exp_06

Default Safari is a mobile App that detects the color white via a smartphone’s camera and replaces it with the color black. It can be used to de-white one’s vision.

movement captured by the app

research questions

  1. How is the color white embedded in the design of architectures, infrastructures and objects?

  2. How biased is technology when it comes to color?

development

  1. We researched how the color white was used in film and how today’s algorithms process and interpret skin colors. We here present only small excerpts of a field of research that we encourage readers to expand on.
Shirley cards are color cards used for calibration in still photography printing and film making. From 1940 to 1995 these cards showed exclusively Caucasian females in colorful dresses, thus cameras and printers were optimized for light skin colors. This white-as-default-logic is evident in photographs of people of color in that time, whose faces are being rendered almost undistinguishable. The producer of these cards, Kodak, only changed their policy when it came to complaints from furniture and chocolate fabricators who wanted to promote their products.
Whitewashing in film describes a practice in which white people take something invented by other cultures and use it as their own, for example when a Caucasian actress plays a person of another skin color. A recent example for whitewashing is the originally Japanese anime Ghost in the Shell which Hollywood adapted in 2016 with a Caucasian lead.
Some Facial Recognition Software is unable or inaccurate in identifying people of color. Because of biased algorithms some humanoid robots are more likely to identify a white mask as a human face rather than the face of a person with dark skin color. In many states, for example in states of the US, police use facial recognition software to scan their surroundings for criminals and some of their software is more prone to error when asked to identify people of color, females and people under the age of 30.
  1. In an early test we programmed an app to calculate the most dominant color of the HTC Vive’s camera images and replaced anything close to that color with the color black. The walls of our university’s rooms are all painted white, hence white was the color that disappeared more so than any other. Because the HTC Vive is connected to a big computer via cables we could not take it outside and decided to translate the experiment into a mobile version.
  1. Initially, we used the game engine Unity to analyze the image of our laptop’s webcam, then transferred this principle to our Android smartphones using shaders to move the calculations to the graphics card. Outside in a park, the color green vanished from the app’s imagery. We were very curious about what would happen if we had an app that exclusively erased white no matter if it was the most prominent color or not. We went back inside and programmed that app.
  1. A first test with the laptop’s webcam image displayed us looking at the app - while the app already interpreted its image of us and completely erased the white room we were in, it also made parts of our faces and bodies disappear.

parts of us that the app interprets as white disappear

try or catch

When using Default Safari inside our apartment 90% of the app image becomes black. Our attentiveness floated towards the objects that were still visible. Dark objects seemed less dark than without the app, probably because white as a contrast color disappeared. We saw the play of light and the many different shades of darker colors. It felt unsettling to us and other people who tested the app to see how much the color white is embedded in our surroundings.

wall with photograph seen on a desk without the app

wall with photograph seen on a desk with the app and thus without the color white

our kitchen wall without the color white our laptop without the color white

Most people with light skin color we looked at were still visible – they were just not as white as their surroundings. How much of their surface was interpreted as white by Default Safari depended on the shade of their skin color as much as the light settings.

a lightskinned hands without white

Outside of houses the color white was less present. In parks, antique statues are white. Only a few animals living at our latitude are white – birds like doves, chickens and swans, some horses and sheep. In the streets, most traffic signs and road markings are white in part or in total.

a street without the color white

conclusion

Default Safari shifted our sensibility towards the presence and absence of the color white. Many things we interact with or encounter on a daily basis are white: Walls inside and outside of houses, a lot of furniture, tableware, the majority of books and paper. Most of these objects would not be as white if they were not painted white or bleached. Digital interfaces could have any color, however most of them are white: operating systems like Mac OS and Windows10 use white, grey and blue as their most prominent colors in the default settings. Most websites - this one included - are white.

Important to note is that a smartphone camera analyzes colors differently than the human eye and in consequence interprets bright light or areas illuminated by light as white which do not seem white to the human eye. On the other hand the camera is more nuanced when it comes to the color of a not-so-bright light source while human eyes adapt to it more quickly. That testers from the fields of design and arts were better at noticing and naming light colors than people from other fields demonstrates that an attentiveness to colors and their implications can be learned.