QiYun

Note

Posted on Edited on In
873 3 mins.
Title MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction
Author Zehao Yu1 Songyou Peng2,3 Michael Niemeyer1,3 Torsten Sattler4 Andreas Geiger1,3
Conf/Jour NeurIPS
Year 2022
Project MonoSDF (niujinshuchong.github.io)
Paper MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction (readpaper.com)
[MonoSDF: Exploring Monocular Geometric Cues for Neural Implicit Surface Reconstruction \		 PDF](https://arxiv.org/pdf/2206.00665)

image.png|666

深度+法向量监督

Read more »

Posted on Edited on In
2.3k 9 mins.
Title Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision
Author Michael Niemeyer1,2 Lars Mescheder1,2,3† Michael Oechsle1,2,4 Andreas Geiger1,2
Conf/Jour CVPR
Year 2020
Project autonomousvision/differentiable_volumetric_rendering: This repository contains the code for the CVPR 2020 paper “Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision” (github.com)
Paper Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervision (readpaper.com)

image.png|666

Read more »

Posted on Edited on In
95 1 mins.
Title Image-based 3D Object Reconstruction: State-of-the-Art and Trends in the Deep Learning Era
Author Xian-Feng Han, Hamid Laga, Mohammed Bennamoun Senior Member, IEEE
Conf/Jour IEEE Transactions on Pattern Analysis and Machine Intelligence
Year 2019
Project
Paper Image-Based 3D Object Reconstruction: State-of-the-Art and Trends in the Deep Learning Era (readpaper.com)
Read more »

Posted on Edited on In
5.7k 21 mins.
Title Voxurf: Voxel-based Efficient and Accurate Neural Surface Reconstruction
Author Tong Wu and Jiaqi Wang and Xingang Pan and Xudong Xu and Christian Theobalt and Ziwei Liu and Dahua Lin
Conf/Jour ICLR
Year 2023
Project wutong16/Voxurf: [ ICLR 2023 Spotlight ] Pytorch implementation for “Voxurf: Voxel-based Efficient and Accurate Neural Surface Reconstruction” (github.Com)
Paper Voxurf: Voxel-based Efficient and Accurate Neural Surface Reconstruction (readpaper.com)

image.png|666

依赖 mask 即 rembg 技术

Read more »

Posted on Edited on In
4.5k 17 mins.
Title Adaptive Shells for Efficient Neural Radiance Field Rendering
Author Zian Wang and Tianchang Shen and Merlin Nimier-David and Nicholas Sharp and Jun Gao and Alexander Keller and Sanja Fidler and Thomas M\”uller and Zan Gojcic
Conf/Jour ACM Trans. On Graph. (SIGGRAPH Asia 2023)
Year 2023
Project Adaptive Shells for Efficient Neural Radiance Field Rendering (nvidia.com)
Paper Adaptive Shells for Efficient Neural Radiance Field Rendering (readpaper.com)

image.png|666

Read more »

Posted on Edited on In
2.5k 9 mins.
Title DMV3D: Denoising Multi-View Diffusion using 3D Large Reconstruction Model
Author Xu, Yinghao and Tan, Hao and Luan, Fujun and Bi, Sai and Wang Peng and Li, Jihao and Shi, Zifan and Sunkavalli, Kaylan and Wetzstein Gordon and Xu, Zexiang and Zhang Kai}
Conf/Jour arxiv
Year 2023
Project DMV3D: Denoising Multi-View Diffusion Using 3D Large Reconstruction Mode (justimyhxu.github.io)
Paper DMV3D: DENOISING MULTI-VIEW DIFFUSION USING 3D LARGE RECONSTRUCTION MODEL (readpaper.com)

需要相机位姿 + 多视图 + Diffusion Model + NeRF Triplane2MLP

image.png|666

不足：

  • 对未见视图的重建质量不高
  • 只支持低分辨率图像和三平面
  • 只支持输入没有背景的物体图像
  • 没用到任何先验知识
Read more »

Posted on Edited on In
4.6k 17 mins.
Title DiffuStereo: High Quality Human Reconstruction via Diffusion-based Stereo Using Sparse Cameras
Author Ruizhi Shao, Zerong Zheng, Hongwen Zhang, Jingxiang Sun, Yebin Liu
Conf/Jour ECCV 2022 Oral
Year 2022
Project DiffuStereo Project Page (liuyebin.com)
Paper DiffuStereo: High Quality Human Reconstruction via Diffusion-based Stereo Using Sparse Cameras (readpaper.com)

image.png|666

THUman2.0数据集demo结果好(除手、脸和脚)，训练代码还未开源
image.png|666

Read more »

Posted on Edited on In
7.5k 27 mins.
Title Human as Points: Explicit Point-based 3D Human Reconstruction from Single-view RGB Images
Author Yingzhi Tang, Qijian Zhang, Junhui Hou, and Yebin Liu
Conf/Jour arXiv
Year 2023
Project yztang4/HaP (github.com)
Paper Human as Points—— Explicit Point-based 3D Human Reconstruction from Single-view RGB Images.pdf (readpaper.com)

image.png|666

深度估计+SMPL 估计+Diffusion Model 精细化(PointNet++)

缺陷：依赖于深度估计方法和 SMPL 估计方法的精度
为了保护隐私不对人脸进行重建

Read more »