RETRACING MUNDANES - Have I Lived Here Before?

This research project rediscovers different places I lived by training AI image generation models on curated datasets of mundane photo snapshots from my everyday routines. As an architect and digital artist, I have been chasing radical ideas of virtual and physical spaces. While dedicating my passion to studio works and inspiration trips, I rarely connected my imagination with the mundane places I lived throughout my journey across China, the US, Japan, and Europe. These places exist as fragmented memories captured by thousands of everyday snapshots from my commuting, shopping, weekend hangouts, and evening walks. On the other hand, I find their mundaneness strangely resonating with the speculative ideas of my artistic journey.

What are my hidden impressions of these places?
Have they been shaping my imagination?
Can I find unexpected inspiration behind their mundaneness?
...
What if AI randomly remix things from my fragmented memories to reimagine the place?

[ Click to Switch the Snapshots ]

Refresh the page if you find trouble viewing the image panels, wait a bit for images to load.

Have I cycled under the metrorail in Miami?
The model is trained on 50 photos of my daily commuting and cycling exercises,
Miami, FL, 2012 - 2017.

Have I walked along the Charles River in Boston?
(The model is trained on 100 photos of my grocery shopping walks,
Boston and Cambridge, MA, 2018 - 2020)

Have I cycled under the metrorail in Miami?
The model is trained on 185 photos of my self-quarantine during the COVID time,
San Leandro, CA, 2020 - 2023

...
Have I wandered between train stations of Tokyo in 2017?
...
Have I cycled across Zhengzhou's old neighborhoods for childhood's memory before 2012?
...
Have I been living in Helsinki since 2023?
...
...to be updated...

As an open-ended research, the project explores different methods of data curation, AI training, and image generation to retrace mundaneness. Current experiments train Stable Diffusion XL models with 50-100 captioned snapshots for each place. Then, I generate hundreds of textual prompts with random combinations of 3-5 keywords from the training image captions to visualize alternative and often unexpected imagery of the place. The next steps will introduce my speculative design as additional data input to reconnect the radical impulses between the imagined and mundaneness of my artistic journey. Stay tuned for updates!