theoretical binary encodification but i can't figure out how to make the images themselves without either corrupting them or cutting their size for some reason??
if the touching color block was exact (it isn't), you could load from color images for much smaller dimensions theoretically, you could load any binary data into scratch with this method, using hacked-in giant images for storage
