BG

Description

The Atlas of the Forgotten is on experiment to visualize how people with dementia recall memories and people around them. By using Gaussian Splatting to capture people, places, and memories, the project emulates the way our brain captures and recalls memories. These scans are then manipulated, distorted, and visualized to represent 5 of the 7 stages of Lewy Body Dementia, by cross-referencing them with the brain and perceptual activity of dementia patients.

Process

While scientifically dementia is characterized by a sustained decline in cognitive functions, it can be described as a death before death. While we never witnessed it happening it directly to a loved one, we’ve seen the impact of this affliction echo through the people around us, only seen through a frown that gradually deepens, a fleeting look that happens more often, or quiet tears mourning a loss that has a beating heart. We began by trying to understand what the person afflicted by dementia was feeling, but realised that their loved ones endure this loss just as deeply, if not more. They are looking at the person they love, but their loved ones are not looking back. They wake up everyday with the small hope that they'll be able to catch a glimpse of them again only to be hurt time and time again. Memory becomes a burden at both ends, as the loss and preservation of it exist in constant tension, in what feels like eternal struggle. After learning this we wanted to represent the loss of human self from the loved one’s perspective. What do they think their loved one is going through? What do they think they are thinking, remembering, feeling? How does it feel to watch the person you love fade away while their physical body is still staring back at you?

Tools

We explored a space between duality and surrealism, using setting to establish mood. Due to the war in Lebanon, each of us worked from different locations, which led us to test various environments and assess our technical limitations. We developed multiple storyboards unified by a surrealist sensibility and began experimenting with Luma, an AI-powered photogrammetry app used to generate photorealistic 3D scans from photos and videos. Through repeated testing, we became more confident in the scanning process and captured multiple scans, including greenery, to create spatial depth. These experimentations clarified the shoot’s direction; although the scans were peaceful and comforting, our intervention introduced a sense of entropy. We carefully art-directed the shoot, selecting a model and capturing two to three scans per shot to ensure detail. We then tested Gaussian splatting in After Effects, a volumetric rendering technique used to represent complex forms as a collection of Gaussian distributions. After several trials, we opted for a subtle background treatment to enhance the sense of place without competing with the visuals. Additional tools included TouchDesigner for real-time interactive multimedia content, as well as Blender and Unreal Engine for 3D rendering and spatial testing. Simple, elegant typography that is also disintegrating, was used to unify the final outcome while mirroring the effects in our visuals.