VR系列——Oculus最佳实践：九、用户输入和导航

对于VR而言，没有任何传统输入方法是完美的，但游戏手柄在当前还是最好的选择；变革和调研是必要的（Oculus也在积极探索中）。

用户从Rift中不能看到他们的输入设备；要让他们使用熟悉的控制器，以便能够在看不到东西的情况下进行操作。

通过操控（Leverage）Rift的传感器来控制输入（例如，使用你的头来进行瞄准），但是要注意避免头部运动和虚拟动作交互产生的晕眩。

移动时，在VR中会产生奇怪的问题。

考虑提供一种用户可以触发的“坦克模式”风格的运动。包括一种重置行进方向为当前目光方向的方式。

鼠标、键盘和控制手柄

一旦用户戴上Oculus Rift，他们不能看见他们的键盘、他们的鼠标、他们的控制手柄，以及他们的显示器，意识到这一点是相当重要的。一旦他们身临其境，将通过独立地触摸来与这些设备进行交互。当然，这种方式并不奇怪；我们习惯于通过触碰来操作我们的输入设备，但是我们使用视觉来进行初始定位并纠正（例如调整我们手指在键盘上的位置）。这对交互设计导致了非常重要的后果。例如，任何使用键盘输入的方式必然是尴尬的，因为用户除了用触碰的方式，将不可能找到独立的按键或者主页位置。相对键盘而言，鼠标可能会比较容易使用，只要用户在戴上设备之前能够清晰地知道他们的鼠标在哪。

尽管不是最佳的解决方案，控制手柄仍是当前最流行的传统控制方案。用户可以使用双手握住手柄并且不会受限于使用那些符合人体工学因素的复杂桌面控制设备。在没有虚拟指引的情况下，对控制器越熟悉，用户用起来就会越舒适。

我们相信控制手柄是比键盘和鼠标输入更好的方式。但是，我们必须强调没有任何输入方法对VR是完美的，并且Oculus正在开展调研，来查找革命性的和直观性的方式，来与相当大一部分的VR内容进行交互。

可选的输入方法

作为一种使用鼠标和控制器进行瞄准的替代方式，一些VR内容让用户使用他们的头进行瞄准；例如，用户通过使用当前正在面对的方向的正中间的十字线或光标来瞄准。在行业内，我们将这种方法称为“光线投射”。Oculus的用户测试建议，光线投射可以作为一种直观的、对用户交互友好的方法，只要用户拥有一个清晰的指向目标的光标（被渲染为跟目标对象在同一深度）并且通过充分的视觉反馈来指明他们凝视方向的效果。例如，如果使用这一方法来对菜单项进行选择，元素应该能够采用一种显著的可视化方式来跟目标十字或光标进行交互（例如，动画和高亮）。同时要注意到，使用头部运动来进行目标定位在精度上会有限制。在这个菜单例子中，菜单项要足够大并且被很好的进行布局，以便用户能够精确地对他们进行定位。进一步地，用户可以移动他们的头而不用改变他们的目标—例如，如果一个提示信息出现在通过使用光线投射进行导航的菜单的外围。要确认光线投射是否适合你的VR内容，用户测试是必要的。

Rift传感器使用方向上、重力感应上，以及相对于初始位置的信息来控制虚拟摄像头，但是这些数据的读取可以通过唯一控制模式来操控，例如目光、头部和控制躯干的运动来进行杠杆放大。例如，用户可能看着他们想要移动的方向，并且向前倾斜来朝着这个方向移动。尽管一些内容已经实现了这些控制方法，他们的舒适程度和可用性跟传统输入方法相比较仍然是未知的。

结果，开发人员必须评估控制模式来确保他们不会不经意阻碍或者让初学用户感到不舒适。例如，在理论上歪头可能是一种合理的控制模式，但是如果一个用户在VR中旋转并且歪头脱离了旋转轴，这个动作创建了一种“伪科里奥利效应”（pseudo coriolis effect）。调查者在测试对象中发现伪科里奥利效应会持续引起晕动症[1]，所以应该在基于歪头的控制模式中避免使用。相似地，其他一些未知的不经意出现的效果可能在新奇的输入方法中存在，从这使得跟用户一起进行测试的需求增强。

