RGB Twin¶

The 'RGB-twin' is an innovative art data visualisation created with Python that takes an image and extracts its RGB channels to create a unique 3D representation. The concept for this project emerged unexpectedly when I was selecting a picture for my home page, and I seized the opportunity to demonstrate my Python skills in a creative and artistic manner.

In [1]:
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
In [5]:
#Colors
background = '#F0F0F0'
red = '#ED462F'
pink = '#F1BBB8'
green = '#4CB866'
blue = '#003D83'
black = '#333333'
grey = '#D0D9DC'
white = 'white'

#Styles
plt.rcParams['axes.facecolor'] = background
plt.rcParams['axes.edgecolor'] = background
plt.rcParams['figure.facecolor'] = background
plt.rcParams['font.family'] = 'monospace'
plt.rcParams['xtick.color'] = background
plt.rcParams['ytick.color'] = background
plt.rcParams['xtick.labelcolor'] = '#8C9496'
plt.rcParams['ytick.labelcolor'] = '#8C9496'
plt.rcParams['legend.labelcolor'] = '#8C9496'
plt.rcParams['grid.color'] = background
plt.rcParams['axes.axisbelow'] = True

Image conversion¶

1. Image as an array of RGB codes¶

In [2]:
#Transform the image into an array of RGB codes
image_as_array = mpimg.imread('Profile picture small.jpg')
image_as_array
Out[2]:
array([[[248, 248, 246],
        [248, 248, 246],
        [248, 248, 246],
        ...,
        [248, 248, 246],
        [248, 248, 246],
        [248, 248, 246]],

       [[248, 248, 246],
        [248, 248, 246],
        [248, 248, 246],
        ...,
        [248, 248, 246],
        [248, 248, 246],
        [248, 248, 246]],

       [[248, 248, 246],
        [248, 248, 246],
        [248, 248, 246],
        ...,
        [248, 248, 246],
        [248, 248, 246],
        [248, 248, 246]],

       ...,

       [[250, 250, 250],
        [250, 250, 250],
        [250, 250, 250],
        ...,
        [250, 250, 250],
        [250, 250, 250],
        [250, 250, 250]],

       [[250, 250, 250],
        [250, 250, 250],
        [250, 250, 250],
        ...,
        [250, 250, 250],
        [250, 250, 250],
        [250, 250, 250]],

       [[250, 250, 250],
        [250, 250, 250],
        [250, 250, 250],
        ...,
        [250, 250, 250],
        [250, 250, 250],
        [250, 250, 250]]], dtype=uint8)
In [3]:
#Checking the shape of the array
image_as_array.shape
Out[3]:
(533, 400, 3)
In [6]:
#Plotting the image
plt.figure()
plt.gca().set_facecolor(grey)
plt.xlim(-50,450)
plt.ylim(583,-50)
plt.imshow(image_as_array)
plt.grid();

2. Explore RGB distribution¶

In [7]:
#Extract dimensional data from the array
(h,w,c) = image_as_array.shape
In [8]:
#Get coordinates for each channel
rs = []
for x in range(0,h):
    for y in range(0,w):
        rs.append(image_as_array[-x,y,0])
gs = []
for x in range(0,h):
    for y in range(0,w):
        gs.append(image_as_array[-x,y,1])
bs = []
for x in range(0,h):
    for y in range(0,w):
        bs.append(image_as_array[-x,y,1])
rgbs = []
for x in range(0,h):
    for y in range(0,w):
        rgbs.append('#%02x%02x%02x' % tuple(image_as_array[-x,y]))
In [9]:
#Plot the colors in a 3D space
fig = plt.figure(figsize=(10,10))
ax = fig.add_subplot(projection='3d')
ax.scatter(rs,gs,bs,s=0.01,c=rgbs)
ax.zaxis.set_pane_color(grey)
ax.yaxis.set_pane_color(grey)
ax.xaxis.set_pane_color(grey)
ax.zaxis.line.set_color(background)
ax.yaxis.line.set_color(background)
ax.xaxis.line.set_color(background)
ax.view_init(elev=0, azim=-45, roll=0)
ax.set_proj_type('persp', focal_length=0.2)
ax.set_title('_Colours distribution in RGB space',pad=-10,fontweight='bold')
ax.set_zlabel('Blue',labelpad=10)
ax.set_xlabel('Red',labelpad=10)
ax.set_ylabel('Green',labelpad=10)
;
Out[9]:
''

