Image Gallery Tests
We are working on the AI image gallery for the ‘Simulated Selves’ installation. We started by testing different generators to come up with ideas. These are some outcomes via Craiyon to prompts for ‘black and white photographs of Brisbane City’ and ‘black and white photographs of Australian animals’ :
The themes were determined in response to common ‘waiting room’ images. As you can see, the ‘Brisbane’ images are quite good at capturing some key aspects of the city including river overpass and some of the buildings. If you look closer, you will notice that the windows are a little wonky and don’t quite align with photographic equivalents. Similar observations apply to the rendering of the animal portraits, where feathers are integrated into some of the features and extra ears and other elements are sometimes added.
We also tested some more ‘artistic’ styles and found that Stable Diffusion was a good text-to-image generator for more painterly outputs.
Stable Diffusion output in response to ‘still life of flowers’.
We also tried to render some painting of Queensland birds, which were quite successful in the painterly style, but often included extra or partial limbs.
As we continued to play with different AI image generators, we decided that we did not just want to use the image gallery as a technological showcase or software comparison. Rather, we want to use the images to prompt reflection on the potential use of AI on our social, cultural and economic spheres. As such, we adapted our focus from more general ‘waiting room’ themes to evoke consideration of how AI may shape our lives in the future. Here are some preliminary outputs via Stable Diffusion to prompts such as “Still life with technology” and “Painting depicting living across real and virtual worlds”:
