Our creative journey began with the idea of going from everyday routine components of existence to a sudden and chaotic crisis message. We were initially inspired by television, specifically the spooky atmosphere of local broadcasts similar to Local 58. (Refer to Storyboard 1)

As our ideas grew, we turned our focus to YouTube, recognising its position as the most popular video site. (Refer to Storyboard 2) However, our final conclusion led us to a mobile interface, specifically TikTok. We chose TikTok because of its enormous popularity among Generation Z. (Refer to Storyboard 3) TikTok's relatability within this group enabled us to effectively tell our narrative and engage our target audience.

ChatGPT

Following that, we delved into creating a narrative of a dystopian future where AI has taken over the human world. Utilising the ability of ChatGPT, we could mould a story that would harmoniously complement our graphics.

AI Voice Manipulation

ㅤ

Sad Talker

The potion pot of our DeepFake Journey. (Refer to figure 1) SadTalker is a free online software that allows users to upload images and audio files separately to generate talking talking portraits/videos. With this workflow, we could create an array of assets to help us make our artefact. (Refer to figure 2)

ㅤ

ㅤ

① Infinite Scrolling Glitch

The first act starts with a mundane act of browsing TikTok and gradually shows signs of interruption. We use the effects of Motion Tile, Modulation and Pixel Sorter to support the narrative. In the picture below, you see our research on the infinite scrolling glitch (Refer to Figure 1) and an example of an exploration of multiplying frames (Refer to Figure 2) to imagine an error in Tiktok's scrolling function. This technique helps us to convey an early sign of AI taking over what we see on the phone.

② Datamoshing

Following the build-up, we introduce Datamoshing to imagine a total invasion of AI. These tools were critical in helping us create a visually compelling effect that keeps audiences on their edge. In the pictures below, we tested with various plugins such as Datamosh2 (Refer to Figure 3) before finalising Pixel Sort(Refer to Figure 4), Modulation(Refer to Figure 5) and Displacer Pro(Refer to Figure 6) on After Effects, as they provided better control over the gradual build-up.

① Phone Notification

A significant motif that we employ is message notifications. Served as a counterpart to the narrative, it allows us to add a sense of humour through incoming messages from imagined friends and family. It also helped transition the video from Tiktok to an emergency broadcast.

During the intermission and the final scene, we employ Phone UI graphics, specifically the Face ID (Refer to Figure 9), Detecting Location (Refer to Figure 10) and Phone Shutdown screen (Refer to Figure 11), to convey a sense of realism. It also helped with the pacing of the gradual build-up.

The second half of the intermission relies on the "found footage" of a local contingency broadcast. By editing the footage, we could weave the video into the second act with a sense of realism.

④ Military Mobilisation Icon

Another significant motif in the third act is the military mobilisation symbol. This graphic brings a localised context to add to the realism and supports the imagery of "Sian Lor". By manipulating the icon's silhouette (Refer to Figure 14), we created the 'women' and 'child' versions (Refer to Figure 15) to tie it back to the narration.

① Phone Vibration

Building on Phone UI, we also introduced phone vibrations, which can be felt when the audience holds the phone. This sensation can also add to the immersive experience.

② Narration

Using the audio we generated from RVC, we stacked various pitches to create an eerie sound, specifically during the third act.

③ Background Track

A significant suspense component is the background track that gradually builds to a peak. This track helps us to tie the video together.

① Making of Booth

As a final touch, we wanted to give viewers and instructors an immersive experience when watching the video. We imagined our artefact to be seen in a semi-enclosed setting, inspired by the normal TikTok watching scenario, in which users frequently scroll through content in an enclosed space, especially at night. To make this vision a reality, we used the L-Shaped columns (Refer to Figure 20) in the classroom to create an environment similar to the TikTok scrolling experience. (Refer to Figure 21)

② Additional Collaterals

We wanted more than just an empty space with a phone. Following that, we included a recaptcha pop-up (Refer to Figure 24), similar to those found on websites, but this time it served as a verification step before entering the viewing booth. We inserted cutouts of unique facial features that are usually deepfaked (Refer to Figure 22), such as the eyes, nose, and mouth, inside the booth. (Refer to Figure 25) A stand alone was placed in the centre with a mobile stand and the mobile device, providing a thorough and engaging experience for individuals interacting with our artefact.