为了解决这些挑战,我们引入了EvaGaussians:一种利用事件相机捕捉的事件流来增强从运动模糊图像中学习高质量3D-GS的新方法。通过利用事件流的高时间分辨率和动态范围,我们将其无缝集成到3D-GS的初始化和优化过程中,从而显著提升高保真新视图的生成质量,并恢复复杂的纹理细节。 EvaGaussians结合事件流数据和相机位姿优化的...
GaussianStego 代表了对新挑战的初步探索,即在当前 3D 生成模型生成的渲染中嵌入可定制、不可察觉和可恢复的信息,同时确保对渲染内容质量的影响最小。 3DGS+大规模场景 VEGS: View Extrapolation of Urban Scenes in 3D Gaussian Splatting using Learned Priors https://arxiv.org/abs/2407.02945 Sungwon Hwang, M...
With the assumption of Gaussian mixture a priori, one improved prestack EVA inversion (elastic impedance variation with angle) conditioned by seismic and well data in mixed-domain is proposed to realize the estimation of discrete lithofacies and continuous geofluid parameters. The peaks number of ...
Libint: high-performance library for computing Gaussian integrals in quantum mechanics - evaleev/libint