Media minimization

Media minimization refers to the minimization of the kind of media that is usually high-bandwidth, such as images and videos.

Media optimization refers to the reduction of file size while keeping the appearance as close to the original as possible. This often involves twiddling with the optimization parameters of the compressor and a moderate use of filters to reduce unnecessary detail.

Another approach starts from stylistic and technical choices that aim at low or moderate file sizes. For instance, low-color styles with large solid-colored areas compress rather well (see posters or black-and-white woodcuts for inspiration). Repetitive pixel patterns may also work well in formats like PNG. High-color pixel art, however, does not compress that well.

If it is reasonable to run arbitrary code, procedural generation and algorithmic art provide a wide variety of stylistic choices.

In the midway between media optimization and style-first approaches are extreme media optimization techniques that lead to particular styles. One possible style is "ditherpunk" where a very small color palette is used in combination with automated dithering.

There is still room for a lot of research in automatic extreme media minimization. Ideally, the result of media minimization should resemble careful artisan work and look better than the original.