Functions
	xyz_D50_to_Yrg (xyz)

	rgb_work_to_Yrg (rgb)

	chroma_hue (Yrg)

	tone_map (rgb)

	bracket_matrices (insets, rotations)

	boundary_colors ()

	_lab_d65_to_work (L, a, b)

	chroma_trajectory (sample, M, M_inv)

	target_profile (baseline)

	residuals (params, drift_target_rad_per_ev)

	brk (p)

	chroma_hue_batch (RGB)

	variant_bracket (v)

	measure_batch (S, C_in, H_in, M, Mo)

	hk_drift_batch (S, M, Mo)

	delta_e_yrg (cr, dh_deg)

	nayatani_hk_excess (RGB)

	scene_ab_target (S_refl, hk_retention=1.0)

	skin_and_reflective_sets ()

	rec2020_worst_boundary_luminance (M, Mo)

	gamut_violation (M, Mo)

	fit_midpoint (lo_key, hi_key, inset_lo, inset_hi, seed_insets)

	print_midpoint_constants (p, mode, endpoints)

	print_variant_entry (name, fit, inset, irot, outset, orot)

	print_c_case (case_name, fit, inset, irot, outset, orot)

	report_variants ()

	per_hue_ab_and_drift (v, S, C, H, bin_idx, nbins)

	report_hue_continuity ()

	print_diagnostics (p, message)

	main ()

Variables
	XYZ_D50_to_D65_CAT16

	XYZ_D65_to_LMS_2006

	filmlightRGB_to_LMS

	LMS_to_filmlightRGB = np.linalg.inv(filmlightRGB_to_LMS)

	Y_2006_COEFFS = np.array([0.68990272, 0.34832189, 0.0])

	REC2020_TO_XYZ_D50

	BLACK_EV

	WHITE_EV

tuple	CURVE_DEFAULTS = (1.18, 10.0, 0.0, 1.5, 1.5)

	CURVE = curve_factory(BLACK_EV, WHITE_EV, *CURVE_DEFAULTS, shoulder_slope_matched=True)

	WHITE_YRG = rgb_work_to_Yrg(np.ones(3))

	EVS = np.arange(-2.0, WHITE_EV + 0.51, 0.5)

int	HUE_STEPS = 24

	SHOULDER_EV = BLACK_EV + CURVE.sh_x * (WHITE_EV - BLACK_EV)

list	_SKIN_LAB

	_D65_WHITE = np.array([0.95047, 1.0, 1.08883])

	IDENTITY = np.eye(3)

dict	BASELINES = {tuple(s): chroma_trajectory(s, IDENTITY, IDENTITY)[0] for s in boundary_colors()}

dict	SHIPPED_VARIANTS

	LUT_X = np.linspace(0.0, 1.0, 8193)

	LUT_Y = np.array([CURVE(v) for v in LUT_X])

list	_BND

tuple	_BND_EV = (-8, -6, -4, -2, -1, 0, 1, 2, 3)

Detailed Description

Fit the inset/rotation anchor constants of filmic RGB's "AgX" color science
(v8) — see doc/filmic-agx.md and filmic_agx_prepare_bracket() in
src/iop/filmicrgb.c, whose PROVISIONAL constants this script is meant to replace.

Model (mirrors the C pixel path, working profile = linear Rec2020):
  work RGB -> inset matrix -> per-channel [log2 encoding -> sigmoid spline ->
  hardness power] -> exact-inverse outset -> measure chroma / hue in Kirk Yrg.

Objectives:
  1. purity-vs-exposure: colors on the working-gamut boundary must reach
     achromatic (chroma ratio ~ 0) at the white end of the curve, monotonically;
  2. hue drift: dh/dEV in Yrg matches a chosen uniform target across the wheel
     (0 for a neutral character, or a uniform warm bias);
  3. hard constraints as penalties: displaced primaries stay inside the working
     gamut triangle with a margin (positivity of the bracket) and away from
     degeneracy (conditioning).

