'Amber Dell' is a VR experience that invites participants to explore embodied performativity in virtual shared spaces. The experience challenges users to inhabit non-human forms within a shared, fantastical virtual environment. Upon entering, visitors select from an array of non-human body types, each possessing distinct locomotion systems and affordances. A surreal landscape serves as a playground for exploration and interaction, challenging participants to navigate it using unfamiliar locomotion systems.

The experience serves a dual purpose: as an artistic exploration of consciousness and embodiment, and as a tool for connecting with our own bodies. I have been interested for a long time in what has been called Homuncular Flexibility in the VR field, i.e. the idea that that users can learn to control bodies different from their own by changing the relationship between tracked and rendered motion. We are quite aware of this effect when it occurs to us in our daily lives, such as when driving a bicycle or a car. Inspired by authors such as James Bridle, Maria Puig de la Bellacasa and Jaron Lanier, It was my initial intuition that this phenomenon could be useful in exploring the more-than-human world and that this combination could produce an engaging non-linear narrative experience.

I have worked with interactive experiences in general and the XR space in particular for quite a while now, and I’ve implemented various avatar systems. One thing that always surprised me (specially in VR) was how similar all those were. I wrote a piece some time ago when Stray came out about how we were seeing new paradigms appear in this regard.

This year, while having a conversation with a Product Designer from an XR company working in an avatar system, the topic came to mind again. After a brief discussion with her and other colleagues regarding the main aspects of an avatar creation user flow, I decided to play around with the idea of a user flow for non-human avatars. I prototyped a simple idea using Bezi, and this served as the starting point for this project.

Even though this first audience was comprised of mostly young tech-savvy designers, it became quite clear that the gesture interaction was quite unintuitive. At the same time, controller tracking turned out to be much more precise than hand tracking, which resulted essential for an experience that gained so much from fast movements.

It also became clear that users had a desire for self perception, and so I sketched a mirror interaction, where they would be able to see themselves. Since this early prototype accomodated only two avatar styles, I simply created a switching button interaction to swap them.

It became clear to me that an avatar creation user flow would have to rely much more on 2D panel interfaces, specially if the experience was to become more complex. More on that later.

In my opinion, developing this with a physics-based approach brings something extra to the table. There is a sense of agency that is created by this that is hard to get when taking a more simulated approach. Sadly, VR is still prone to glitching, and I find that relying on the physics engine works best with handling these cases.

The challenge in designing this kind of locomotion system is finding uncommon yet easy to understand movements. I wanted to balance the alien nature of movements with human performability, demanding physical commitment from players while still incorporating learning curves to allow for skill mastering.

A big part of this experience relies on self-perception. During initial testing it became quite self apparent that one of the key challenges when onboarding users to the experience is getting them to construct an accurate mental model of their avatar and a proper mapping of how their body movements translate into avatar motions.

So when building an avatar with multiple bones & joints but only three control points (head and controllers) I needed a way to procedurally generate motion across the rig. For this purpose I utilized Unity's Animation Rigging tool for procedural animations. This helps users create a self image of how they look to an outside viewer. In the game, this is achieved through some mirror objects placed in the scene and on the avatar creation flow.

At the same time, this can not be the image users have when they look at their own hands. Users should be able to see their hands in this new skin but not more, or it breaks the immersion & the mental model of the self.

The visual aspect of the experience was low poly from the beginning, and this decision has been tested and validated throughout the development process. The obvious reason for this is that the focus of the experience should not be on its visuals, but rather its mechanics and gameplay. Low poly of course allows for larger map sizes without the need for advanced hardware or complicated asset streaming.

But also, low poly models are quite clear on their collider rules: they provide clear visible physical boundaries, clear planes with clear normals. So it plays well with a physics based approach. This requires crafting low poly assets with primitive colliders aligned with their model faces, since this is much more performant than mesh colliders.

On top of this, I implemented various optimization techniques including:

While VR is usually employed as a visual narration medium, Amber Dell focus is not what happens in the screens, but what happens in the body.

Virtual Reality’s capacity to trick our perception presents immense opportunities for us to explore our perceptions and reevaluate our fundamental understanding of the human condition. Used with care, XR can be an amazing tool to unlock positive change in our cognition, something I believe we as a species are in dire need of.

It is for this reason that projects like Amber Dell benefit from constant exposure: the cycle of development and presentation reinforce each other and help me get make sure development focuses in relevant features of the experience.