Visual Design Methods for Virtual Reality

Mike Alger
MA Moving Image

September 2015

Abstract: This paper presents some pre-visualization design methods for volumetric user
interfaces and experiences within the larger scope of a virtual reality operating system.
 !2

Initial Manifesto:
(11:05)

https://vimeo.com/116101132

Final Summary:
(17:47)

https://vimeo.com/141330081
 !3

Table of Contents

4 Introduction
4 Personal Motivation
6 Primary Question
7 Research Process
9 Context
15 Opportunity for the Workplace
17 Theory
19 Input
21 Input UI
26 Content
27 Environment
29 Icons
31 Buttons
36 Content Zones
51 Practical Application
53 Use Case: Animation Prototyping
59 Use Case: Zone and Environment Prototyping
61 Adding Depth to Monoscopic Photo Spheres
66 Putting these Concepts Together for a VR OS Design
75 Prototype and Evaluation
76 Whats Next?
76 Conclusion
79 References
90 Appendix: Avatars
92 Avatar Creation Methods
92 3D Modelling
93 Photogrammetry
95 3D Scanning
97 Hybrid Solutions
 !4

Introduction

Personal Motivation

I nd myself sitting cross-legged, zipping around on a golden cloud above a small

sand island with a single building. Godzilla emerges from the water and joins

several characters, mostly from pop culture, either standing on the island or ying

on their own clouds. Most of them are laughing and commenting on how ridiculous

the oversized Iron Giant looks perched cross-legged on his own oating yellow poof.

Their accents are Dutch, French, British, and American. As a group, weve come here

through a portal from one world and will depart through another when we get

bored.

Its virtual reality, or colloquially, VR. In actual reality, Im sitting at my desk

holding a game controller. My hearing is replaced by the sound of headphones and

my sight is replaced by the light of a head mounted display. Im playing what is

essentially a rst-person-shooter video game connected with people online, except

the shoot function is replaced by talk. The application is called VR Chat and each

week a group of people will log on to see the latest avatars and tour the latest

worlds created by each other (Gaylor and Joudrey, 2015). There are other social VR

applications such as ConVRge and AltSpace, but as Virtual Reality as a medium is

still in its infancy, only regaining popularity in recent years (Nelson, 2013), the

attendance tends to be between ten and twenty weekly.

Each participant is interested in consuming or contributing in their own way

and many have their own projects. Gunter, for example, leads the tours and hosts a

weekly podcast interviewing VR content creators (Gunter, 2015). Tom creates

 !5

avatars and worlds heavily themed with the cross section of biblical and digital

concepts (Tom23, 2015). Ggodin develops an application that allows people to use

their Windows interface in another environment (Godin, 2015). Jesse and Graham

are the creators of VR Chat itself (Gaylor and Joudrey, 2015). But what do I do?

In the past, I have focused on digital media with an emphasis on video

production and graphic design. The marriage of these mediums is motion graphics

or motion design - my favorite area. As I continued to create videos and design

websites, I couldnt help but feel that I was producing content that would soon

become irrelevant, lost forever in the ether with the myriads of preceding lms and

designs. I yearned to be part of the creation of something larger, something with a

more signicant and lasting impact on mankind. It would seem that thing could be

virtual reality.

As a medium, there are very few existing standards, protocols, and

workows. Its like participating in lm before cinematography had shot names, or

like participating in the internet before web designer was a position. The things

you do and techniques you use to do them for virtual reality can become the

standards that people later think of as commonplace. Although a fairly weak and

still eeting form of immortality, participating in the developmental stages of a

medium with such potential makes me feel like Im contributing to something that

will have a lasting impact, for worse and better. Deciding to work with the human

perception of the moving image as it relates to virtual reality led me to ask how can

I contribute?
 !6

Primary Question

It would seem that a large portion of virtual reality content being produced

currently is intended purely for consumption (Bye, 2015). Very few virtual reality

experiences allow the user to create something with it. I initially thought it was

strange that developers would write code, pull the headset down over their eyes to

test it, take it o to adjust their work, put it back on to test again it seemed that

this all should be done with the headset on. If it is such a powerful and versatile

medium, why not use it to build the experiences themselves?

Even as the concept is suggested, the worms start to exit the can. Creating a

game, for example, may require 3D modeling, image editing, and coding. While

each have tools on traditional 2D computer screens, no interfaces have been

created that are designed for these tasks in the volume of virtual reality. A user

cant even see their keyboard to type. Theyre now moving their body in ways that

werent the case with the previous mouse and keyboard system. It would seem

initially that everything we know about digital interaction design is thrown out the

window as the majority of our current interfaces are perceived as the two

dimensional screens of computer monitors and smartphone screens. This thought

process led me to my initial question:

How can the two dimensional paradigm of operating systems be most

ergonomically redesigned for head mounted displays?

One of the interesting caveats of this question is the prospect of design. It proposes

the creation of workow - to design a process for design. In order to create this new

medium of interaction, a method of creation must rst be invented.

 !7

Research Process

Virtual reality is currently a very fast evolving topic, seeming to change by the week.

There are several research methods I have adopted so as to attempt a well-rounded

approach. While traditional methods of reading books and papers is necessary and

helpful, many of the concepts described in older publications have evolved in

recent years. In order to remain up to date, it is necessary to subscribe to blogs and

discussions online and read the problems and ndings of developers as they

experience and document them. These sources are, of course, typically anecdotal

and purely qualitative without controlled studies or quantitative rigor. There remain

many guidelines that are generally agreed upon by the VR community without

particular proof from a study, but are apparent in practice. This is one of the

reasons that my research process has also included trying out as many experiences

as possible for myself, allowing me to form my own opinions about the validity and

eectiveness of techniques and implementations. Making my own experiences and

testing them as well as showing them to other people has been another part of my

research process. Understanding what other people are thinking and feeling about

VR is helpful for the user experience design process for obvious reasons. Another

part of this has been my participation in meetups, hackathons, and game jams -

physical gatherings where groups discuss and create content. I have also been

giving talks presenting my own ndings along the way at such events

(documentation of these events can be found here: (Alger, 2015a,b,c,d)). Another

source of information for me has been to ask questions and interview experts via

email, video conference, or even in virtual reality itself. Participating in social VR

applications has also aorded me the opportunity to converse with the spectrum of
 !8

developers and hobbyists participating in the medium rst-hand. As a result of

these research practices, many of my resources come from recent years and

internet-available resources, which to me seems appropriate to me for a digital

information age.

As I have progressed through the discovery and understanding process of

virtual realitys current state and potential, I have found myself pursuing and

answering tangential questions to my original focus. For example, I have found

myself interested in the avatar as it relates to identity and visual representation of

self, for which I present a few thoughts and methods for creation of personalized

avatars in the Appendix. Within the original question, however, I found that I would

rst need to create methods and guidelines for VR design. I found that I needed to

start with ergonomically responsible zones for content mixed with design

workows modied from existing mediums. It is this process that this manuscript

intends to describe.
 !9

Context

When attempting to understand and develop relevant ideas for subjects, its

necessary to reiterate our understanding of their current denitions and historical

context. With regard to virtual reality, one of the basic things to consider is the way

that our senses serve as the input our brain uses to construct an understanding of

the world around us. Sight, hearing, touch, smell, and taste are the most widely

accepted set of external stimulus that the human body perceives (Sense, 2015).

These senses and our reactions to them are the result of millennia of natural

selection (Darwin, 1859)* and there are several consequences of this built into our

instinct. This is all relatively common knowledge and seems like it may not need to

be reiterated here, but the important thing is to state that we, as humans, have

certain predictable outputs based on certain sets of inputs. Essentially, its instinct.

Human nature. Certain behaviors are hard-wired into us from our ancestors as

much as a dog wants to dig and a bird wants to migrate.

A common word in interface design is intuitive: basically, knowing

something without having to be told (Intuitive, 2015). Interfaces are praised for

being intuitive because people will know how to use them without any time or

energy spent on training. Often, intuitive interfaces rely on some aspects of

human instinct. A planned exhibition may use light, motion, sound, and space to

draw a persons attention through an area. A well designed website will similarly

use color, distance, and typography to clearly communicate a purpose and often

persuade some sort of action. Contrasting elements of sight like light, color, and

motion naturally draw our attention because they were necessary for our own

survival as animals, whether we were hunting or avoiding being killed ourselves.

 !10

The same applies for contrast in sound, touch, smell, and taste. The process of

design is often the creation of methods to coax an automatic response from end

users by way of these natural tendencies (Sundstrom 2015; Alger 2015e).

The evolution of the graphical user interface (GUI) for computers is an

interesting vignette in the enormous subject of visual design. The GUI evolved as a

solution to understand the data and processes taking place in a computers system.

By visually representing programs and information using text and shapes, an

interface is created. Its medium allows input and output that the computer

understands, but also that a human can understand, though typically with some

amount of training. The rst graphical user interface to use the now-common

desktop concept was created by Xerox PARC and set the precedent for desktop

metaphors that are still used today (Thacker et al., 1979; Koved and Selker, 1999).

The main 2D analogy was a desktop with pieces of paper sitting on top of each

other and consisted of the now common elements of windows, icons, menus, and

pointer (Preece et al. 1994; Hinckley 2002). This same structure has proven useful in

personal computing for decades now, based on rectangle sections of content within

a rectangle screen. Of course, several other GUIs now exist like iPods or ATMs... the

most widely adopted of which recently may be the multitouch smartphone. Each of

these has a tailored user interface to accept physical input from a human and

display feedback clearly to complete tasks.

Of course, this is all mentioned to come back to virtual reality and the way we

can interact with it. There are several forms of virtual reality including the cave

with projections on walls (Cruz-Neira et al., 1992; 1993) and the workbench with

stereoscopic desk projections (Kreuger et al., 1994; 1995). This paper relates to

virtual reality through head mounted displays (HMD). An HMD is like headphones
 !11

for your eyes. Headphones give your ears articial sounds, HMDs give your eyes

articial light (Shibata, 2002). Just like headphones can be designed to block out

outside sound, HMDs can either let light through or replace your vision entirely. At

the time of this writing, replacing vision entirely is referred to as virtual reality (VR),

while mixing real light with articial is referred to as augmented reality (AR)

(Agarwal and Thakur 2014). Head mounted displays for VR and AR, like other forms

of virtual reality, work by presenting stereoscopic images to each eye and updating

those images as the user moves their head (Cakmakci and Rolland, 2006). As long

as the hardware are software are performing their tasks correctly, virtual objects

appear to remain in a xed position relative to the user.

What is particularly impressive in use are the illusions of scale, space, and

depth that naturally occur with HMDs. The way we perceive the world optically can

be dened by a few variables describing the way our eyes receive light in what is

called the plenoptic function (Adelson and Bergen, 1991). We are accustomed to

moving through and perceiving light elds as light will be entering our pupils from

every direction no matter where we place our head. Our degrees of freedom to

perceive a scene can be dened by 10 variables: x, y, and z position; pitch, yaw, and

roll rotation; distance, horizontal, and vertical position to the point of convergence

and focus; and the size of the pupils aperture (McGinity, 2014). Virtual reality head

mounted displays are currently able to replace 8 of these convincingly: The 3

position, 3 rotation, and 2 convergence variables. As the user moves and rotates

their head while looking around a scene, the image is updated for each eye

accordingly. However, the nal two variables are not typically accounted for in VR

systems: focus and aperture. The eyes experience some strain over time as they

attempt to focus as they normally would by compressing/expanding in accordance

 !12

with their convergence (Homan et al., 2008). The 2D display screen doesnt allow

for this, though. Displays are also currently restricted to a dynamic range dictated

by their electronics, and dont aord the high dynamic ranges of light found in the

real world that would aect the pupils aperture normally (Reinhard et al., 2010). In

the future, plenoptic lenses (Lanman and Luebke, 2013), light eld tensors

(Wetzstein et al., 2012), and high dynamic range screens (Seetzen et al., 2004) may

solve these problems, but they are not currently consumer product solutions. Even

still, the eect of perception to the user with current HMDs is enough to convince

them of the spatial location and scale of a virtual object.

This is where a few common buzzwords of virtual reality come in: immersion

and presence. With mediums of entertainment like theater or cinema, there is a

certain suspension of disbelief aorded by an audience participant. The viewer can

choose to accept the presented reality as plausible for the sake of the experience.

This may even reach visceral level in which the bodys subconscious reactions are

triggered based on human nature: jumping, screaming, laughing, crying, etc.

