Palette of Place

Completed: May 2014
Course: Capstone Project
Advisor: Tad Hirsch
Role: Concept, Code and Visual Design

 
 
 
 

"What are the colors in a place and how do those colors change over time?"

 

This project was inspired by the above question; as someone who loves to be outdoors, I wanted to explore ways of visualizing the palette of places that I care about. 

 
 
 
 
 
 
 
 

With my background in photography, using photos as a source of color seemed like a natural choice. Comparing two images of the West Point Light at Seattle's Discovery Park, taken in different seasons, reveals differences that while striking, are not entirely surprising.

 
 
 
 
The range of photos of Discovery Park on Instagram

The range of photos of Discovery Park on Instagram

 
 

I began to wonder what could be revealed by looking at and working with many more photos. Could I learn about the colors and scenes in a place that matter to people? Could I learn something about that place's role in people's lives? My original question began to evolve:

 

"What can the images people take of a place say about that place and the colors in it?"

 
 
 
 
 
 
 

Palette of Place uses custom software, written in a Java based language called Processing, to build visualizations of locationally tagged images. Images are collected through the Instagram API, abstracted to isolate color from the context of the original photograph, and plotted by their upload date, and printed out as 8" x 8" posters.

January occupies the 12:00 position, with the remaining months progressing clockwise around.

Details like a strikingly regular weekly cycle, holiday spikes and changes in season are now clearly visible. Discovery Park, for example appears to be slightly less popular in the winter months, get's a spike around Memorial Day, Labor Day and again at Thanksgiving. 

 
 
 
 
 
 

At close range other details become visible. Great sunsets, blue skies and more impressionistic representations of color. The color of the place, as captured by those who've photographed it, begins to emerge.

 
 

By comparing these visualizations differences begin to emerge that offer a unique sort of visual fingerprint of a place. Some places are visited and photographed on a seasonal basis while others are more consistent. In many there are spikes that relate to holidays or special events.

By applying this technique to images collected from places around Seattle, insights are now available that simply weren't possible through the standard Instagram feed.

 
 
 
 
 
 
 
 
 
 

I also created a larger, 24" x 36" poster version for a pair of northwest landmarks that display significant seasonal swings. The small data panel highlights some numerical characteristics and some dates of interest. 

 
 
 
 

A close crop of the info block for Stevens Pass

 
 
 

 

 

 

 
A close crop of the info block for Alki Beach

A close crop of the info block for Alki Beach

 
 
 
 
 

Palette of Place was part of the UW Design 2014 Senior Show, and was accepted as a featured project in the Jacob Lawrence Gallery.

 
 
 
 
 
 

histogram to instagram

PROCESS

Initially interested in how to distill a photograph into it's most prominent colors, I quickly discovered how tricky it is to bucket colors when the computer understands them as RGB values. I became more interested in large sets of images after seeing the project Spotmaps, by Andy Willis, and began to explore the possibilites available to me via the instagram API as a way to get large numbers of consistently tagged and formatted images.

 
A screenshot from one of my early Processing sketches, where I was trying to count the frequency of colors in an image. 1,671 different colors in this image.

A screenshot from one of my early Processing sketches, where I was trying to count the frequency of colors in an image. 1,671 different colors in this image.

 
 
 
 

An early attempt at scanning the frequency of colors in an image and presenting it in a meaningful way. At best a confusing histogram. BUT I was learning a lot about code and working with images.

 

A screenshot of the interactive app I built to show the sorted and counted colors from four different images of Discovery Park, one from each season.

 

After clicking on a season, it would expand and show frequency as well as range of color. There was something slightly interesting happening here, but I came to the realization that to take this direction father I needed more math and programming skill than I could hope to learn in the time allotted.

 
This is from an early attempt  at distilling and presenting multiple images by only showing the single most predominant color in each. Nothing too thrilling here, but I was starting to learn to work with multiple files in the same Processing sketch.

This is from an early attempt  at distilling and presenting multiple images by only showing the single most predominant color in each. Nothing too thrilling here, but I was starting to learn to work with multiple files in the same Processing sketch.

 
A screen shot from an early composition, just starting to get into applying order and organization. Several iterations (and lots of code) later I arrived at the radial format that I ultimately went for.

A screen shot from an early composition, just starting to get into applying order and organization. Several iterations (and lots of code) later I arrived at the radial format that I ultimately went for.