Exploring Pointer Enhancement Techniques for Target Selection on Large Curved Display
Extracted source text · arXiv 2511.01826
%%
%% This is file `sample-sigchi.tex',
%% generated with the docstrip utility.
%%
%% The original source files were:
%%
%% samples.dtx (with options: `sigchi')
%%
%% IMPORTANT NOTICE:
%%
%% For the copyright see the source file.
%%
%% Any modified versions of this file must be renamed
%% with new filenames distinct from sample-sigchi.tex.
%%
%% For distribution of the original source see the terms
%% for copying and modification in the file samples.dtx.
%%
%% This generated file may be distributed as long as the
%% original source files, as listed above, are part of the
%% same distribution. (The sources need not necessarily be
%% in the same archive or directory.)
%%
%% The first command in your LaTeX source must be the \documentclass command.
%\documentclass[manuscript,review,anonymous]{acmart}
\documentclass[acmsmall]{acmart}
%%% Amanat Package imports below
\usepackage{adjustbox}
\usepackage{xr}
\usepackage{hyperref}
%\usepackage[table,xcdraw]{xcolor}
%\usepackage{multirow}
\usepackage[normalem]{ulem}
\usepackage{amsmath}
\newcommand{\mygreen}{rgb,180:red,180;green,120;blue,40}
% %hello
% \newcommand{\amanat}[1]{{\textcolor{\mygreen}{[AMANAT: #1]}}}
% \newcommand{\update}[1]{\textcolor{blue}{#1}}
% \newcommand{\kh}[1]{\textcolor{red}{\bf [KH: #1]}}
% \newcommand{\dhruv}[1]{{\textcolor{green}{[Dhruv]: #1}}}
% %\newcommand{\amanat}[1]{{\textcolor{brown}{#1}}}
% \newcommand{\co}[1]{{\textcolor{purple}{[Charlo]: #1}}}
% \newcommand{\remove}[1]{{\sout{#1}}}
% \newcommand{\outline}[1]{{\textcolor{pink}{\bf [OUTLINE: #1]}}}
% \newcommand{\degree}{\textbf{$^{\circ}$}}
% \newcommand{\p}[1]{{$p < #1$}}
%hello2
\newcommand{\amanat}[1]{}
\newcommand{\update}[1]{\textcolor{black}{#1}}
\newcommand{\kh}[1]{\textcolor{red}{\bf [KH: #1]}}
\newcommand{\dhruv}[1]{}
%\newcommand{\amanat}[1]{{\textcolor{brown}{#1}}}
\newcommand{\co}[1]{}
\newcommand{\remove}[1]{}
\newcommand{\outline}[1]{{\textcolor{pink}{\bf [OUTLINE: #1]}}}
\newcommand{\degree}{\textbf{$^{\circ}$}}
\newcommand{\p}[1]{{$p < #1$}}
\usepackage{listings}
\usepackage{xcolor}
\lstset{
language=C++,
basicstyle=\ttfamily\small,
keywordstyle=\color{blue},
commentstyle=\color{gray},
stringstyle=\color{green},
numberstyle=\tiny\color{gray},
numbers=left,
stepnumber=1,
numbersep=5pt,
showstringspaces=false,
breaklines=true,
frame=single,
captionpos=b,
tabsize=2
}
% \usepackage[normalem]{ulem}
% \usepackage{amsmath}
% \newcommand{\kh}[1]{\textcolor{blue}{\bf [KH: #1]}}
% \newcommand{\dhruv}[1]{{\textcolor{RedOrange}{#1}}}
% \newcommand{\amanat}[1]{{\textcolor{brown}{#1}}}
% \newcommand{\co}[1]{{\textcolor{purple}{[Charlo]: #1}}}
% \newcommand{\remove}[1]{{\textcolor{red}{\sout{#1}}}}
% \newcommand{\outline}[1]{{\textcolor{pink}{\bf [OUTLINE: #1]}}}
% \newcommand{\degree}{\textbf{$^{\circ}$}}
% \newcommand{\p}[1]{{$p < #1$}}
%%
%% \BibTeX command to typeset BibTeX logo in the docs
\AtBeginDocument{%
\providecommand\BibTeX{{%
\normalfont B\kern-0.5em{\scshape i\kern-0.25em b}\kern-0.8em\TeX}}}
% \setcopyright{acmlicensed}
% \acmJournal{PACMHCI}
% \acmYear{2024} \acmVolume{8} \acmNumber{ISS} \acmArticle{535} \acmMonth{12}\acmDOI{10.1145/3698135}
% \acmConference[CHI '24]{CHI 2024: ACM CHI Conference on Human Factors in Computing Systems}{May 11--15, 2024}{Honolulu, Hawai'i}
% \acmBooktitle{CHI 2024: ACM CHI Conference on Human Factors in Computing Systems,
% May 11--15, 2024, Honolulu, Hawai'i}
% \acmPrice{15.00}
% \acmISBN{978-1-4503-XXXX-X/18/06}
%\newcommand{\kh}[1]{\textcolor{blue}{\bf [KH: #1]}}
%%
%% Submission ID.
%% Use this when submitting an article to a sponsored event. You'll
%% receive a unique submission ID from the organizers
%% of the event, and this ID should be used as the parameter to this command.
%%\acmSubmissionID{123-A56-BU3}
%%
%% The majority of ACM publications use numbered citations and
%% references. The command \citestyle{authoryear} switches to the
%% "author year" style.
%%
%% If you are preparing content for an event
%% sponsored by ACM SIGGRAPH, you must use the "author year" style of
%% citations and references.
%% Uncommenting
%% the next command will enable that style.
%%\citestyle{acmauthoryear}
%%
%% end of the preamble, start of the body of the document source.
%%% The following is specific to ISS '24 and the paper
%%% 'Exploring Pointer Enhancement Techniques for Target Selection on Large Curved Display'
%%% by Dhruv Bihani, A. K. M. Amanat Ullah, Charles-Olivier Dufresne-Camaro, William Delamare, Pourang Irani, and Khalad Hasan.
%%%
\setcopyright{acmlicensed}
\acmDOI{10.1145/3698135}
\acmYear{2024}
\acmJournal{PACMHCI}
\acmVolume{8}
\acmNumber{ISS}
\acmArticle{535}
\acmMonth{12}
\received{2024-02-22}
\received[accepted]{2024-05-30}
\begin{document}
%%
%% The "title" command has an optional parameter,
%% allowing the author to define a "short title" to be used in page headers.
\title[Exploring Pointer Enhancement Techniques for Target Selection on Large Curved Display]{{Exploring Pointer Enhancement Techniques for Target Selection on Large Curved Display}}
%%
%% The "author" command and its associated commands are used to define
%% the authors and their affiliations.
%% Of note is the shared affiliation of the first two authors, and the
%% "authornote" and "authornotemark" commands
%% used to denote shared contribution to the research.