Because the user of VR appears to be surrounded by the virtual objects and

environment, it is much easier for them to accept these surroundings as fact, both

consciously and subconsciously (Abrash, 2014). While the terms immersion and

presence are increasingly used for hype and marketing purposes, I personally like

the way that Michael Abrash describes presence as it relates to virtual reality. He

describes the human perceptual system through the use of optical illusions, in

which our understanding of raw image data is clearly being fooled. He describes

that our other systems (hearing, proprioception, touch, etc.) are also susceptible to

illusions. It is this fooling of these subconscious perceptual systems that he denes

 !13

as presence. Immersion is being presented only with data from a false

environment, but presence is having your body believe it on a fundamental level.

It is now that all of those millennia of evolution nally come into play with

computers. It is now that the human perceptual system and the graphical user

interface really meet. Whats particularly interesting here is that virtual reality as a

technology is actually older than the graphical user interface. Ivan Sutherland

created the rst virtual reality and augmented reality head mounted displays in the

late 1960s (Sutherland, 1968). The rst GUIs were created by Xerox in the early

1970s (Thacker et al., 1979). However, shortcomings of technological capabilities

prevented the adoption of head mounted displays for consumers. In order to

maintain the illusion of a presented reality, an image must be updated a minimum

of about 75 times per second (Oculus, 2015a). Others may describe it as a 20

millisecond motion-to-photon latency, where the time between a users head

movement and the updated image must meet that benchmark (Abrash, 2013).

Valve requires a frame rate of 90 frames per second for their Vive system (Faliszek,

2015). Resolution of screens for HMDs also must be extremely high to not be

noticeable/distracting; higher than high denition television packed into the size of

a smartphone screen (Forums.oculus.com, 2015). The processing and rendering

power required to meet these requirements with quality experiences was

unavailable or unaordable in previous decades. Lower frame rates result in

nausea for users. The brain senses a mismatch between the optical and vestibular

systems, assumes the body has been poisoned, and makes the user ill to eject

whatever substance has been consumed (Kennedy and Frank, 1985). Many

attribute the commercial failure of virtual reality as a consumer industry in the

1990s to the fact that computers were not fast enough to overcome this obviously
 !14

serious problem (Barras, 2014). However, the advance of Moores law (Moore, 1965)

in conjunction with the proliferation of the smartphone market has facilitated the

creation of small, fast computers with extremely high resolution displays including

miniaturized accelerometers and gyroscopes (Poeter, 2015). This allows us to create

high quality environments capable of rendering in real-time to be experienced in

head mounted displays.

That doesnt mean there arent still limitations. In order to maintain an

acceptable frame rate, developers must budget the number of polygons, scripts,

materials, etc. used in their scenes. The resolution of the displays have reached a

level that I would personally say appears like illuminated sand, but thats still not

particularly clear for reading small or far away text. There are also diering

opinions on solutions to user movement and text input. A users viewpoint cannot

rotate or accelerate independently from their head without risking some degree of

motion sickness (Oculus, 2015a). Each of these are challenges to be aware of and

worked around or solved in the design process of individual experiences.

At the time of this writing, there are a few existing and announced consumer

head mounted displays and input devices that can be targeted for development.

The community tends to categorize these as being either virtual reality or

augmented reality, but as time goes on it becomes clear that there will be no

hardware distinction between the two eventually. Putting a camera (or two) on a VR

headset makes it AR. Covering the whole eld of view with pixels in an AR device

makes it VR. So, from the design perspective of a graphical user interface, many of

the same tactics can be applied to both. Many elements of an operating system

interface design for the Oculus Rift or Valve Vive could be used with the Microsoft

Hololens. The methods and types of input would be the main things that would
 !15

change the interfaces design, but most seem to be adopting some form of motion

controllers for both hands in 3D space. So, while I will more often use the term

virtual reality throughout this manuscript, the same principles will typically be true

for augmented reality.

As a summary of the concepts introduced so far: we try to design interfaces

for humans that will be easy to use based on their instincts. Virtual reality interfaces

with the human perceptual system to a more intense degree, providing a more

eective opportunity for computer to human communication. By designing

interfaces appropriately, we now have the opportunity to facilitate more robust

human computer interaction.

Opportunity for the Workplace

A well designed operating system specically for virtual reality has the potential to

be revolutionary for the digital workplace. There are droves of people who go to

work and sit in front of a computer screen. They use their operating systems to

complete tasks throughout the day. In traditional two dimensional computer

interfaces, screen space is at and fairly limited. This means users must spend

some of their brain power navigating the abstraction between tasks that is inherent

to the display method (Medich, 2015). The interruption of doing this breaks thought

continuity and decreases productivity (Shamim, Islam and Hossain, 2012). One way

people deal with this is by getting multiple monitors. The increased screen space

real estate of larger monitors or multiple monitors serves to diminish navigational

interruptions and increase productivity (NEC, 2010; Ball and North, 2005; Kang and

Stasko, 2008). That is to say, when you dont have to organize windows on top of

each other, you can get more done. The immersive volume of virtual reality entirely
 !16

surrounding the user 360 seems a natural inevitability for a maximized working

canvas. That canvas has the added benet of z-depth. In addition to this, letting

users spatially organize their tasks can lead to an increase in productivity upwards

of 40% (Colgan, 2015). And if that werent enough, workers in more pleasant

surroundings tend to be happier (Fisher, 2010; Gallagher, 2007) and users have the

ability to customize their environment in virtual reality, like one customizes a

desktop background on their computer. The combination of an innite working

volume with spatially organized tasks in an ideal environment results in people

being both happier and more productive.

 !17

Theory

In order for any of this to be eective, reasonable design principles need to be both

implemented and discovered. There are several existing principles for design which

can be translated from other mediums. Print design, web design, architecture,

interior design, theater, motion graphics, etc. all have elements that can be seen as

relevant and adopted. At the same time, the medium of virtual reality has

properties, like the ability for content to intersect, that are unique.

I decided that it would be necessary to dene some common user tasks. I

thought the easiest way to gure out what people want to do with their computers

would be to look at what the most common applications and websites are currently,

and what their primary tasks are.

Looking at Alexa (Alexa, 2015) for the most visited websites and online lists for the

most popular mobile and desktop applications, I compiled this list, organizing them

by task genre according to my own understanding of each:

Task Genre Example Services

Status Indicators time, battery, wi strength, bluetooth

Schedule Planning calendar, alarms, weather, event notications, ights, Fandango

Note Taking Notepad, Textedit, Evernote, stickies

Reference Google, Wikipedia, IMDb, Ask, Yelp, Wikia, recipes

Calculation calculator, unit converstion, stopwatch, timer

Watching Video youtube, Netix, Amazon Prime, twitch, QuickTime, VLC, Windows Media Player, Plex

Listening to Audio Pandora, Spotify, Soundcloud, Vevo, iTunes, audiobooks, Podcasts

Shopping Amazon, Taobao, Tmall, Apple, Craigslist, Flipkart, Adobe, eBay, Etsy, Walmart, Ikea

Email mail, Outlook, Gmail, Thunderbird

Instant Message WhatsApp, Facebook Messenger, iMessage, MSN

 !18

Audio/Video Chat Skype, FaceTime, Google Hangout

File Management Windows Explorer, Finder, iPhoto, Dropbox, Google Play, App Store, iTunes Store, zip
extractors, Contacts

Computer Management System Preferences, Microsoft System Center, Symantec, McAfee, Remote Desktop,
1Password

Finance Quicken, stocks, online banking, BankofAmerica, Chase, WellsFargo, Mint, TurboTax,
currency conversion

Location Finding Google Maps, Yelp, Apple Maps

Games Steam, Candy Crush, Angry Birds, Temple Run, Bioshock, Solitaire, Portal 2, Counter-Strike,
Skyrim, Half-Life, GTA V, Minecraft, Sims, World of Warcraft, etc.

Reading Adobe Reader, Kindle, Preview, Document Reader

Pornography xvideos, xhamster, pornhub, redtube, youporn, webcam sites

Talking to Strangers forums, Tinder, MMO games, Match, OkCupid

Capturing Pictures/Video camera, Photo Booth

Development MySql, stackoverow, github, text apps, dreamweaver, OS X server, Unity

Contacts Linkedin, Salesforce

Training Youtube, Adobe, Lynda

Money Making Swagbucks, Ibotta, Stock portfolio managers

Word Processing Word, Pages, Google Docs, Open Oce, DayOne, Adobe Acrobat

Data Analysis/Visualisation Excel, Fitbit, Google Analytics, Mint

Presentations Powerpoint, Keynote, Presi, Google

Image Creation Photoshop, Gimp, Lightroom, Aperture, Illustrator

Audio Creation Garage Band, Logic, ProTools, FruityLoops

3D Creation Maya, 3DSMax, Cinema4D, Zbrush, Blender, CAD

Video Creation Final Cut Pro, Premiere, After Eects

Viewing subscribed content

Specic News Cnet, ESPN, the Verge, Wired, Engadget

General News Yahoo, Live, Bing, MSN, BBC, Xinhua, NYtimes, HungtonPost

Friends/Contacts Facebook, Twitter, Instagram, Snapchat, LinkedIn, Tumblr, Blogspot, Wordpress,

Livejournal, Google Plus, Vine

Content Discovery youtube, Reddit, Pinterest, Imgur, Diply, Youkou, Tudou, Vimeo, DeviantArt, Buzzfeed
 !19

What surprised me in doing this practice was that, even though they are my

own categories, I could only come up with these 35. There werent any other things

that I could think of that people use computers for on a regular basis. Based on

this, it seems like you could provide an operating system that does everything

someone would want to do by designing 35 applications thoughtfully and

comprehensively. At the same time, I understand that designing any one of these to

that level is a large task and people may want to do entirely dierent things in

virtual reality that have yet to be invented. I tried starting with some of the things I

thought might be easier rst:

A media viewer for basic content like documents, images, video, & audio

status indicators

a calendar with weather and alarms.

But how are video and audio represented in 3D? Where do status indicators go that

they can be checked at a glance? Can a calendar communicate more information in

a 3D design than its traditional 2D form? How would one interact with it? It became

very quickly clear that there were some much more basic questions to answer

before individual applications could be designed. Namely, where to put content and

how to interact with it. I was going to have to start at square one and think about

the basics of layout and design workows repurposed for three dimensions.

Input

One of the rst questions for interface design is how the user is going to be giving

information back to the system. This is typically dictated by the hardware available.
 !20

The mouse, keyboard, joystick, ATM buttons, iPod clickwheel, etc. all suggest

dierent types of interaction and the feedback represented on screen will change

accordingly. The most common interface for personal computing is the keyboard

and mouse combination. Before touch screens, the mouse provided a way for users

to indicate an X and Y location on a screen using a cursor to choose and modify

items. The keyboard, adopted from the typewriter, provided a familiar text input

method. As developers created applications, the functionality of the keyboard

expanded to include hotkeys to choose tools and perform actions quickly as well as

modiers like the control or alt keys. This allows users to use both hands: one

modifying the tools while the other performs the tasks.

The virtual reality community remained undecided on an input

standardization for a very long time and is still arguably unsettled. Google

Cardboard uses a single button on the side of the handheld viewing box (Google,

2015). Gear VR uses a trackpad and back button (Samsung, 2015). Recently, both

Valves Vive and Oculus Rift HMDs have been announced to have separate motion

tracked controllers (Valve, 2015; Oculus, 2015b). Development kits of the Oculus Rift

headset had no input solution so developers mostly relied on existing keyboards,

mice, trackpads, and game controllers. In the absence of inputs in the beginning,

several startup companies oered solutions including their own motion tracked

controllers, omnidirectional treadmills, and hand tracking (VRWiki, 2015). Hands

seem like a particularly ideal solution because anecdotally, people tend to put on a

VR headset and raise their hands automatically in my experience and those I have

talked to. I have witnessed many people try to touch invisible things and have been

guilty several times myself of attempting to touch or lean on non-existent objects.

However, for detailed and intricate tasks, hand tracking and gesture recognition
 !21

technology is often still too primitive to use consistently without becoming

frustrated (Plafke, 2013). There is also no haptic feedback for hands in space. For

these reasons, motion tracked controllers, or wands, have gained preference.

They provide normal range of movement and rotation for hands as well as physical

buttons to press.

A hybrid solution may be an option in the future. Just as the left hand used a

keyboard while the right hand used a mouse, a nal solution might be to have a

controller in the left hand while the right hand is tracked in space. Or a wand in the

right hand as the mouse cursor while the left hand presses modier buttons using

a multitouch surface. These systems would be inherently more ambidextrous than

the current solutions, which are optimized for right-handed users. Left-handedness

is a supportable setting in a virtual interface. With any solution, there must be

minimal motion to avoid fatigue. Also, if we have an opportunity to eradicate carpal

tunnel, we should take it by designing inputs to ergonomically account for repeated

use over years. For this project, I will assume the case of a motion tracked controller

in the dominant hand with a free tracked non-dominant hand. This will allow design

and analysis for each.

