PanoSynthVR: View Synthesis From A Single Input Panorama with Multi-Cylinder Images

PanoSynthVR: View Synthesis From A Single Input Panorama with Multi-Cylinder Images
复制标题

PanoSynthVR:从单输入全景图与多圆柱图像进行视图合成

DOI:
10.1145/3450618.3469144
复制
发表时间:
2021
期刊:
SIGGRAPH '21: ACM SIGGRAPH 2021 Posters
影响因子:
--
通讯作者:
Ventura, Jonathan
Ventura, Jonathan
中科院分区:
--
文献类型:
--
作者:
Gadgil, Richa;John, Reesa;Zollmann, Stefanie;Ventura, Jonathan

文献摘要

参考文献

相似文献

我们介绍了一种方法,自动转换成一个多缸的图像表示,支持实时,自由视点视图合成的虚拟现实的一个单一的全景输入。我们将现有的在针孔图像上训练的卷积神经网络应用于具有包裹填充的圆柱形全景图,以确保左右边缘之间的一致性。该网络输出一堆不同深度的半透明贴图,这些贴图可以很容易地渲染和合成。初步实验表明,该方法产生令人信服的视差和清洁的物体边界比纹理网格表示。
We introduce a method to automatically convert a single panoramic input into a multi-cylinder image representation that supports real-time, free-viewpoint view synthesis for virtual reality. We apply an existing convolutional neural network trained on pinhole images to a cylindrical panorama with wrap padding to ensure agreement between the left and right edges. The network outputs a stack of semi-transparent panoramas at varying depths which can be easily rendered and composited with over blending. Initial experiments show that the method produces convincing parallax and cleaner object boundaries than a textured mesh representation.
DOI: 10.1109/cvpr42600.2020.00805
发表时间: 2020-04
期刊: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
影响因子: --
作者:
Meng-Li Shih;Shih-Yang Su;J. Kopf;Jia-Bin Huang
通讯作者: Meng-Li Shih;Shih-Yang Su;J. Kopf;Jia-Bin Huang