Usage:
  python3 tools/derive_filmic_agx_primaries.py --max-desat FRAC   (recommended)
  python3 tools/derive_filmic_agx_primaries.py --fit-priority
  python3 tools/derive_filmic_agx_primaries.py [--drift-target DEG_PER_EV]

The recommended mode is --max-desat : you state the maximum average chroma loss
over the priority set (skin + reflective) you accept, and the solver returns the
bracket with the best hue match at or below that budget. Sweep it to trace the
hue/chroma frontier. See its --help for the full method.

Prints the C arrays to paste into filmic_agx_prepare_bracket(). Requires
numpy + scipy (run with python3.12). Constants below are copied verbatim from the
C code so the model matches the pipeline bit-for-bit in spirit (float64 here,
float32 there).

Function Documentation

◆ _lab_d65_to_work()

derive_filmic_agx_primaries._lab_d65_to_work	(	L,
		a,
		b
	)

protected

Definition at line 165 of file derive_filmic_agx_primaries.py.

References f.

Referenced by skin_and_reflective_sets().

◆ boundary_colors()

derive_filmic_agx_primaries.boundary_colors ( )

saturated colors on the working-gamut boundary, one channel at zero

Definition at line 135 of file derive_filmic_agx_primaries.py.

Referenced by residuals().

◆ bracket_matrices()

derive_filmic_agx_primaries.bracket_matrices	(	insets,
		rotations
	)

inset matrix M (work->rendering) and its exact inverse, exactly as the C code.

Definition at line 108 of file derive_filmic_agx_primaries.py.

References xyz_D50_to_Yrg().

Referenced by brk(), print_midpoint_constants(), residuals(), and variant_bracket().

◆ brk()

derive_filmic_agx_primaries.brk ( p )

Definition at line 276 of file derive_filmic_agx_primaries.py.

References bracket_matrices().

Referenced by main().

◆ chroma_hue()

derive_filmic_agx_primaries.chroma_hue ( Yrg )

Definition at line 92 of file derive_filmic_agx_primaries.py.

Referenced by chroma_trajectory().

◆ chroma_hue_batch()

derive_filmic_agx_primaries.chroma_hue_batch ( RGB )

(N,3) working-linear-Rec2020 -> (chroma, hue) in Kirk Yrg, relative to white.

Definition at line 281 of file derive_filmic_agx_primaries.py.

Referenced by main(), measure_batch(), per_hue_ab_and_drift(), report_hue_continuity(), and report_variants().

◆ chroma_trajectory()

derive_filmic_agx_primaries.chroma_trajectory	(	sample,
		M,
		M_inv
	)

chroma ratio and hue drift vs exposure, through the bracketed curve

Definition at line 173 of file derive_filmic_agx_primaries.py.

References chroma_hue(), max, rgb_work_to_Yrg(), and tone_map().

Referenced by residuals().

◆ delta_e_yrg()

derive_filmic_agx_primaries.delta_e_yrg	(	cr,
		dh_deg
	)

Perceptual color-shift distance in the chroma-NORMALIZED Yrg plane : the input
sits at (1, 0) — chroma ratio 1, zero hue drift — and the output at (cr*cos, cr*sin),
so the distance between them folds chroma loss AND hue drift into one number.
0 = colour unchanged ; ~1 = fully bleached ; up to 2 = hue-flipped at full chroma.

NOTE: subtracting the (1, 0) reference is the whole point. An earlier version
computed hypot(cr*cos, cr*sin), which reduces algebraically to just cr — the output
vector's LENGTH, not its distance from the input — so it was not a real delta-E and
cancelled the desaturation term to a constant. Both --min-bleach and --max-desat now
use THIS function.

Definition at line 317 of file derive_filmic_agx_primaries.py.

Referenced by main(), and report_variants().

◆ fit_midpoint()

derive_filmic_agx_primaries.fit_midpoint	(	lo_key,
		hi_key,
		inset_lo,
		inset_hi,
		seed_insets
	)

Solve for the bracket that best reproduces the AVERAGE of two variants' processed
outputs — the perceptual midpoint. Target = 0.5*(post_bracket(lo) + post_bracket(hi))
over a skin + reflective + Rec2020-boundary sample set (skin weighted x2), fit in
least squares under Rec2020 gamut safety, positivity and conditioning. Returns the
10-parameter bracket (uniform inset, inset rot[3], outset[3], outset rot[3]) or None.