Input UI

As virtual reality has existed for nearly fty years, several interaction interface

concepts have already been created. One is to cast a ray like a laser pointer for a

cursor (Sherman and Craig, 2003). It can be dicult to hit a target at a distance, so

sometimes a cone is introduced, becoming larger as the ray extends.

Menus in the past have often mimicked the 2D dropdown style, but radial

designs are also becoming more common. Which choice is best tends to depend on
 !22

where the menu is originating from. Options around a hand will often be radial

because they can be accessed quickly and with muscle memory.

An existing way to customize and interact with an environment is the World

in Miniature (Stoakley, Conway, and Pausch, 1995). Instead of traveling to or

directly aecting objects at a distance, the user has a small map of the

environment. Object properties in the larger world and the world in miniature are

bound such that changing things on the map changes them in the larger

environment as well.

Some of the less solved problems are locomotion and text input. Several

hardware and software strategies for moving a user through an environment in

virtual reality have been presented, but none of them have been standardized by

unanimous or even majority adoption. The major problem is the previously

mentioned simulator sickness. The experiences view cant appear to accelerate

dierently from the users head without risking the possibility of vestibulo-ocular

mismatch, and therefore nausea (Yao, 2014). Many experiences will just ignore this

and allow users to walk their game view in a rst-person-shooter style. Others

attempt to mitigate the problem through the use of a cockpit, occluding a portion of

the view with more static surroundings. Variations of teleportation are also used,

either moving the user to another location very quickly or immediately. Finally, the

last way to solve abstracted locomotion is to ignore it altogether and only allow the

user to navigate the area they have available for tracking, one-to-one as it is called.

This particular solution would be ne for a room-scale tracking system like Valves

lighthouse in an oce space which is both physical and virtual, in my opinion.

Text input remains another unsolved mystery. Like the other user interfaces,

some solutions have been created already. Every solution relies on input method,
 !23

however. The best user to keep in mind is probably a code writing developer,

because they rely on text input. The main goals of text input would be accuracy,

speed, and comfort. If a proposed text input solution is less accurate, slower, or less

comfortable than current keyboards, people will prefer to go back. A physical

tracked keyboard can be represented in the virtual space so that the user can

touch-type as they currently do on real buttons (Sleight, 2014). The same could be

done with a multitouch surface like an iPad. But if the user is carrying a motion

controller or two, switching back and forth will not be as simple as moving from a

mouse to the keyboard as I think putting down the controllers would be annoying.

Voice recognition is possible, though not yet accurate enough for a task like coding

as I imagine saying slash, colon, bracket wouldnt be ideal. Writing letters in the air

would also be possible, but not as fast as 60 words per minute (Brown, 1988). Hand

and gesture tracking is also not robust enough to recognize sign language at that

speed. Another text input would be to use the controllers buttons and analog sticks

or trackpads like the Steam Controller interface (Plunkett, 2015). One solution is to

provide both hands with radial menus. If each hand has six options, rotating the

wrists in combinations makes 36 possible entries. These can be added to with

depth of multiple rings and modied with controller buttons or gestures. It would

require signicant training as keyboarding does currently. Someone typing would

look like they were twitching their wrists.

 !24

a radial typing interface concept

Another text input solution may be what I think of as an orchestra conductor

interface. Using the primary hand as a raycast cursor, the user could swipe the

beam through a virtual keyboard, similar to the Swype keyboard style (Swype Inc.,

2015). The keyboard mode and beam on/o could be controlled by the left hand,

with buttons on a motion tracked controller or simple gestures with a tracked hand.

conductor typing interface concept

 !25

These are some of the text input options Ive considered, but more are sure to

appear over time.

Tool shortcuts and modiers can also be useful for virtual reality. Currently

you may hold the Shift key while dragging to keep alignment or hold the Alt key to

maintain symmetry in some applications. These same tool modications could be

applied either as buttons on a motion controller or gestures with a tracked hand,

like the text input. Simple gestures like a C symbol, pointing, or extended thumb

would be most reliable and allow a user to modify their tools and actions on the y.

Many of our interaction design processes include contact with a surface.

Objects intersecting or users intersecting with objects is sometimes seen as a

mistake, but this behavior can be embraced as a part of the mediums strength. For

example, the slider is a common user interface, allowing a user to adjust a variable

within a range, like volume. In reality, it is typically a knob, either rotating or in a

slot. In 2D and touch interfaces, this is represented somewhat skeuomorphically

with an icon that can be dragged through a range. In virtual reality, the zone can be

represented as a cylinder. The user can intersect their hand at whatever value they

want.

types of range selection sliders

 !26

Content

Lets build an image of what the things in a virtual reality operating system might

look like. At rst, I made a concept art illustration to communicate the idea:

Initial conceptual illustration

It depicts a user standing in a beach environment using their arms in space as

applications and icons surround them. As I continued through the design process, I

realized this concept has some major fallacies. For one, if this is an operating

system designed for a work environment, 40 hours per week is an awfully long time

to be standing. Its also a lot of time to be raising and moving your arms. This would

be uncomfortable after just two hours. The initial concept also places all content at

the same distance from the user, foregoing the opportunity to use z-depth for

hierarchical understanding in a volumetric interface.

I realized that a primarily seated experience with minimal wrist movement in

an interface volume would be preferable. That doesnt mean users are prohibited

from standing up or using their arms more, just that they arent required to. In fact,
 !27

I think that room scale tracking has opened up the opportunity for the virtual oce

space where you can stand up and walk around your customized environment if

you want to. This is probably good as either sitting or standing for long periods

becomes uncomfortable.

Virtual reality may also not be the best medium yet for many types of

content. Low resolution makes text dicult to read at a distance and, as mentioned

before, vergence-accommodation conict causes eye strain and fatigue over time.

For these reasons, mobile and PC will still be preferable for tasks like reading emails

or books. Extruding text to make it 3D doesnt help with the silhouette recognition

of letters, so text would remain an inherently 2D medium even in VR, anyway.

What VR would be a better medium for, rather, are the types of content that

are inherently 3D, but are traditionally abstracted to 2D forms out of display format

necessity. Blueprints would be a classic example of this. Buildings are 3D forms, but

we have to communicate their design in 2D mediums using at blueprints. Medical

imaging, ski resort maps, mechanical schematics, and geology lessons all have 3D

content that we attempt to communicate with 2D mediums. It is in the

consumption and creation of these content types that virtual reality would be the

arguably better format, with stereoscopic spatial presentation. This eliminates the

abstraction of 2D diagrams and maps trying to describe 3D forms, which can be

advantageous (SpaceX, 2013).

Environment

The most obvious 3D content initially may be the environment that you are in. As

stated, the user has the ability to customize their environment like a current user

can change their PCs desktop background. However, the types of customization
 !28

available in VR are dierent. Any type of 2D or 3D content can be placed in the

environment around the user. They can sit underwater, atop a skyscraper, on the

moon, in a Frank Lloyd Wright building, or amidst elds of abstract color and

particles. The environment can also be from a spherical panorama (like Google

Maps street view photos) or use elements from real life scanned objects using laser

scanners or photogrammetry.

Users also have the ability to place content within their environment, which

brings up an interesting design point. In physical reality, all objects must exist in a

single instance taking up space in an actual physical location. If you have a

document that is a piece of paper, it will exist somewhere and you will have to go to

that place if you want to retrieve that document. In traditional computer interfaces,

les and applications tend to remain hidden until summoned or navigated to

through the folder structure. The only persistent objects are those which are

located on the desktop, either as their original le or as a shortcut pointing to the

actual les location. A common science ction representation of virtual reality is to

have the user walking or ying through interconnected systems, locating a le by

traversing distance to its location. This concept is not particularly practical for an

actual le system or database of any realistic size.

Consider the hypothetical prospect of digitizing items in a museum exhibit.

By doing so, the exhibit space can be recreated and the museums experience is

available to people worldwide. You can walk around the virtual space and nd the

piece you are looking for. The concept is such a success that the entire museums

collection is digitized. You can now virtually walk through the entire museum, but it

takes several minutes to get from one end to the other to nd a desired piece. This

is virtual reality, however, so naturally ying or teleportation become part of the

 !29

navigational interface. Finally, every museum participates. The Smithsonian, Louvre,

MoMA all museums have their entire collections digitized making a virtual reality

super-museum. At this point the concept of physically navigating to the correct

location to nd a piece is obviously unsustainable. It will be much easier to nd an

object based on name, date, collection, medium, artist, or any dimension of

searchable criteria. What you have is a database of content with metadata. You

dont go to the content, you bring the content to you.

Thus you end up naturally with the same kind of paradigm as the current PC

desktop. Things that the user wants immediately visible or accessible exist as

physically located objects in the environment around them. Everything else exists in

an immediately accessible, though not persistently visible le system. The location

and placement of these things is decided by the user, and I would expect that some

people would prefer as little as possible while others prefer what amounts to

clutter in the physical world very much like the dierences seen in both physical

desk spaces and computer desktops currently. This also lends itself to the spatial

cognition Im sure weve all heard referred to: I have it how I like it! I know where

everything is!

Icons

What do those objects look like, though? Initially, the GUI relied on icons: thumbnail

images to represent le types. This has largely remained the case, but content types

like images and video will now oer a thumbnail preview of their actual content in

more recent operating systems. It would make sense to quickly nd an image

visually this way in virtual reality, too. But at the same time, two-dimensional

symbolic icons are more quickly recognized than three-dimensional realistic icons
 !30

(Smallman et al. 2000). The styles of icons will likely need to be reimagined to

represent the newer content types. Images, for example, can be monoscopic or

stereoscopic and intended for mapping to a plane, cube, sphere, or mesh. Video

can be all the same. Essentially, they are all textures for models. Models could be

displayed without texture, but textures cant be displayed without some sort of

model. Some sort of iconography is still necessary for abstract content types like

applications or project les. In these cases, understandable silhouettes and colors

would be most helpful for quick identication.

The size of icons and objects is another interesting opportunity. Currently,

computer interfaces typically show all le types as the same visual size, regardless

of their size on the disk. When manually cleaning up a hard drive, nding the larger

les or sections is more dicult without a data visualization program. One idea for

a virtual reality operating system would be to have a view mode where les

volumes are actually representative of their physical disk space. In thinking about it,

though, this probably wouldnt be good as the default view because a text

document might appear physically tiny compared to a body scan. But it would be a

good mode to have as data visualization for disk space management.

 !31

Buttons

Buttons allow a user to initiate an action. Typically, a label accompanies the button

indicating its function.

examples of physical buttons

Buttons have been implemented digitally in many forms. Most obviously, they are

represented as labeled regions with beveled skeuomorphic lighting. Recently, the

design trend has favored minimalism and simple text or colored regions (Turner,

2014). In order to communicate their functionality to users more eectively,

additional visual states were created like hover and pressed. These tell the user

that it is an interactable button and simulates the z-depth action of pressing it.

examples of two dimensional buttons

 !32

Button style for VR depends on the interaction method. A raycast cursor pointing at

a button in the distance will need dierent feedback from a button being pressed

by a nger using motion tracking. I decided to design a button intended to be

placed within arms reach and physically pressed. The implementation would be

either in the area of a console around the user, or attached to the users arms or

controllers.

I began with some visual style mockups. A button would need to be easily

legible whether showing text or a symbol. It would need to be opaque for this

legibility as the background is unknown. It would also need to communicate its

functionality as press-able and not just an image or text region. Initially, I created

these prototype 2D images.

initial button style concepts

The last two seemed legible while communicating the functionality simply. Next I

thought it would be necessary to communicate the states of the button to the user.

There is no haptic feedback, but visual cues and sound can be triggered. In fact,

sound may help users feel like they have touched something when they actually

havent through synesthesia (Ward, 2007). To visualize dierent options, I created

several animations in After Eects. I purposely simulated the nger going beyond

the bounds of the button, intersecting it to see how this might be perceived. The

main states to communicate were:

 !33

i. The ngers distance to the button

ii. When they have contacted the button

iii. When they have pressed the button to become active

iv. When the button is reset and again press-able

still frames from button test animations

By seeing the nger go through the button, we are reminded that this isnt entirely

unnatural. It looks like a nger dipping into water. This seemed like an opportunity

to incorporate the concept of human nature in the computer interface. I decided to

design the button to mimic the act of submersion. The user pushes it through a

faux-liquid surface and the state changes are inspired by such.

