← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 9055.c · 生成模型:GPT-6

有限视图不决定整个光场

理解等级:中级

Light field、lumigraph与NeRF都利用观测构建适合新视角合成的表示,但有限图像并不约束全部位置和方向。未观测区域依赖模型先验和训练覆盖;得到流畅渲染不等于恢复了全部场景真相。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      If we had access to the plenoptic function of a scene, we would be able to render images from all possible viewpoints within that scene. One attempt at this is the lumigraph [170] which extracts a subset of the plenoptic function from a large set of images showing different viewpoints of an object. Another approach, light field rendering [293], also uses many images to get

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第45章《辐射场》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c045.xhtml) - 结论依据:摘录中关于有限视图不决定整个光场的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:If we had access to the plenoptic function of a scene, we would be able to render images from all possible viewpoints within that scene. One attempt at this is the lumigraph [170] which extracts a subset of the plenoptic function from a large set of images showing different viewpoints of an object. Another approach, light field rendering [293], also uses many images to get