37. Computer art

[status: content-partly-written]

Sadly the fractal flame screensaver ElectricSheep was made proprietary several years ago, so it cannot grace our computer monitors anymore. But there is much good computer art that can be generated using free software tools.

Here are some references to explore to see if it is possible to develop a chapter for this book on computer art. At this time these references are focused on evolutionary art, but other possibilities should be examined. Connections to 1d and 2d (like Conway’s life) cellular automata should be explored.

Preqrequisites:

sudo apt install -y geeqie imagemagick
sudo apt install -y gimp

37.1. Understanding photos and images

37.1.1. Discussion of graphics formats

Show a PDF and a .png or .jpg and try to zoom in on them. Talk about vector formats vs. raster formats.

37.1.2. Photo collection management

Introduce shotwell and digikam. Urge students to take pictures with a camera or phone, but bring a cheap camera to have students take some 80 photos at the start of class, then download them to the computer with a USB cable.

Here is a file that I have put up:

# prepare a space under ~/Pictures for stock photos
cd ~/Pictures
mkdir stock
cd stock
# get a couple of sets of images
wget https://www.galassi.org/mark/.tmp/for-courses/eric-kim-stock-photos.zip
unzip eric-kim-stock-photos.zip

And one from NASA:

# prepare a space under ~/Pictures for other NASA photos
cd ~/Pictures
mkdir nasa
cd nasa
wget https://solarsystem.nasa.gov/system/downloadable_items/1304_JPG.zip
unzip -d NASA_1304 1304_JPG.zip

This should be enough to get started. But we can get more from the https://creativity103.com/ web site, using a tip on recursive wget from stackoveflow:

https://stackoverflow.com/questions/4602153/how-do-i-use-wget-to-download-all-images-into-a-single-folder-from-a-url

37.1.3. Image manipulation: command line and GUI

37.1.3.1. Introduce ImageMagick

Work through some of the filters shown in the ImageMagick cookbook at:

https://www.imagemagick.org/Usage/transform/#art

Start with everyone taking their favorite image. They should name it myimage-original.jpg

If you don’t have a favorite image handy then you can download NASA’s “Pillars of Creation”:

wget https://upload.wikimedia.org/wikipedia/commons/6/68/Pillars_of_creation_2014_HST_WFC3-UVIS_full-res_denoised.jpg
mv Pillars_of_creation_2014_HST_WFC3-UVIS_full-res_denoised.jpg myimage-original.jpg

Then give it a workable size so things dont’ take too long for the purpose of this class. I recommend it be approximately 1500x700 pixels in size so that it’s not too slow to process, but if students want to use a photo with a modern phone camera (which have many more pixels than that) they can do so and then reduce them with something like:

convert -resize 1500x myimage-original.jpg myimage.jpg

So from here on let’s call our picture myimage.jpg, so rename your picture to that filename.

Next we see how the pixel spread works:

convert myimage.jpg -spread 5 myimage-spread-005.jpg
convert myimage.jpg -rotate 5 myimage-rotate-005.jpg

Now we’ll create a sequence of files with different pixel spreads using a quick inline shell script:

for i in `seq -w 0 100`
do
    echo $i
    convert myimage.jpg -spread $i myimage-spread${i}.jpg
done

Now make a film with:

ffmpeg -framerate 3 -i myimage-spread%03d.jpg spread-movie.mp4

and view it with

totem spread-movie.mp4 &

for i in `seq -w 0 360`
do
    echo $i
    convert myimage.jpg -rotate $i myimage-rotate${i}.jpg
done

ffmpeg -framerate 20 -i myimage-rotate%03d.jpg rotate-movie-fr20.mp4

37.1.3.2. Introduce the Gimp

Everyone should know how to start and use the GIMP. Here is a collection of 40 tutorials:

https://www.noupe.com/inspiration/tutorials-inspiration/best-of-gimp-40-professional-tutorials-to-level-up-your-skills-79428.html

We pick one of them and work through part of it with the students.