% \author{Dhruv Bihani}
% \authornote{Both authors contributed equally to this research.}
% \email{dhruv109@student.ubc.ca}
% \orcid{0009-0006-6498-766X}
% \author{A. K. M. Amanat Ullah}
% \authornotemark[1]
% \email{amanat7@student.ubc.ca}
% \orcid{0000-0001-5402-0160}
% \affiliation{%
% \institution{University of British Columbia, Okanagan}
% \city{Okanagan}
% \state{BC}
% \country{Canada}
% }
% \author{Dhruv Bihani}
% \orcid{0009-0006-6498-766X}
% \affiliation{%
% \institution{University of British Columbia}
% \city{Okanagan}
% \country{Canada}
% }
% \email{dhruv109@student.ubc.ca}
% \authornote{Both authors contributed equally to this research.}
% \author{A. K. M. Amanat Ullah}
% \orcid{0000-0001-5402-0160}
% \affiliation{%
% \institution{University of British Columbia}
% \city{Okanagan}
% \country{Canada}
% }
% \email{amanat7@student.ubc.ca}
% \authornotemark[1]
% \author{Charles-Olivier Dufresne-Camaro}
% \orcid{0000-0002-8563-7523}
% \affiliation{%
% \institution{University of British Columbia}
% \city{Okanagan}
% \country{Canada}
% }
% \email{chcamaro@student.ubc.ca}
% \author{William Delamare}
% \orcid{0000-0002-1830-4294}
% \affiliation{%
% \institution{Univ. Bordeaux, ESTIA-Institute of Technology}
% \city{Bidart}
% \country{France}
% }
% \email{w.delamare@estia.fr}
% \author{Pourang Irani}
% \orcid{0000-0002-7716-9280}
% \affiliation{%
% \institution{University of British Columbia}
% \city{Okanagan}
% \country{Canada}
% }
% \email{pourang.irani@ubc.ca}
% \author{Khalad Hasan}
% \orcid{0000-0002-4815-5461}
% \affiliation{%
% \institution{University of British Columbia}
% \city{Okanagan}
% \country{Canada}
% }
% \email{khalad.hasan@ubc.ca}
\author{Dhruv Bihani}
\affiliation{%
\institution{University of British Columbia, Okanagan}
%\streetaddress{Street address}
\city{Okanagan, BC}
\country{Canada}}
\email{dhruv109@student.ubc.ca}
\orcid{0009-0006-6498-766X}
\authornote{Both authors contributed equally to this research.}
\author{A. K. M. Amanat Ullah}
\affiliation{%
\institution{University of British Columbia, Okanagan}
%\streetaddress{1 Th{\o}rv{\"a}ld Circle}
\city{}
\country{Canada}}
\email{amanat7@student.ubc.ca}
\orcid{0000-0001-5402-0160}
\authornotemark[1]
\author{Charles-Olivier Dufresne-Camaro}
\affiliation{%
\institution{University of British Columbia, Okanagan}
%\streetaddress{Street address}
\city{Kelowna, BC}
\country{Canada}}
\email{chcamaro@student.ubc.ca}
\orcid{0000-0002-8563-7523}
\author{William Delamare}
\affiliation{%
\institution{Univ. Bordeaux, ESTIA-Institute of Technology, Bidart}
%\streetaddress{1 Th{\o}rv{\"a}ld Circle}
\city{}
\country{France}}
\email{william.delamare@acm.org}
\orcid{0000-0002-1830-4294}
\author{Pourang Irani}
\affiliation{%
\institution{University of British Columbia, Okanagan}
%\streetaddress{1 Th{\o}rv{\"a}ld Circle}
\city{}
\country{Canada}}
\email{pourang.irani@ubc.ca}
\orcid{0000-0002-7716-9280}
\author{Khalad Hasan}
\affiliation{%
\institution{University of British Columbia, Okanagan}
%\streetaddress{1 Th{\o}rv{\"a}ld Circle}
\city{}
\country{Canada}}
\email{khalad.hasan@ubc.ca}
\orcid{0000-0002-4815-5461}
%\author{* These authors contributed equally}
%%
%% By default, the full list of authors will be used in the page
%% headers. Often, this list is too long, and will overlap
%% other information printed in the page headers. This command allows
%% the author to define a more concise list
%% of authors' names for this purpose.
\renewcommand{\shortauthors}{Bihani, Ullah, Dufresne-Camaro, Delamare, Irani, and Hasan }
%%
%% The abstract is a short summary of the work to be presented in the
%% article.
\begin{abstract}
\input{01_abstract.tex}
\end{abstract}
%%
%% The code below is generated by the tool at http://dl.acm.org/ccs.cfm.
%% Please copy and paste the code instead of the example below.
%%
\begin{CCSXML}
<ccs2012>
<concept>
<concept_id>10003120.10003121.10003128.10011754</concept_id>
<concept_desc>Human-centered computing~Pointing</concept_desc>
<concept_significance>500</concept_significance>
</concept>
<concept>
<concept_id>10003120.10003121.10003125.10010591</concept_id>
<concept_desc>Human-centered computing~Displays and imagers</concept_desc>
<concept_significance>300</concept_significance>
</concept>
<concept>
<concept_id>10003120.10003121.10003122.10003334</concept_id>
<concept_desc>Human-centered computing~User studies</concept_desc>
<concept_significance>100</concept_significance>
</concept>
</ccs2012>
\end{CCSXML}
\ccsdesc[500]{Human-centered computing~Pointing}
\ccsdesc[300]{Human-centered computing~Displays and imagers}
\ccsdesc[100]{Human-centered computing~User studies}
%%
%% Keywords. The author(s) should pick words that accurately describe
%% the work being presented. Separate the keywords with commas.
\keywords{Pointing, Large Curved Display, Control-Display Gain, Cursor Enhancement}
\begin{teaserfigure}
\includegraphics[width=\textwidth]{figures/new-teaser.png}
\caption{(a-b) A user selects targets on an immersive 3D large curved display. (c) \textbf{Interaction Distance and Lateral Offset}: Interaction Distance represents user positions along the main axis, while Lateral Offset represents users positions with a deviation of R/2 towards the left from the main axis.}
%\label{fig:teaser}
\Description{(a-b) A user selects targets on an immersive 3D large curved display. (c) The user is positioned at different distances from the centre of the display. }
\label{fig:teaser}
\end{teaserfigure}
%%
%% This command processes the author and affiliation and title
%% information and builds the first part of the formatted document.
\maketitle
% sections are each in separate files
\input{1_introduction}
\input{2_background}
\input{3_user_study1}
\input{3.5_exp-results}
\input{4_user_study2}
\input{4.5_exp2-results}
%\input{3_technique}
%\input{4_implementation}
%\input{5_exp-protocol}
%\input{6_exp-results}
\input{7_discussion}
\input{8_conclusion}
\phantomsection
\addcontentsline{toc}{section}{Acknowledgements}
\section*{Acknowledgements}
This research was partially funded by NSERC Discovery Grant \#RGPIN-2019-05211 and the Canada Foundation for Innovation (CFI) Project number \#40440.
