A Room of One's Own

Over the past five months, A Room of One’s Own evolved both creatively and technically. In the process of building a collective virtual archive of individual narratives, I’ve collected 3D models of residential spaces as well as audio recordings from twelve people who were experiencing the lockdown in Shanghai. After fine-tuning and relighting the 3D models, I designed a user journey map and assembled the models in a timeline on Cinemachine. While setting predetermined camera movements to motivate the view of the users, I also incorporated a custom script onto the virtual cameras for users to look around in the scene with mouse drag. Balancing the level of control and autonomy, I aim to make the navigation process in the virtual environment simple, intuitive, yet not overly restrictive. As a result, the experience is a journey through the abstract landscape and soundscape infused with 3D scanned interiors and fragments of individuals’ reflection, contemplation, dreams, and imagination.    

 

The surreal, desolate texture of non-professional photogrammetry intrigued me deeply. If environmental storytelling manifests through displaying traces of use, photogrammetry accentuates that very aspect of architecture: it captures the fragmented, used, and imperfect essence resulting from human usage and technological limitation. The willing acceptance and tolerance of such fragmentation, deformity, and distortion echoes the philosophical concept of “Wabi-Sabi,” which embraces beauty in the incomplete, impermanent, and imperfect.       

 

In the process of refining 3D models and incorporating them into the scene, I found that the quality of the scans differed greatly depending on whether their device was equipped with the Lidar sensor as well as their individual fluidity with 3D capturing. To compensate for that difference, I experimented with adding volumetric fog, baking the lightmap, and altering the style and tone of their narration to adhere to the visual aesthetics of the model. 

 

Initially, when collecting the audio sources from my subjects, I asked for audio clips ranging from 5 to 20 minutes. However, after creating a demo scene and doing initial user testing, I discovered that the visuals do not match the audio in the density of information it provides. I also found that in a virtual and interactive setting, users tend to get bored and distracted more easily if there’s not enough visual content to keep them engaged. Eventually, I found that 2 minutes is the longest time a user can linger in the same 3D scanned room without losing interest. Based on this feedback, I trimmed the audio clip of each person down to under two minutes and made sure to provide different camera perspectives while in the same virtual space. 

 

In future editions, I plan to work with a sound designer to refine the narrations and sound design of the project. Furthermore, I will re-edit or re-scan some of the models that weren’t well scanned. Eventually, after incorporating all the elements and content, I will deploy a 3D web page version of the project and possibly also develop a VR version.