Cartographic Software Art
- John Craig
- Dec 22, 2022
- 2 min read
Updated: Jan 24
Working with ArcMap I have noticed that there are many tools within the software that can produce vibrant, aesthetically pleasing effects from map data. The resulting representations reflect the nature of the data used of course, but with some criteria of data and selection of classification, the imagery can be quite abstract and beautiful.
For example many road data resources come with a "road_dir" data field, which indicates which of eight compass directions a segment of roadway is facing. By mapping that data with a choropleth classification, each direction is assigned a color and the map is totally populated by those colors. Where these colors meet forms a psuedo-geography and an appearance of depth, roads become terraces, valleys, or cliff sides.

Before any post-processing.
These colors can also be adjusted to fit a palette, or to reduce or enhance the depth effect. I wanted to achieve a more painterly look from these images, but didn't want to use a filter in photoshop or some other trick that could affect the image too much and stray from the original weirdness of it.
So I used a trick I had heard about for testing AI image upscalers. I first downsampled the original images to roughly one third of their original resolution, compressing all the details of the images. I then ran the now low-resolution images through an online AI image upscaler, making it upscale the images to be larger than the originals, and forcing the AI to rebuild all of the lost details from the little information available to it.
The result is a softer and more painterly image, that still has detail, but isn't as pixelly feeling as the original images. Some colors are blended now creating the appearance of shading details where there were originally none.

After downsampling and upscaling.

Another image after downsampling and AI upscaling, notice how the roads create the illusion of cliffsides or terraces, no topographic data was used, only roadway vector data.
I am fascinated by the prospect of using, what is essentially a utilitarian software package, for the creation of abstract but grounded art. I hope to do more of these with more thematic intent soon. The addition of the AI upscaling method makes for interesting results, and adds another software-centric variable to the stew, which I find compelling.
Comentários