导航

对大多数用户，运动将通过一些输入形式被执行，而不是实际的站立和行走。通用方法简单地使用当前传统单机游戏 — 使用控制手柄或者键盘或者鼠标。不幸的是，传统的控制方法—当有效地用于在一个视频游戏环境进行导航时—有时可能会在沉浸式VR中引起不舒适感。例如，在机枪扫射和向后走动产生的之前提到的虚拟环境眩晕问题，就不会在控制台和PC游戏中出现。我们当前正在致力于新的VR导航控制模式的调研。

可选的控制模式已经被考虑用来改进用户在运动中的舒适程度。典型地，点击“前进”在传统控制模式下，会使得用户朝着摄像机被指向的方向移动。但是，开发人员可能同时使用“坦克模式”或者“坦克视图”来进行导航，这时输入方式控制着运动的方向，并且用户通过头部运动独立控制摄像机。例如，一个用户可能保持朝着相同的路径只要他们只按着向前的按钮，并且移动他们的头将允许他们向四周环顾而不影响行进。一个人可能喜欢这种方式来浏览一个商店的过道—你的大腿沿着过道的笔直的路径，但是你的头转来转去来寻找，与你当前的走动相互独立。

这种可选的控制模式有优点，也有缺点。一些用户在Oculus办公室（并且假设开发人员已经在现有的内容中实现了该模式）发现这些控制方法比起传统的导航控制更加舒适。但是，这个可能引入新的令人不舒适的和用户体验的问题，特别是用户的头部和运动的方向可能变得无法对齐的时候—一个用户想要朝着他看的方向前进有可能因为他的头和身体在椅子上转了个身而变成朝着对角线前进。因此，任何人使用这种方法进行导航时，应该有一种简单的方式来让用户重置“坦克”的头部，以匹配用户的凝视方式，例如点击模拟的摇杆或者点击一个按钮。

进一步的研究是有必要的，这样可以完全确定在不同的使用场景下“坦克模式”的舒适与有效性。但是研究表明，开发者认为传统的可选控制模式是用户的可选择项。

现在，只要开发人员能够注意避免我们在这份指南中提到的各种问题，传统输入方法对于大多数用户来说，是一种安全的和易接触的选择。

一些内容同时借用其他方式在虚拟空间运动。例如，一个用户可能在不同的层次操作，每个都使用新的位置开始。一些游戏被渐变成黑色来传达一个玩家进行了睡眠或者散失了意识，然后让他们在某个地方醒来，作为叙事的一部分。这些约定可以被在VR中执行，而不会产生大的问题；然而，需要注意到的是，在用户可控范围之外改变用户的位置（例如，向右90度视角的跳跃，将他们移动到相同地图的另一个位置）会让人感到迷惑，并且潜在地伴随着视觉上的不舒适。

[1] Dichgans, J. & Brandt, T.(1973).由移动虚拟刺激引起的视觉晕眩感和伪科里奥利效应。

原文如下

No traditional input method is ideal for VR, but gamepads are currently our best option; innovation and research are necessary (and ongoing at Oculus).

Users can’t see their input devices while in the Rift; let them use a familiar controller that they can operate without sight.

Leverage the Rift’s sensors for control input (e.g., aiming with your head), but be careful of nauseating interactions between head movements and virtual motion.

Locomotion can create novel problems in VR.

Consider offering a “tank mode” style of movement that users can toggle. Include a means of resetting heading to the current direction of gaze.

Mouse, Keyboard, Gamepad

It’s important to realize that once users put on the Oculus Rift, they can’t see their keyboard, their mouse, their gamepad, or their monitor. Once they’re inside, interacting with these devices will be done by touch alone. Of course, this isn’t so unusual; we’re used to operating our input devices by touch, but we use sight to perform our initial orientation and corrections (such as changing hand position on a keyboard). This has important ramifications for interaction design. For instance, any use of the keyboard as a means of input is bound to be awkward, since the user will be unable to find individual keys or home position except by touch. A mouse will be a bit easier to use, as long as the user has a clear idea of where their mouse is before they put on the headset.

Although still perhaps not the ultimate solution, gamepads are the most popular traditional controller at this time. The user can grip the gamepad with both hands and isn’t bound to ergonomic factors of using a more complicated control device on a desktop. The more familiar the controller, the more comfortable a user will be when using it without visual reference.

We believe gamepads are preferable over keyboard and mouse input. However, we must emphasize that neither input method is ideal for VR, and research is underway at Oculus to find innovative and intuitive ways of interacting with a wide breadth of VR content.