3D projections of RGB channels¶

1. Coordinates for height, width and each RGB channel¶

In [10]:
#Create X coordinates using the height dimension
xs = list(range(0,w))*h
In [11]:
#Create Z coordinates using the width dimension
zs = []
for i in range(0,h):
    zs += [i]*w
In [12]:
#Create Y coordinates for the red channel
ys_r = []
for x in range(0,h):
    for y in range(0,w):
        ys_r.append(image_as_array[-x,y,0])
In [13]:
#Create cropped coordinates for the red channel to avoid plotting a flat background
xs_r_cropped = []
ys_r_cropped = []
zs_r_cropped = []
for i,y in enumerate(ys_r):
    if y < 200:
        xs_r_cropped.append(xs[i])
        ys_r_cropped.append(ys_r[i])
        zs_r_cropped.append(zs[i])
In [14]:
#Create Y coordinates for the green channel
ys_g = []
for x in range(0,h):
    for y in range(0,w):
        ys_g.append(image_as_array[-x,y,1])
In [15]:
#Create cropped coordinates for the green channel to avoid plotting a flat background
xs_g_cropped = []
ys_g_cropped = []
zs_g_cropped = []
for i,y in enumerate(ys_g):
    if y < 200:
        xs_g_cropped.append(xs[i])
        ys_g_cropped.append(ys_g[i])
        zs_g_cropped.append(zs[i])
In [16]:
#Create Y coordinates for the blue channel
ys_b = []
for x in range(0,h):
    for y in range(0,w):
        ys_b.append(image_as_array[-x,y,2])
In [17]:
#Create cropped coordinates for the blue channel to avoid plotting a flat background
xs_b_cropped = []
ys_b_cropped = []
zs_b_cropped = []
for i,y in enumerate(ys_b):
    if y < 200:
        xs_b_cropped.append(xs[i])
        ys_b_cropped.append(ys_b[i])
        zs_b_cropped.append(zs[i])

2. 3D plot for all channels¶

In [18]:
#Plotting in 3D height along Z axis, width along X axis and all three channels along Y axis
fig = plt.figure(figsize=(10,10)) 
ax = fig.add_subplot(projection='3d')
ax.scatter(xs_r_cropped,ys_r_cropped,zs_r_cropped,s=0.01,color=red,alpha=0.3)
ax.scatter(xs_g_cropped,ys_g_cropped,zs_g_cropped,s=0.01,color=green,alpha=0.3)
ax.scatter(xs_b_cropped,ys_b_cropped,zs_b_cropped,s=0.01,color=blue,alpha=0.3)
ax.zaxis.set_pane_color(grey)
ax.yaxis.set_pane_color(grey)
ax.xaxis.set_pane_color(grey)
ax.zaxis.line.set_color(background)
ax.yaxis.line.set_color(background)
ax.xaxis.line.set_color(background)
ax.view_init(elev=0, azim=-45, roll=0)
ax.set_proj_type('persp', focal_length=0.2)
ax.set_aspect('equal')
ax.set_title('_RGB channels depth',pad=-10,fontweight='bold')
ax.set_zlabel('Height',labelpad=10)
ax.set_xlabel('Width',labelpad=10)
ax.set_ylabel('RGB code',labelpad=10) 
;
Out[18]:
''

3. Channels distribution¶

In [19]:
#Plotting the distribution of all three channels present in the image
fig2,axes = plt.subplots(nrows=3,ncols=1,figsize=(5,10))
plt.subplots_adjust(hspace=0.5) 
axes[0].hist(rs,color=red,rwidth=0.7)
axes[0].set_facecolor(grey)
axes[0].grid()
axes[0].set_title('_Red channel distribution',pad=10,fontweight='bold')
axes[0].set_xlabel('R code')
axes[0].set_ylabel('Amount of points')
axes[1].hist(gs,color=green,rwidth=0.7)
axes[1].set_facecolor(grey)
axes[1].grid()
axes[1].set_title('_Green channel distribution',pad=10,fontweight='bold')
axes[1].set_xlabel('G code')
axes[1].set_ylabel('Amount of points')
axes[2].hist(bs,color=blue,rwidth=0.7)
axes[2].set_facecolor(grey)
axes[2].grid()
axes[2].set_title('_Blue channel distribution',pad=10,fontweight='bold')
axes[2].set_xlabel('B code')
axes[2].set_ylabel('Amount of points');

