we develop a benchmark to evaluate the NeRF understanding capability of our method. Results show that processing NeRF weights performs favourably against extracting 2D or 3D representations from NeRFs.