Alternative input methods

As an alternative to aiming with a mouse or controller, some VR content lets users aim with their head; for example, the user aims a reticle or cursor that is centered in whatever direction they’re currently facing. Internally, we currently refer to this method as “ray-casting.” User testing at Oculus suggests ray-casting can be an intuitive and user-friendly interaction method, as long as the user has a clear targeting cursor (rendered at the depth of the object it is targeting) and adequate visual feedback indicating the effects of their gaze direction. For example, if using this method for selecting items in a menu, elements should react to contact with the targeting reticle/cursor in a salient, visible way (e.g., animation, highlighting). Also keep in mind that targeting with head movements has limits on precision. In the case of menus, items should be large and
ad5e
well-spaced enough for users to accurately target them. Furthermore, users might move their heads without intending to change their target—for instance, if a tooltip appears peripherally outside a menu that is navigated by raycasting. User testing is ultimately necessary to see if ray-casting fits your content.

The Rift sensors use information on orientation, acceleration, and position primarily to orient and control the virtual camera, but these readings can all be leveraged for unique control schemes, such as gaze- and head-/torso-controlled movement. For example, users might look in the direction they want to move, and lean forward to move in that direction. Although some content has implemented such control methods, their comfort and usability in comparison to traditional input methods are still unknown.

As a result, developers must assess any novel control scheme to ensure they do not unintentionally frustrate or discomfort novice users. For example, head tilt can seem like a reasonable control scheme in theory, but if a user is rotating in VR and tilts their head off the axis of rotation, this action creates a “pseudo coriolis effect.” Researchers have found the pseudo coriolis effect to consistently induce motion sickness in test subjects,[1] and therefore should be avoided in any head-tilt-based control scheme. Similar unintended effects may exist unknowingly inside your novel input method, highlighting the need to test it with users.

Navigation

For most users, locomotion will be carried out through some form of input rather than actually standing up and walking around. Common approaches simply carry over methods of navigation from current gen first-person games, either with a gamepad or keyboard and mouse. Unfortunately, traditional controls—while effective for navigating a video game environment—can sometimes cause discomfort in immersive VR. For example, the simulator sickness section above described issues with strafing and backwards walking that do not affect console and PC games. We are currently engaged in research into new control schemes for navigation in VR

Alternative control schemes have been considered for improving user comfort during locomotion. Typically, pressing “forward” in traditional control schemes leads to moving in whatever direction the camera is pointed. However, developers might also use a “tank mode” or “tank view” for navigation, where input methods control the direction of locomotion, and the user controls the camera independently with head movements. For example, a user would keep walking along the same straight path as long as they are only pressing forward, and moving their head would allow them to look around the environment without affecting heading. One might liken this to browsing an aisle in a store—your legs follow a straight path down the aisle, but your head turns side to side to look around independently of where you are walking.

This alternative control scheme has its pros and cons. Some users in the Oculus office (and presumably the developers who have implemented them in extant content) find this method of control to be more comfortable than traditional navigation models. However, this can also introduce new issues with discomfort and user experience, particularly as the direction of the user’s head and the direction of locomotion can become misaligned—a user who wants to move straight forward in the direction they are looking may actually be moving at a diagonal heading just because their head and body are turned in their chair. Anyone using this method for navigation should therefore include an easy way for users to reset the heading of the “tank” to match the user’s direction of gaze, such as clicking in an analog stick or pressing a button.

Further research is necessary to fully determine the comfort and effectiveness of “tank mode” under different use cases, but it represents an alternative to traditional control schemes that developers might consider as a user-selectable option.

For now, traditional input methods are a safe and accessible option for most users, as long as developers are mindful of avoiding known issues we have described in this guide.

Some content also lends itself to alternative means of moving the player around in a virtual space. For instance, a user might progress through different levels, each of which starts in a new location. Some games fade to black to convey the player falling asleep or losing consciousness, and then have them awake somewhere else as part of the narrative. These conventions can be carried over to VR with little issue; however, it is important to note that applying changes to the user’s location in the virtual space outside their control (e.g., a jump in perspective 90° to the right, moving them to another location in the same map) can be disorienting and, depending on the accompanying visuals, potentially uncomfortable.

[1] Dichgans, J. & Brandt, T. (1973). Optokinetic motion sickness and pseudo-coriolis effects induced by moving visual stimuli. Acta Oto-laryngologica, 76, 339-348.