% This work was made possible by NSERC Discovery Grant \#2018-05187, the Canada Foundation for Innovation Infrastructure Fund ``Facility for Fully Interactive Physio-digital Spaces'' (\#33151), and Ontario Early Researcher Award \#ER16-12-184.
\appendix
\section{C\# Unity Code Snippet for Cursor Speed Enhancement in 3D Space via Game Object Rotation} \label{appendix} When using the Pointer Acceleration-based (PA) technique, the method ``PA()'' is called every frame inside the Update() function. The method for other techniques was similar, but the calculation of ``PAgetSpeed()'' differed based on each technique.
The current game object (in the code, ``transform.rotation'') was rotated in 3D space based on the movement of the controller (in the code, ``CGS.ControllerPosition.transform''). A cursor was placed in the orthogonal direction of the current game object. We rotate the current game object more for a faster speed than the controller, and rotate the current game object slower if we want to reduce the speed compared to the controller’s movement. This speed is determined by the function ``PAgetSpeed()''.
\begin{lstlisting}[
caption={C\# Unity Method to Rotate the Current Game Object Based on the Controller's Rotation in Front of a Curved Display},
label={lst:rotation-code},
xleftmargin=1.5em, % Adjust this value to increase the left margin
xrightmargin=1.5em % Adjust this value to increase the right margin
]
void PA()
{
Quaternion rotationDifference = CGS.ContollerPosition.transform.rotation * Quaternion.Inverse(lastcontrollerrotation);
// Extract the angle and axis from the rotation difference.
rotationDifference.ToAngleAxis(out float angle, out Vector3 axis);
// current gameobject's rotation = rotation difference * speed factor.
Quaternion tempRotation = transform.rotation * Quaternion.AngleAxis(angle * PAgetSpeed(), transform.InverseTransformDirection(axis));
// Convert the rotation to Euler angles.
Vector3 euler = tempRotation.eulerAngles;
// Unity uses degrees from 0 to 360. Convert them to -180 to 180.
if (euler.x > 180)
euler.x -= 360;
if (euler.y > 180)
euler.y -= 360;
// based on our display xMax = 90, xMin = -90, yMax = 70, and yMin = -70
if ((euler.y <= xMax && euler.y >= xMin) && (euler.x <= yMax && euler.x >= yMin))
{
transform.rotation = tempRotation; // cursor always in display
}
// Store the controller's current rotation for the next frame.
lastcontrollerrotation = CGS.ContollerPosition.transform.rotation;
}
\end{lstlisting}
\begin{lstlisting}[
caption={Movement Speed of the Game Object rotation method is determined by the PAgetSpeed() Method},
label={lst:rotation-code},
xleftmargin=1.5em, % Adjust this value to increase the left margin
xrightmargin=1.5em % Adjust this value to increase the right margin
]
float PAgetSpeed()
{
float speed = 0;
Vector3 mouseDelta = CGS.ContollerPosition.transform.position;
Quaternion currentControllerRotation = CGS.ContollerPosition.transform.rotation;
// Calculate the difference in position between the current and last frame
Vector3 diff = mouseDelta - lastMousePosition;
// Calculate the speed of the controller's movement (magnitude of diff / time since last frame)
float controllerSpeed = diff.magnitude / Time.deltaTime;
// The values below are constant values which are set based on pilot studies
float x = controllerSpeed;
float vInf = vRatioInf * (vMax - vMin) + vMin;
speed = ((cdMax - cdMin) / (1 + Mathf.Exp(-lambda * (x - vInf)))) + cdMin;
CGS.debug = "Controller speed: " + controllerSpeed + "\n" + "Cursor Speed: " + speed;
lastMousePosition = mouseDelta;
return speed;
}
\end{lstlisting}
% reference section
\bibliographystyle{ACM-Reference-Format}
\bibliography{main.bib}
\end{document}
\endinput
%%
%% End of file `sample-sigchi.tex'.
Large curved displays are becoming increasingly popular due to their ability to provide users with a wider field of view and a more immersive experience compared to flat displays. Current interaction techniques for large curved displays often assume a user is positioned at the display's centre, crucially failing to accommodate general use conditions where the user may move during use. In this work, we investigated how user position impacts pointing interaction on large curved displays and evaluated cursor enhancement techniques to provide faster and more accurate performance across positions. To this effect, we conducted two user studies. First, we evaluated the effects of user position on pointing performance on a large semi-circular display (3m-tall, 3270R curvature) through a 2D Fitts' Law selection task. Our results indicate that as users move away from the display, their pointing speed significantly increases (at least by 9\%), but accuracy decreases (by at least 6\%). Additionally, we observed participants were slower when pointing from laterally offset positions. Secondly, we explored which pointing techniques providing motor- and visual-space enhancements best afford effective pointing performance across user positions. Across a total of six techniques tested, we found that a combination of acceleration and distance-based adjustments with cursor enlargement significantly improves target selection speed and accuracy across different user positions. Results further show techniques with visual-space enhancements (e.g., cursor enlargement) are significantly faster and more accurate than their non-visually-enhanced counterparts. Based on our results we provide design recommendations for implementing cursor enhancement techniques for large curved displays.
% %\co{I have a hard time finding this figure insightful. Can we revise what we'd like readers to take away from this, and maybe take new pictures?}
% %\amanat{I made changes. the new teaser should be insightful}
% %\co{Low priority: scale of the people in b) is off. Gives the impression than the curved display is gigantic.}
% Large curved displays are becoming increasingly popular due to their ability to provide users with a wider field of view and a more immersive experience compared to flat displays.
% Current interaction techniques for large curved displays often assume a user is positioned at the display's centre, crucially failing to accommodate general use conditions where the is located at a non-center position.
% In this work, we investigate how user position impacts pointing interaction on large curved displays and propose techniques to provide faster and more accurate performance across positions.
% %It is crucial to understand how user position impacts interactions, like pointing, and to develop methods to minimize these effects for consistent performance.
% %\kh{tell why this is important to explore users position around the screen}
% To this effect, we conducted a user study to evaluate the effect of user position on pointing performance on a large semi-circular display (3m-tall, 3270R curvature) through a Fitts' Law selection task. % a Fitts law study to first investigate the effects of user position on pointing performance using a large semi-circular display (3m-tall, 3270R curvature).
% Our results indicate that as users move away from the display, their pointing speed significantly increases (at least by 9\%), but accuracy decreases (by at least 6\%).
% %However, being off-centered (i.e., displacement towards the left) did not significantly affect performance in terms of movement time and accuracy.
% Additionally, we did not find lateral displacements to significantly affect pointing performance.
% %This leads to difference in performance based on the user's position.
