lf_plenoptic_design — LIGHTFIELD depth op

Data kinds: nonetable (an op determined by its arguments alone — it takes no image or data input)

Call: import lightfield; lightfield.lf_plenoptic_design(focal_mm=50.0, f_number=8.0, object_mm=300.0, pixel_um=3.45, mla_pitch_um=27.6, sensor_px=(2048, 2448), *, subpixel_px=0.1) (or opslightfield.get("lf_plenoptic_design"))

Usage

Size a plenoptic camera: what angular/spatial resolution and depth range you buy.

The plenoptic trade in one table. A microlens spanning ``mla_pitch_um /

pixel_um`` pixels turns that many pixels into that many *directions*, so the

sensor's pixel count is unchanged but the image is `U*V` times smaller and

carries `U*V viewpoints. This operator composes :mod:optics` rather than

re-deriving it: `optics.thin_lens` places the image, and

`optics.depth_of_field` is called twice — once with the pixel pitch as

the circle of confusion (the depth of field of a single refocused slice) and

once with the *microlens* pitch (the range over which refocusing can still

recover a sharp image). Their ratio is the refocusing gain, and it comes out

at the angular resolution, which is the textbook result — measured

2026-09-01 at `f = 50 mm, N = 8, s_o = 300 mm`: an 8x8 angular

grid gives `refocus_gain = 8.0038, 10x10 gives 10.0075` and 6x6 gives

`6.0016`.

Returns a dict — `angular_u / angular_v` (whole pixels per microlens,

from `floor) · angular_exact` (the unrounded ratio) and

`pitch_is_integer` (whether the MLA pitch is a whole number of pixels; it

usually is not, which is why real decoding needs sub-pixel calibration) ·

`spatial_w / spatial_h` (microlenses = sub-aperture image size) ·

`n_views · resolution_loss (U*V) · image_mm` /

`magnification / working_distance_mm` from the thin lens ·

`aperture_mm (f/N) · baseline_mm` (viewpoint spacing across the

pupil, `aperture / (U - 1)) · focal_px_subaperture` (focal length in

units of the *microlens* pitch, the pixel of a sub-aperture image) ·

`dof_pixel_mm / dof_refocus_mm and refocus_gain` · and

`depth_precision_mm`, the object-side distance change that moves the

disparity by *subpixel_px* pixels (`Z^2 * dp / (focal_px * baseline)`) —

the honest depth resolution at *object_mm*.

Raises `ValueError`: any non-positive or non-finite length, an

`mla_pitch_um` smaller than two *pixel_um* (fewer than 2 directions is not

a light field), a sensor smaller than one microlens, a non-positive

*subpixel_px*, `object_mm == focal_mm` (propagated from

`optics.thin_lens`: the object images at infinity), and an angular

resolution of 1 in either axis, where the baseline would be a 0/0.

Detailed usage guide

lightfield_depth family guide

References (sample data, literature)

• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).

• Operator provenance and references — the sources of the research/methods this op family came from.

• The canonical algorithm (author, year) and its uses are named in the family usage guide above.

Runnable examples (verified samples that actually call this op)

lightfield_depthpy -3.11 examples/lightfield_depth.py

Ops the type connects to (they accept table as input)

Same category (depth)

lf_depth_from_focus · lf_epi_slope · lf_disparity_to_depth · lf_all_in_focus


*Provenance: lightfield.py — LIGHTFIELD operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*

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