raw_black_level — GFX2D isp op

Data kinds: image2dimage2d

Call: import fullseye as fs; fs.ledger.raw_black_level(raw, offsets, white=1.0, pattern='RGGB') (to call the implementation directly, import gfx2d; gfx2d.raw_black_level(raw, offsets, white=1.0, pattern='RGGB'); from the registry, opsgfx2d.get("raw_black_level"))

Usage

Black-level compensation on a Bayer frame: `(raw - offset) / (white - offset) per colour channel, clipped to [0, 1]`.

A sensor's dark output is not zero (pedestal + dark current); every later

stage that multiplies (gains, CCM) would scale that pedestal into a colour

cast, so this runs first. *offsets* is a scalar, 3 values `(R, G, B)`

or 4 values `(R, G1, G2, B) in the same [0, 1]` units as *raw*; *white*

is the saturation level, so a pixel at *white* maps to exactly 1.

Ground truth: a frame built as `offset + k * signal` comes back as

`k * signal / (white - offset)` to 1e-15; a pixel at the offset maps to 0.

Raises `ValueError: non-finite input, an offset >= white`, or an

unknown *pattern*.

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)

raw_to_display_isppy -3.11 examples/raw_to_display_isp.py

Ops the type connects to (they accept image2d as input)

shadow_cast_2d · dither · raw_dead_pixel_mask · raw_dead_pixel_correct · lens_shading_gain · lens_shading_correct · raw_apply_gains · raw_demosaic_bilinear

Same category (isp)

raw_dead_pixel_mask · raw_dead_pixel_correct · lens_shading_gain · lens_shading_correct · awb_gains · rgb_apply_gains · raw_apply_gains · raw_demosaic_bilinear


*Provenance: gfx2d.py — GFX2D 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.