Definition at line 450 of file derive_filmic_agx_primaries.py.

References max, min, rec2020_worst_boundary_luminance(), skin_and_reflective_sets(), and variant_bracket().

Referenced by main().

◆ gamut_violation()

derive_filmic_agx_primaries.gamut_violation	(	M,
		Mo
	)

Definition at line 438 of file derive_filmic_agx_primaries.py.

References max.

Referenced by main().

◆ hk_drift_batch()

derive_filmic_agx_primaries.hk_drift_batch	(	S,
		M,
		Mo
	)

Helmholtz-Kohlrausch apparent-brightness drift : excess(output) - excess(input),
per sample, through the same inset -> curve -> outset path as measure_batch. Positive
= the bracket made the colour read brighter-for-its-luminance than the original (H-K
inflation) ; negative = deflation. The look wants this as close to 0 as it can, so the
rendered colours keep the SAME apparent-brightness balance as the scene.

Definition at line 306 of file derive_filmic_agx_primaries.py.

References nayatani_hk_excess().

Referenced by report_variants().

◆ main()

derive_filmic_agx_primaries.main ( )

Definition at line 692 of file derive_filmic_agx_primaries.py.

References brk(), chroma_hue_batch(), delta_e_yrg(), fit_midpoint(), gamut_violation(), main(), max, measure_batch(), min, nayatani_hk_excess(), print_c_case(), print_diagnostics(), print_midpoint_constants(), print_variant_entry(), rec2020_worst_boundary_luminance(), report_hue_continuity(), report_variants(), scene_ab_target(), skin_and_reflective_sets(), and variant_bracket().

Referenced by main().

◆ measure_batch()

derive_filmic_agx_primaries.measure_batch	(	S,
		C_in,
		H_in,
		M,
		Mo
	)

(N,3) samples through inset -> per-channel curve -> outset ; returns
(chroma_ratio post output<=input clamp, hue_drift_deg), both (N,).

Definition at line 295 of file derive_filmic_agx_primaries.py.

References chroma_hue_batch().

Referenced by main(), and report_variants().

◆ nayatani_hk_excess()

derive_filmic_agx_primaries.nayatani_hk_excess ( RGB )

Helmholtz-Kohlrausch apparent-brightness excess (Gamma - 1), Nayatani (1997) VAC
model, for a batch of working-linear-Rec2020 colours. This is the FRACTIONAL amount
by which a chromatic colour looks brighter than an equally-luminous grey — a real
perceptual effect that per-channel tone mapping amplifies unevenly by hue. It is
~0 for neutrals and yellow-greens and largest for saturated blue / red / magenta
(measured : gray 0.00, red 0.32, green 0.12, blue 0.43 at equal luminance), which is
exactly why an over-saturated red reads brighter than an equally-bright green.
Luminance-independent (a fractional boost), so penalizing it targets chroma+hue only.

Definition at line 331 of file derive_filmic_agx_primaries.py.

Referenced by hk_drift_batch(), main(), per_hue_ab_and_drift(), and scene_ab_target().

◆ per_hue_ab_and_drift()

derive_filmic_agx_primaries.per_hue_ab_and_drift	(	v,
		S,
		C,
		H,
		bin_idx,
		nbins
	)

For one variant, return (apparent_brightness[nbins], signed_hue_drift_deg[nbins],
output_chroma[nbins]) binned over the reflective hue circle. APPARENT BRIGHTNESS = output
luminance x (1 + Nayatani H-K excess) — how bright a colour READS, not just its luminance.
OUTPUT CHROMA = Yrg saturation — must DECREASE monotonically no->extra (the non-monotone-
saturation bug lived here). Hue drift is signed, only where output chroma survives (ratio > 0.2).

Definition at line 615 of file derive_filmic_agx_primaries.py.

References chroma_hue_batch(), nayatani_hk_excess(), and variant_bracket().

Referenced by report_hue_continuity().

◆ print_c_case()

derive_filmic_agx_primaries.print_c_case	(	case_name,
		fit,
		inset,
		irot,
		outset,
		orot
	)

Print a C switch-case block for the fitted bracket constants.

Definition at line 542 of file derive_filmic_agx_primaries.py.