4. 3D plots for each channel separately¶

In [20]:
#Plotting in 3D height along Z axis, width along X axis and the red along Y axis
fig = plt.figure(figsize=(10,10)) 
ax = fig.add_subplot(projection='3d')
ax.scatter(xs_r_cropped,ys_r_cropped,zs_r_cropped,s=0.01,color=red,alpha=0.3)
ax.zaxis.set_pane_color(grey)
ax.yaxis.set_pane_color(grey)
ax.xaxis.set_pane_color(grey)
ax.zaxis.line.set_color(background)
ax.yaxis.line.set_color(background)
ax.xaxis.line.set_color(background)
ax.view_init(elev=0, azim=-45, roll=0)
ax.set_proj_type('persp', focal_length=0.2)
ax.set_aspect('equal')
ax.set_title('_Red channel depth',pad=-10,fontweight='bold')
ax.set_zlabel('Height',labelpad=10)
ax.set_xlabel('Width',labelpad=10)
ax.set_ylabel('R code',labelpad=10)
;
Out[20]:
''
In [21]:
#Plotting in 3D height along Z axis, width along X axis and the green along Y axis
fig = plt.figure(figsize=(10,10)) 
ax = fig.add_subplot(projection='3d')
ax.scatter(xs_g_cropped,ys_g_cropped,zs_g_cropped,s=0.01,color=green,alpha=0.3)
ax.zaxis.set_pane_color(grey)
ax.yaxis.set_pane_color(grey)
ax.xaxis.set_pane_color(grey)
ax.zaxis.line.set_color(background)
ax.yaxis.line.set_color(background)
ax.xaxis.line.set_color(background)
ax.view_init(elev=0, azim=-45, roll=0)
ax.set_proj_type('persp', focal_length=0.2)
ax.set_aspect('equal')
ax.set_title('_Green channel depth',pad=-10,fontweight='bold')
ax.set_zlabel('Height',labelpad=10)
ax.set_xlabel('Width',labelpad=10)
ax.set_ylabel('G code',labelpad=10)
;
Out[21]:
''
In [22]:
#Plotting in 3D height along Z axis, width along X axis and the blue along Y axis
fig = plt.figure(figsize=(10,10)) 
ax = fig.add_subplot(projection='3d')
ax.scatter(xs_b_cropped,ys_b_cropped,zs_b_cropped,s=0.01,color=blue,alpha=0.3)
ax.zaxis.set_pane_color(grey)
ax.yaxis.set_pane_color(grey)
ax.xaxis.set_pane_color(grey)
ax.zaxis.line.set_color(background)
ax.yaxis.line.set_color(background)
ax.xaxis.line.set_color(background)
ax.view_init(elev=0, azim=-45, roll=0)
ax.set_proj_type('persp', focal_length=0.2)
ax.set_aspect('equal')
ax.set_title('_Blue channel depth',pad=-10,fontweight='bold')
ax.set_zlabel('Height',labelpad=10)
ax.set_xlabel('Width',labelpad=10)
ax.set_ylabel('B code',labelpad=10)
;
Out[22]:
''

5. Ready art works¶

In [23]:
#3D plot of the red and the green channels
fig = plt.figure(figsize=(10,10))
ax = fig.add_subplot(projection='3d')
ax.scatter(xs_r_cropped,ys_r_cropped,zs_r_cropped,s=0.01,color=red,alpha=0.3)
ax.scatter(xs_g_cropped,ys_g_cropped,zs_g_cropped,s=0.01,color=green,alpha=0.3)
ax.view_init(elev=0, azim=-75, roll=0)
ax.set_aspect('equal')
ax.axis('off');
In [24]:
#3D plot of the red and the blue channels
fig = plt.figure(figsize=(10,10))
ax = fig.add_subplot(projection='3d')
ax.scatter(xs_r_cropped,ys_r_cropped,zs_r_cropped,s=0.01,color=red,alpha=0.3)
ax.scatter(xs_b_cropped,ys_b_cropped,zs_b_cropped,s=0.01,color=blue,alpha=0.3)
ax.view_init(elev=0, azim=-60, roll=0)
ax.set_aspect('equal')
ax.axis('off');
In [25]:
#3D plot of the blue and the green channels
fig = plt.figure(figsize=(10,10))
ax = fig.add_subplot(projection='3d')
ax.scatter(xs_g_cropped,ys_g_cropped,zs_g_cropped,s=0.01,color=green,alpha=0.3)
ax.scatter(xs_b_cropped,ys_b_cropped,zs_b_cropped,s=0.01,color=blue,alpha=0.3)
ax.view_init(elev=0, azim=-45, roll=0)
ax.set_aspect('equal')
ax.axis('off');

The variety of artworks is limitless; the possibilities for different angles and color combinations are virtually infinite. With every new piece, unique visions and perspectives can be explored, creating something entirely new.

In [ ]: