def test_n_dimensional_RGB_luminance(self):
"""
Tests:func:`colour.models.rgb.derivation.RGB_luminance` definition
n_dimensional arrays support.
"""
RGB = np.array([50.0, 50.0, 50.0]),
P = np.array([0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
W = np.array([0.32168, 0.33767])
Y = 50
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
RGB = np.tile(RGB, (6, 1))
Y = np.tile(Y, 6)
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
RGB = np.reshape(RGB, (2, 3, 3))
Y = np.reshape(Y, (2, 3))
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
def tonemapping_operator_exponential(RGB,
q=1,
k=1,
colourspace=RGB_COLOURSPACES['sRGB']):
"""
Performs given *RGB* array tonemapping using the exponential method.
Parameters
----------
RGB : array_like
*RGB* array to perform tonemapping onto.
q : numeric, optional
:math:`q`.
k : numeric, optional
:math:`k`.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
ndarray
Tonemapped *RGB* array.
References
----------
:cite:`Banterle2011k`
Examples
--------
>>> tonemapping_operator_exponential(np.array(
... [[[0.48046875, 0.35156256, 0.23632812],
... [1.39843753, 0.55468757, 0.39062594]],
... [[4.40625388, 2.15625895, 1.34375372],
... [6.59375023, 3.43751395, 2.21875829]]]),
... 1.0, 25) # doctest: +ELLIPSIS
array([[[ 0.0148082..., 0.0108353..., 0.0072837...],
[ 0.0428669..., 0.0170031..., 0.0119740...]],
<BLANKLINE>
[[ 0.1312736..., 0.0642404..., 0.0400338...],
[ 0.1921684..., 0.1001830..., 0.0646635...]]])
"""
RGB = as_float_array(RGB)
q = 1 if q < 1 else q
k = 1 if k < 1 else k
L = RGB_luminance(RGB, colourspace.primaries, colourspace.whitepoint)
L_a = log_average(L)
L_d = 1 - np.exp(-(L * q) / (L_a * k))
RGB = RGB * L_d[..., np.newaxis] / L[..., np.newaxis]
return RGB
def tonemapping_operator_logarithmic(RGB,
q=1,
k=1,
colourspace=RGB_COLOURSPACES['sRGB']):
"""
Performs given *RGB* array tonemapping using the logarithmic method.
Parameters
----------
RGB : array_like
*RGB* array to perform tonemapping onto.
q : numeric, optional
:math:`q`.
k : numeric, optional
:math:`k`.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
ndarray
Tonemapped *RGB* array.
References
----------
:cite:`Banterle2011k`
Examples
--------
>>> tonemapping_operator_logarithmic(np.array(
... [[[0.48046875, 0.35156256, 0.23632812],
... [1.39843753, 0.55468757, 0.39062594]],
... [[4.40625388, 2.15625895, 1.34375372],
... [6.59375023, 3.43751395, 2.21875829]]]),
... 1.0, 25) # doctest: +ELLIPSIS
array([[[ 0.0884587..., 0.0647259..., 0.0435102...],
[ 0.2278222..., 0.0903652..., 0.0636376...]],
<BLANKLINE>
[[ 0.4717487..., 0.2308565..., 0.1438669...],
[ 0.5727396..., 0.2985858..., 0.1927235...]]])
"""
RGB = as_float_array(RGB)
q = 1 if q < 1 else q
k = 1 if k < 1 else k
L = RGB_luminance(RGB, colourspace.primaries, colourspace.whitepoint)
L_max = np.max(L)
L_d = np.log10(1 + L * q) / np.log10(1 + L_max * k)
RGB = RGB * L_d[..., np.newaxis] / L[..., np.newaxis]
return RGB
def tonemapping_operator_logarithmic_mapping(RGB,
p=1,
q=1,
colourspace=RGB_COLOURSPACES[
'sRGB']):
"""
Performs given *RGB* array tonemapping using the logarithmic mapping
method.
Parameters
----------
RGB : array_like
*RGB* array to perform tonemapping onto.
p : numeric, optional
:math:`p`.
q : numeric, optional
:math:`q`.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
ndarray
Tonemapped *RGB* array.
References
----------
- :cite:`Schlick1994`
Examples
--------
>>> tonemapping_operator_logarithmic_mapping(np.array(
... [[[0.48046875, 0.35156256, 0.23632812],
... [1.39843753, 0.55468757, 0.39062594]],
... [[4.40625388, 2.15625895, 1.34375372],
... [6.59375023, 3.43751395, 2.21875829]]])) # doctest: +ELLIPSIS
array([[[ 0.2532899..., 0.1853341..., 0.1245857...],
[ 0.6523387..., 0.2587489..., 0.1822179...]],
<BLANKLINE>
[[ 1.3507897..., 0.6610269..., 0.4119437...],
[ 1.6399638..., 0.8549608..., 0.5518382...]]])