First, I would need a color palette to represent these elements. Because the

ethos was human nature, I decided to sample images of things found in nature that

may inspire some sort of instinctive reaction. The sight of blood is naturally

alarming, while campres and sunsets are naturally calming, in my opinion. Purple

rarely occurs naturally and draws a lot of attention when it does. I created this

sample set of images with descriptors and subsequent palette:

 !34

color palette inuences

With the hope of inspiring some amount of intimacy with the interface for

the user, I chose the skin on the palms of hands as the main inuence for the color

of touched surfaces. It was the combination of this and the water color that I used

to create some new button designs:

 !35

button design exploration based on natural palette

I then tested several more animations based on the concept of submersion.

Animated versions can be seen at http://aperturesciencellc.com/vr/button-

iteration2.gif and http://aperturesciencellc.com/vr/button-iteration4.gif:

still frames from iterative button animations

 !36

I think the nal iteration serves as a good simple example of a

communicative button for a hand tracked interface. Its legible, it communicates its

functionality and states to the user, and it feels natural because its based on

nature.

Content Zones

The areas in space to put menus and content becomes another question. As an

example, video editors have title safe and action safe zones to make sure their

content would be seen on older screens (NAB, 2010). Theater employs the use of

foreground, midground, and background areas for artistic purposes and to help

audiences understand zones of action (Malloy, 2014). With the assumption that the

virtual reality operating system is being designed as a seated experience, we can

dene some guideline zones for types of content. Nuances of the zones

measurements will, however, be variable based on the specications of the device.

For this practice, I am dening zones based on the Oculus Rift Developer Kit 2 (DK2)

(Oculus, 2015c). These zones will assume the user to be in a non-rotating chair

because HMDs like the Rift and Vive have wires that a user will get wrapped up in if

they rotate.

Firstly, we can dene the eld of view if the user is looking straight forward.

The DK2s horizontal eld of view is 94.2, based on the camera settings of Oculus

Unity assets (Oculus, 2015d). Alex Chu of Samsung research gave some useful

measurements related to the perception of depth at dierent distances in VR (Chu,

2014). Your eyes strain more to focus on objects as they get closer to your face until

you are eventually cross-eyed. The distance that he gives where this starts to

become noticeable is about 0.5 - 1 meters. Oculus recently began to recommend a

 !37

minimum distance of 0.75 meters (Oculus, 2015e). Between there and 10 meters is

a strong sense of stereo depth and separation between elements. This gradually

fades o and is less noticeable up to 20 meters away. After 20 meters, the stereo

separation is essentially imperceptible, partially due to the resolution of the screen.

As objects approach innite distance, they approach a limit at which the two

screens would be identical, pixel for pixel. Innite distance is, essentially

monoscopic. I will explain the mathematical reasoning for this far depth horizon

soon, but this diagram illustrates the perception of depth as it relates to the DK2s

eld of view, based on Alex Chus presented measurements:

Field of view and depth perception in Oculus DK2

According to that same presentation, people can comfortably rotate their heads

horizontally 30 from the center and have a maximum rotation of 55. I concluded

that rotation of 30 combined with the devices eld of view gives an area in which a

user can comfortably rotate their head and see elements, 77 to the side (94/2 +
 !38

30). Beyond that, combining the maximum rotation of 55 with the eld of view

gives an additional area where people can strain to see things in their peripheral,

but persistent content would not be comfortable to see on a regular basis, 102 to

the side (94/2 + 55). After that, content behind the user could only be seen if they

physically rotate their body, likely out of curiosity about the environment. By

combining all of these measurements, we can begin to create a diagram of content

zones.

Left: Seated angles of neck rotation

Right: Combining rotation with FOV results in beginning zones for content
 !39

Because persistent content begins to be uncomfortable at distances less than 0.5

meters, that radius around the user can be deemed as an area devoid of

permanent interface elements. While it is actually something of a gradient of

discomfort which increases closer to the user, 0.5 meters is chosen as an easily

understood measurement. Ive labeled it somewhat unscientically the no-no

zone.

The no-no zone comprises the area directly around the users head at a radius of ~0.5

meters
 !40

The far boundary for content can be determined by the physical properties of the

head mounted display. The convergence angle of the eyes is a primary reason for

the perception of stereoscopic depth (Banks et al., 2012). As objects appear farther

away from the observer, the angle to which the eyes must rotate inwards for that

convergence becomes more discreet.

Perceived distance is directly correlated to convergence rotation angle

 !41

Head mounted displays have exact resolutions, so each pixel represents a xed

incremental change in rotation degree. Near the center of the display, where the

image is most clear, the incremental angle of rotation can be estimated by dividing

a single eyes horizontal eld of view by the number of horizontal pixels

encompassed within it. In the case of the DK2, the horizontal resolution of 1920 can

be divided by two to get 960 pixels per eye. Dividing the eld of view of 94.2 by the

960 pixels distributed through it yields a rotation of approximately 0.1 per pixel, on

average.

Eye rotation per pixel near the center of the display can be estimated

based on the eld of view and device resolution

 !42

The degree of convergence rotation for the eye is directly related to an objects

distance, so we can trigonometrically calculate a perceived distance based on that

rotation using the inter-pupillary distance. Using a fairly standard inter-pupillary

distance of 63mm (Dodgson, 2004), we get an asymptotic curve relating rotation

angle to perceived distance, where a rotation angle of 90 (eyes straight forward)

approaches a perceived distance of innity. It is at this innite distance that both

left and right eye displays render the exact same image pixel for pixel, being

essentially monoscopic.

Left: Equation to calculate distance as a function of IPD and convergence rotation angle

Right: Graph of asymptote approaching innite distance at 90 rotation. All distances

beyond 1 meter are perceived within two degrees of rotation.

By subtracting the average rotation angle of a single pixel, we can estimate the

maximum perceivable depth for a head mounted display. In the case of the DK2,

subtracting 0.1 yields a perceived distance of 20.34 meters for this inter-pupillary
 !43

distance. The IPD range of 52mm to 78mm yields perceived distances of 14.9 to

22.3 meters, respectively.

Left: A pixel rendered at the same coordinates for each eye is perceived at an innite distance

Right: Moving the pixel inward by one increment yields a maximum perceivable distance of
~20 meters for this device

By combining the equations of depth calculation and resolution as it relates to eld

of view, we can create a comprehensive equation to quickly estimate the maximum

perceivable distance for any given head mounted display. In this equation, the

devices resolution is treated as the full resolution across both eyes, assuming no

pixels are either shared or wasted.

Equation for estimating the maximum perceivable distance for a head mounted display
 !44

Using the same equation for other head mounted displays such as the

consumer Rift, Vive, or Gear VR, yields nearly the same 20 meter distance every

time (Further information can be found on my blog here: (Alger, 2015f)). All other

distances exist within the anti-aliasing and interpolation of a single pixel. Content

beyond this distance of approximately 20 meters loses the benet of depth

perception and can thus be deemed the far horizon at which meaningful content

should be placed. This results in the following zones diagram:

But, of course, this is only the horizontal plane. This diagram is again an

abstraction of what is actually a 3-dimensional concept. These zones have to be

three dimensional volumes. First, there is that no-no zone extending at a 0.5

meter radius from an average adult height users eyes, but as a sphere. The DK2s

screen is rectangular and the default game camera has a vertical FOV of 106.1;

however, the eld of view is functionally the same vertically as it is horizontal

because it has circular lenses. Im choosing to use the narrower of the two as the
 !45

safer measurement. Vertical head movement upwards is 20 comfortably with a 60

maximum, as per Mr Chus presentation. Downwards is less with 12 comfortably

and 40 maximum because our necks get in the way. In 3D, that zone between 0.5

and 20 meters looks like this:

Oce ergonomics with relation to computers has been around long enough

for some more clearly dened numbers to emerge. The recommended angle of

viewing for longer working periods tends to be between 15-50 downwards and at

a distance outside of 0.6 meters as is illustrated in the diagram on the following

page from Dennis Ankrums Visual Ergonomics in the Oce (Ankrum, 1999). We

can slice that section out of our content zone to get an area most comfortable for

permanent content. Text for longer reading would be most comfortably placed in

this area at a distance that matches the focal distance of the devices lenses. This

would be 1.3 meters for the DK2 and likely 2.5 meters for future devices

(Answers.oculus.com, 2014).
 !46

We now have zones for content surrounding the user. Next we will dene

zones of touchable interaction. This would only be useful for interface elements

that are meant to be touched with hand tracking or motion controllers using

collisions of some sort. They would naturally be within arms reach. If the solution

were one tracked hand and one motion controller, then the reaching area of the

tracked non-dominant hand is the one touch buttons would be used for. Reaching

the arms to their full extent on a regular basis would likely result in fatigue, we well

reduce it to a two-thirds extension. By subtracting the no-no zone again, where

persistent content would be uncomfortable, we are left with an area ideal for touch

interaction.
 !47

The touch UI zone is comfortably reachable without causing eye strain

This zone interestingly includes the users thighs. With body or surface tracking,

virtual interface elements could be placed on a users thighs in addition to the

hands or arms. Actual physical controllers could be placed there as well like buttons

or multitouch surfaces. The same can be applied to the users forearms. This is

currently possible with hand tracking, but not with motion controllers because the

position of the elbow is unknown. For tracked hands, the area just outside of the

silhouette is available for interface elements, but intersecting the silhouette causes
 !48

interference and compromises the input, based on my own experience. An example

of UI just outside the arm can be seen in Leap Motions Planetarium application

(Planetarium, 2015). For motion controllers, the area immediately around them is

available, though the radius of that is likely a design decision. An example of this is

the Tilt Brush interface for the Vive (Tiltbrush, 2015).

Left: Hand tracking can use the areas around the hand and forearm silhouette as long as
the left and right silhouettes dont intersect.

Right: Motion controllers can have UI all around them

By combining all of the zones together, I was able to create a 3D template,

useable as a guide for VR applications. The asset can be imported and dropped into

a project and interface elements can be placed in the zones. Then the guide can be

hidden or deleted for the projects export.

In order to gauge the viability of this concept, I added signs at various

distances color coded with the zones labels, paying attention to the extremities

where they would be less likely to work. I also added several at varying distances

beyond 20 meters to test the maximum distance for depth separation.

 !49

Left: Template 3D le for placing content and UI. Content and workspace zones are shown

for areas of strong 3D between 0.5 and 10 meters.

Right: Testing zones applicability with multiple labels at varying locations for each

I found that there may be some caveats. The angles of the content, peripheral, and

curiosity zones seem appropriate. However, I could tell that the nearer labels at 20

meters were closer than others. This is probably because in the testing done by

Samsung, the compared elements occluded the same area regardless of distance

(Chu, 2014). Distances were easier to see in this application because my signs were

the same size but extending into the distance with perspective, getting smaller. I

also noticed that the hand UI zone occludes a main portion of the workspace zone -

that is to say, your buttons could get in the way of the thing youre trying to look at.

Depending on the nature of the UI, having an overlaid element may actually be of

use, but more likely, designers would usually avoid this corner of the zone.

Another caveat to this method of zones is the existence of a oor. Users feel

o balance if they appear to be oating, but oor beneath them and a static
 !50

horizon line help stop this from happening (Ludwig, 2013; Cleworth, Horslen and

Carpenter, 2012; Messing and Durgin, 2005). Adding a oor that extends from the

users feet to the horizon turns the entire sphere to a dome and cuts o most of the

workspace zone. The result is the user looking directly at the virtual oor in front of

you with only small elements existing in the workspace. One possible solution is to

design the working environment to have the user on a peak, slope, or cli of some

sort, extending to the horizon. This would likely be uncomfortable for people with a

fear of heights without conditioning, though (Opdyke, Williford and North, 1995).

Having the slope angle at 50 downward to match the bottom of the main content

zone may be the best middle-ground, though I havent tested this yet.

These zones presented are specically for a non-rotating, seated design with

the Oculus Rift DK2. Changing the criteria results in dierent zones. If the user is

intended to be able to rotate all the way around in a swivel chair, then the content

zones will wrap all the way around, and the main content zone will exist as a ring. If

the device is a later model with higher resolution, the 20 meter maximum for depth
 !51

information may increase. Obviously this also changes dramatically for a room-

scale walking experience or for social multi-person implementations. I do believe

that creating a starting point for content zone guidelines for this application is

helpful, though, particularly with regards to an operating system interface.

Practical Application

Outside of the operating system concept, I have tried putting some of these

principles into application for various projects. The primary usefulness for this is

that I have come up with some design workows that others may be able to utilize.

