Multi-times Monte Carlo Rendering for Inter-reflection Reconstruction
Authors: Tengjie Zhu, Zhuo Chen, Jingnan Gao, Yichao Yan, Xiaokang Yang
NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments show that our method outperforms other inverse rendering methods on various object groups. We also show downstream applications, e.g., relighting and material editing, to illustrate the disentanglement ability of our method. |
| Researcher Affiliation | Academia | Tengjie Zhu Zhuo Chen * Jingnan Gao Yichao Yan Xiaokang Yang Mo E Key Lab of Artificial Intelligence, AI Institute, Shanghai Jiao Tong University {zhutengjie, ningci5252, gjn0310, yanyichao, xkyang}@sjtu.edu.cn |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | No | Our project page: https://zhutengjie.github.io/Ref-MC2/. Justification: We will release the data and code in a few weeks. |
| Open Datasets | No | We construct a dataset of multiple reflective objects based on the existing single objects to evaluate the performance. Our dataset consists of 16 groups of object compositions, most of which contain indirect illumination between reflective objects. We render the composed objects with various environmental lighting in the Blender engine. Justification: We will release the data and code in a few weeks. |
| Dataset Splits | Yes | Each group contains 300 images, with 200 for the training set and 100 for the test set. |
| Hardware Specification | Yes | We optimize the 3D model on 1 RTX 3090 GPU with 24G memory. |
| Software Dependencies | No | The paper mentions rendering with 'Blender engine' but does not specify versions for any key software components or libraries used in their method implementation. |
| Experiment Setup | Yes | We use the Adam optimizer for the material and the environment map with an initial learning rate of 0.03. The coefficients of loss function ω1, ω2, and ω3 are set to 0.1, 0.05, and 1, respectively. The rate of Monte Carlo sampling is commonly set to 128 consistent with the setting in Nvdiffrecmc [9]. |