depth op• データ種: lightfield × image2d → image2d
• 呼び出し: import lightfield; lightfield.lf_all_in_focus(lf, slope_map, levels=None, *, n_levels=16, interp='linear', edge='nearest') (または opslightfield.get("lf_all_in_focus"))
画素ごとに自分の傾きで再合焦し、全面がシャープな 1 枚を合成する。
> 以下の詳細説明は原文のままです —— 要約と見出しは訳出済み。
Builds a focal stack at *levels* and, for every pixel, takes the slice whose
level is closest to `slope_map` there. This is the "2-D image" half of what
a plenoptic camera delivers, at the full depth range rather than the depth of
field of one focus setting.
*levels* controls which refocus planes are actually rendered, and the
`None` default has two documented branches — a continuous slope map has
one distinct value per pixel, and rendering a refocus plane for each of
those would be tens of thousands of full-frame shifts:
• if *slope_map* holds at most *n_levels* distinct values (the case when
it came from :func:lf_depth_from_focus with `subpixel=False`), those
exact values are used and the composite is exact;
• otherwise the map's range is quantised to *n_levels* evenly spaced
planes (default 16), which is a real approximation: a pixel whose true
slope falls between two planes is refocused at the nearer one. Pass
*levels* explicitly — e.g. the sweep you handed
:func:lf_depth_from_focus — when that matters.
Returns a `(H, W)` 2-D image.
Raises `ValueError`: *lf* not a valid light field, a *slope_map* that
is not 2-D / not `(H, W)` / non-finite, *levels* empty or over the stack
caps, *n_levels* outside `[1, MAX_STACK_SLICES]`, unknown *interp* /
*edge*.
• サンプルデータ カタログ(DL URL / ライセンス) — 2-D は skimage.data(BSD/public)+ 合成、3-D は実データ源(Stanford/PDS 等)の DL URL。
• 演算子の来歴・参考文献 — この op 族の元になった研究/手法の出典。
• アルゴリズムの正典(著者・年)と用途は上記ファミリ使い方ガイドに記載。
• lightfield_depth — py -3.11 examples/lightfield_depth.py
image2d を入力に取れる)lf_from_mla · lf_disparity_to_depth
depth)lf_depth_from_focus · lf_epi_slope · lf_disparity_to_depth · lf_plenoptic_design
*Provenance: lightfield.py — LIGHTFIELD 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.