2 # -*- coding: utf-8 -*-
5 Implements support for *ARRI* colorspaces conversions and transfer functions.
8 from __future__ import division
14 import PyOpenColorIO as ocio
16 import aces_ocio.generate_lut as genlut
17 from aces_ocio.utilities import ColorSpace, mat44_from_mat33, sanitize
19 __author__ = 'ACES Developers'
20 __copyright__ = 'Copyright (C) 2014 - 2015 - ACES Developers'
22 __maintainer__ = 'ACES Developers'
23 __email__ = 'aces@oscars.org'
24 __status__ = 'Production'
26 __all__ = ['create_log_c',
30 def create_log_c(gamut,
45 Parameter description.
50 Return value description.
53 name = '%s (EI%s) - %s' % (transfer_function, exposure_index, gamut)
54 if transfer_function == '':
55 name = 'Linear - ARRI %s' % gamut
57 name = 'Curve - %s (EI%s)' % (transfer_function, exposure_index)
62 cs.equality_group = ''
63 cs.family = 'Input/ARRI'
66 if gamut and transfer_function:
67 cs.aces_transform_id = (
68 'IDT.ARRI.Alexa-v3-logC-EI%s.a1.v1' % exposure_index)
70 # A linear space needs allocation variables
71 if transfer_function == '':
72 cs.allocation_type = ocio.Constants.ALLOCATION_LG2
73 cs.allocation_vars = [-8, 5, 0.00390625]
76 IDT_maker_version = '0.08'
79 black_signal = 0.003907
80 mid_gray_signal = 0.01
81 encoding_gain = 0.256598
82 encoding_offset = 0.391007
85 return (math.log(EI / nominal_EI) / math.log(2) * (
86 0.89 - 1) / 3 + 1) * encoding_gain
88 def log_c_inverse_parameters_for_EI(EI):
90 slope = 1 / (cut * math.log(10))
91 offset = math.log10(cut) - slope * cut
92 gain = EI / nominal_EI
93 gray = mid_gray_signal / gain
94 # The higher the EI, the lower the gamma.
95 enc_gain = gain_for_EI(EI)
96 enc_offset = encoding_offset
98 nz = ((95 / 1023 - enc_offset) / enc_gain - offset) / slope
99 enc_offset = encoding_offset - math.log10(1 + nz) * enc_gain
102 b = nz - black_signal / gray
103 e = slope * a * enc_gain
104 f = enc_gain * (slope * b + offset) + enc_offset
106 # Ensuring we can return relative exposure.
116 'cut': (cut - b) / a,
122 def normalized_log_c_to_linear(code_value, exposure_index):
123 p = log_c_inverse_parameters_for_EI(exposure_index)
124 breakpoint = p['e'] * p['cut'] + p['f']
125 if code_value > breakpoint:
126 linear = ((pow(10, (code_value - p['d']) / p['c']) -
129 linear = (code_value - p['f']) / p['e']
132 cs.to_reference_transforms = []
134 if transfer_function == 'V3 LogC':
135 data = array.array('f', '\0' * lut_resolution_1d * 4)
136 for c in range(lut_resolution_1d):
137 data[c] = normalized_log_c_to_linear(c / (lut_resolution_1d - 1),
140 lut = '%s_to_linear.spi1d' % (
141 '%s_%s' % (transfer_function, exposure_index))
146 os.path.join(lut_directory, lut),
153 cs.to_reference_transforms.append({
156 'interpolation': 'linear',
157 'direction': 'forward'
160 if gamut == 'Wide Gamut':
161 cs.to_reference_transforms.append({
163 'matrix': mat44_from_mat33([0.680206, 0.236137, 0.083658,
164 0.085415, 1.017471, -0.102886,
165 0.002057, -0.062563, 1.060506]),
166 'direction': 'forward'
169 cs.from_reference_transforms = []
173 def create_colorspaces(lut_directory, lut_resolution_1d):
175 Generates the colorspace conversions.
180 Parameter description.
185 Return value description.
190 transfer_function = 'V3 LogC'
193 # EIs = [160, 200, 250, 320, 400, 500, 640, 800,
194 # 1000, 1280, 1600, 2000, 2560, 3200]
195 EIs = [160, 200, 250, 320, 400, 500, 640, 800,
196 1000, 1280, 1600, 2000, 2560, 3200]
201 log_c_EI_full = create_log_c(
208 ['%sei%s_%s' % ('logc3', str(EI), 'arriwide')])
209 colorspaces.append(log_c_EI_full)
213 log_c_EI_linearization = create_log_c(
220 ['crv_%sei%s' % ('logc3', str(EI))])
221 colorspaces.append(log_c_EI_linearization)
224 log_c_EI_primaries = create_log_c(
231 ['%s_%s' % ('lin', 'arriwide')])
232 colorspaces.append(log_c_EI_primaries)