Referenced by main(), and print_diagnostics().

◆ print_diagnostics()

derive_filmic_agx_primaries.print_diagnostics	(	p,
		message
	)

Definition at line 671 of file derive_filmic_agx_primaries.py.

References print_c_case(), print_variant_entry(), and report_variants().

Referenced by main().

◆ print_midpoint_constants()

derive_filmic_agx_primaries.print_midpoint_constants	(	p,
		mode,
		endpoints
	)

Print the C constant block for a fitted midpoint bracket (or a failure note).

Definition at line 513 of file derive_filmic_agx_primaries.py.

References bracket_matrices(), max, and rec2020_worst_boundary_luminance().

Referenced by main().

◆ print_variant_entry()

derive_filmic_agx_primaries.print_variant_entry	(	name,
		fit,
		inset,
		irot,
		outset,
		orot
	)

Print a SHIPPED_VARIANTS entry that can be pasted directly into the script.

Definition at line 532 of file derive_filmic_agx_primaries.py.

Referenced by main(), and print_diagnostics().

◆ rec2020_worst_boundary_luminance()

derive_filmic_agx_primaries.rec2020_worst_boundary_luminance	(	M,
		Mo
	)

Worst output luminance over the Rec2020 primaries and secondaries across the
tonal range. Rec2020 IS the working space, so its primaries are the most saturated
colors that can occur — the worst case. A strongly over-expanding outset can push
them to NEGATIVE luminance (the pixel renders BLACK); this must stay > 0. This is
the gamut-safety check that caught the no-bleach blue-goes-black bug.

Definition at line 408 of file derive_filmic_agx_primaries.py.

References min.

Referenced by fit_midpoint(), main(), print_midpoint_constants(), and report_variants().

◆ report_hue_continuity()

derive_filmic_agx_primaries.report_hue_continuity ( )

Per-hue colour-CONTINUITY diagnostic across the bleach ladder (reflective set). For each
of 12 hue bins it prints, for every shipped variant, the APPARENT BRIGHTNESS (output luminance
x (1 + H-K excess)) and the SIGNED hue drift, with the per-step deltas. A well-behaved ladder
is MONOTONE with EVEN steps per hue ; a big jump (e.g. reds/magentas over-brightening at one
step) or a sign reversal (a hue rotating against the ramp) marks a variant that left the ramp.
This is the tool behind the no-bleach red-darkening + low-bleach hue-divergence fixes.

Definition at line 635 of file derive_filmic_agx_primaries.py.

References chroma_hue_batch(), per_hue_ab_and_drift(), and skin_and_reflective_sets().

Referenced by main().

◆ report_variants()

derive_filmic_agx_primaries.report_variants ( )

Print the {avg,max} x {desaturation, hue drift, delta-E} x {skin,reflective} table
for the shipped variants, plus a Rec2020 gamut-safety check. Desaturation is over all
colors that carry chroma ; hue drift is measured where chroma survives (ratio > 0.2 —
a bleached color has no meaningful hue) ; delta-E is the combined chroma+hue fidelity
(delta_e_yrg) over ALL samples, the single tie-breaker metric. Source of the tables in
doc/filmic-agx.md and the user docs.

Definition at line 552 of file derive_filmic_agx_primaries.py.

References chroma_hue_batch(), delta_e_yrg(), hk_drift_batch(), measure_batch(), rec2020_worst_boundary_luminance(), row, skin_and_reflective_sets(), and variant_bracket().

Referenced by main(), and print_diagnostics().

◆ residuals()

derive_filmic_agx_primaries.residuals	(	params,
		drift_target_rad_per_ev
	)

Definition at line 197 of file derive_filmic_agx_primaries.py.

References boundary_colors(), bracket_matrices(), chroma_trajectory(), max, and target_profile().

◆ rgb_work_to_Yrg()

derive_filmic_agx_primaries.rgb_work_to_Yrg ( rgb )

Definition at line 87 of file derive_filmic_agx_primaries.py.

References xyz_D50_to_Yrg().

Referenced by chroma_trajectory().