37.2. metapixel and photomosaics

• Collect a bunch of photos in a hurry, possibly with some cool use of wget and an image search. For example, creativity103.com has archives of images licensed under the “creative commons “attribution” license: https://creativecommons.org/licenses/by/3.0/

• I already showed how to beef up your Pictures folder with commands like wget -r -np -nc https://creativity103.com/

• Install metapixel on Linux

• the metapixel tutorial I have found has pointless racy images, so I need to find another one.

• install the webcollage screensaver: sudo apt install xscreensaver-screensaver-webcollage

• we have beefed up our personal ~/Pictures directory so we can use that to make the photomosaic database

mkdir ~/mp
metapixel-prepare -r ~/Pictures ~/mp --width 32 --height 32

once the database is ready we can make our metapixel photo with:

metapixel --metapixel --library ~/mp --scale=2 myimage.jpg myimage-mp2.jpg
# and a scaled-by-4 version with:
metapixel --metapixel --library ~/mp --scale=4 myimage.jpg myimage-mp4.jpg

Now use geeqie with its “control+scrollwheel” zooming feature to zoom in on the individual tiles in the photomosaic.

37.3. ASCII art

What is ASCII? Explain encoding of characters, and show man ascii.

What is ASCII art?

Tools for drawing ASCII art.

Fun and amusing tools. Some of these can be found at https://www.binarytides.com/linux-fun-commands/

echo an example of figlet | figlet
banner "have a nice day"
cowsay hey dude
cowsay -f dragon "Run for cover, I feel a sneeze coming on."
cowsay -l
cowsay -f ghostbusters Who you Gonna Call
sl

The grand old UNIX fortune command is still available, an dyou can pipe it to others:

fortune -s     # repeat a few times
fortune -s | cowsay

cmatrix uses the UNIX curses library to draw a matrix animation on your terminal.

Converting raster images to ASCII art. jp2a on GNU/Linux systems: let us take pictures of famous computer scientists Margaret Hamilton and Dennis Ritchie:

wget https://upload.wikimedia.org/wikipedia/commons/6/6f/Margaret_Hamilton_-_restoration.png
wget https://upload.wikimedia.org/wikipedia/commons/2/23/Dennis_Ritchie_2011.jpg


## make your terminal very big and try
jp2a -f Margaret_Hamilton_-_restoration.png
jp2a -f --color Margaret_Hamilton_-_restoration.png
jp2a -f Dennis_Ritchie_2011.jpg
jp2a -f --color Dennis_Ritchie_2011.jpg

To see jp2a work better, make your terminal have really tiny fonts (in many terminal programs you can do this with ctl+-), then resize the window to be really big, and then run the jp2a commands again. You might be astonished at the results.

emacs has an ASCII art drawing mode. It might be worth exploring.

Joyce Levine has pointed me to tilde town: https://tilde.town/ which might be worth exploring.

There was a famous ascii art movie of Star Wars that one could reach with

telnet towel.blinkenlights.nl

but it has gone offline recently. You can find videos of it on youtube. A real tour-de-force. According to this thread on reddit it has been taken down:

https://www.reddit.com/r/sysadmin/comments/pbv7xm/is_towelblinkenlightsnl_dead/

but you can get it (without the fancy ascii colors) by typing a telnet command followed by typing starwars:

telnet telehack.com
starwars

A discussion of the ascii animation used there can be found at:

https://www.asciimation.co.nz/index.php

https://www.asciimation.co.nz/asciimation/ascii_faq.html

A few more things to type:

sl
fortune
factor 12103  # factoring numbers? can we use this to search for Mersenne primes?
factor `echo "2^7-1" | bc` ; factor `echo "2^11-1" | bc` ; factor `echo "2^13-1" | bc`
pi 50
espeak "Hello Linux, where are the penguins"
aafire -driver curses

37.4. Evolutionary art

• https://en.wikipedia.org/wiki/Electric_Sheep

• https://en.wikipedia.org/wiki/Software_art

• https://en.wikipedia.org/wiki/Evolutionary_art

• https://iasl.uni-muenchen.de/links/GCA-IV.3e.html

• transcend https://transcend.sourceforge.net/

37.5. Image manipulation from your own Python program

We will learn to use Pillow, the python imaging library (PIL).