% %\kh{what does that mean? explain clearly}, where we show how current interaction techniques fail to provide a consistent user experience as the users moves away from the display's centre.
% %\co{squeeze in one interesting number/result from study 1 if possible.}
% We then conducted a second user study to explore which pointing techniques providing motor- and visual-space enhancements best afford effective pointing performance across locations.
% %\co{Remove "consistent" unless we actually tested for that?} \amanat{removed consistent}
% %To improve users pointing performance regardless of their location, we explored six techniques focusing on motor space enhancement (i.e., hand movement acceleration and user-screen distance) and visual space enhancements (i.e., cursor enlargement).
% Across a total of six techniques tested, we found that a combination of acceleration and distance adjustments with cursor enlargement significantly improves target selection speed and accuracy across different user positions.
% Also, results showed that techniques with visual-space enhancements (e.g., cursor enlargement) are significantly faster and more accurate than their non-visually-enhanced counterparts.
% Based on our results we provide design recommendations for implementing cursor enhancement techniques for large curved displays.
%!TEX root = paper.tex
\section{Introduction}
Large displays, whether flat or curved, are becoming increasingly popular in fields like data visualization \cite{novak2008designing, satyanarayan2013chi, birnholtz2007exploratory}, entertainment \cite{ardito2015interaction, scheible2005mobilenin, khoo2009designing}, and collaborative applications \cite{andrews2011information, card1999information, baudisch2001focus, cockburn2009review}. %\kh{Amanat: update this, we are not only focusing on large curved display now - immersive is a key feature here}
Curved displays have been found to enhance user satisfaction and task performance compared to large flat displays, because they offer a wider field of view \cite{ahn2014research, kyung2021curved, sultana2024exploring}.
Pointing using ray-casting is a common interaction technique on large displays because it allows for rapid cursor movements and for users to move freely in front of the screen \cite{kopper2008increasing}. %\co{fill-in the blanks and add ref.}
User pointing performance on large flat displays has been shown to vary in speed and accuracy based on notably the user's \textit{Interaction Distance}, i.e., the distance from the user perpendicular to the centre of the display \cite{kopper2010human,janzen2016modeling,kovacs2008perceptual,tao2021freehand, sindhupathiraja2024exploring}.
The size of the selection target, as well as the distance to the target, further affects performance, with small distant targets leading to slower selection times and lower accuracy \cite{tao2021freehand, babic2018pocket6, siddhpuria2018pointing, nancel2013high}. %\co{break refs into multiples, or just remove some of them.} \amanat{reduced some citations}
Pointing on curved displays introduces unique challenges due to the geometry of the displays altering how users may interact with them as they move around (e.g., laterally from the display's centre) \cite{zannoli2017perceptual}.
% , such as pointing from off-centered positions (i.e., lateral displacements, and \textit{Interaction Distance} in the forward-backward direction relative to the center). %\co{Need to clarify what this means in terms of unique challenges. What makes this such a big deal?} \amanat{I have somewhat addressed it @CHARLO have a look.}
%\remove{Curved displays notably require users to rotate their heads more, but reduce movements in front of the display \cite{shupp2009shaping}.}
For instance, interacting with a curved display often requires more head rotation to engage with different areas, compared to flat displays \cite{shupp2009shaping}.
Users' field-of-view during interactions also varies with user position when the display is curved \cite{zannoli2017perceptual}.
%\co{Add one sentence focusing on field-of-view as users move away from centre.}\amanat{added}
%\update{Curved displays result in perceptual differences, such as a varying field-of-view compared to flat displays, with these perceptual variations intensifying as the user shifts from the center of the curved display \cite{zannoli2017perceptual}.}
%\update{Zannoli and Banks \cite{zannoli2017perceptual} found that a major advantage of curved screens minimize the distortion of the image surface near the screen's boundaries.
%For this reason, they found that curved screens can accommodate a wider range of user positions compared to Flat screens.
%However, Zannoli and Banks acknowledged that to fully realize this advantage, the projection of images should match the screen curvature instead of being based on the planar projection which is intended for the flat display.
%Furthermore, they found that display curvature leads to perceptual consequences such as having different field-of-view and seeing reduced distracting reflection. Zannoli and Banks further stated perceptual consequences based on the user's horizontal and vertical position relative to the screen.}
%\update{In contrast, on a flat display, a user positioned off-center to the left would maintain a uniform distance to all regions across the display's surface, without the varying proximity caused by curvature.}
Furthermore, curved displays cannot be represented as 2D planes, which limits the applicability of commonly-used pointer enhancement techniques \cite{casiez2008impact,nancel2013high} which rely on planar geometry to adjust the effects of motor movement.
%\update{Most pointer enhancement techniques \cite{casiez2008impact,nancel2013high} relied on 2D transformation of the cursor based on the motor movement, however these techniques cannot be applied on a curved display as unlike the Flat display, the interaction surface of the curved display is not 2D.}
%\co{Fill rest of paragraph with motivation to really sell that large curved displays are a unique design space. Pointing techniques should be a separate paragraph.}
%\amanat{filled the text above. feel free to use parts you like.}
% Shupp et al. \cite{shupp2009shaping} have found flat displays forced users to walk more and while curved displays require the users to constantly rotate their head.
%Moreover, the focus point of curved while viewing content varies significantly between flat and curved displays, with curved screens promoting a more centralized focus while on a flat display, it tends to be at the lower third of the screen \cite{ahn2014research}. \co{I don't know what this sentence adds to the introduction. Commenting it.}
%\co{How so?} \amanat{added}
%For instance, a user standing close to the left side of the display has a smaller distance from the display's left side than from the display's right side.
%\co{This seems to be true regardless of if the display is flat or curved, no? I dont remember what we're trying to say here.} \amanat{CHANGED: commented out old justification and added a new justification}
%\co{review paragraph below to see if it reads well.}
\update{Handheld controller-based ray-casting is a widely used technique for pointing on large displays, where users manipulate a ray in 3D space by moving the controller with their hand. Although other research leverages head and eye movements for pointing at large displays \cite{khamis2017eyescout, James_2020_personal}, hand-based pointing remains the most prevalent technique due to its intuitive nature and ease of learning, which lead to faster and more precise selections \cite{lin2015investigation, bernardos2016comparison}. Nevertheless, hand-based pointing is susceptible to precision errors due to hand tremors and involuntary movements \cite{myers2002interacting}. %Therefore, we focus on handheld controller-based ray-casting in our study.
}
User position has additionally been shown to significantly affect pointing performance on flat displays \cite{kopper2010human,janzen2016modeling,kovacs2008perceptual,tao2021freehand, hourcade2012small, Ullah2023Pointing}, resulting in less accurate selections at larger pointing distances.
Numerous techniques have been developed over the years to improve pointing performance by modifying the ratio between cursor movement and hand movement \cite{casiez2008impact, nancel2013high, chowdhury2022wriarm, chowdhury2023paws}, and modifying the visuals \cite{GrossmanCHI2005}.
