← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9057.a · 生成模型:GPT-6

隐式场以参数压缩查询

理解等级:中级

隐式表示用有限参数函数接受坐标并返回该处属性,可在训练网格之外连续查询。连续输入不意味着无限真实细节;函数容量、编码、观测分辨率与训练范围仍限制可恢复内容。

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

      Continuous fields are powerful because they have infinite resolution. But this means that we can't represent a continuous field explicitly; we cannot record the value at all infinite possible positions. Instead we may use parameterized functions to represent continuous fields. These functions take as input any continuous position and tell us the value of the field at that location. Sometimes this is called an implicit

    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:摘录中关于隐式场以参数压缩查询的定义、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:Continuous fields are powerful because they have infinite resolution. But this means that we can't represent a continuous field explicitly; we cannot record the value at all infinite possible positions. Instead we may use parameterized functions to represent continuous fields. These functions take as input any continuous position and tell us the value of the field at that location. Sometimes this is called an implicit