depth op• 데이터 종류: lightfield × image2d → image2d
• 호출: import lightfield; lightfield.lf_all_in_focus(lf, slope_map, levels=None, *, n_levels=16, interp='linear', edge='nearest')(또는 opslightfield.get("lf_all_in_focus"))
화소마다 자기 기울기로 재초점하여 전면이 선명한 한 장을 합성합니다.
> 아래 상세 설명은 원문입니다 —— 요약과 제목은 번역되어 있습니다.
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*.
• 샘플 데이터 카탈로그(DL URL / 라이선스) —— 2-D 는 skimage.data(BSD/public)+ 합성, 3-D 는 실데이터 소스(Stanford/PDS 등)의 DL URL.
• 연산자의 내력·참고문헌 —— 이 연산자 족의 바탕이 된 연구/기법의 출처.
• 알고리즘의 정전(저자·연도)과 용도는 위의 패밀리 사용 가이드에 적혀 있습니다.
• lightfield_depth — py -3.11 examples/lightfield_depth.py
image2d 를 입력으로 받는 것)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 연산자 레지스트리. 이 op 노트는 tools/opdocs.py md 가 자동 생성합니다(직접 편집하지 마세요).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.