However, we find no prior work on designing techniques that leverage \textit{Interaction Distance} for large displays, whether flat or curved. %\dhruv{we do have work for large flat displays right amanat ?}
%(i.e., Control-to-Display (CD) gain), can improve selection accuracy by slowing down the cursor .
%\co{How is the previous sentence relevant? How is lowering CD gain potentially useful?} \amanat{slower speed leads to easy to select smaller targets as with slow speeds cursors are easier to control.}
%However, slow cursor speeds can cause the ``clutching'' problem, where the user must reposition the input device to further move the cursor along a direction\cite{nancel2013high}, which may be addressed via cursor enhancement techniques. \co{cite a couple papers, if possible.}
%\dhruv{I tried to find more papers but couldnt find any that related to what we are looking for.}
%To avoid such issues, researchers have proposed cursor enhancement techniques that adjust the CD gain ratio based only on the movement of the input device \cite{casiez2008impact}. \co{not needed, in my opinion}
%Nevertheless, user position has been shown to significantly affect pointing performance on flat displays \cite{kopper2010human,janzen2016modeling,kovacs2008perceptual,tao2021freehand, hourcade2012small}, highlighting a need for user position to be considered in pointer enhancement techniques to improve pointing performance.
%then the right part of the screen will be far away for him to point at. %Optimal selection in this scenario, requires different cursor speeds depending on whether the person is pointing toward the left or right.
More importantly, the performance of commonly-used cursor enhancement techniques such as cursor acceleration \cite{casiez2008impact,nancel2013high} and enlargement \cite{GrossmanCHI2005} have not been investigated for curved displays.
Considering the unique design space of large curved displays, existing techniques may not generalize well as-is to this space.
In this work, we aim to understand 1) how user position affects pointing performance on large curved displays, and 2) which cursor enhancement techniques best afford users with optimal pointing performance across positions by ensuring faster selections and higher accuracy. %\co{I still have doubts regarding "across positions". Is there nothing to discuss in our results about how results vary between positions?}
%\dhruv{So there are results to discuss various positions for our first study. But for the second study we are just comparing different techniques based on all positions . So I dont there is a difference across any positions.}
%\co{Change objective 2) if we dont actually test for consistency} \amanat{removed consistency}
To this effect, we conduct two user studies.
First, we evaluate how \textit{Interaction Distance}, and \textit{Lateral Offset} (i.e., lateral distance from the display's centre) affect pointing performance on a large curved display through a Fitts' Law selection task.
%Our results confirm that pointing performance varies primarily with \textit{Interaction Distance}: larger distances increase pointing speed but lower pointing accuracy \cite{ens2016moving}.
We confirm the effect of \textit{Interaction Distance} is similar to large flat displays: larger distances increase pointing speed but lower pointing accuracy \cite{kopper2010human}.
%\co{This is already known in the literature, right? if so, we might want to say instead "We confirm the effect of X is similar to large flat displays: larger distances... [cite]".} \amanat{done as you suggested.}%\amanat{dont want to reduce insights from our study by recognizing previous literature.} \co{This isn't reducing insights. It's being transparent and acknowledging where our work stands.}
Furthermore, we observed a significant effect of \textit{Lateral Offset} on pointing speed, where selections are slower from offset positions. %\dhruv{I think there is a need to change this statement also now . I have updated it.}
%\update{Furthermore, unlike large flat displays where there was no significant effect of \textit{Lateral Offset} on pointing speed or accuracy\cite{Ullah2023Pointing}, we actually found a significant effect of \textit{Lateral Offset} on pointing speed or accuracy for large curved display.}
Secondly, we explore six cursor enhancement techniques defined along two axes --- motor space (i.e., motor-space acceleration and distance-based augmentations) and visual space (i.e., cursor enlargement) --- to identify which technique best reduces performance imbalances across different user positions (i.e., \textit{Interaction Distances} and \textit{Lateral Offsets}). %\co{Again, what do we mean by performance imbalances?} \amanat{added}
%\update{Secondly, we explore six cursor enhancement techniques defined along two axes --- motor-space enhancement, i.e., motor-space acceleration and distance-based augmentations, and visual-space enhancement, i.e., cursor enlargement --- to identify the technique that best reduces efficiency gaps across different user positions.}
%\dhruv{I changed the sentence so that it makes more sense and is understandable}
We found that \textit{Interaction Distance} combined with controller-acceleration-based techniques led to superior pointing performance. % over only \textit{Interaction-Distance}-based techniques.
Furthermore, augmenting motor-space enhancements with visual-space enhancements improves pointing accuracy and speed.
%xxcomebackxx
The main contributions of our work are: %\kh{please update the contribution statements}:
\begin{itemize}
\item Empirical support for the effects of user position (i.e., \textit{Interaction Distance} and \textit{Lateral Offset}) on pointing performance on large curved displays.
\item Evidence of the benefits of combining acceleration-based, and \textit{Interaction-Distance}-based cursor enhancement techniques on pointing performance on large curved displays across user positions.
\item Design recommendations on improving pointing performance on large curved displays using motor-space and visual-space enhancements.
%\co{Can we maybe add design guidelines as a third contribution based on the results of study 2?}
\end{itemize}
%Fitts law \cite{fitts1954information} is widely used to predict and evaluate the pointing performance of different input modalities in different environments.
%Fitts law defines task difficulty or Index of difficulty (ID) of pointing at targets to be a function of two factors: target width and target amplitude (i.e. distance between two targets).
%Based on Fitts law \cite{fitts1954information}, researchers \cite{tao2021freehand, babic2018pocket6, lischke2016magic, liu2015gunslinger, haque2015myopoint, siddhpuria2018pointing, nancel2013high} have explored pointing performance on large physical displays and found that higher task difficulty leads to higher movement/selection time and lower accuracy.
%\kh{reference needed and the next sentence should at least refer to a paper and discuss the results} \amanat{added}.
%These studies revealed that both increasing target amplitude and decreasing target width lead to higher tasks difficulty (i.e., increased movement time and decreased accuracy).
%However there has been limited research on pointing performance with varying user position and distance on large curved displays.
%Consequently, researchers have explored acceleration-based techniques to improve pointing at targets with higher Fitts IDs where the cursor speed changes dynamically based on the speed of the controller/hand movement \cite{casiez2008impact, nancel2013high}.
%Another promising cursor enhancement technique involves enlarging the cursor as it nears its target, a strategy found to enhance both pointing speed and accuracy \cite{grossman2005bubble, delamare2022multifingerbubble}.
%While the impact of \textit{Interaction Distance} has been well-established for flat displays \cite{kopper2010human, janzen2016modeling, kovacs2008perceptual, tao2021freehand}, no study to date has explored its implications for pointing on curved displays.
%Adjusting the cursor speed based on \textit{Interaction Distance} also remains unexplored for both flat and curved displays.