Building assets and coding virtual reality experiences takes a lot of time. If it is done

without planning or pre-visualization, unexpected challenges are likely to arise. It is

the same case with workows for other designed digital mediums, especially those

with interface elements. The now common design to development workow of

websites is a good example of this. It is a method of prototyping. The design can

then be rened and tweaked before a lot of valuable time is spent in development.

Of course, these design workows either havent been standardized or dont exist

for virtual reality.

Some designers have talked about their processes. In Alex Chus talk, he talks

about the system of greyboxing when designing for the Gear VR headset (Chu,

2014). He places primitive polygon shapes without textures in the environment and

tests how their placement and size feel in VR. Josh Carpenter of Mozilla described

his process of designing an interface for WebVR (Carpenter, 2015). He designs the

layout as a at 360cm x 90cm Illustrator canvas, intended to be put on a cylinder

 !52

surface with a 0.5m radius around the user. He keeps the background transparent

with a blurred image of the VR environment to see legibility. While these are

possible design practices, I thought it would be appropriate to create a workow

that incorporated my personal background and understanding. I can create

animated mockups to show motion design in interfaces as well as test the

previously mentioned content zones using video production tools.

As mentioned, designs depend entirely upon input methods and distribution

platform. The example I will describe as well as case studies I include in Appendices

III and IV were for the Google Cardboard, Oculus Rift Developer Kit 2 with Leap

Motion controller, and Vive. Google Cardboard is a handheld VR device where a

user puts their smartphone into a cardboard box with lenses (Google, 2015). The

image is updated using the gyroscope and accelerometer information from phone

and the user can press one button. Google Cardboard has rotation tracking, but no

position tracking. The Rift DK2 was introduced earlier in this manuscript as a head

mounted display. Its position is tracked within the range of a camera facing the

user. The Leap Motion controller is a hand tracking input device that uses infrared

to track hands silhouettes (Leapmotion.com, 2015). By mounting the controller to

the front of an HMD, hands can be presented in the virtual world, as long as the

silhouette remains recognizable by the software. Vive is a virtual reality system

developed by Valve Software and manufactured by HTC (htcvr.com, 2015). It uses a

head mounted display and two motion controllers which are tracked in a room-

scale volume by two laser emitters in the rooms corners (Valve, 2015). This means

users can walk across the oor within the boundaries of the room.

Designing for these devices is interesting because the distribution medium is

very dierent from the creation medium. For example, when creating a painting,
 !53

you make it on a canvas and it will be seen on that canvas. When designing a

website, you create it on a 2D screen and it is experienced on a 2D screen. But with

VR, robust creative tools dont yet exist within the medium itself. We have to design

on a 2D screen what will be experienced in a volume. That is to say, we have to use

existing creative technologies to design the next creative technology.

Use Case: Animation Prototyping

I did VR interface design work for a Berlin based virtual reality company. Their

intention is to create a mobile application for Google Cardboard. This application

will allow the creation and customization of environments to be viewed and

navigated between with a VR interface. Environments are based on spheres

because the primary content is photos and videos captured with smartphones.

Features of the application are similar to YouTube, with custom content uploading,

viewing, commenting, sharing, categorization, etc. but there are other features like

multiplayer shared viewing and talking. I am not at liberty to divulge the nal

concepts, methods, or interfaces currently as the application is still in development.

I will instead describe the workow process of rapid prototyping an animated UI.

My involvement in the design process began with a weekend hackathon to

create a browsing interface. I worked with Chris Manseld to brainstorm and create

options to prototype. Google Cardboard as a delivery device has fairly limited

inputs. There is the yaw, pitch, and roll rotations, and a single button. This

interaction can be expanded slightly by considering that the user can do things with

the button like holding or double clicking. A knock on the device could also be

picked up by the accelerometers. We began with typical design processes of

 !54

brainstorming and sketching ideas. We used the greyboxing technique, creating

simple scenes quickly to test our impression of them when experienced with depth.

Initial media browsing interface prototypes

After establishing some spatial interaction guidelines using greyboxing, I moved on

to create another prototyping method for visualizing the interface as an animated

video mockup. To do this, I used After Eects because it is useful for animating

motion graphics quickly. After eects is a compositing program in which two

dimensional assets can be placed and animated in either a two or three

dimensional space (Brinkmann, 1999). It does not do vertex and polygon 3D

rendering of meshes. It can, however, link with Cinema 4D which is a full 3D

modeling and rendering application (Kopriva, 2013). To create this animation, I

began by creating 3 compositions to represent the foreground, midground, and

background; again borrowing from the popular concept in theater and cinema.

Each composition was intended to be wrapped around the user on a sphere, so I

set the width-to-height ratio to 2:1 for equirectangular mapping. Design guidelines
 !55

often reference the use of a grid to align elements (Mller-Brockmann, 1996). I

created a grid with vertical and horizontal lines for the sphere to represent every 5

around the user, 360 around and 180 vertically. For the actual visualization

composition, I intended to mimic the delivery screen, which is typically around a

16:9 ratio, like the HD standard resolution of 1920x1080. I divided the width in half

because it is separated between the two eyes, and its much quicker to simulate one

for a 2D mockup, giving a resolution of 960x1080. I embedded the three

compositions within this nal composition. I applied the CC Sphere eect to all of

them, changing the settings to render only the inside and removing lighting eects.

The midground composition had a radius such that the angle of the sphere that

was visible matched the HMDs eld of view. This can be seen with the grid in place,

since every line represents 5. The foreground had a slightly smaller radius and

the background slightly larger. When all three spheres are rotated together, this

oset gives a sense of parallax in the 2D video which helps viewers understand the

depth relationships better. I also added a vignette to simulate the fallo in eld of

view toward the edges.

 !56

Left: Equirectangular grid with each line representing 5 rotation

Right: Spheres with dierent radius represent foreground, midground, and background

Composition settings with sphere placement

 !57

Something I realized was that the angles of the content zones can be directly

translated to 2D areas on the grid. These grid and area guides can be turned on and

o for quick reference by showing and hiding the layers in their compositions.

Zone angles applied to an equirectangular grid

Using this template, I was able to simulate a hypothetical user navigating through

hypothetical content with interface elements and animations. Individual elements

like colors, fonts, and animation speeds remain changeable to test dierent styles.

The interface design is based on a locked reticle following the users rotation.

The user can point their head at dierent options and a clear hover state is shown

for selectable elements, similar to the Oculus Home application for the Gear VR

HMD (Oculus Home, 2015). The user can press the button once to bring up a

content browsing menu, or hold the button down to bring up another radial

navigation interface. Letting the button go then chooses the option the users reticle

is hovered on. There are several other aspects of the interface that the creators

have asked me not to share yet because its not out yet, so theyve been omitted.
 !58

To communicate nuances of the interface not understandable from the video

by itself, I made another composition with a section for notes to the side. It has a

representation of the Google Cardboards button. This allows me to show when the

button is being pressed/held and leave notes for the developers about what is

happening. These are some example screenshots of the still in-process mockup:

Still frames from animated pre-visualization

Animated pre-visualization with button indication and notes for developers

 !59

Use Case: Zone and Environment Prototyping

The Ravensbourne postgraduate students were expected to showcase their in-

progress works for an event in June 2015 called Utopia. Because most people

havent experienced virtual reality, I decided to showcase some of the technologys

simple but impressive capabilities while testing some environment design

workows. I chose to create a simple button pressing interface using hand tracking.

The interface elements positions would be based on testing the content zones

template previously described. Users could toggle environments, 3D models, and a

text document to look at by compressing buttons in the touch interface zone in

front of them. Video of the interface can be found at http://bit.ly/1KNctRT.

Left: Screenshots from within application

Right: Demonstration environment at Utopia exhibition

In use, I noticed some caveats of the tracked hand button interface. For one,

the users head has to be pointed in the direction of the button they are pressing.

The hand tracking device is mounted to the front of the HMD, so it cant track
 !60

something outside of its own infrared cameras eld of view. My demonstration had

an option to turn on a chair mesh with a button. Some people would look behind

themselves at the chairs location and use their proprioception to press in the air

where they remembered the button to be, in the opposite direction. The action

wouldnt take place and they would appear disappointed or confused. I also saw in

my own testing and with other people that it is a bad idea to have buttons of this

style aligned vertically. In order to press the button, many users would swing their

arm downward through it. Theres no normal force to stop their hand or give

feedback, so their arm would continue in an arc through the motion past the

button. If there was another button beneath, it would often be triggered, too. Again,

the extra result would appear to confuse the user.

One of the items the users could toggle on and o was a scaled down

photogrammetric scan of the Ravensbourne building (This was as asset from a

hackathon project called Museum of Lies for which the summary can be seen

here: (Alger, 2015c)). I purposely placed this within the no-no zone near the users

face. Not unexpectedly, its appearance was startling to many. Some attempted to

lean back away from it immediately. What was unexpected was how much people

liked to put their head inside of it. Similar to the world in miniature concept, this

appeared to be a miniature building model that they could go into and enjoyed.

One part of the process while making this experience was my use of

skyspheres and skydomes. The environments that users could switch between in

this experience were primarily equirectangular photos. The simplest way to display

these in virtual reality is to map them to either a cube skybox or a sphere with

normals facing the user. Single equirectangular panoramas are monoscopic by

nature and typically captured by taking photos in every direction from around a
 !61

single point. Besides the lack of stereoscopic depth, these images suer from

distortions in perception of scale when viewed in virtual reality. Objects nearer to

the camera appear too large. This is because they take up a certain angle in the

eld of view which is also typically associated with a certain convergence related to

the objects distance. When the eyes are viewing a monoscopic equirectangular

photo on a sphere, they are converging at the same distance for every point in the

photo. If the skybox is set to render at an innite distance, objects in the scene will

appear to be the correct scale as their actual distance approaches innity. Objects

that are closer to the camera will look too big as they take up a wider eld of view

but also appear to be at innite distance. This is particularly obvious often when

looking down at the ground below the photos capture position. It will appear to be

far away with large features.

Size distortion in monoscopic photo spheres. Objects appear larger as the eld of view angle

is preserved but their distance is increased to innity. Decreasing the spheres radius results in

the opposite eect - far objects appear too small.

 !62

One way to mitigate this is to move the geometry of the sphere closer to the user

for specic locations in the photo. This brings the convergence back to the correct

depth and eliminates the disproportionate scale illusion. I tested this by creating a

dome mesh for some of the photos to be mapped to. In order for the photo to

appear correctly, the textures UVs must remain at the same viewing angle from the

user while moving closer to the eyes. One way to do this is to select the vertices to

be transformed and scale them downward toward the point of origin at the center

of the sphere. I tried this with a at ground only, assuming the cameras distance to

the ground during capture to be 1.6 meters.

By moving points closer to the user, but keeping their angle, we can both reintroduce

convergence and reduce the size distortion. This diagram shows how it can be applied to

only the ground extending to the horizon. This doesnt account for other objects besides

ground and sky, though.

 !63

The main problem with this that I encountered in testing was UV mapping artifacts.

As the mesh gets pulled, the squares become trapezoids and the texture, mapped

to triangles, is distorted across the hypotenuse. This can be mitigated by modifying

the shader (Northway, 2013) or more clearly dening the texture space coordinates

(Everitt, 1997). One way to try to work around it is to increase the polygons. This

doesnt actually solve the problem, but makes it less noticeable.

Top: Low and high poly dome mesh

Bottom: With resulting texture distortion from spherical UV mapping

View from inside the dome comparing low and high poly UV distortion. While increasing the

geometry helps, it doesnt solve the problem and a custom shader to eliminate the artifact is

a better solution.
 !64

For the exhibition, I had environment photos mapped to both spheres and

domes. People who were afraid of heights were often uncomfortable with the

spheres because the ground appeared far away. Some would pull their feet up or

brace themselves when saying they didnt like it. They would often quickly switch

the environment back to one of the dome options. The dome solution was clearly

imperfect for many types of photographic content, however. For example, a beach

scene has palm trees extending from the sand. While the sand and ocean appear to

correctly map into the distance, the palm trees do not. They extend across the

ground to the horizon where they then move up the sky, again looking too large.

Ideally, every pixel of the photo would be mapped to its correct depth. The

current polygon sphere model is not ideal for this because one vertex per pixel

would be millions of faces and far too many for a VR experience to process fast

enough. Software could be written to display each pixel at a stereoscopic

convergence depth according to a depth map channel on the image, though.

