Stadtmuseum Berlin

The semester project was organized in cooperation with the Märkisches Museum, which is part of the Stadtmuseum Berlin Foundation, one of the largest historic museum societies in Germany.

It’s various sites house exhibitions on Berlin culture and history, ranging from pre- and early to contemporary history. The Märkisches Museum on the banks of the Spree River, presents a host of artefacts related to the culture and history of the city.

The Challenge

The Märkisches Museum is in the process of renovating for the 21st century. Alongside new space for temporary exhibitions, this process includes a new concept for the permanent exhibition. The museum management is aiming to create an inviting and participatory space.

With this in mind, the theoretical focus of the seminar centered on personalization. The challenge was to develop a concept that addresses the visitors personally and thus involves them in an interaction with the exhibits.

Ideation

In the first ideation phase, the main objective was to think as broadly as possible. In weekly sprints, I therefore developed five varying mini-concepts, which were then presented to the group.

To be able to share my ideas more clearly with others, I used digital sketching (SketchBook from Autodesk on iPad) to create those visual discussion artifacts. For me it was a rather new medium and I enjoyed trying it out.

Ideation The Wonderpose

I wanted to work with the latest technologies. However, it was essential for the project to set the right focus. The museum, the collection and the visitors had to come first in the concept and only then the technical means, be they cutting edge new like code or classic like a glass showcase. This challenge was overcome in a supportive and instructive dialog with the lecturer.

The final concept for The Wonderpose only came about as a combination after a round of ideas with fellow students. One of the approaches dealt with berlin photo booths, which then had an inspiring effect on my own work.

Design Research

In developing my own concept, I was influenced by three important model projects: Move Mirror, the Strike a Pose kulturBdigital project, and the Body Scan installation at the Cooper Hewitt Museum.

Exhibit Concept

A slightly altered replica of a photo booth is set up inside the museum. Visitors can enter the booth one at a time. As soon as one is inside the photo booth, the interface interaction begins.

First visitors are asked to take a remote control. With its help they can adjust the angle of the camera. Once the video input begins, the PoseNet algorithm compares the visitor’s poses to the image database of the museum’s photo collection and displays the closest photo.

One can take as many poses as one likes and in this way navigate through the collection.

Prototype

To convey the concept visually and content wise to potential users, I decided to create a website with the option to try out the prototype code in one's own browser.

The website is still under construction, but the Figma dummy can already be viewed here.

The video shows the process in the browser.

Implementation

Code Setup

First of all, it is essential to mention that the code part of the project could not have been done without the hands-on support of Sebastian. I forked his original folder on GitHub into my own GitHub repository. After that, the joint code adaptation for my project took place. To understand the code I wrote down the algorithm in simple human language for myself. Finally, I designed the layout of the prototype page in the code, added metadata and saved the data sample of 113 photos into a json file.

The code setup consists of ml5js, an algorithm that builds on the PoseNet and TensorFlow libraries. It is a pretrained Machine Learning (ML) modell. The browser experience is created by using the pose estimation model PoseNet which runs in the browser using TensorFlow.js, itself a library that allows web developers to train and run machine learning models in the browser.

Useful Links

Move Mirror Code on GitHub

PoseNet tutorials by the Coding Train

Good read Move Mirror article

Data Sample

I manually collected the data sample consisting of 113 photographs and their associated metadata. I searched the museum's online collection using keywords such as „woman, man, dance, sport, etc.“ to find suitable images. It is important, for example, that the people in the images are alone and ideally assume a full-body pose.

Concept Website

Illustrations

The illustrations serve a clear visualization of the concept. They show the automat-human proportions and the overall structure of the exhibit as a whole. Furthermore, accessibility was an important part of the course, so I integrated it into my concept and visualizations as well.

Interface

The interface in the Wonderpose exhibit is interactive. First it serves to convey and display information. Thereafter it provides an alternative navigation within the data sets.

Typography

Based on the typeface of the iconic photo booths, I decided to use Apparat as the text font. I chose Dico Mono as the interface typeface. The light handwritten touch coupled with the good legibility seemed to me to appropriately convey the approach that I intended with the exhibit: playfully light and technically refined at the same time.

Conclusion

For me personally, this was one of the best concept and prototyping courses I have ever taken. From the introduction to the topic, to the collaboration with the museum management, to the excellent and tireless support – in terms of creation and technical skills – from Sebastian, everything had its time and place.

With this in mind, my learning curve on concept creation, visualization and prototyping was tremendous. The course opened up a new potential field of work for me in which I would really enjoy working.

To say it with Johnny Cashe's unparalleled lyrics, I completed the course with a Satisfied Mind. Thank you Sebastian.