fullseye

tb_specular_free_transform — 2D typed op

tb_specular_free_transform: input → output

図は合成の入力 128×128 で実際に走らせた出力。左が入力、右が出力。点群は上から見た散布(明るさ = z)、1-D 列は折れ線、体積は z 方向の最大値投影、動画は中央フレーム、複素画像は振幅、絵にならない返り値は値そのもの。

つまみ a は出力を変えない(実測: 0.1 / 0.5 / 0.9 で同一)。

つまみ b は出力を変えない(実測: 0.1 / 0.5 / 0.9 で同一)。

段階(前置きの op → この op。左から順):

tb_specular_free_transform: stages

別の画像でも(合成シーン / 写真 / 硬貨。上段が入力、下段がその出力。つまみは既定):

tb_specular_free_transform: other inputs

使い方

Project out the illuminant direction: the part of the image a highlight cannot touch. → (H, W, 3).

``I - (I.G) G`` for the unit illuminant colour ``G``. Under the dichromatic
model the interface term is ``m_s * G``, so it lies entirely in the removed
direction and the result is **invariant to any specular term whatsoever** —
exactly, for any lobe shape, any strength, any spatial pattern. That is the
specular-invariant subspace of Mallick et al. (2005); this operator is the
projection itself, with no rotation into named channels, so it stays in RGB
and composes with the rest of the family.

Use it when the *shape* of the specular lobe is unknown or the surface is
textured — feature matching, edge detection and correlation all work in this
subspace without any of the assumptions
:func:`specular_diffuse_split` needs.

**This is a projection, not a picture.** The result loses one of three
degrees of freedom (its component along ``G`` is exactly zero everywhere)
and, for an image with negative values after black-level subtraction, keeps
them. It is not a displayable "highlight-removed photo" and does not claim
to be; for that, use :func:`specular_diffuse_split`.

**Raises** ``ValueError``: *image_rgb* is not a valid ``(H, W, 3)`` linear
RGB image (see :func:`specular_diffuse_split`); *illuminant_rgb* is not a
non-zero 3-vector.

Typed bridge of the specular op specular_free_transform into the 2-D evolution registry: the same implementation, called under the op(v, a, b) convention. This op has no tunable parameter; a and b are unused.

参考(サンプルデータ・文献)

Studio で試す

下のプログラムは実際に走ることを確かめてある(図と同じ入力)。Studio のヘルプではこのブロックがボタンになり、その場で読み込んで実行できる。

img_to_rgb 0.50 0.50
tb_specular_free_transform 0.50 0.50

実行できる例(この op を実際に呼ぶ検証済みサンプル)

次の例は元の台帳 op specular_free_transform を呼ぶもの。この橋渡し op は同じ実装を fn(v, a, b) 規約に合わせただけなので、挙動はそのまま当てはまる(呼び出し形だけ違う)。

型が繋がる次の op(rgbimage を入力に取れる)

identity · tb_wetness · tb_sensor_capture · tb_specular_diffuse_split · tb_specular_coefficient_map · tb_rgb_to_quaternion

同カテゴリ(typed)

tb_points_to_voxel · tb_estimate_point_normals · tb_iss_keypoints · tb_project_points · tb_render_point_depth · tb_statistical_outlier_removal · tb_radius_outlier_removal · tb_voxel_grid_downsample


Provenance: ops.py — 2D operator registry. この per-op ノートは tools/opdocs.py md が自動生成(手編集しない)。

© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.