PrimDiffusion: Volumetric Primitives Diffusion for 3D Human Generation

Authors: Zhaoxi Chen, Fangzhou Hong, Haiyi Mei, Guangcong Wang, Lei Yang, Ziwei Liu

NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments validate that Prim Diffusion outperforms state-of-the-art methods in 3D human generation.
Researcher Affiliation Collaboration Zhaoxi Chen1 Fangzhou Hong1 Haiyi Mei2 Guangcong Wang1 Lei Yang2 Ziwei Liu1,B 1S-Lab, Nanyang Technological University 2Sensetime Research
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper mentions supplementary material for more details and video results, but does not explicitly state that the source code for their method is released or provide a direct link to it.
Open Datasets Yes We obtain 796 high-quality 3D humans from Render People [55] with diverse identities and clothes. ... [55] https://renderpeople.com/3d-people/. Renderpeople, 2018. ...finetuned EVA3D that is first pre-trained on Deep Fashion [26] dataset
Dataset Splits No The paper states: 'All methods are trained from scratch (except mentioned) on this dataset for fair comparisons.' and 'The metrics PSNR, SSIM [58], and LPIPS [63] are averaged across all training identities, views, and poses.' It does not explicitly mention training/validation/test splits for the dataset used in the main experiments.
Hardware Specification No The paper does not provide specific hardware details used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers, such as Python, PyTorch, or CUDA versions.
Experiment Setup No The paper briefly mentions the number of primitives (K=1024) and the denoiser architecture (2D U-Net) but does not provide specific experimental setup details such as hyperparameters, learning rates, batch sizes, or optimizer settings in the main text, deferring to supplementary material for 'more details'.