Capturing the depth map would be a matter of 3D scanning techniques like infrared

or laser distance measurement (Bernardini and Rushmeier, 2002), or parallax

analysis of multiple photos using photogrammetry (Walford, 2007; Matthews, 2008).

 !65

Left: Equirectangular photo sphere

Right: Depth map. White is 0 distance, black is innite distance. Its a bit dicult to see, but
the nearest lamp post is the closest distinguishable feature.

The problem with this solution is that it lacks the occluded stereo parallax

information where one eye would be seeing things that the other cannot because

of its positional distance. So, where a pixels position was shifted to accommodate

for convergence, its original position would be left blank. There would appear to be

a shadow tear between every near object and its backdrop. Repeating the further

pixels for one eye would be possible, but not ideal as its not the true information

from the environment. Capture methods incorporating lateral movement for stereo

information such as photogrammetry and light elds remain ideal (Wilburn et al.,

2005).
 !66

Putting these Concepts Together for a VR OS Design

This manuscript has presented several guidelines and workows related to virtual

reality experience and interface design. The larger goal was to apply these methods

to the use of an operating system environment. This concept can easily become

one of hypothetical futurism based on conceptual software and hardware

technologies. It would be possible, for example, to implement several Microsoft

Kinects in a room to build an augmented reality system with a consumer version

Oculus Rift and Touch controller mounted with a Leap Motion version 2 hand

tracker (Kreylos, 2014; Bedikian, 2015). It would also be possible for the software

content to include 3D mesh video (Collette et al., 2015) or a multi-user shared

working environment (Underkoer, 2010). However, to avoid assumptions, I will

only present a concept for an operating system for a simple hardware setup using

only existing content.

The example I will present is a distributable operating system design for the

Vive system in an oce workplace. To recap, Vive is a head mounted display with

two motion controllers tracked in a room-scale space of up to approximately

4.5x4.5 meters (Htcvr.com, 2015). This concept would work exactly the same with

the Oculus Rift and Touch controllers because it is also capable of room-scale

tracking (Lang, 2015); however, I havent yet had the opportunity to try the

consumer Oculus headset or controllers myself, so I am opting to use the system

with which I have more experience for this thought experiment. I will also present

this as a hypothetical environment where the user has one motion controller in

their dominant right hand and a tracked left hand as mentioned before. This way, I

can show interaction methods for both at the same time.

 !67

The hypothetical user has a business that provides 3D scanning services for

museums. She has an oce space with a chair, in which the Vives lighthouse

tracking system has been set up. Her current task is researching a 3D scanner to

see if her business should invest in it. Wearing the headset, she watches a 2D video

explaining the project. She is in the VR equivalent of Full screen mode, with just

the video and controls. The video itself is taking up 54 of her horizontal viewing

angle, the same as a standard movie theater (Imax, 2013), and it is at a distance of

20 meters, the distance described earlier at which stereoscopy is negligible.

 !68

She scrubs ahead in the video by intersecting her hand with a cylinder representing

the videos timeline. Removing her hand from the cylinder resumes the video

playback from the exit point.

She presses the pause button, which is near her left hand and styled as described

earlier. She exits out of full screen by pressing a button oating next to her

controller.
 !69

The video returns to a smaller size amidst her customized environment and

applications. She now appears to have a mountain valley in front of her. To her

sides are some of her favorite sculptures and paintings in and on a partial building.

She points the motion controllers cursor at the scanners user manual document,

represented as a preview thumbnail of its content. When hovered, its name appears

and the thumbnail edges glow. She presses the thumbpad button to open it and

its position and scale transform to the ideal content area for reading: 1.3 meters

away between 15 and 50 downward.

 !70

Using the thumbpad on the controller, she scrolls through the document. She

selects a point with the trigger, scrolls ahead, and makes a gesture, pointing her left

thumb to the side while selecting the second point. This is akin to holding the shift

key while clicking on a computer. When the gesture is recognized, the cursor

pointer changes shape. The gesture modies the cursors action, highlighting the

text between the rst and second points.

 !71

She brings up the documents menu by squeezing the controller. The radial menu

protrudes from the space around it. She selects a sub-menu by moving the

controller into the correct options and releasing the squeeze, choosing the option

to speed read.

The words ash rapidly one at a time, and she adjusts the speed using the haptic

thumbpad on the motion controller. While this is going on, a radial light begins to

pulse in her peripheral vision. It is an alarm she had set earlier as a calendar alert to

remind her to meet with a prospective client in an hour. She stops the speed

reading function by pressing a button with a stop icon similar to the video pause

earlier. Looking in the direction of the alarms beacon reveals its title and time, as

well as options for dismiss (left), snooze (right), or details (down). Pointing the

raycast cursor at it, she pulls the trigger and swipes it to the left, dismissing it.
 !72

She looks at her left wrist, which has the current time. Her forearm shows the

currently allocated resources per application, and the back of her hand shows her

wi connectivity, bluetooth status, and peripheral device battery charge levels.

Shell need to grab lunch before the meeting. She turns to a personal assistant

application, similar to the concepts of Apples Siri or Microsofts Cortana. Because

articial intelligence assistance is often incorrect, the virtual representation, while

still minimalistic, has elements inspired by the proportions of human children. The
 !73

goal of this is to reduce user frustration for unexpected responses. Upon directing

her gaze at it, two dash icons widen to simulate eye contact showing that the

application is now listening to the microphone. She says Wheres a good place to

eat near the Museum of Art? and the words she speaks appear above the assistant

as they are recognized.

The assistant replies Heres restaurants with more than ve stars near Museum of

Art. and a list appears. The user says Show them on a map. Again, the words

appear above the animated assistant icon. A 3-dimensional map of the museums

neighborhood appears with location indicators for points of interest. The higher

rated restaurants labels are displayed at a higher altitude. She points the raycast

cursor at it and pulls the trigger, holding the map as she repositions it. While she

holds it, moving her left hand further left scales it larger, and towards and away

from her rotates it. She selects a restaurant by pointing her cursor at it and

clicking. It shows further information like descriptions, reviews, and photos.

 !74

Each time she interacts with content, the button interface in front of her changes to

the relevant application. She presses a button with her hand to get directions. The

map displays a path from her oce to the restaurant. She presses another button

to send the directions to her smartphone. Removing the head mounted display, she

now has the mobile directions to use for her drive.

 !75

What I have just described is one potential solution for this series of tasks.

There is a huge number of changeable variables involved, and the proposed

methods I depicted may not be, and indeed very likely arent the most ideal. Even

changing individual aspects of the narrative alters the probable interface elements.

For example, a multi-touch surface tracked in the environment instead of hand

tracking would mean all touched elements being constrained to that surface. This

narrative thought experiment simply serves as a starting point from which user

experience testing and iteration can begin. There is no doubt that unexpected

caveats and best practices would emerge from repeated meaningful feedback.

Prototype and Evaluation

As a personal starting point for the evaluation and progressive iteration of these

concepts, I have created my own beginning prototypes with the equipment

available to me. The prototype application which is intended to accompany this

paper is a modied extension of the interaction application used for the Utopia

exhibition. Its purpose is to illustrate the concepts described in this document

within the medium itself. This is available for download at http://

aperturesciencellc.com/vr/application.zip.

Whats good about this is that several of the concepts I describe are

volumetric and three dimensional by nature. Viewing them in virtual reality as 3D

objects makes them more understandable. Its quite obviously a far shout from the

robust narrative just described, though. Hand tracking using the current Leap

Motion controller is not accurate enough for raycasting cursors without frustration,

for example. As a user experiences the project, they will undoubtedly understand
 !76

the concepts with greater depth by feeling the interactions that they nd more and

less comfortable.

With a mass of resources and time including teams of developers, I have no

doubt that a full virtual reality operating system could be created using existing

technology. This prototype and the concepts described and illustrated are merely a

small stone near the beginning of a path. Its just the beginning, but it is a solid

start, I believe.

Whats Next?

The most obvious next step is testing and adjustment for each element as it is

implemented. The goal would be to build applications that I would want to use.

Where and how to do that are unknowns. It does seem to me that positions like

graphic designer, web designer, or game designer will be joined by VR Designer.

Elements of user interface are only small parts of the larger and more important

user experience. The design process for head mounted displays is particularly

important because of the immense power the designer wields to create

experiences that are magical, that reshape a users view of the world, or that can

literally make a user ill.

Conclusion

Throughout this manuscript I have explained why virtual reality harbors a

promising opportunity for human productivity. I explained what types of content it

is best suited for and the most likely ideal locations for that content, both

ergonomically and understandably. I have also detailed workow methods through

 !77

which we can design interfaces to interact with and modify that content using

existing technology, expounded upon signicantly in the appendices of this

manuscript.

On a personal note, I do hope that this work will prove useful to others. I

explained in the introduction that I have a desire to work on something with a more

lasting application and value. I also explained that virtual reality is a rapidly evolving

topic, changing frequently. It is very possible that much of these concepts and

workows will be rendered obsolete within even a years time. But it does seem that

concepts like ergonomic considerations for zones of content will be necessary, even

as the borders of those zones change depending on the hardware specications

and use cases.

For a year, I wasnt sure if evolving personally from a focus in motion design

to virtual reality would be a wise decision. After spending hundreds of hours

reading, experiencing, creating, and talking to others about and in virtual reality, I

still wasnt completely sold myself on a career shift. It was at a weekend game jam

using the Vive system that my mind was made up. This was the rst time I used a

consumer resolution display, room-scale tracking, and motion controllers. Walking

around the digital objects that we had just created I couldnt help but gain a sense

of boundless creativity. Experiencing the polished demos of professional teams

really hit home the concept of the most powerful and versatile storytelling medium

ever created.

Whats particularly interesting about this section of time is that digital

volumetric interfaces do not yet have established conventions. Where writing, lm,

television, radio, theater, graphic design, etc. have expected elements, head

mounted displays remain conceptually open-ended. As a community, we are

 !78

discovering the mediums unexpected strengths and weaknesses. In coming years

the consumer market will run virtual reality through the rening crucible of ethics,

etiquette, and social acceptance. Rating systems, legislation, and standards

committees will form to ensure the mitigation of social risks. We will soon see the

rst VR related death, claims of head mounted displays causing cancer, blaming the

medium for causing violence, social detachment, psychologically or physically

melting the brains of its users. Alongside this will be the immersive storytelling,

compelling experiences, and discussions of human bodily transcendence by way of

technological augmentation. And, of course, there is the prospect of heightened

productivity and happiness which I so editorially focused on in context of

opportunity for the workplace.

It is VRs medium dening process.

Its going to be interesting, and Im going to be there.

 !79

References

Aberdeen Group (2013). Co-Browsing in Customer Service: The Path to Just-in-Time

Customer Engagement. [online] Available at: http://www.oracle.com/us/products/
applications/aberdeen-cobrowse-2348021.pdf

Abrash, M. (2013). Why Virtual Reality Is Hard (and where it might be going). Game
Developers Conference 2013 [presentation] Available at: http://
media.steampowered.com/apps/abrashblog/MAbrash%20GDC2013.pdf

Abrash, M. (2014). Oculus Connect Keynote. [presentation] Available at: https://

www.youtube.com/watch?v=knQSRTApNcs

Adelson, E. and Bergen, J. (1991). The Plenoptic Function and the Elements of Early
vision. Computational Models of Visual Processing. Cambridge, MA: MIT Press.
pp. 3-20.

Agarwal, C. and Thakur, N. (2014). The Evolution and Future Scope of Augmented
Reality. IJCSI International Journal of Computer Science Issues. Volume 11, Issue
6, No 1. pp. 59.

Agisoft (2015). PhotoScan. v 1.1.6. [software] Available at: http://

www.agisoft.com/

Alexa (2015). The Top 500 Sites on the Web. [website] Available at: http://
www.alexa.com/topsites

Alger, M. (2015a). Designing VR for Humans. C-Base Raumstation, Berlin,

Germany. [presentation] Available at: http://www.meetup.com/Berlin-Virtual-
Reality-Meetup/events/220294981/

Alger, M. (2015b). Take Care of Your Humans (with Virtual Reality!). Hemnes, Warsaw,
Poland. [presentation] Available at: http://www.meetup.com/GoMobile-with-
Design/events/220210315
 !80

Alger, M. (2015c). Museum of Lies. [video] Available at: https://vimeo.com/

130581302

Alger, M. (2015d). Insider Notes from the HTC Vive VR Jam at Londons Playhubs.
[online] Road to VR. Available at: http://www.roadtovr.com/insider-notes-from-
the-htc-vive-vr-jam-at-londons-playhubs/

Alger, M. (2015e). VR Interface Design Manifesto. 1:15 - 2:30 [presentation]

Available at: https://vimeo.com/116101132

Alger, M. (2015f). HMD resolution and maximum depth perception. [blog] Available
at: http://mikealgermovingimage.tumblr.com/post/127113260256/hmd-
resolution-and-maximum-depth-perception

AltspaceVR, Inc. (2015). AltspaceVR. [software] Available at: http://altvr.com/

Ankrum, D. R. (1999). Visual Ergonomics in the Oce - Guidelines. Occupational

Health & Safety. pp. 68.

