SpectralNeRF: Physically Based Spectral Rendering with Neural Radiance Field
Authors: Ru Li, Jia Liu, Guanghui Liu, Shengping Zhang, Bing Zeng, Shuaicheng Liu
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Comprehensive experimental results demonstrate the proposed Spectral Ne RF is superior to recent Ne RF-based methods when synthesizing new views on synthetic and real datasets. |
| Researcher Affiliation | Academia | Ru Li1, Jia Liu2, Guanghui Liu2 , Shengping Zhang1 , Bing Zeng2, Shuaicheng Liu2* 1Harbin Institute of Technology, Weihai, China 2University of Electronic Science and Technology of China, Chengdu, China {liru, s.zhang}@hit.edu.cn, {liujia21@std., guanghuiliu@, eezeng@, liushuaicheng@}uestc.edu.cn |
| Pseudocode | No | The paper describes the architecture and processes, but it does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | The codes and datasets are available at https://github.com/liru0126/Spectral Ne RF. |
| Open Datasets | Yes | The codes and datasets are available at https://github.com/liru0126/Spectral Ne RF. ... We render 8 spectral datasets and capture 2 real-world scenes with spectrum maps and RGB images... |
| Dataset Splits | No | The paper describes the synthetic and real-world datasets used and mentions sampling points along rays for coarse and fine models. However, it does not specify explicit train/validation/test splits (e.g., percentages or exact counts) for these datasets, nor does it refer to standard predefined splits with citations. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used for running experiments, such as GPU models, CPU types, or memory specifications. |
| Software Dependencies | No | The paper mentions using Ne RF (Mildenhall et al. 2020) and Adam optimizer (Kingma and Ba 2015). However, it does not provide specific version numbers for any software, libraries, or frameworks (e.g., PyTorch version, Python version, CUDA version). |
| Experiment Setup | Yes | We implement the Spectral MLP on top of Ne RF (Mildenhall et al. 2020), which uses an eight-layer MLP with 256 channels and Re LU activation to predict the density σ, and following two fully-connected layers with 128 and 3 snum channels to obtain the spectral radiance. We sample 64 points along each ray in the coarse model and 128 points in the fine model on the dataset. Adam optimizer (Kingma and Ba 2015) is used for the Spectral MLP and the SAUNet, and their learning rate is set to 5 10 4 and 0.001, respectively. ... λRGB is a hyper-parameter to balance the contributions of the two losses. We empirically set it to 1.1. |