Create your own Random Dot Stereogram (RDS) in Python.

Random Dot Stereogram (RDS)

The original uploader has been moved from Los Hawlos on German Wikipedia to Commons from de.wikipedia. , CC display-inheritance 3.0, by https://commons.wikimedia.org/w/index.php?curid=1966500

Can you see it? By shifting the focus of the eyes back and forth, pictures and letters emerge (look three-dimensional). I'm sorry for those who are watching from a smartphone. Look from your computer.

I want to make this myself.

There was an implementation example in Python on the following site, so I made an original random dot stereogram by referring to it.

• Making Your Own Autostereograms using Python

What to use

import numpy as np
import matplotlib.pyplot as plt
import cv2

I've been experimenting with Notebook format and used the familiar matplotlib to display images, but anything is fine as long as you can see the images.

procedure

1. Create a random pattern

def make_pattern(shape=(16, 16)):
    return np.random.uniform(0, 1, shape)

Try to run it.

pattern = make_pattern((400,400))
plt.imshow(pattern, cmap='gray')

Something is already about to emerge.

2. Create a pattern that emerges

def make_depthmap(shape=(400, 600)):
    depthmap = np.zeros(shape, dtype=np.float)
    cv2.circle(depthmap, (int(shape[1]/2), int(shape[0]/2)), 100, (255 ,255, 255), -1)
    return depthmap

Try to run it.

depthmap = make_depthmap()
plt.imshow(depthmap, cmap='gray')

This one should come to the fore.

3. I tried it

def make_autostereogram(depthmap, pattern, shift_amplitude=0.1, invert=False):
    "Creates an autostereogram from depthmap and pattern."
    depthmap = normalize(depthmap)
    if invert:
        depthmap = 1 - depthmap
    autostereogram = np.zeros_like(depthmap, dtype=pattern.dtype)
    for r in np.arange(autostereogram.shape[0]):
        for c in np.arange(autostereogram.shape[1]):
            if c < pattern.shape[1]:
                autostereogram[r, c] = pattern[r % pattern.shape[0], c]
            else:
                shift = int(depthmap[r, c] * shift_amplitude * pattern.shape[1])
                autostereogram[r, c] = autostereogram[r, c - pattern.shape[1] + shift]
    return autostereogram

def normalize(depthmap):
        return depthmap/255

The main process is quoted from the reference site mentioned at the beginning. Try to run it.

autostereogram = make_autostereogram(depthmap, pattern, 0.3)
plt.imshow(autostereogram, cmap='gray')

** A circle is clearly visible. ** ** (The circle mark appears to be dented in the crossing method, which brings the focus of the eyes closer.)

Because it's a big deal

I tried to make the text a random dot stereogram.

def make_text_depthmap(shape=(400, 600), text='Q i i t a'):
    font = cv2.FONT_HERSHEY_SIMPLEX
    cv2.putText(depthmap, text, (50, 250), font, 4, (255,255,255), 12, cv2.LINE_AA)
    return depthmap

Something that emerges.

depthmap_text = make_text_depthmap()
plt.imshow(depthmap, cmap='gray')

Try to run it.

autostereogram = make_autostereogram(depthmap_text, pattern, 0.05)
plt.imshow(autostereogram, cmap='gray')

It's a little vaguer than the circle, but you can see it.

Quiz for engineers

What is it written on? ?? The text.

Answers are welcome in the comments ^^

Postscript

I wrote a sequel.

-In search of the best random dot stereogram (RDS).

I wrote a sequel.

-Create an original random dot stereogram (RDS) from 2D photos with depth estimation by machine learning