"""
RGB = np.asarray(RGB)
L = RGB_luminance(RGB, colourspace.primaries, colourspace.whitepoint)
L_max = np.max(L)
L_d = (np.log(1 + p * L) / np.log(1 + p * L_max))**(1 / q)
RGB = RGB * L_d[..., np.newaxis] / L[..., np.newaxis]
return RGB
def tonemapping_operator_exponentiation_mapping(RGB,
p=1,
q=1,
colourspace=RGB_COLOURSPACES[
'sRGB']):
"""
Performs given *RGB* array tonemapping using the exponentiation mapping
method.
Parameters
----------
RGB : array_like
*RGB* array to perform tonemapping onto.
p : numeric, optional
:math:`p`.
q : numeric, optional
:math:`q`.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
ndarray
Tonemapped *RGB* array.
References
----------
- :cite:`Schlick1994`
Examples
--------
>>> tonemapping_operator_exponentiation_mapping(np.array(
... [[[0.48046875, 0.35156256, 0.23632812],
... [1.39843753, 0.55468757, 0.39062594]],
... [[4.40625388, 2.15625895, 1.34375372],
... [6.59375023, 3.43751395, 2.21875829]]])) # doctest: +ELLIPSIS
array([[[ 0.1194997..., 0.0874388..., 0.0587783...],
[ 0.3478122..., 0.1379590..., 0.0971544...]],
<BLANKLINE>
[[ 1.0959009..., 0.5362936..., 0.3342115...],
[ 1.6399638..., 0.8549608..., 0.5518382...]]])
"""
RGB = np.asarray(RGB)
L = RGB_luminance(RGB, colourspace.primaries, colourspace.whitepoint)
L_max = np.max(L)
L_d = (L / L_max)**(p / q)
RGB = RGB * L_d[..., np.newaxis] / L[..., np.newaxis]
return RGB
def test_RGB_luminance(self):
"""
Test :func:`colour.models.rgb.derivation.RGB_luminance`
definition.
"""
self.assertAlmostEqual(
RGB_luminance(
np.array([0.18, 0.18, 0.18]),
np.array(
[0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]
),
np.array([0.32168, 0.33767]),
),
0.18000000,
places=7,
)
self.assertAlmostEqual(
RGB_luminance(
np.array([0.21959402, 0.06986677, 0.04703877]),
np.array(
[0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]
),
np.array([0.32168, 0.33767]),
),
0.123014562384318,
places=7,
)
self.assertAlmostEqual(
RGB_luminance(
np.array([0.45620519, 0.03081071, 0.04091952]),
np.array([0.6400, 0.3300, 0.3000, 0.6000, 0.1500, 0.0600]),
np.array([0.31270, 0.32900]),
),
0.121995947729870,
places=7,
)
def light_probe_sampling_variance_minimization_Viriyothai2009(
light_probe, lights_count=16, colourspace=RGB_COLOURSPACES['sRGB']):
"""
Sample given light probe to find lights using *Viriyothai (2009)* variance
minimization light probe sampling algorithm.
Parameters
----------
light_probe : array_like
Array to sample for lights.
lights_count : int
Amount of lights to generate.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
list
list of
:class:`colour_hdri.sampling.variance_minimization.Light_Specification`
lights.
References
----------
:cite:`Viriyothai2009`
"""
light_probe = as_float_array(light_probe)
iterations = np.sqrt(lights_count).astype(np.int_)
if iterations**2 != lights_count:
warning(
'{0} lights requested, {1} will be effectively computed!'.format(
lights_count, iterations**2))
Y = RGB_luminance(light_probe, colourspace.primaries,
colourspace.whitepoint)
regions = find_regions_variance_minimization_Viriyothai2009(Y, iterations)
lights = []
for region in regions:
y_min, y_max, x_min, x_max = region
c = centroid(Y[y_min:y_max, x_min:x_max])
c = (c + np.array([y_min, x_min]))
light_probe_c = light_probe[y_min:y_max, x_min:x_max]
lights.append(
Light_Specification((c / np.array(Y.shape))[::-1],
np.sum(np.sum(light_probe_c, 0), 0), c))
return lights
def tonemapping_operator_normalisation(RGB,
colourspace=RGB_COLOURSPACES['sRGB']):
"""
Performs given *RGB* array tonemapping using the normalisation method.
Parameters
----------
RGB : array_like
*RGB* array to perform tonemapping onto.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
ndarray
Tonemapped *RGB* array.
