Scene base

You're in a near future where life almost exclusively happens online. Jobs are worked remotely, friends are met in chat rooms or via augmented/virtual reality tools, and algorithms are hyper-controlled, leading to the development of super niche sub-cultures, similar to hyper-specific subreddits nowadays. The difference is that it permeates your entire day-to-day life. Groups with niche hyper-fixations form 'Bubbles', such as one for nature enthusiasts working on reactivating extinct flora and fauna to address climate change, and others like theater enthusiasts enjoying old-school poetry like Shakespeare. Thousands of bubbles exist in this alternate futuristic universe, some good and some questionable. To keep minds open to new ways of thinking and prevent extremist forms, a specific day has been introduced to get out of your comfort zone and 'Burst your Bubble'.

𝕰𝖝𝖕𝖑𝖔𝖗𝖊 𝖙𝖍𝖊 𝖜𝖊𝖇𝖘𝖎𝖙𝖊

// due to licensing conditions the used font is not yet implemented in the web version

CONTENT GEN

Texts

Using OpenAI's 'generative pre-trained transformer' software 'ChatGPT' was a no-brainer for us. Its natural language processing capabilities offered instant high-end responses for a wide variety of text based input. The generated outcome span from one-word descriptions to up to 200.000 character long, astonishingly well written, running texts.

Image

Working mostly with the software 'Midjourney,' introduced by David Holz in 2022, we used prompts that were partly self-written, partly generated by ChatGPT, and sometimes a combination of both methods. 

Aesthetically inspired by the illustrations from the episode 'The Witness' in the series 'Love, Death & Robots,' written and directed by Alberto Mielgo, we aimed to achieve a cyber-goth look to complement the futuristic but still non-dystopian concept of the project. This was technically possible using the '--sref' prompt-function and also by not hesitating to go all-in on phrasing super direkt and sometimes even bluntly, what outcomes we wanted from the software.

Another crucial part of getting the results that matched our creative vision was generating, destroying, re-matching, generating again and destroying again the original images. This gave us more options for specifying the outcome and ensuring a unique result.

Programming

Using the final generated texts and images we created the code for the website to make an interactive near future experience. We used common web technologies like HTML and CSS to create elements and style them. While styling we heavily relied on our own knowledge, because using AI to generate code often lead to typography lacking in variety and excitement.

To really nail down on the experimental factor we used multiple CSS properties to break the lines, words and paragraphs apart. Our initial approach of an endless zooming canvas or space, through which the user is passing, was realisied by capturing the users scroll position with JavaScript. The script then translates the elements forwards and backwards, depending on the scroll state, which comes out as a flying motion.

With this base set, we started creating canvases stacked on top of each to create the illusion of depth and elements being broken up into layers.

Highly experimental typography and layout create a futuristic dynamic while still maintaining a near-future feel to the project itself.

This was probably one of the only parts of the project where no AI tools where used, which has shown us that AI is definitively a very powerful tool that can result in unexpected outcomes, while still not being able to match the randomness of real human ideas and thought processes, always hanging on to repetitive behaviours and predictability.

Demo