Answers.oculus.com (2014). How do I nd a comfortable distance "zone of comfort"

in VR?. [online] Available at: https://answers.oculus.com/questions/41/how-do-i-
nd-a-comfortable-distance-zone-of-comfo.html

Autodesk (2015a). 123D Catch. [software] Available at: http://www.

123dapp.com/catch

Autodesk (2015b). Maya. [software] Available at: http://www.autodesk.com/

products/maya/overview

Ball, R., North, C. (2005). Eects of tiled high-resolution display on basic visualization
and navigation tasks. CHI '05 Extended Abstracts on Human Factors in
Computing Systems. pp. 1196-1199.

Banks, M., Read, J., Allison, R. and Watt, S. (2012). Stereoscopy and the Human
Visual System. SMPTE Motion Imaging Journal, 121(4), pp.24-43.
 !81

Barras, C. (2014). How Virtual Reality Overcame its 'Puke Problem'. BBC. [online]
Available at: http://www.bbc.com/future/story/20140327-virtual-realitys-puke-
problem

Bedikian, R. (2015). AR Screen hackathon project. [video] Available at: https://

www.youtube.com/watch?v=zxM4vN_4jJY

Bernardini, F. and Rushmeier, H. (2002) The 3D Model Acquisition Pipeline.

Computer Graphics Forum. Vol. 21, No. 2. pp. 149-172.

Brinkmann, R. (1999). The art and science of digital compositing. San Diego: Morgan
Kaufmann.

Brown, C. M. (1988). Human-computer interface design guidelines. Norwood, NJ:

Ablex Publishing.

Bye, K. (2015). VRJam Submissions - Unocial List by Voices of VR Podcast. [online]

Available at: https://docs.google.com/spreadsheets/d/
1mPtT7ZBeK4VANtKwHIFCJY_N3BSYeptC3yy3nlnZCoc/edit#gid=1077132853

Cakmakci, O., and Rolland, J. (2006). Head-Worn Displays: A Review. Journal of

Display Technology. Vol. 2, No. 3. pp. 199-216.

Carpenter, J. (2015). UI/UX design for WebVR with Josh Carpenter. SFHTML5.
[presentation]

Chu, A. (2014). VR Design: Transitioning from a 2D to a 3D Design Paradigm.

Samsung Developer Conference. [presentation] Available at: http://alexchu.net/
Presentation-VR-Design-Transitioning-from-a-2D-to-a-3D-Design-Paradigm

Cleworth, T., Horslen, B. and Carpenter, M. (2012). Inuence of Real and Virtual
Heights on Standing Balance. Gait & Posture. 36(2), pp.172176.

Colgan, A. (2015). Designing VR Tools: The Good, the Bad, and the Ugly. Leap Motion.
[online] Available at: http://blog.leapmotion.com/designingvrtoolsgoodbadugly/
 !82

Collet, A., Chuang M., Sweeney P., Gillett D., Evseev D., Calabrese D., Hoppe H.,
Kirk A., Sullivan. S. (2015). High-quality streamable free-viewpoint video. ACM
Transactions on Graphics. 34(4).

Cruz-Neira, C., Sandin, D. and DeFanti, T. (1993). Surround-screen projection-based

virtual reality. Proceedings of the 20th annual conference on Computer graphics
and interactive techniques - SIGGRAPH '93.

Cruz-Neira, C., Sandin, D., DeFanti, T., Kenyon, R. and Hart, J. (1992). The CAVE:
audio visual experience automatic virtual environment. Communications of the ACM,
35(6), pp.64-72.

Darwin, C. (1859). On the Origin of Species. London: John Murray.

*Note: I hope citing Darwin is more humorous than necessary.

Dodgson, N. (2004). Variation and Extrema of Human Interpupillary Distance.

Proceedings of SPIE: Stereoscopic Displays and Virtual Reality Systems XI. Vol.
5291. pp. 36-46.

Everitt, C. (1997). Getting to know the Q texture coordinate. [online] Available at:
http://www.xyzw.us/~cass/qcoord/

Faliszek, C. (2015). Vive Game Jam. Playhubs, London. [in person] 11 July 2015.

Fisher, C. (2010). Happiness at Work. International Journal of Management

Reviews, 12(4), pp.384-412.

Forums.oculus.com, (2015). 1080p is not enough. [online] Available at: https://

forums.oculus.com/viewtopic.php?t=3552

G. Wetzstein, D. Lanman, M. Hirsch, R. Raskar. (2012). Tensor Displays:

Compressive Light Field Synthesis using Multilayer Displays with Directional Backlighting.
Proc. of SIGGRAPH 2012. ACM Transactions on Graphics 31, 4. Available at:
http://web.media.mit.edu/~gordonw/TensorDisplays/TensorDisplays.pdf
 !83

Gallagher, W. (2007). The Power of Place. New York: Harper Perennial.

Gaylor, G. and Joudrey, J. (2015). VR Chat. v 0.8.7. [software] Available at: http://
www.vrchat.net

Godin, G. (2015). Virtual Desktop. [software] Available at: http://

www.vrdesktop.net

Google (2015) Google Cardboard. [website] Available at: https://www.google.com/

get/cardboard/

Gunter, (2015). Gunters Universe. [online] Available at: http://

guntersuniverse.com

Hinckley, K. (2002). Input Technologies and Techniques. The human-computer

interaction handbook: fundamentals, evolving technologies and emerging
applications. pp. 151-168.

Homan, D., Girshick, A., Akeley, K. and Banks, M. (2008). Vergence-

accommodation conicts hinder visual performance and cause visual fatigue. Journal
of Vision, 8(3), pp.33-33.

htcvr.com (2015). HTC Vive. [website] Available at: http://www.htcvr.com/

Imax (2013). IMAX 101: Theatre Geometry. [online] Available at: http://
www.imax.com/community/blog/imax-101-theatre-geometry-video/

Innite-Realities (2015). [website] Available at: http://ir-ltd.net/

Intuitive. (2015). In: Merriam-Webster [online] Available at: http://www.merriam-

webster.com/dictionary/intuitive

": having the ability to know or understand things without any proof or
evidence"
 !84

Kang, Y. and Stasko, J. (2008). Lightweight task/application performance using single

versus multiple monitors: a comparative study. Proceedings of Graphics Interface
2008. pp. 17-24.

Kennedy, R. and Frank, L. (1985). A Review of Motion Sickness with Special Reference
to Simulator Sickness. Westlake Village, CA: Canyon Research Group Inc.

Kopriva, T. (2013). Details of Cinema 4D Integration with After Eects CC (12.0).

Adobe. [online] Available at: http://blogs.adobe.com/aftereects/2013/04/
details-of-cinema-4d-integration-with-after-eects.html

Koved, L. and Selker, T. (1999). Room With a View (RWAV): A metaphor for interactive
computing. Yorktown Heights, N.Y.: IBM T.J. Watson Research Center.

Kreuger, W. and Foechlich, B. (1994). The Responsive Workbench. IEEE Computer

Graphics and Application. Vol. 14, No. 3. pp. 12-15.

Kreuger, W. Bohn, C., Froehlich, B., Schueth, H., Strauss, W., and Wesche, G.
(1995). The Responsive Workbench: A Virtual Work Environment. IEEE Computer. Vol.
28, No. 7. pp. 42-48.

Kreylos, O. (2014). 3D Video Capture with Three Kinects. Doc-Ok.org [online]

Available at: http://doc-ok.org/?p=965

Lang, B. (2015). Oculus Demonstrates Their Own Room-scale Tracking Capability at E3

2015. Road to VR. [online] Available at: http://www.roadtovr.com/oculus-rift-
room-scale-tracking-volume-e3-2015-oculus-touch/

Lanman, D. and Luebke, D. (2013). Near-Eye Light Field Displays. NVIDIA Research.
Available at: https://research.nvidia.com/sites/default/les/publications/NVIDIA-
NELD_0.pdf

Leapmotion.com (2015). Leap Motion for Virtual Reality. [website] Available at:
https://www.leapmotion.com/product/vr
 !85

Lee, H., Whiting, S. (2015). ConVRge. [software] Available at: http://

www.convrge.co/

Lessig, L. (1999). Code and Other Laws of Cyberspace. New York: Basic Books.

Ludwig, J. (2013). Lessons Learned Porting Team Fortress 2 to Virtual Reality. Game
Developers Conference. [presentation]

MakeHuman (2015). v 1.0.2. [software] Available at: http://

www.makehuman.org/

Malloy, K. (2014). The Art of Theatrical Design: Elements of Visual Composition,

Methods, and Practice. Hoboken; Taylor and Francis. pp. 53-55.

Matthews, N. (2008). Aerial and Close-Range Photogrammetric Technology: Providing

Resource Documentation, Interpretation, and Preservation. Technical Note 428. U.S.
Department of the Interior, Bureau of Land Management, National Operations
Center. Denver, Colorado. 42pp.

McGinity, M. (2014). Presence, immersion and the panorama - a theoretical, technical

and artistic inquiry into the nature of presence and immersion in virtual reality. Ph.D.
College of Fine Arts, UNSW. pp. 61-69.

Medich, J. (2015). What Would a Truly 3D Operating System Look Like?. Leap Motion.
[online] Available at: http://blog.leapmotion.com/truly3doperatingsystemlook-
like/

MeshLab (2015). v 1.1.3. [software] Available at: http://

meshlab.sourceforge.net/

Messing, R. and Durgin, F. (2005). Distance Perception and the Visual Horizon in
Head Mounted Displays. TAP, 2(3), pp.234-250.

Moore, G. (1965). Cramming More Components onto Integrated Circuits. Electronics.

pp. 114-117.
 !86

Mller-Brockmann, J. (1996). Grid Systems in Graphic Design. Sulgen: Verlag Niggli.

NAB (2010). Television Safe Areas Redened. TV TechCheck. [online] Available at:
http://www.nab.org/xert/scitech/pdfs/tv031510.pdf

NEC, (2010). Monitor Size and Aspect Ratio Productivity Research. [presentation]

Nelson, N. (2013). Is Virtual Reality Gaming Destined For A Comeback?. All Tech
Considered. NPR. Available at: http://www.npr.org/sections/alltechconsidered/
2013/06/12/191067676/is-virtual-reality-gaming-destined-for-a-comeback

Northway, C. (2013). Drawing a Trapezoid with Stage3d. Northway Games. [online]

Available at: http://northwaygames.com/drawing-a-trapezoid-with-stage3d/

Oculus (2015a). Oculus Best Practices. [online] pp. 15-16. Available at: http://
static.oculus.com/documentation/pdfs/intro-vr/latest/bp.pdf

Oculus (2015b) Oculus Rift. [website] Available at: https://www.oculus.com/en-

us/rift/

Oculus (2015c). Oculus Rift Development Kit 2. [website] Available at: https://
www.oculus.com/en-us/dk2/

Oculus (2015d). Oculus Utilities for Unity 5. [website] Available at: https://
developer.oculus.com/downloads/game-engines/0.1.0-beta/
Oculus_Utilities_for_Unity_5/

Oculus (2015e). Oculus Best Practices. [online] Available at: https://

developer.oculus.com/documentation/intro-vr/latest/concepts/bp_app_ui/

Oculus Home (2015). Oculus VR. [software]

Opdyke, D., Williford, J. and North, M. (1995). Eectiveness of computergenerated

(virtual reality) graded exposure in the treatment of acrophobia. A JP. 152(4), pp.626-
628.
 !87

Plafke, J. (2013) Leap Motion review: Is it time to replace the mouse?. Extreme Tech.
[online] Available at: http://www.extremetech.com/extreme/161813-leap-
motion-review/3

Planetarium (2015). Leap Motion. [software] Available at: https://

developer.leapmotion.com/gallery/planetarium

Plunkett, L. (2015). Valve Thinks It's Cracked Typing With A Controller. Kotaku.
[online] Available at: http://kotaku.com/valve-thinks-its-cracked-typing-with-a-
controller-1709175825

Poeter, D. (2015). How Moore's Law Changed History (and Your Smartphone). PC
Mag. [online] Available at: http://uk.pcmag.com/cpus-components-products/
41195/news/how-moores-law-changed-history-and-your-smartphone

Preece, J., Rogers, Y., Sharp, H., Benyon, D., Holland, S., Carey, T. (1994). Human-
Computer Interaction, Addison Wesley.