◆ scene_ab_target()

derive_filmic_agx_primaries.scene_ab_target	(	S_refl,
		hk_retention = `1.0`
	)

PRINCIPLED uniform apparent-brightness target for the apparent-brightness stabilizers
(--ab-pull on min-bleach, and the reference level for --ab-stabilize on extra), replacing a
hand-tuned constant. = mean over the reflective set of
    curve(L_in) * (1 + hk_retention * H-K_excess(input))
i.e. the apparent brightness a colour gets if the ACHROMATIC tone curve maps its luminance and
it keeps `hk_retention` of its OWN natural (input) Helmholtz-Kohlrausch excess :
    hk_retention = 1   -> SCENE-PRESERVING ceiling (full natural H-K pop, ~0.379)
    hk_retention = 0   -> GRAY-EQUIVALENT floor    (H-K fully neutralized, ~0.336)
    hk_retention = 0.5 -> midway (~0.357) ; LINEAR in hk_retention, so 0.5 IS the average of the two.
Holding every hue at this value preserves the scene's apparent-brightness STRUCTURE (no hue
over/under-brightened relative to the others) at the chosen H-K-retention level. Hue-independent
(the set places every hue at luminance GREY*2^EV, matched), curve-relative (recomputes if the
default curve or the set changes). A chroma-preserving end (min-bleach) wants retention 1 ; a
bleaching end (extra), whose colours lose chroma/H-K, sits lower in the [floor, ceiling] band.

Definition at line 358 of file derive_filmic_agx_primaries.py.

References nayatani_hk_excess().

Referenced by main().

◆ skin_and_reflective_sets()

derive_filmic_agx_primaries.skin_and_reflective_sets ( )

The SINGLE canonical colour-constancy evaluation set — shared by --report AND every
fit mode, so a "desaturation %" (or hue / delta-E / H-K drift) means the same thing
everywhere. Two disjoint arrays, each swept over tonal placements : the skin-tone
database, and the reflective hue circle. The reflective circle spans a purity sweep
from LOW (0.3 : diffuse matte reflectances) to HIGH (0.9 : high-chroma memory colours) —
these used to live in two separate builders (this one and priority_samples), which is
what let the modes disagree on what "reflective" meant ; they are merged here.

Definition at line 379 of file derive_filmic_agx_primaries.py.

References _lab_d65_to_work().

Referenced by fit_midpoint(), main(), report_hue_continuity(), and report_variants().

◆ target_profile()

derive_filmic_agx_primaries.target_profile ( baseline )

Design target for the bleaching : transparent (baseline) below the shoulder,
then a smooth ramp to full achromatic exactly at the white endpoint. This pins
the bleach *rate* through the shoulder — without it the fit has no optimum and
rails the insets (endpoint-only objectives are monotone in the inset amount).

Definition at line 188 of file derive_filmic_agx_primaries.py.

Referenced by residuals().

◆ tone_map()

derive_filmic_agx_primaries.tone_map ( rgb )

Definition at line 98 of file derive_filmic_agx_primaries.py.

References CURVE, and max.

Referenced by chroma_trajectory().

◆ variant_bracket()

derive_filmic_agx_primaries.variant_bracket ( v )

(inset matrix M, applied outset matrix) for a SHIPPED_VARIANTS entry.

Definition at line 289 of file derive_filmic_agx_primaries.py.

References bracket_matrices().

Referenced by fit_midpoint(), main(), per_hue_ab_and_drift(), and report_variants().

◆ xyz_D50_to_Yrg()

derive_filmic_agx_primaries.xyz_D50_to_Yrg ( xyz )

Definition at line 80 of file derive_filmic_agx_primaries.py.

Referenced by bracket_matrices(), and rgb_work_to_Yrg().

Variable Documentation

◆ _BND

list derive_filmic_agx_primaries._BND

protected

Initial value:

1= [np.maximum(np.array(c, float), 1e-6) for c in

2 ([1, 0, 0], [0, 1, 0], [0, 0, 1], [0, 1, 1], [1, 0, 1], [1, 1, 0])]

Definition at line 434 of file derive_filmic_agx_primaries.py.

◆ _BND_EV

tuple derive_filmic_agx_primaries._BND_EV = (-8, -6, -4, -2, -1, 0, 1, 2, 3)

protected

Definition at line 436 of file derive_filmic_agx_primaries.py.

