watermedia op• 数据种类:无 → table(仅由参数决定的算子 —— 不接受图像或数据输入)
• 调用: import fullseye as fs; fs.ledger.refracted_view(points_xyz, eye_xyz, surface='flat', n_water=None, wavelength_um=0.55) -> 'dict'(要直接调用实现,import watermedia; watermedia.refracted_view(points_xyz, eye_xyz, surface='flat', n_water=None, wavelength_um=0.55) -> 'dict';从台账取用则 opsoptics.get("refracted_view"))
• GPU:无(仅 CPU 参考实现)
水下的点在平坦或有波的水面上方的相机中出现在哪里(费马原理)-> 字典。
> 以下的详细说明为原文 —— 摘要与标题已翻译。
points_xyz is (M, 3) (or one point), z up and the mean water level z = 0; eye_xyz is the camera, above the surface. A point with z >= surface height is in the air and is returned unchanged (in_air True, no refraction point). For an underwater point the path from the point to the eye is the one of stationary optical length n |R - P| + |E - R| over the crossing point R on the surface; at that point Snell's law n sin(theta_w) = sin(theta_a) holds. Flat surface: the crossing lies in the vertical plane through the point and the eye and is the unique root of the Snell condition (found by bisection to machine precision); the apparent position is the point seen along the eye ray continued below the surface to the horizontal position of the true point: its depth is x h / (s - x) and tends to d / n for a camera straight above. Wavy surface: surface is a table with height (square, periodic) and extent_m, as returned by ocean_height_field; the crossing is found by Newton iteration on the gradient of the optical path with bicubic heights and normals, started from the flat solution; the vector Snell condition n (d_w x N) = d_a x N is verified at the result and a point that fails it (or does not converge) is returned as NaN with converged False. The table has apparent (M, 3), refraction_point (M, 3), converged, in_air, snell_residual (flat: |n sin(theta_w) - sin(theta_a)|, wavy: the norm of the vector residual), apparent_depth_m and n_water. A caustic can make several paths possible; the one found is the stationary point nearest the flat solution, or (when the Newton iteration fails on steep ripples) the shortest optical path on a local grid.
optics 的每个算子都先校验输入再计算(不让任何东西无声通过):
• 单位写进参数名 —— _mm / _um / _deg / _mrad。把 mm 和 µm 弄混不会崩溃,而是给出「看着合理却是错的答案」,所以用命名来防。这里绝不从数值大小去猜单位。
• 字符串一律 ValueError —— float('50') 会成功,于是未解析的配置值会被当成长度混进来(实测:thin_lens('50', '200') 曾返回看着合理的 66.667 mm)。bool 也按 True == 1 的隐式提升拒绝。
• complex / masked array 一律 ValueError(仅接受实数槽位;拒绝无声丢弃虚部或剥掉掩码)。所有输入中的 NaN/Inf 一律 ValueError。但已声明的缺测除外:bottle_fit_from_pattern 的 u_map 以 NaN 接收未能测得的像素(∞ 仍被拒绝)。
• 逐项点名拒绝除零及其近亲:焦距 0、曲率半径 0、折射率 <= 0、全不透明光阑(全为 0,归一化变成 0/0)、总和 <= 0 的 PSF、S0 = 0 的 Stokes 矢量、物体位于前焦点(像在无穷远)。
• 只有两个算子会返回非有限值,而且都写进了契约:depth_of_field 在超焦距以外返回 far_mm = inf(这正是超焦距的定义),gaussian_beam 在束腰处返回 wavefront_radius_mm = inf(平面波前的曲率半径)。两者都同时返回一个有限的搭档(far_is_infinite / curvature_per_mm)。除此之外的无声 NaN/Inf 都在内部检出并 ValueError ——「float64 溢出了」和「答案是无穷大」是两种不同的主张,不能拿后者的脸去交付前者。但已声明的缺测除外:bottle_background_map 对光线不可用的像素(全反射等)在 u_mm 中返回 NaN,并同时返回 valid 掩码和数量。
• 尺寸上限:生成网格受 optics.MAX_GRID(4096)限制,传入的场/PSF/光阑受 optics.MAX_FIELD_ELEMENTS(2^24),ABCD 元件序列受 optics.MAX_SYSTEM_ELEMENTS(1024),Zernike 受 MAX_ZERNIKE_TERMS(512)/ MAX_ZERNIKE_ORDER(40)/ MAX_ZERNIKE_BASIS(2^25)。以 fail-closed 堵住「小参数引发巨大内部分配」的路径(实测:n_max=40 × 4096² 需要 108 GB)。
• 物理上不可能的状态同样拒绝:偏振度 > 1 的 Stokes 矢量、负透过率、负强度、n-|m| 为奇数等非法 Zernike 指标。
• 示例数据目录(下载 URL / 许可证) —— 2-D 用 skimage.data(BSD/公有领域)加合成图,3-D 给出真实数据源(Stanford/PDS 等)的下载 URL。
• 算子来历与参考文献 —— 该算子族所依据的研究/方法出处。
• 算法的正典(作者、年份)与用途见上面的族使用指南。
• poc_water_transparent_media — py -3.11 examples/poc_water_transparent_media.py
table 作为输入)abcd_matrix · wavefront_stats · sfr_from_edge · paraxial_trace · seidel_coefficients · spot_stats · tolerance_analysis · wavefront_from_opd
watermedia)ocean_spectrum · ocean_height_field · fresnel_water_air · snell_window · water_attenuation · underwater_image · phase_function_hg · scattering_phase_petzold
*Provenance: watermedia.py — OPTICS 算子登记表。本条目由 tools/opdocs.py md 自动生成(请勿手工编辑)。*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.