Bricolaging software for taking source images and converting them into assemblages of 88x31px banners from geocities archive
If the filehosting of the data zips goes down, the scraper.py script can be used, it's a lot slower though.
Otherwise, grab the deduped zip file linked on the hellnet.work site. put it in the root of this repo.
python -m venv venv
on osx/linux:
source ./venv/bin/activate
on windows:
./venv/Scripts/activate
and
pip install -r requirements.txt
Pass the image you want to convert into a call to image_converter.py
.
python image_converter.py imagefile.png
wait
done!
- load in all immages
- Go through each image, and calculate mean color of image.
- for animated images, there should also be a stddev calc, to determine if they are flashy. High stddev intervals means lots of variation, and therefore more likely to be a flashy animation.
- Some images are absolutely massive. One at least is too large for me to process, causes a memory error. Don't use those.
- Take in a source image, break into pixels.
- Replace pixels with banners that have closest match in rgb value. Random pick if identically close to 2 different banners.
- optionally squanch banners to squares.
- compose this all into one image, animation included.