At the shell:

sudo apt install python3-pil

Also give yourself a photo to work with. For this tutorial let’s keep it small so it’s faster and views better. You can use a command like:

convert -resize 800 my_big_picture.jpg myimage.jpg

In the python interpreter, following the tutorial at https://pillow.readthedocs.io/en/stable/handbook/tutorial.html

## import the Image portion of the PIL library
from PIL import Image
## load an image
im = Image.open('myimage.jpg')
## show its information
print(im.format, im.size, im.mode)
## show the picture itself
im.show()
## then close that window

We now have an image stored in the object im and we can work on manipulating it.

37.5.1. Geometric transformations

import os, sys
from PIL import Image

size = 128, 128

infile = 'myimage.jpg'
outfile = os.path.splitext(infile)[0] + '.thumbnail' + '.jpg'
print(outfile)
thumb = Image.open(infile)
thumb.thumbnail(size)
thumb.save(outfile, 'JPEG')

## copy a rectangle from the image
box = (300, 300, 600, 600)
region = im.crop(box)

## now play with that rectangle, then paste it back in
region = region.transpose(Image.ROTATE_90)
## now paste it back in
im.paste(region, box)
## now box is the image with the small region we chose
im.show()
im.save('myimage_pasted.jpg', 'JPEG')

You will now see that the region we cut out of the photo has been rotated 90 degrees.

## reload the image from scratch
im = Image.open('myimage.jpg')
## rotate 45 degrees counterclockwise
rotated = im.rotate(45)
rotated.save('myimage_rotated_45.jpg', 'JPEG')
## now make an animation
various_rotations = []
for i in range(360):
    rotated = im.rotate(i)
    fname = 'myimage_rotated_%03d.jpg' % i
    print('writing out %s' % fname)
    rotated.save(fname, 'JPEG')

If you type ls you will see that you now have many files with names like myimage_rotated_17.jpg. You can use geeqie to view them and even get a crude animation by holding the space bar down. You can also use what we learned in Section 12.13 and try this:

## make a movie with:
ffmpeg -framerate 24 -i myimage_rotated_%03d.jpg myimage_animated.mp4
## view it with:
vlc myimage_animated.mp4 &

37.5.2. Filters and enhancement

from PIL import Image
from PIL import ImageFilter
## load an image
im = Image.open('myimage.jpg')
## out = im.filter(ImageFilter.DETAIL)
## out.save('myimage_detail.jpg', 'JPEG')
out = im.filter(ImageFilter.CONTOUR)
out.save('myimage_contour.jpg', 'JPEG')
out = im.filter(ImageFilter.EMBOSS)
out.save('myimage_emboss.jpg', 'JPEG')
out = im.filter(ImageFilter.FIND_EDGES)
out.save('myimage_edges.jpg', 'JPEG')
out = im.filter(ImageFilter.SMOOTH)
out.save('myimage_smooth.jpg', 'JPEG')
out = im.filter(ImageFilter.BLUR)
out.save('myimage_blur.jpg', 'JPEG')

## NOTE: in the next one we start from "out", the smoothed photo
restored = out.filter(ImageFilter.EDGE_ENHANCE)
restored.save('myimage_restored.jpg', 'JPEG')

out = im.filter(ImageFilter.SMOOTH_MORE)
out.save('myimage_smooth_more.jpg', 'JPEG')

Now compare all the images we produced. In particular, look carefully at myimage_smooth.jpg and myimage_restored.jp to see how the EDGE_ENHANCE transformation takes away the fuzziness that had been introduced by SMOOTH.

37.6. Topics for further study

Another Pillow tutorial which starts slowly but gets into interesting examples:

https://www.tutorialspoint.com/python_pillow/python_pillow_quick_guide.htm

https://realpython.com/fingerprinting-images-for-near-duplicate-detection/

https://www.imagemagick.org/script/compare.php

https://askubuntu.com/questions/209517/does-diff-exist-for-images

https://jeremykun.com/2012/01/01/random-psychedelic-art/

Take one of those classic old “over the top” movie transition effects, or a modern parody of one, like this one from Spinal Tap: https://youtu.be/QrJlyapt6OY?t=114

Try to take an image and create your own over the top transition using Pillow.