I'm very fascinated by autostereograms, especially single image random dot (auto)stereograms (SIRDS) where no equipment is needed to view a 3d image within a single image by using random dots (pixels).

But only just recently I discovered single image random text (auto)stereograms (SIRTS), which are of course also very fun and nerdy to look at 😉

With the small program aa3d it's possible to convert 3d txt files into a SIRTS. A few examples for such txt files are listed here.

Hint: I created a copy of aa3d for further improvements.

But it's quite inconvenient to create these txt files, so I created a small python script (my first!) based on asciify to convert a depth map image like dm.png to a 3d txt map to be used by aa3d.

Supported are only Arch Linux (yay) and Debian/Ubuntu (apt-get).
First install aa3d, python and pip:

yay -S aa3d python python-pip # Arch Linux with yay as AUR helper
apt-get -y install aa3d python3 python3-pip # Debian/Ubuntu

Afterwards install pillow via pip:

pip install pillow

With docker installed, we can use it and the provided Dockerfile to install python, pip, pillow, aa3d and dm2txt.py and tag the image with aa3d.
I used python:buster (Debian 10 with Python 3) as the base image.

docker build -t aa3d-tools .

Alternatively you can use the provided docker image thanks to this GitHub workflow:

docker pull ghcr.io/tohn/aa3d-tools

Now you can use dm2txt.py to convert an image (dm.png) to a 3d txt map and use aa3d to convert the map to a SIRTS.

dm2txt.py expects an image file as the first parameter.

The second parameter can be used to set the width of the resulting "txt image" (in the range of 1-500) since an image with a width of 1920 wouldn't look good on a terminal. The default width is 80.

The third parameter can be used to set the number of layers used by aa3d. Only 10 are possible with 0 as the lowest layer (aka the background) and 9 as the highest one. The input has to be in the range 0-9, the default is 9.

The parameters of aa3d are described in man 1 aa3d.

aa3d <heart.txt
./dm2txt.py dm.png | aa3d -d
./dm2txt.py dm.png 80 3 | aa3d -t tohn

With the docker image aa3d-tools we can now mount our current folder pwd (use (pwd) instead of $(pwd) in fish) and run the same commands as above.

# create a SIRTS from heart.txt with `aa3d`
docker run --rm -v $(pwd):/opt -i ghcr.io/tohn/aa3d-tools aa3d <heart.txt
# show the output of `dm2txt.py`
docker run --rm -v $(pwd):/opt -i ghcr.io/tohn/aa3d-tools dm2txt.py dm.png
# create SIRTS from dm.png with `dm2txt.py` and `aa3d`
docker run --rm -v $(pwd):/opt -i ghcr.io/tohn/aa3d-tools bash -c 'dm2txt.py dm.png | aa3d -d'
docker run --rm -v $(pwd):/opt -i ghcr.io/tohn/aa3d-tools bash -c 'dm2txt.py dm.png 80 3 | aa3d -t tohn'

I wrote another python script after I saw this 3D Stereogram Ascii Movie Generator, which should create a WebM file consisting of some SIRTS showing the movement of a given txt depth map from left to right. More movements are planned in later versions.

The script is included in the aa3d-tools Dockerfile. If you want to run it natively, the following programs are necessary:

yay -S aa3d imagemagick ffmpeg # Arch Linux with yay as AUR helper
apt-get -y install aa3d imagemagick ffmpeg # Debian/Ubuntu

Now you can run it with a txt depth map (maybe generated by dm2txt.py? :wink:)

docker run --rm -v $(pwd):/opt -i ghcr.io/tohn/aa3d-tools txt2webm.py heart.txt

• dm.png
• heart.txt (heart_n_heart)