无偏差近摄像机训练的辐射场梯度缩放（英文版）

英文标题：Radiance Field Gradient Scaling for Unbiased Near-Camera Training中文摘要：本文提出了一种基于梯度缩放的方法，以解决 NeRF 采集中的采样偏差问题，避免了在不需要使用近平面的情况下出现背景坍塌问题。英文摘要：NeRF acquisition typically requires careful choice of near planes for thedifferent cameras or suffers from background collapse, creating floatingartifacts on the edges of the captured scene. The key insight of this work isthat background collapse is caused by a higher density of samples in regionsnear cameras. As a result of this sampling bias, near-camera