"""
Metaparameter functions allow for the presentation of complex patterns that are
coordinated across different inputs (controlled via the input name) or
controlled by a higher-level parameter.
Current examples are transformations of contrast between the stimulus and the
background based on a contrast parameter, which can be set to implement
Michelson or Weber contrast or the coordinated presentation of input patterns
to each eye, which assigns the correct input pattern based on whether an input
source contains 'left' or 'right'.
"""
import numpy as np
from colorsys import hsv_to_rgb
import param
import imagen
from imagen import Sweeper
from imagen.patterngenerator import ComposeChannels
from imagen.image import ScaleChannels
[docs]class contrast2centersurroundscale(param.ParameterizedFunction):
"""
Allows controlling the contrast in orientation contrast patterns, where the
contrast of the central stimulus has to be set independently from the
surround annulus and the background. Can be controlled with the
contrast_parameter, which can be set to one of three values:
'michelson_contrast', 'weber_contrast' or simply 'scale'.
"""
contrast_parameter = param.String(default='weber_contrast')
def __call__(self, inputs, features):
if "contrastcenter" in features:
if self.contrast_parameter == 'michelson_contrast':
for g in inputs.itervalues():
g.offsetcenter = 0.5
g.scalecenter = 2*g.offsetcenter * g.contrastcenter/100.0
elif self.contrast_parameter == 'weber_contrast':
# Weber_contrast is currently only well defined for
# the special case where the background offset is equal
# to the target offset in the pattern type
# SineGrating(mask_shape=Disk())
for g in inputs.itervalues():
g.offsetcenter = 0.5 #In this case this is the offset of
# both the background and the sine grating
g.scalecenter = 2*g.offsetcenter * g.contrastcenter/100.0
elif self.contrast_parameter == 'scale':
for g in inputs.itervalues():
g.offsetcenter = 0.0
g.scalecenter = g.contrastcenter
if "contrastsurround" in features:
if self.contrast_parameter == 'michelson_contrast':
for g in inputs.itervalues():
g.offsetsurround = 0.5
g.scalesurround = 2*g.offsetsurround * g.contrastsurround/100.0
elif self.contrast_parameter == 'weber_contrast':
# Weber_contrast is currently only well defined for
# the special case where the background offset is equal
# to the target offset in the pattern type
# SineGrating(mask_shape=Disk())
for g in inputs.itervalues():
g.offsetsurround = 0.5 #In this case this is the offset
# of both the background and the sine grating
g.scalesurround = 2*g.offsetsurround * g.contrastsurround/100.0
elif self.contrast_parameter == 'scale':
for g in inputs.itervalues():
g.offsetsurround = 0.0
g.scalesurround = g.contrastsurround
[docs]class contrast2scale(param.ParameterizedFunction):
"""
Coordinates complex contrast values in single and compound patterns.
To change the contrast behavior change the contrast_parameter to one of
three values: 'michelson_contrast', 'weber_contrast' or simply 'scale'.
"""
contrast_parameter = param.String(default='michelson_contrast')
def __call__(self, inputs, features):
if "contrast" in features:
if self.contrast_parameter == 'michelson_contrast':
for g in inputs.itervalues():
g.offset = 0.5
g.scale = 2*g.offset * g.contrast/100.0
elif self.contrast_parameter == 'weber_contrast':
# Weber_contrast is currently only well defined for
# the special case where the background offset is equal
# to the target offset in the pattern type
# SineGrating(mask_shape=Disk())
for g in inputs.itervalues():
g.offset = 0.5 #In this case this is the offset of both
# the background and the sine grating
g.scale = 2*g.offset * g.contrast/100.0
elif self.contrast_parameter == 'scale':
for g in inputs.itervalues():
g.offset = 0.0
g.scale = g.contrast
[docs]class direction2translation(param.ParameterizedFunction):
"""
Coordinates the presentation of moving patterns. Currently
supports an old and new motion model.
"""
def __call__(self, inputs, features):
if 'direction' in features:
import __main__ as main
if '_old_motion_model' in main.__dict__ and main.__dict__['_old_motion_model']:
for name in inputs:
try:
offset = int(name[-1])
except:
if not hasattr(self, 'direction_warned'):
self.warning('Assuming step is zero; no input lag'
' number specified at the end of the'
' input sheet name.')
self.direction_warned = True
offset = 0
setattr(inputs[name], 'orientation', features['direction']-np.pi/2)
inputs[name] = Sweeper(generator=inputs[name],
step_offset=offset,
speed=features['speed'],
reset_period=1.0)
else:
for name in inputs:
setattr(inputs[name], 'orientation', features['direction']+np.pi/2)
inputs[name] = Sweeper(generator=inputs[name],
speed=features['speed'],
reset_period=features['duration'])
[docs]class phasedisparity2leftrightphase(param.ParameterizedFunction):
"""
Coordinates phase disparity between two eyes, by looking for
the keywords Left and Right in the input names.