◆ _D65_WHITE

derive_filmic_agx_primaries._D65_WHITE = np.array([0.95047, 1.0, 1.08883])

protected

Definition at line 163 of file derive_filmic_agx_primaries.py.

◆ _SKIN_LAB

list derive_filmic_agx_primaries._SKIN_LAB

protected

Initial value:

=  [  # L avg,std, a avg,std, b avg,std — see color_vocabulary.c for sources
    (60.9, 3.4, 7.0, 1.7, 15.0, 1.8), (61.9, 3.7, 7.1, 1.7, 17.4, 2.0),
    (60.6, 4.8, 6.5, 1.6, 16.4, 2.3), (63.0, 5.5, 5.6, 1.9, 14.0, 2.9),
    (56.4, 3.2, 11.7, 2.1, 16.3, 1.4), (56.8, 4.1, 11.6, 2.2, 17.7, 1.8),
    (56.1, 4.5, 11.3, 2.1, 16.4, 2.2), (59.2, 5.1, 11.6, 2.8, 15.1, 2.3),
    (44.0, 2.0, 14.0, 1.0, 19.0, 1.0), (58.0, 1.0, 15.0, 1.0, 21.0, 1.0),
    (58.9, 3.1, 11.4, 2.1, 14.2, 1.5), (60.7, 4.0, 10.5, 2.3, 17.2, 2.1),
    (58.0, 4.4, 11.7, 2.3, 15.8, 2.1), (59.6, 5.5, 11.8, 3.1, 14.6, 2.6),
    (48.0, 1.0, 15.0, 1.0, 20.0, 1.0), (63.0, 1.0, 16.0, 1.0, 21.0, 1.0)]

Definition at line 153 of file derive_filmic_agx_primaries.py.

◆ BASELINES

dict derive_filmic_agx_primaries.BASELINES = {tuple(s): chroma_trajectory(s, IDENTITY, IDENTITY)[0] for s in boundary_colors()}

Definition at line 186 of file derive_filmic_agx_primaries.py.

◆ BLACK_EV

derive_filmic_agx_primaries.BLACK_EV

Definition at line 74 of file derive_filmic_agx_primaries.py.

◆ CURVE

derive_filmic_agx_primaries.CURVE = curve_factory(BLACK_EV, WHITE_EV, *CURVE_DEFAULTS, shoulder_slope_matched=True)

Definition at line 76 of file derive_filmic_agx_primaries.py.

Referenced by tone_map().

◆ CURVE_DEFAULTS

tuple derive_filmic_agx_primaries.CURVE_DEFAULTS = (1.18, 10.0, 0.0, 1.5, 1.5)

Definition at line 75 of file derive_filmic_agx_primaries.py.

◆ EVS

derive_filmic_agx_primaries.EVS = np.arange(-2.0, WHITE_EV + 0.51, 0.5)

Definition at line 132 of file derive_filmic_agx_primaries.py.

◆ filmlightRGB_to_LMS

derive_filmic_agx_primaries.filmlightRGB_to_LMS

Initial value:

=  np.array([
    [0.95, 0.38, 0.00],
    [0.05, 0.62, 0.03],
    [0.00, 0.00, 0.97]])

Definition at line 53 of file derive_filmic_agx_primaries.py.

◆ HUE_STEPS

int derive_filmic_agx_primaries.HUE_STEPS = 24

Definition at line 133 of file derive_filmic_agx_primaries.py.

◆ IDENTITY

derive_filmic_agx_primaries.IDENTITY = np.eye(3)

Definition at line 185 of file derive_filmic_agx_primaries.py.

◆ LMS_to_filmlightRGB

derive_filmic_agx_primaries.LMS_to_filmlightRGB = np.linalg.inv(filmlightRGB_to_LMS)

Definition at line 57 of file derive_filmic_agx_primaries.py.

◆ LUT_X

derive_filmic_agx_primaries.LUT_X = np.linspace(0.0, 1.0, 8193)

Definition at line 263 of file derive_filmic_agx_primaries.py.