%More crucially, cursor enhancement techniques have only been investigated for flat displays, leaving the effects of different cursor enhancement on curved displays unexplored. \co{See 2nd paragraph question, what makes curved displays unique?}
%This is problematic because there is no guideline on how to implement existing cursor-curved displays and no guarantee these enhancement techniques will work on curved displays.
\section{Background and Related Work}
%\co{This section has not been revised in quite a while. Please ping me once new information are included, and I will work on it.}
%\co{ Considering the problem we are exploring only applies to curved displays, I don't think we need an entire subsection dedicated to justifying why we focus on curved displays. Maybe combining subsections 2.1 and 2.2 together would tell a nicer story: i.e., we first need to explain what's a curved display, and why should people care about them before jumping into pointing issues on curved displays.}
%\amanat{I would argue to keep the three sections.
%1. why explore curved?
%2. why explore pointing ?
%3. why explore cursor enhancement?
%we approved this talking to khalad
%}
%\amanat{@DHRUV you have to work on section 2.3 which is cursor enhancement. I can work on the first two subsections}
\subsection{Large Displays: Flat vs Curved}
%\co{This section needs to discuss (repeat) the differences between flat and curved displays in terms of design spaces. Right now, it merely compares the performance on tasks, but this does not justify the research project. Leave the pointing discussion for the next subsection though.}
%\textcolor{red}{NOT ADDRESSED}
Comparative studies between flat and curved large displays have been conducted across various tasks such as visual searches \cite{kyung2021curved, shupp2009shaping, sultana2024exploring}, data visualization \cite{liu2020design, shupp2009shaping} and reading \cite{wei2020reading, park2019effects, sultana2024exploring}.
Curved displays have been shown to be more effective on graph comparison and map search tasks \cite{shupp2009shaping} and video watching \cite{urakami2021comparing, choi201553, ahn2014research}, notably due to their improved ergonomics. %\cite{urakami2021comparing, ahn2014research, kyung2021curved, choi201553, shupp2009shaping}.
Zannoli et al. \cite{zannoli2017perceptual} have additionally been found to afford a wider range of user positions for interactions and reduce reflections, but require more complex rendering techniques to support their geometry \update{for a 65-inch with large display curvature of 4.18m}.
%\co{"certain tasks": What are they? Can we combine the first two sentences together to focus on the important outcomes?} \dhruv{We are talking about the tasks in the upcoming sentence for example the search task. I also combined the first two sentences}
%\amanat{I did not combine but added what are the certain tasks.}
%\update{Zannoli and Banks \cite{zannoli2017perceptual} found that a major advantage of curved screens minimize the distortion of the image surface near the screen's boundaries.
%For this reason, they found that curved screens can accommodate a wider range of user positions compared to Flat screens.
%However, Zannoli and Banks acknowledged that to fully realize this advantage, the projection of images should match the screen curvature instead of being based on the planar projection which is intended for the flat display.
%Furthermore, they found that display curvature leads to perceptual consequences such as having different field-of-view and seeing reduced distracting reflection. Zannoli and Banks further stated perceptual consequences based on the user's horizontal and vertical position relative to the screen.}
%\update{In contrast, on a flat display, a user positioned off-center to the left would maintain a uniform distance to all regions across the display's surface, without the varying proximity caused by curvature.}
%Curved displays were found to outperform flat displays in 2D graph comparison and visual searches on maps \cite{shupp2009shaping}.
%This can be attributed to the constant pixel-visual-angle resolution afforded only by curved displays when the user is positioned at the display's centre.
Interacting with curved displays also require more head movements, and are less suited for overviews \cite{shupp2009shaping} \update{for large curved (having 0.76m radius) and flat display with the combination of 24
monitors arranged in 8 × 3 grid approximately 9 feet wide and 3 feet tall}.
Display types have also been compared based on content density (e.g., the number of charts displayed simultaneously) \cite{liu2020design, shupp2009shaping}.
Users performed better on flat displays for small numbers of data visualizations rendered on a 2D grid \cite{liu2020design}, but preferred curved displays when content density increased.
Furthermore, Kuyung and Park \cite{kyung2021curved} found that large semi-circular \update{displays (33" and 50") with a curvature radius of 60cm} lead to improved performance in visual search tasks, compared to flat and other curved displays.
Additionally, reading on \update{27" displays} was found to be faster when the users were positioned at a distance equal to the curved display curvature's radius \cite{park2019effects}.
VR studies further revealed horizontally curved virtual displays to be more comfortable than vertically curved, flat and spherical displays \cite{wei2020reading}.%\co{Amanat, I think some words are missing in this sentence} \amanat{added missing words.}
%Moreover, they found 4000R had significantly lower physical demand than the other radii.
%\co{missing sentence relating their work to ours. Since they focused on small regular displays, can we really assume the results will generalize?} \amanat{removed that sentence.}
%Although they found these results for regular-sized curved displays, it is important to test user performance for large-scale curved displays.
%Wei et al. \cite{wei2020reading} explored how various 3D surface shapes affect reading experiences in virtual reality (VR) by conducting two user studies.
%Their findings indicated that reading on a horizontally curved surface was more comfortable compared to vertically curved, Flat, and spherical displays.
%\co{This work seems irrelevant to our paper. I suggest removing this paragraph.} \amanat{i changed the wording now it should fit}
%Therefore, in this study we chose a horizontally curved display
%Their findings further highlighted that as the size increased, curved displays continued to perform efficiently, while flat displays encountered challenges.
%\co{This last paragraph should briefly summarize the benefits of curved displays, but also the unique design space it involves which might affect performance, and has not be studied before. Then, the current sentence can end the paragraph.} \amanat{added}
%\dhruv{Amanat can you addess this}
%Since previous studies have highlighted the benefits of semi-circular curved displays compared to conventional flat screens,
Overall, curved displays offer immersive viewing experiences and can enhance task performance.
However, they introduce unique and open challenges compared to flat displays in terms of supporting user interactions across various positions, which have yet to be solved.
%\update{While curved displays offer immersive viewing experiences and enhanced task performance, they also introduce a unique design space that can impact user performance in ways not yet fully explored.}
Therefore, in this work, we investigate how the performance of pointing --- a general task --- varies on large curved displays, and what sets of techniques can best improve user performance in this context.
\subsection{Pointing on \update{Large Curved} Displays }
%\amanat{These two research papers are the only previous work done on large curved displays apart from my two works which are in press. We would like to highlight that. I feel } \co{Fair, but then we need to get that point across in the subsection.}
Pointing is considered one of the most common metaphors for selection and interaction with user interfaces \cite{bowman20043d, argelaguet2009efficient}.
Ray-casting (i.e., a ray projected from the tip of an input device) is the most widely used method for pointing at targets on large \update{flat} displays \cite{cavens2002interacting, davis2002lumipoint, oh2002laser, jota2010comparisonGIpaper}.
Research on pointing on large displays has primarily focused on large flat displays \cite{sluyters2023consistent, jota2010comparisonGIpaper, nancel2013high, siddhpuria2018pointing, haque2015myopoint}.
We thus include in this section relevant work on pointing at \remove{non-large} \update{large curved} displays.
%\co{FYI, the most recent paper in that list is from 2019. Is there not any more recent work?} \amanat{added a 2023 paper}
We found limited prior work on pointing at \update{large} curved displays \cite{Ullah2023Pointing, Ullah2023LargeCurved, hennecke2013investigating, ens2016moving}.
Pointing performance has been shown to not vary between physical and virtual \update{large} curved displays \cite{Ullah2023Pointing}, facilitating the exploration of this design space.
Fitts' Law was found to be applicable for 2D pointing tasks on \update{large} curved displays using direct touch or mouse \cite{hennecke2013investigating}.
Similarly, 1D pointing performance has been shown to vary on \update{large} virtual curved displays based on curvature and improve when the display is less curved \cite{Ullah2023LargeCurved}.
Additionally, the speed-accuracy trade-off in pointing has been shown to apply to 1D pointing on \update{large} curved displays, but lateral displacements were not found to affect performance \cite{Ullah2023Pointing}.
\remove{Finally, larger field-of-views (e.g., having the user stand closer to the curved display) may improve pointing performance \cite{ens2016moving} for CAVE environment but require further investigation regarding user position and display properties.} %\update{on how changes in the user's distance to the display and variations in display sizes impact pointing accuracy }
%\update{Researchers \cite{Ullah2023Pointing} have also compared differences in performance for virtual and physical curved displays. They found no difference in pointing performance between the two platforms.}
%\remove{Finally, 1D pointing performance has been found to vary based on the user's distance to the large curved display, but not on their lateral deviation from the display's centre (equivalent to half of the display curvature's radius).}
%\co{Amanat, I wasnt aware of your paper beforehand. I think it would be worthwhile to stress that there exists prior work looking into lateral offsets, and they hadnt found any differences. Otherwise, we're one google search away from losing most of our motivation.}
%\dhruv{Amanat pls look into this}
%Henneke et al. \cite{hennecke2013investigating} investigated whether a 2D pointing task on a 42" curved display using direct touch and mouse can be modeled using Fitts' Law.
%In both instances, they found Fitts' Law to be applicable.
%However, their results may not be generalized to curved displays with large radius (e.g., 3.27m).
%Ens et al. \cite{ens2016moving} explored how the field-of-view of a curved display may affect pointing performance in a CAVE setting using a Fitts' Law pointing task.
%provided a different approach by conducting a Fitts' law study where targets were set in a circular pattern using projectors in a CAVE setting.
%Their study explored various field-of-view conditions, from 8\degree to 128\degree.
%They showed that although pointing performance improved as the field-of-view expanded, this change was not statistically significant.
Taken together, our understanding of the effects of factors unique to curved displays (e.g., user position relative to display curvature) on pointing performance is limited.
In this work, we first validate the effects of \textit{Interaction Distance} and \textit{Lateral Offset} on 2D pointing task performance on large curved displays, and then investigate cursor enhancement techniques to best support pointing across varied user positions.
\subsection{Cursor Enhancement}
We explore in this work which cursor enhancement techniques best improve the accuracy and selection times for pointing on a large curved display from various positions.
To the best of our knowledge, no prior work has explored techniques for this setting.
We thus review existing literature on cursor enhancement techniques on large flat displays to identify candidate techniques. %\co{Please clarify the scope of our review: large flat displays, vs flat displays vs all displays. NThis paper is not for large displays, right? \cite{NancelCHI2013} } \amanat{The scope of the section is "pointer enhancement on large flat displays." we do not consider all displays. But our paper is on only "Curved displays."}%From the motivation of previous cursor enhancement techniques, we investigate how pointing performance changes when considering different cursor enhancement techniques.
We categorize cursor enhancement techniques under two groups: \textbf{motor-space} enhancements, and \textbf{visual-space} enhancements.
%\co{Is this correct? Are there references supporting this?}
%\dhruv{so I checked but there is no literature review but this is an assumption we made since right now there is only two ways for cursor enhancement, either by changing the speed or the size therefore we mentioned these two sub categories}
\textbf{Visual-space} enhancements act on cursor and target sizes to facilitate target selection.
For example, \textit{bubble cursor} which dynamically updates its size based on the proximity of surrounding targets \cite{GrossmanCHI2005}.
%\co{If there are many papers stating bubble cursor is the best, let's mention bubble cursor here. Otherwise, let's not dwell on the details given how well known the technique is.}
%The technique was found to significantly outperform point cursor and object pointing techniques on 1D and 2D pointing tasks. %\co{Are the two compared techniques relevant to name? Do we refer to them elsewhere?} \amanat{We take the idea of the cursor enlargement in bubble cursor and use it as our visual-space enhancements. I added the visual-space enhancement in to the sentence. it now should be good.}
%They found out that bubble cursor could be accurately modeled and predicted using Fitts’ law.
\textbf{Motor-space} enhancements instead act on the cursor's speed to improve pointing performance.
Such techniques often modify the Control-to-Display (CD) gain to adjust the ratio between cursor movement and input device movement to require smaller input movements to traverse large distances, or larger input movements when the cursor is near a target to facilitate accurate selections \cite{casiez2011no}.
Common motor enhancement techniques involve Constant Gain (CG), Pointer Acceleration (PA) which dynamically adjusts CD gain based on input device movement (e.g., a mouse) \cite{casiez2008impact}, and Adaptive Pointing (AP) which adjusts the CD gain based on input device speed and device-cursor offset \cite{konig2009adaptive}.
PA has notably been shown to lead to faster selections when their CD gain levels are not precisely optimized \cite{casiez2008impact}.
AP further leads to more accurate and faster selections than direct pointing techniques \cite{konig2009adaptive}.
Techniques combining ray-casting (absolute pointing) with relative pointing have also been proposed \cite{VogelUIST2005,McCallumUIST2009,ForlinesUIST2006,GalloIEEE2010}.
Finally, cursor enhancements can be designed for high-precision settings, such as RelaSmall and RelaLarge which adjust CD gain based on input device acceleration \cite{nancel2013high}.
%Casiez et al. \cite{CasiezHCI2008} explored motor-space enhancement by examining the impact of control-display gain on mouse pointing performance for regular, and very large displays.
%They used two techniques for modifying the CD gain: Constant Gain (CG) which uniformly adjusts the CD gain using a constant multiplier and Pointer Acceleration (PA) where CD gain is dynamically adjusted using a function based on motor-space movement characteristics.
%\co{Are CD and CG both novel techniques proposed by Casiez? Would like to know to organize this paragraph better.} \amanat{not CD, PA is novel while constant gain (CG) has been previously used by others.}
%They found that both the choice of CD gain level and the technique (CG vs. PA) can impact pointing performance.