"""
def __call__(self, inputs, features):
if "phasedisparity" in features:
temp_phase1 = features['phase'] - features['phasedisparity']/2.0
temp_phase2 = features['phase'] + features['phasedisparity']/2.0
for name in inputs.keys():
if (name.count('Right')):
inputs[name].phase = temp_phase1 % 2*np.pi
elif (name.count('Left')):
inputs[name].phase = temp_phase2 % 2*np.pi
else:
if not hasattr(self, 'disparity_warned'):
self.warning('Unable to measure disparity preference, '
'because disparity is defined only when '
'there are inputs for Right and Left '
'retinas.')
self.disparity_warned = True
[docs]class hue2rgbscale(param.ParameterizedFunction):
"""
Coordinates hue between inputs with Red, Green or Blue in their
name.
"""
def __call__(self, inputs, features):
if 'hue' in features:
for name in inputs.keys():
r, g, b = hsv_to_rgb(features['hue'], 1.0, 1.0)
if (name.count('Red')):
inputs[name].scale = r
elif (name.count('Green')):
inputs[name].scale = g
elif (name.count('Blue')):
inputs[name].scale = b
else:
# Note: this warning will be skipped if any retina
# has 'Red', 'Green', or 'Blue' in its
# name. E.g. if there is only a 'RedRetina', the
# warning will be skipped and a hue map will be
# measured without error.
if not hasattr(self, 'hue_warned'):
self.warning('Unable to measure hue preference,'
' because hue is defined only when'
' there are different input sheets'
' with names with Red, Green or Blue'
' substrings.')
self.hue_warned = True
#TFALERT: Should be merged with hue2rgbscale and checking whether
#there are multiple channels or not.
class hue2rgbscaleNewRetina(param.ParameterizedFunction):
"""
Coordinates hue within the new multichannel retina object.
"""
def __call__(self, inputs, features):
if 'hue' in features:
hue_to_rgb = imagen.colorspaces.color_conversion.analysis2working
sat_for_analysis = 1.0
r,g,b = hue_to_rgb((features['hue'],sat_for_analysis,1.0))
for name in inputs.keys():
inputs[name] = ComposeChannels(
generators=[inputs[name]]*3,
channel_transforms = [ScaleChannels(channel_factors = [r, g, b])])
[docs]class ocular2leftrightscale(param.ParameterizedFunction):
"""
Coordinates patterns between two eyes, by looking for the
keywords Left and Right in the input names.
"""
def __call__(self, inputs, features):
if "ocular" in features:
for name in inputs.keys():
if (name.count('Right')):
inputs[name].scale = 2 * features['ocular']
elif (name.count('Left')):
inputs[name].scale = 2.0 - 2*features['ocular']
else:
self.warning('Skipping input region %s; Ocularity is defined '
'only for Left and Right retinas.' % name)
#TFALERT: Should be merged with hue2rgbscale and checking whether
#there are multiple channels or not.
class ocular2leftrightscaleNewRetina(param.ParameterizedFunction):
"""
Coordinates patterns between two eyes, by looking for the
keywords Left and Right in the input names. This class is
intended for the new multichannel retina object.
"""
def __call__(self, inputs, features):
if "ocular" in features:
oc = features['ocular']
for name in inputs.keys():
if (name.count('Right')):
inputs[name] = ComposeChannels(generators=[inputs[name]]*3,
channel_transforms =
[ScaleChannels(channel_factors=[oc, oc, oc])])
elif (name.count('Left')):
inputs[name] = ComposeChannels(generators=[inputs[name]]*3,
channel_transforms =
[ScaleChannels(channel_factors=[1.0-oc, 1.0-oc, 1.0-oc])])
else:
self.warning('Skipping input region %s; Ocularity is defined '
'only for Left and Right retinas.' % name)
__all__ = ['contrast2centersurroundscale',
'contrast2scale',
'hue2rgbscale',
'phasedisparity2leftrightphase',
'direction2translation',
'ocular2leftrightscale']