References
----------
:cite:`Banterle2011k`
Examples
--------
>>> tonemapping_operator_normalisation(np.array(
... [[[0.48046875, 0.35156256, 0.23632812],
... [1.39843753, 0.55468757, 0.39062594]],
... [[4.40625388, 2.15625895, 1.34375372],
... [6.59375023, 3.43751395, 2.21875829]]])) # doctest: +ELLIPSIS
array([[[ 0.1194997..., 0.0874388..., 0.0587783...],
[ 0.3478122..., 0.1379590..., 0.0971544...]],
<BLANKLINE>
[[ 1.0959009..., 0.5362936..., 0.3342115...],
[ 1.6399638..., 0.8549608..., 0.5518382...]]])
"""
RGB = as_float_array(RGB)
L = RGB_luminance(RGB, colourspace.primaries, colourspace.whitepoint)
L_max = np.max(L)
RGB = RGB / L_max
return RGB
class TestRGBLuminance(unittest.TestCase):
"""
Defines :func:`colour.models.rgb.derivation.RGB_luminance` definition
unit tests methods.
"""
def test_RGB_luminance(self):
"""
Tests:func:`colour.models.rgb.derivation.RGB_luminance`
definition.
"""
self.assertAlmostEqual(RGB_luminance(
np.array([50.0, 50.0, 50.0]),
np.array([0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
np.array([0.32168, 0.33767])),
50.00000000,
places=7)
self.assertAlmostEqual(RGB_luminance(
np.array([74.6, 16.1, 100.0]),
np.array([0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
np.array([0.32168, 0.33767])),
30.17011667,
places=7)
self.assertAlmostEqual(RGB_luminance(
np.array([40.6, 4.2, 67.4]),
np.array([0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
np.array([0.32168, 0.33767])),
12.16160184,
places=7)
def test_n_dimensional_RGB_luminance(self):
"""
Tests:func:`colour.models.rgb.derivation.RGB_luminance` definition
n_dimensional arrays support.
"""
RGB = np.array([50.0, 50.0, 50.0]),
P = np.array([0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
W = np.array([0.32168, 0.33767])
Y = 50
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
RGB = np.tile(RGB, (6, 1))
Y = np.tile(Y, 6)
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
RGB = np.reshape(RGB, (2, 3, 3))
Y = np.reshape(Y, (2, 3))
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
@ignore_numpy_errors
def test_nan_RGB_luminance(self):
"""
Tests :func:`colour.models.rgb.derivation.RGB_luminance`
definition nan support.
"""
cases = [-1.0, 0.0, 1.0, -np.inf, np.inf, np.nan]
cases = set(permutations(cases * 3, r=3))
for case in cases:
RGB = np.array(case)
P = np.array(np.vstack((case[0:2], case[0:2], case[0:2])))
W = np.array(case[0:2])
try:
RGB_luminance(RGB, P, W)
except np.linalg.linalg.LinAlgError:
import traceback
from colour.utilities import warning
warning(traceback.format_exc())
class TestRGBLuminance(unittest.TestCase):
"""
Defines :func:`colour.models.rgb.derivation.RGB_luminance` definition
unit tests methods.
"""
def test_RGB_luminance(self):
"""
Tests:func:`colour.models.rgb.derivation.RGB_luminance`
definition.
"""
self.assertAlmostEqual(
RGB_luminance(
np.array([0.18, 0.18, 0.18]),
np.array(
[0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
np.array([0.32168, 0.33767])),
0.18000000,
places=7)
self.assertAlmostEqual(
RGB_luminance(
np.array([0.21959402, 0.06986677, 0.04703877]),
np.array(
[0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
np.array([0.32168, 0.33767])),
0.123014562384318,
places=7)
self.assertAlmostEqual(
RGB_luminance(
np.array([0.45620519, 0.03081071, 0.04091952]),
np.array([0.6400, 0.3300, 0.3000, 0.6000, 0.1500, 0.0600]),
np.array([0.31270, 0.32900])),
0.121995947729870,
places=7)
def test_n_dimensional_RGB_luminance(self):
"""
Tests:func:`colour.models.rgb.derivation.RGB_luminance` definition
n_dimensional arrays support.
"""
RGB = np.array([0.18, 0.18, 0.18]),
P = np.array([0.73470, 0.26530, 0.00000, 1.00000, 0.00010, -0.07700]),
W = np.array([0.32168, 0.33767])
Y = 0.18000000
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
RGB = np.tile(RGB, (6, 1))
Y = np.tile(Y, 6)
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
RGB = np.reshape(RGB, (2, 3, 3))
Y = np.reshape(Y, (2, 3))
np.testing.assert_almost_equal(RGB_luminance(RGB, P, W), Y)
@ignore_numpy_errors
def test_nan_RGB_luminance(self):
"""
Tests :func:`colour.models.rgb.derivation.RGB_luminance`
definition nan support.