Reinhard, E., Ward, G., Pattanaik, S., Debevec, P., Heidrich, W., Myszkowski, K.
(2010). High Dynamic Range Imaging. Burlington, MA: Morgan Kaufmann/Elsevier.
pp. 239.

Romo, E. (2015). Altspace VR. [interview] 5 February.

Samsung (2015) Samsung Gear VR. [website] Available at: http://

www.samsung.com/global/microsite/gearvr/

ScanLab (2015). [website] Available at: http://www.scanlab.ca/about/

Seetzen, H., Heidrich, W., Stuerzlinger, W., Ward, G., Whitehead, L., Trentacoste,
M., Ghosh, A. and Vorozcovs, A. (2004). High Dynamic Range Display Systems. TOG,
23(3), p.760.
 !88

Sense. (2015). In: Merriam-Webster [online] Available at: http://www.merriam-

webster.com/dictionary/sense

": one of the ve natural powers (touch, taste, smell, sight, and hearing)
through which you receive information about the world around you"

Shamim, S., Islam, M. and Hossain, M. (2012). A study on unplanned interruptions

in software development. [online] https://novicearshad.wordpress.com/
2012/01/24/unplannedinterruptionsinsoftwaredevelopment/

Sherman, W. and Craig, A. (2003). Understanding Virtual Reality. San Francisco, CA:
Morgan Kaufmann, pp.310-325.

Shibata, T. (2002). Head Mounted Display. Displays. 23. pp. 57-64.

Sleight, L. (2014). VRO (VR Objects). AnyAll. [website] Available at: http://
www.anyall.net/#!vro/c1igr

Smallman, H., John, M., Oonk, H. and Cowen, M. (2000). When Beauty is Only Skin
Deep: 3-D Realistic Icons are Harder to Identify than Conventional 2-D Military Symbols.
Proceedings of the Human Factors and Ergonomics Society Annual Meeting,
44(21), pp.3-480-3-483.

Space X (2013). The Future of Design. [video] Available Online: https://

www.youtube.com/watch?v=xNqs_S-zEBY

Stoakley, R., Conway, M., Pausch, R. (1995). Virtual Reality on a WIM: Interactive
Worlds in Miniature. University of Virginia.

Sundstrom, M. (2015). Immersive Design. Backchannel. [online] Available at

https://medium.com/backchannel/immersive-design-76499204d5f6

Sutherland, I. E. (1968). A head-mounted three dimensional display. In Proceedings

of the December 9-11, 1968, fall joint computer conference, part I, pp. 757-764.
ACM.
 !89

Swype Inc. (2015). Swype. [software] Available at: http://www.swypeinc.com/

Thacker, C., McCreight, E., Lampson, B., Sproull, R. and Boggs, D. (1979). Alto: A
personal computer. Computer Structures: Principles and Examples, second
edition, pp.549-572.

Tiltbrush (2015). [software]. Available at: http://www.tiltbrush.com/

Tom23, (2015). Tom23DigitalLife. [online] Available at: http://worlds.vrchat.net

Turner, A. (2014). The History of Flat Design: How eciency and minimalism turned the
digital world at. The Next Web. [online] Available at: http://thenextweb.com/dd/
2014/03/19/history-at-design-eciency-minimalism-made-digital-world-at/

Underkoer, J. (2010). Pointing to the Future of UI. TED. [presentation] Available

online: http://www.ted.com/talks/
john_underkoer_drive_3d_data_with_a_gesture?language=en

Valve (2015) Steam VR. [website] Available at: http://store.steampowered.com/

universe/vr

VRWiki (2015) Input Controllers. [website] Available at: http://

vrwiki.wikispaces.com/Input+controllers

Walford, A. (2007). What is Photogrammetry?. [online] Available at: http://

www.photogrammetry.com/

Ward, J. (2007). Acquired Auditory-Tactile Synesthesia. Annals of Neurology. Vol. 62,

No. 5. pp. 429-430.

Wilburn, B., Joshi, N., Vaish, V., Talvala, E., Antunez, E., Barth, A., Adams, A.,
Horowitz, M., Levoy, M. (2005). High Performance Imaging Using Large Camera
Arrays. ACM Transactions on Graphics. Vol. 24, No. 3. pp. 765-776

Yao, R. (2014). The Human Visual System and the Rift. Oculus Connect.
[presentation]
 !90

Appendix: Avatars

Im going to change gears to talk about avatars. An avatar in virtual reality is a

representation of the user (Lessig, 1999). This is usually for the purpose of

representing the user to other participants in a multiplayer setting, but it can also

just be the body or reection the user sees as belonging to themselves. An avatar

isnt necessary and many experiences choose to forego the representation of the

users body in order to avoid another perceptive conict called proprioceptive

disparity (Romo, 2015). This is the mismatch of signals to the brain between the

internal sense of knowing where the body parts are not matching up with where

they look like they are. Most systems dont track the whole users body, so an avatar

wont be moving in unison with the user. Multiplayer experiences still need to

represent the user, though, so dierent applications address this to varying

degrees. VR Chat allows any kind of avatar (Gaylor and Joudrey, 2015). ConVRge

allows a single object that rotates with the users head motion (Lee and Whiting,

2015). AltSpace only allows their own avatars which will only represent the aspects

that are known (AltspaceVR Inc., 2015). For example, if the user has eye tracking,

eye representations are shown. If they have hand tracking, hands are shown,

otherwise they are not. By default, the avatars are all similar looking torsos with

heads inspired by the Eve character from Wall-E (Romo, 2015).

Within the concept of an operating system, multiple users is an interesting

prospect. Multiple people can be in the same space working on the same content at

the same time. This does already exist in 2D. For example, support services may use

a remote desktop functionality to control a clients computer and shared web

navigation is already called co-browsing (Aberdeen Group, 2013). Users in these

 !91

situations are shown as cursors essentially their avatar. In virtual reality, the

representation tends to be more robust.

The concepts of self identity and social representation of self are interesting

to me, personally. I nd it interesting that our identities are completely independent

of our physical appearances and we choose to represent ourselves outwardly in

certain ways. One could write thousands of papers on the topic and indeed

thousands may have been written, but virtual reality provides a particularly

interesting niche for this.

Participating in the VR Chat community, it would seem like every day was

Halloween; Like every gathering is a low-key masquerade. Each user can choose any

name they want and don any appearance they want. A majority of users then

choose the appearance of favorite characters from pop culture, video games, comic

books, television, and movies. While most of these are humans, there are currently

no restrictions of any kind on the types of avatars allowed. A miniature cartoon

rabbit with a rainbow trail will jump around the feet of a 20 meter tall giant. One

user stuck to the identity of Mr. Whiskers, a black cat who never says anything, but

will meow or hiss and defecate on things he doesnt like. He hasnt broken character

once in the months Ive seen him. Another would switch rapidly in sessions

between being Batman, a human centipede zombie, a blaze of ame, Mario, a

honking van with headlights, being completely invisible, or even being an entire

room himself.

What I found in returning to talk to these people each week was that I could

remember people I had met more easily if they had the same name and

appearance. This may seem obvious, but while most people keep the same name,

they change what they look like on a regular basis. The experience would be
 !92

analogous to attending a party where everyone wears nametags. You meet people,

have conversations, and relationships are established. The next week, you go back

to the same party with the same people, but theyve switched skin. The personality

is attached to the nametag and not the persons appearance. You could imagine

how you would have a more dicult time remembering who someone was,

particularly if someone else now looks like they used to. Basically, the continuity of

relationships is more dicult to maintain. I realized this was the case for me, too, as

people wouldnt remember me as much if I switched avatars. I decided it would be

necessary, therefore, to have a single identity. It would also make sense to

represent myself in a way that people could maintain their relationship continuity

with me outside of VR. Representing myself as I actually appear in reality makes

sense as the option that fullls the criteria.

Avatar Creation Methods

I tried out three methods for avatar creation from actual humans: 3D Modeling,

photogrammetry, and 3D scanning. The purpose of each here is to obtain geometry

and textures which can be viewed and animated in a game engine. Each option has

pros and cons and I will describe my conclusions on the strengths and weaknesses

as I encountered them in the process.

3D Modelling

Basic 3D modelling is the rst option. I used a free program called MakeHuman to

begin (MakeHuman, 2015). It allows a user to use sliders to modify a characters

features. Using reference photos, I attempted to get the model as close as I could.

Then, after exporting the result, I used Maya to tweak vertices and Photoshop to
 !93

edit textures, but the same tasks could likely be completed with free softwares like

Blender and Gimp. The main problem with this method, in my opinion, is my own

inability to be objective. We tend to see and process humans, and especially faces,

and even more especially our own faces with distinct scrutiny and replicating their

exact image requires skilled artists. This was the process and result:

Photogrammetry

The second option is photogrammetry. This method uses several traditional 2D

photos taken around a subject at varying angles (Walford, 2007). Detail points are

then analyzed for dierences between images and a point cloud is generated from

their parallax. This point cloud then is used to create a mesh and the original

photos are applied as the texture. Agisofts Photoscan and Autodesks 123D Catch

are examples of programs that can do this (Agisoft, 2015; Autodesk, 2015a). I tested

the ability of 123D Catch to generate a mesh of my face using only a smartphone

camera and app. While it would also need some mesh and texture cleanup, it

provided a very good starting point, particularly having originated solely from

sele style photos with a smartphone pointed at myself.

 !94

Sele style scanning using 123D Catch smartphone app

I continued to test the viability of photogrammetry by having a subject stand in a T-

pose for an attempt at a full body scan. I took photos all around the subject, again

using only a smartphone to see if it would be an easy consumer solution with a

fairly uncontrolled environment. After running those photos through the Photoscan

analysis process, the results had quite a few problems.

Full body photogrammetry results from a smartphone in a purposely non-ideal environment

- testing for consumers since most would not have a custom studio
 !95

Entire sections of geometry are missing where photo edges cropped the subject.

The partially translucent clothing isnt understandable by the software, as it can

only determine a single face of geometry in a plane. The legs were also not well

determined by the software because they were not far apart and there was very

little contrast provided by the dynamic range of the camera. In order to have the

best tracking, camera settings like exposure need to be constant between photos.

The environment tested here had a high dynamic range that caused the

smartphone camera to automatically adjust. All of these issues can be remedied by

controlling the environment and taking photos carefully. Including the whole

subject in every photo with a high resolution camera locked at a specic exposure,

and having the subject wear opaque clothing in an evenly lit environment with legs

and arms apart would be ideal. It is no surprise, then, that this is the process for

professional photogrammetry studios (ScanLab, 2015; Innite-Realities, 2015). You

can pay to have a photogrammetric scan or use these techniques to do it yourself.

3D Scanning

The third option is 3D scanning using computer peripherals. The scanners I used for

this were the Artec MH-T, Eva, and Spider. The rst problem with these is that they

are cost-prohibitive for home or hobby use. They are very detail accurate, though.

The post-process of aligning scans, forming meshes, and texturing takes a

considerable amount of time, to the order of several hours or days depending on

the computing power. The scans yielded these results.

 !96

3D scan data, point cloud, and mesh.

To test avatar creation from the scan, I isolated the head to create an object for use

with ConVRge. I used a free program called Meshlab to reduce the geometry in

steps using the quadratic edge collapse decimation command (MeshLab, 2015).

Many of the UVs needed to be adjusted so the texture would look more correct.

Then, using Maya, I created a normal map of the high-poly model for the low-poly

version (Autodesk, 2015b). I created a custom reectivity map for unity that mimics

the increased specularity of eyes, lips, forehead, nose, cheeks and hairs. The nal

product seems acceptable.

Albedo (color), normal (bump), and metallic (specular) maps

 !97

Pros Cons

3D Modelling Some free softwares Subjective and inaccurate

Photogrammetry Free Prone to anomalies without

Relatively easier controlled environment

Common equipment

3D Scanning Extremely accurate Expensive, uncommon equipment

Long post-processing workow

Hybrid Solutions

Solutions that combine these methods would also be viable. For example, the body

could be created photogrammetrically and the face 3D scanned. Or the body could

be a customizable 3D model that users can attach a photogrammetry scan to.

Thanks for checking this paper out. Heres a bonus silly illustration (or as is its

scientic name: a sillustration) about VR design. If you want to talk about the things

in this paper or anything at all, feel free to email me:

[email protected]