Player Embodiment in VR
Successfully establishing Player presence in VR involves many hardware and software factors ranging from system latency to how the environment and other characters respond. Of these, one of the critical elements that can promote or hinder suspension of disbelief is what form the Player takes, for example what their body looks like. There isn’t a single embodiment that is suitable for every experience. A deliberate set of creative choices must drive the design of player embodiment based on specific goals of the experience; player’s tasks & abilities; and affordances provided by the hardware setup.
The selection of embodiment in turn, creates different challenges for technical implementation, some of which we will discuss in this blog. We also discuss the specific choices we made for our character centric VR experience “Emily & Me” based on the design framework we will present.
Design Factors
The form and abstraction of Visual Representation should be tied to the goals of the VR experience. Having a physical body can be very effective in experiences that seek to invoke empathy or sense of play and freedom that comes from assuming an avatar. However, it can also raise dissonance in the Player’s mind by trying to nail down gender, body type, age, ethnicity, etc. Naturally as with games, Players can be provided multiple, customizable avatars and also visual representations can be designed that are “neutral” such as robots or aliens. These avatar options that are neutral are void of gender, race and age making them ideal for general purpose characters. It is worth noting that sense of embodiment is much stronger in VR compared to screen based games since Player is much more aware of their body and Player controls are less abstracted.
Depending on the goals of the experience, using a more abstract representation can actually help Players suspend disbelief and instead focus on the experience itself. This is similar to the uncanny valley in robotics and animation where enhanced realism after a point can actually reduce empathy before the representation actually becomes indistinguishable from physical reality.
Movement is a function of Player’s tasks and abilities; visual representation; and input affordances. If the experience can rely on touch controllers, direct manipulation is the best choice. However, if Player is using a gamepad, discreet event actions, similar to console games, may work better. This also applies to exactly when and what movement is triggered. For example, Player’s walk could translate into simple root motion or actually having swinging arms and head bobbing. Especially when combined with a realistic physical representation, the latter can quickly slide into the uncanny valley.
While on the subject of movement, it is worth giving special consideration to the head since that is a unique input stream for VR. Any programmed player movement must never take control of the head. It has to be driven by the head mounted display (HMD) otherwise immersion completely breaks down. Depending on the hardware affordances, it is certainly possible to take control of or modulate other parts of the representational body or even Player’s location in virtual space as long as it doesn’t conflict with the physical feedback from the hardware setup. This is the underlying observation that makes our work on physiologically motivated navigation to reduce nausea possible. The head can also be used for input gestures such as nodding, shaking and tilting.
Lighting Cues include shadows and reflections. This can be a bigger challenge (or creative opportunity) for abstract representations. Audio cues such as breathing or footsteps can be very important in maintaining a sense of embodiment in situations where most or all of the Player’s visual representation is missing.
Embodiment Symmetry is a less obvious concept and only applies to multiplayer experiences. In Hover Junkers the Players see themselves as abstract, dis-connected hands to keep controls uncluttered and focus on the game action while other players see them as different animated avatars. In other words, its possible to differentiate in a social context how a Player may feel or see themselves versus how others view them.
Technical Considerations
Full body IK makes it possible to adjust Player’s pose to conform to HMD position in addition to simple head rotation. Final IK from Root Motion provides a fast and robust way of implementing this in Unity3D . A demo of using a character rig with a camera mounted as the head can be viewed in this brief video.
Camera parenting should technically be offset from the head joint. This provides the head bobbing one would get from walking. For more abstract representations and to reduce viewer discomfort, it maybe better to offset from the Player’s root joint.
For full body representations Clipping Plane of the camera can become a serious issue as character’s hands and other body parts pop in and out of the viewing plane. In addition to reducing the near clipping plane as much as possible and increasing object bounds, a couple of other tricks that can reduce this problem are to fade the image to black as objects get very close to the clipping plane or fade the objects themselves using transparency to at least avoid a sharp pop, a trick that I first used for bubbles in Finding Nemo.
Choices for “Emily & Me”
Emily & Me is a character centric VR experience that we are building for Oculus HMD with an Xbox controller. Emily is a young girl that invites you to play in her backyard. For our initial demo we have focused the core experience on how the Player can build a relationship with Emily rather than narrative, game play or visual fidelity.
Since our experience is primarily about building a relationship with Emily through shared activities, we wanted to minimize Player’s awareness of their own specific body by using an abstract representation of just the hands. Further more, they only fade in when Player triggers specific actions and fade away when they are no longer needed. In keeping with the theme of kids playing outdoor, we use wool mittens. We also have a Cookie Monster watch in one hand and a friendship bracelet and Kids & Art band in the other to give the model some character and to ease the hard transition from the cut off mittens.
We use the physiologically motivated WhamixPlayerController for Player navigation. It is parented to Player’s root joint to provide smooth motion and avoid unnecessary head movement. The mittens cast shadows whenever they are visible and with our outdoor set there is no call for reflections. We do employ breathing as well as footstep sound effects to keep continuity of embodiment even when the mittens themselves are not visible. Breathing is modulated between rest state and movement.
Conclusion
In dialing in the appropriate Player embodiment for “Emily & Me” we have realized how impactful this can be on immersion as well as achieving our goal of creating a character led VR experience. In general, there are so many exciting creative possibilities including empathy, role play and perception in a social context that Player embodiment can unlock. In this blog we have tried to present a framework with different attributes that can be consulted in designing Player embodiment. We also discussed how full body IK, camera parenting and managing the viewer clipping plane can support these design goals. In future blogs we hope to share more of our ongoing learnings on how the environment and character performance can further Player presence.