"""
cases = [-1.0, 0.0, 1.0, -np.inf, np.inf, np.nan]
cases = set(permutations(cases * 3, r=3))
for case in cases:
RGB = np.array(case)
P = np.array(np.vstack([case[0:2], case[0:2], case[0:2]]))
W = np.array(case[0:2])
try:
RGB_luminance(RGB, P, W)
except np.linalg.linalg.LinAlgError:
pass
def tonemapping_operator_Reinhard2004(RGB,
f=0,
m=0.3,
a=0,
c=0,
colourspace=RGB_COLOURSPACES['sRGB']):
"""
Performs given *RGB* array tonemapping using *Reinhard and Devlin (2004)*
method.
Parameters
----------
RGB : array_like
*RGB* array to perform tonemapping onto.
f : numeric, optional
:math:`f`.
m : numeric, optional
:math:`m`.
a : numeric, optional
:math:`a`.
c : numeric, optional
:math:`c`.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
ndarray
Tonemapped *RGB* array.
References
----------
:cite:`Reinhard2005c`
Examples
--------
>>> tonemapping_operator_Reinhard2004(np.array(
... [[[0.48046875, 0.35156256, 0.23632812],
... [1.39843753, 0.55468757, 0.39062594]],
... [[4.40625388, 2.15625895, 1.34375372],
... [6.59375023, 3.43751395, 2.21875829]]]),
... -10) # doctest: +ELLIPSIS
array([[[ 0.0216792..., 0.0159556..., 0.0107821...],
[ 0.0605894..., 0.0249445..., 0.0176972...]],
<BLANKLINE>
[[ 0.1688972..., 0.0904532..., 0.0583584...],
[ 0.2331935..., 0.1368456..., 0.0928316...]]])
"""
RGB = as_float_array(RGB)
C_av = np.array((np.average(RGB[..., 0]), np.average(RGB[..., 1]),
np.average(RGB[..., 2])))
L = RGB_luminance(RGB, colourspace.primaries, colourspace.whitepoint)
L_lav = log_average(L)
L_min, L_max = np.min(L), np.max(L)
f = np.exp(-f)
m = (m if m > 0 else (0.3 + 0.7 * (
(np.log(L_max) - L_lav) / (np.log(L_max) - np.log(L_min)) ** 1.4)))
I_l = (c * RGB + (1 - c)) * L[..., np.newaxis]
I_g = c * C_av + (1 - c) * L_lav
I_a = a * I_l + (1 - a) * I_g
RGB = RGB / (RGB + (f * I_a) ** m)
return RGB
def tonemapping_operator_Tumblin1999(RGB,
L_da=20,
C_max=100,
L_max=100,
colourspace=RGB_COLOURSPACES['sRGB']):
"""
Performs given *RGB* array tonemapping using
*Tumblin, Hodgins and Guenter (1999)* method.
Parameters
----------
RGB : array_like
*RGB* array to perform tonemapping onto.
L_da : numeric, optional
:math:`L_{da}` display adaptation luminance, a mid-range display value.
C_max : numeric, optional
:math:`C_{max}` maximum contrast available from the display.
L_max : numeric, optional
:math:`L_{max}` maximum display luminance.
colourspace : `colour.RGB_Colourspace`, optional
*RGB* colourspace used for internal *Luminance* computation.
Returns
-------
ndarray
Tonemapped *RGB* array.
References
----------
:cite:`Tumblin1999c`
Examples
--------
>>> tonemapping_operator_Tumblin1999(np.array(
... [[[0.48046875, 0.35156256, 0.23632812],
... [1.39843753, 0.55468757, 0.39062594]],
... [[4.40625388, 2.15625895, 1.34375372],
... [6.59375023, 3.43751395, 2.21875829]]])) # doctest: +ELLIPSIS
array([[[ 0.0400492..., 0.0293043..., 0.0196990...],
[ 0.1019768..., 0.0404489..., 0.0284852...]],
<BLANKLINE>
[[ 0.2490212..., 0.1218618..., 0.0759427...],
[ 0.3408366..., 0.1776880..., 0.1146895...]]])
"""
RGB = as_float_array(RGB)
L_w = RGB_luminance(RGB, colourspace.primaries, colourspace.whitepoint)
def f(x):
return np.where(x > 100, 2.655,
1.855 + 0.4 * np.log10(x + 2.3 * 10 ** -5))
L_wa = np.exp(np.mean(np.log(L_w + 2.3 * 10 ** -5)))
g_d = f(L_da)
g_w = f(L_wa)
g_wd = g_w / (1.855 + 0.4 * np.log(L_da))
mL_wa = np.sqrt(C_max) ** (g_wd - 1)
L_d = mL_wa * L_da * (L_w / L_wa) ** (g_w / g_d)
RGB = RGB * L_d[..., np.newaxis] / L_w[..., np.newaxis]
RGB = RGB / L_max
return RGB
