def add(rgb, severity, jab_set, cvd_type='deuteranomaly'):
if severity is None: # Full trichromatic vision
jab = cmu.convert(rgb, 'sRGB1', UNIFORM_SPACE)
else: # Simulate CVD
jab = cmu.convert(cvu.get_cvd(rgb, severity=severity, cvd_type=cvd_type), 'sRGB1', UNIFORM_SPACE)
# Add to set
jab_set.add(tuple(jab))
def get_cvd(data, cvd_type=CVD_TYPE, severity=100):
'''
Converts RGB values to CVD space RGB values.
Parameters
----------
data: string or 3 x 256 array
Colormap name OR array with complete color data. Invalid
colormap names throw a ValueError. Refer to _check_cmap for
more information.
cvd_type: string
Type of CVD to be simulated. Options: deuteranomaly or
protanomaly. Default: deuteranomaly.
severity: int
Severity of CVD to be simulated. Can be any integer between 0
and 100. Default: 100
Returns
----------
cvd: 3 x 256 array
Colormap data in CVD space
'''
rgb, _ = cmu.get_rgb_jab(data, calc_jab=False)
cvd_space = {
'name': 'sRGB1+CVD',
'cvd_type': cvd_type,
'severity': severity
}
cvd = cmu.convert(rgb, cvd_space, CSPACE1)
return cvd
def _iter_make_linear(jab):
# Linearize a' and b' changes
jab1 = cmu.make_linear(np.copy(jab))
# Convert J'a'b' values to RGB values and clip
rgb1 = np.clip(cmu.convert(jab1, CSPACE2, CSPACE1), 0, 1)
# Convert RGB to CVD RGB space
rgb2 = get_cvd(rgb1)
# Bring back to to J'a'b'
jab2 = cmu.convert(rgb2, CSPACE1, CSPACE2)
# Force to have same J' as original J'a'b' array
jab2[0, :] = jab1[0, :]
return jab2
def iter_make_linear(jab, l=100000):
'''
Takes J'a'b' array and performs the following two times: linearizes
the change of a' and b' then converts to CVD space and back (to
ensure colormap is still CVD compatible).
Parameters
----------
data: 3 x 256 array
J'a'b' values for colormap
Returns
----------
rgb: 3 x 256 array
RGB data
jab: 3 x 256 array
J'a'b' data
'''
jab = np.copy(jab)
jab = _iter_make_linear(_iter_make_linear(jab))
rgb = cmu.convert(jab, CSPACE2, CSPACE1)
return rgb, jab
FAX = 16
#%% Get iteration data
cmap = 'viridis' # The optimized version of this colormap is cividis!
# Get original RGB/Jab
t = time()
rgb1, jab1 = cmu.get_rgb_jab(cmap)
# Get CVD RGB/Jab
rgb2, jab2 = cmu.get_rgb_jab(cvu.get_cvd(rgb1, severity=100))
# Make perceptual deltas (for a' vs. b') straight
# Need to set a and b before straightening J
jab3 = cmu.make_linear(jab2)
rgb3 = cmu.convert(jab3, cmu.CSPACE2, cmu.CSPACE1)
# Linearize J'
jab4, jab5 = cmu.correct_J(
jab3) # This will create a figure showing the J bounds
if jab4 is not None:
rgb4 = cmu.convert(jab4, cmu.CSPACE2, cmu.CSPACE1)
rgb4 = np.clip(rgb4, 0, 1)
if jab5 is not None:
rgb5 = cmu.convert(jab5, cmu.CSPACE2, cmu.CSPACE1)
rgb5 = np.clip(rgb5, 0, 1)
#%% Figure: Iterations for optimization
fig = plt.figure(figsize=(25, 12), dpi=500)
# Globals FLABEL = 20 FAX = 16 # Input colormap name cmap = 'viridis' # Optimize rgb1, jab1 = cmu.get_rgb_jab(cmap) # Original colormap rgb2, jab2 = cmu.get_rgb_jab(cvu.get_cvd(rgb1)) # CVD colormap jab3 = cmu.make_linear(jab2) # Uniformize hue (a' vs. b') _, jab4 = cmu.correct_J(jab3) # Linearize J' # Convert back to sRGB rgb4 = cmu.convert(jab4, cmu.CSPACE2, cmu.CSPACE1) rgb4 = np.clip(rgb4, 0, 1) # Resimulate CVD in case corrections took the map outside CVD-safe space rgb4 = cvu.get_cvd(rgb4) #%% Creat CDPS plots (shown in Fig 1 and 5 of the paper) # Import test data img_name = 'example_nanosims_image.txt' # Image used for paper img = np.loadtxt(img_name)[45:, :-45] # Make square high = 3 low = -1 # Std. Dev bounds for normalizing img = cmu.bound(cmu.normalize(img), high, low) # Normalize slice_img = np.array(img[265, 50:-50], ndmin=2) # Slice used in paper gslope = 27.4798474 # Slope for gray. Used to normalize CDPS slopes.