nextwall

This is the README file for nextwall - A wallpaper rotator for the GNOME desktop.

Nextwall is a small application that changes the background of the GNOME desktop to a random image. Nextwall has the following features:

• Operates as a command-line tool. Run nextwall --help for usage information.
• The fit time of day feature automatically sets backgrounds that fit the time of the day (dark backgrounds at night, bright backgrounds at day, intermediate at twilight). It uses an Artificial Neural Network to determine the brightness of a wallpaper.

Nextwall also comes with a GNOME Shell extension.

Setup

To compile nextwall you need to have gcc and automake as well as the following development libraries:

• check
• help2man
• libfann
• libglib2.0
• libmagic
• libmagickwand
• libreadline
• libsqlite3
• libbsd

On Debian and relatives, run this command to install the dependencies:

apt install build-essential automake check help2man libfann-dev \
libglib2.0-dev libmagic-dev libmagickwand-dev libreadline-dev \
libsqlite3-dev libbsd-dev

If you're building nextwall from the Git repository, you first need to use GNU Autotools to make the GNU Build System files before the below commands work. This can be done with the shell command autoreconf --install.

You can then build and install nextwall. See the INSTALL file for building and installation instructions. Briefly, the shell commands ./configure; make; make install should configure, build, and install this package.

Usage

Execute nextwall --help to see usage information for nextwall.

When running nextwall for the first time, you need to scan your wallpaper directory for wallpapers:

nextwall -sr /path/to/wallpapers/

Then nextwall can be used as follows:

nextwall [OPTION...] PATH

See man nextwall for details.

ANN trainer

Nextwall comes with a trainer for the Artificial Neural Network (ANN). Run nextwall-trainer --help to see usage information for the trainer. To use this trainer, first create a directory with image files that you want to train the ANN on. Then you can run the trainer as follows:

nextwall-trainer -p30 ~/Pictures/train/

This will train the ANN on 30 image files from the directory ~/Pictures/train/. Then follow the instructions on the screen to start training. Once completed, two files will be created:

• nextwall.dat - The training data
• nextwall.net - The artificial neural network

The ANN is created from the training data. Now copy the ANN (nextwall.net) to ~/.local/share/nextwall/ to have nextwall use this ANN the next time it is executed with the --scan option.

You can update the ANN by using an existing training data file as follows:

nextwall-trainer -r

This will create/update nextwall.net from an existing nextwall.dat file. This is useful if you want to create a new ANN with different parameters, without having to rebuild the training data.

License

Nextwall is free software. See the file COPYING for copying conditions.