I wanted to exhibit this work in a tent or cave-like space with unconventional lighting and hanging. Time constraints meant I wouldn’t be able to do too much more building, however, and I didn’t want to reduce the accessibility of the space. Fortunately I found a cave-like area in the building, making use of the narrow and angular coat check area of TMAC, which made seeing the pictures a bit like exploring a cave. It also provided a smaller space suited to these small panels, and offered four walls of different sizes to group the works, as well as subdued lighting.
People looked in and discussed from the outside, and often came around to the inside.
I printed a book called “Contribution” for visitors to examine. The first page is an introduction, then there is a list of about 4500 paintings used to make the images, and credits around 24000 icons used with a GAN to make the glyphs that appear on the cover. The last section of the book contains the various GitHub repositories I used, and those they built on. My contribution is acknowledged among the others in this last section.
The glyphs appeared on the cover of the book, on the image captions, and on the temporary tattoos.
Some of the later images were created with a neural network trained on head and shoulders to create busts. These became the Four Saints.
The fourth one (top left) is right on the edge of mode collapse. After this training epoch, the network deteriorated into repetitive garbled nonsense.
The two panels lit by flickering candles are portraits on gold panels
The two largest images were also offered as postcards
Lots of good discussion at the show. People were intrigued by the idea of machines dimly remembering humanity, and most were interested to know how the images were made. The work was also shown at the 2019 FITC conference in Toronto and at GradEx 104 in 2019.
At this show, some VC types were asking “yes, but what are the commercial applications?” I have an allergy to VCs and told them this was not intended to be a commercial effort. One visitor overheard and told me when they left that in her work doing VFX and compositing, they have a need for producing numerous individual faces. She explained that she was putting together crowd scenes for The Expanse, and they digitally copy and paste the same extras over and over to make a crowd, but the problem is the faces are repeated. It was interesting to hear that some of my Future Renaissance people could find work in TV and film if they get tires of hanging around galleries.