◆ LUT_Y

derive_filmic_agx_primaries.LUT_Y = np.array([CURVE(v) for v in LUT_X])

Definition at line 264 of file derive_filmic_agx_primaries.py.

◆ REC2020_TO_XYZ_D50

derive_filmic_agx_primaries.REC2020_TO_XYZ_D50

Initial value:

=  np.array([
    [0.6734241, 0.1656411, 0.1251286],
    [0.2790177, 0.6753402, 0.0456377],
    [-0.0019300, 0.0299784, 0.7973330]])

Definition at line 64 of file derive_filmic_agx_primaries.py.

◆ SHIPPED_VARIANTS

dict derive_filmic_agx_primaries.SHIPPED_VARIANTS

Definition at line 221 of file derive_filmic_agx_primaries.py.

◆ SHOULDER_EV

derive_filmic_agx_primaries.SHOULDER_EV = BLACK_EV + CURVE.sh_x * (WHITE_EV - BLACK_EV)

Definition at line 145 of file derive_filmic_agx_primaries.py.

◆ WHITE_EV

derive_filmic_agx_primaries.WHITE_EV

Definition at line 74 of file derive_filmic_agx_primaries.py.

◆ WHITE_YRG

derive_filmic_agx_primaries.WHITE_YRG = rgb_work_to_Yrg(np.ones(3))

Definition at line 90 of file derive_filmic_agx_primaries.py.

◆ XYZ_D50_to_D65_CAT16

derive_filmic_agx_primaries.XYZ_D50_to_D65_CAT16

Initial value:

=  np.array([
    [9.89466254e-01, -4.00304626e-02, 4.40530317e-02],
    [-5.40518733e-03, 1.00666069e+00, -1.75551955e-03],
    [-4.03920992e-04, 1.50768030e-02, 1.30210211e+00]])

Definition at line 43 of file derive_filmic_agx_primaries.py.

◆ XYZ_D65_to_LMS_2006

derive_filmic_agx_primaries.XYZ_D65_to_LMS_2006

Initial value:

=  np.array([
    [0.257085, 0.859943, -0.031061],
    [-0.394427, 1.175800, 0.106423],
    [0.064856, -0.076250, 0.559067]])

Definition at line 48 of file derive_filmic_agx_primaries.py.

◆ Y_2006_COEFFS

derive_filmic_agx_primaries.Y_2006_COEFFS = np.array([0.68990272, 0.34832189, 0.0])

Definition at line 59 of file derive_filmic_agx_primaries.py.

Functions

Variables

Detailed Description

Function Documentation

◆ _lab_d65_to_work()

◆ boundary_colors()

◆ bracket_matrices()

◆ brk()

◆ chroma_hue()

◆ chroma_hue_batch()

◆ chroma_trajectory()

◆ delta_e_yrg()

◆ fit_midpoint()

◆ gamut_violation()

◆ hk_drift_batch()

◆ main()

◆ measure_batch()

◆ nayatani_hk_excess()

◆ per_hue_ab_and_drift()

◆ print_c_case()

◆ print_diagnostics()

◆ print_midpoint_constants()

◆ print_variant_entry()

◆ rec2020_worst_boundary_luminance()

◆ report_hue_continuity()

◆ report_variants()

◆ residuals()

◆ rgb_work_to_Yrg()

◆ scene_ab_target()

◆ skin_and_reflective_sets()

◆ target_profile()

◆ tone_map()

◆ variant_bracket()

◆ xyz_D50_to_Yrg()

Variable Documentation

◆ _BND

◆ _BND_EV

◆ _D65_WHITE

◆ _SKIN_LAB

◆ BASELINES

◆ BLACK_EV

◆ CURVE

◆ CURVE_DEFAULTS

◆ EVS

◆ filmlightRGB_to_LMS

◆ HUE_STEPS

◆ IDENTITY

◆ LMS_to_filmlightRGB

◆ LUT_X

◆ LUT_Y

◆ REC2020_TO_XYZ_D50

◆ SHIPPED_VARIANTS

◆ SHOULDER_EV

◆ WHITE_EV

◆ WHITE_YRG

◆ XYZ_D50_to_D65_CAT16

◆ XYZ_D65_to_LMS_2006

◆ Y_2006_COEFFS