%PA was found to be faster compared to CG only when their CD gain levels were not precisely optimized.
%Kim et al. \cite{kim2019motion} created a new position-to-position CD mapping technique, named motion-display(MD) gain by connecting motor space (e.g., user's hand movements) and visual space (display) based on the empirical data collected. This technique reflects natural human pointing gestures for distal pointing at large displays. They used a direct linear transformation (DLT) method, considering user's stature and eye height for mapping. Participants were asked to do pointing task at different distances and target locations. They found that MD gain compared to existing techniques like Laser pointing, showed superiority in performance and subjective satisfaction.
%Konig et al. \cite{konig2009adaptive} introduced a new technique called Adaptive pointing that dynamically adjusts the Control-Display (CD) gain depending on the movement velocity and the current offset between the motor-space position and display-space position.
%This enhanced pointing performance without disrupting the user's understanding of how absolute devices operate.
%The technique aimed to maintain the user's perception of a 1:1 mapping between device movement and pointer movement, while subtly adjusting the CD-gain for higher precision when needed.
%They found that Adaptive Pointing showed a significant improvement over traditional absolute pointing, with a 19\% improvement in movement time and a 63\% reduction in error rate.
%Participants also rated the Adaptive Pointing technique significantly better than traditional absolute pointing
%Casiez et al. \cite{casiez2011no} compared the default transfer functions used by Windows, OS X, and Xorg (the X.Org Foundation server) in their experiment.
%They also added a constant Control-Display (CD) gain function for comparison purposes. These default transfer functions were chosen because they are widely used. These default transfer functions were compared to the CD gain function.
%They found that default transfer functions improved performance by up to 24\% compared to a constant CD gain function. There were also significant differences found between the functions, with the one from OS X improving performance for small target widths but reducing performance by up to 9\% for larger ones compared to Windows and Xorg.
%Nancel et al. \cite{NancelCHI2013} investigated five high-precision pointing techniques (Conthead, LaserGyro, RelaLarge, RelaSmall ad SmoothPoint) on targets as small as 4 millimeters on a 5.5 meters wide display.
%The authors argued that portable devices can accommodate both high-precision remote pointing and object manipulation simultaneously.
%They compared their high-precision pointing techniques to state-of-the-art distant pointing techniques.
%The RelaSmall and RelaLarge were motor-space-based adaptive pointer techniques that adjusted CD gain based on the acceleration of the finger movement on a touchpad.
%These two adaptive techniques match or perform better than the other three state-of-the-art pointing techniques.
%Head-orientation-based, and relative pointing outperformed pointing-only technique performance, and required significantly less effort. \co{Could you define "interaction resources"? I assume it's tied to focus and effort, but I would like a confirmation.}
%\co{Also, what is the "purely relative" technique mentioned in commented sentenced below?} \amanat{RelaSmall uses a small action or touch-pad space which was referred to as limited resources.}
%Two of their techniques, a purely relative one and one that uses head orientation, perform as well or better than the best pointing-only input techniques while using a fraction of the interaction resources.
To the best of our knowledge, no prior work has investigated cursor enhancement on large curved displays. %\co{references?} \amanat{changed to no work so no references.}
We argue relying on performance and guidelines for large flat displays to be insufficient due to the unique design space defined by large curved displays: for instance, changes in field-of-views as users move away from the display's centre, and non-planar geometry.
We additionally note that although user position has been found to impact interaction on curved displays \cite{zannoli2017perceptual, Ullah2023Pointing}, no cursor enhancement techniques for large displays have been designed and evaluated considering user position.
In this work, we conduct an investigation of six techniques to support pointing on large curved displays inspired by existing motor-space and visual-space techniques.
%\co{Why does this matter? Reiterate the value of our work by stating the main gap that we are addressing. Novelty is not enough.} \amanat{i added that it has major influence, now it should be good.}
to\section{User Study 1 - Pointing Performance across User Positions}
Pointing performance on flat displays has been shown to vary based on the user's position with respect to the display \cite{kopper2010human,janzen2016modeling,kovacs2008perceptual,tao2021freehand, hourcade2012small}.
The effects on large curved displays have also been studied, and only the the distance of the user from the display, and not any lateral displacements was shown to impact performance \cite{Ullah2023Pointing}.
However, the task (1D pointing with reciprocal targets at fixed amplitudes) may have been too simple, and limited the impacts of user position on performance.
In order to design cursor enhancement techniques that best support users on large curved displays, we see a need to first verify the effects of user position on pointing performance.
%However, this has not been demonstrated for large curved displays. \co{Not up to date due to \cite{Ullah2023Pointing}.}
%Furthermore, this performance change may not generalize to curved displays from off-centered positions due to their unique geometry,
%Therefore, the effect of different positions on flat displays have not been tested on curved displays which have unique challenges of position the user at different \textit{Interaction Distance} and \textit{Lateral Offsets} . \co{Change this previous sentence with motivation used in intro (beyond novelty), e.g. "...which have unique challenges of X".} \amanat{added more info.}
%complicating designing relevant pointing augmentations for such displays. \co{Revise previous sentence after motivation is fixed (Intro and rel.) One option would be to say: We wish to validate the results of \cite{Ullah2023Pointing} because they do not match our anecdotal experiences using large curved displays. We posit the results may be caused by (?????) which we rectify here.} %as the impact of user position on pointing performance has not been quantified.
We therefore conducted a within-subject user study to evaluate how the user's position, defined here as a 2D grid around the display (see Fig. \ref{fig:teaser}c), affects their pointing performance on a large curved display.
Specifically, we employed a 2D Fitts' Law selection task to investigate the effects of two factors on pointing performance: 1) the user's \textit{Interaction Distance} (i.e., their perpendicular distance to the centre of the large curved display), and 2) the user's \textit{Lateral Offset} from the centre of the display.
%We also consider the effects of selection target distances (\textit{Amplitude)}) and \textit{Width} for completeness.
Formally, we formulate the following Hypotheses around the two position-based factors:
\begin{itemize}
\item \textbf{H1}) Pointing speed will increase as the \textit{Interaction Distance} increases, as users will require smaller movements to cover large distances.
\item \textbf{H2}) Pointing performance (speed and accuracy) will be optimal when the user is aligned with the display's centre, compared to when their position includes a \textit{Lateral Offset}. %as both sides of the display are uniformly distant from the user.
\item \textbf{H3)} Pointing accuracy will decrease as \textit{Interaction Distance} increases, as moving farther away amplifies hand jitter \cite{kopper2011understanding}.
%\amanat{because we added this reference the reviewers said that our project does not have any new insights as these have been observed before. I feel we should remove the citation and only state the reason.} \co{Hard disagree. We just need to make it clear we are testing hypotheses in a different context (where prior findings may not extend).}
%\co{Can we cite a reference supporting this justification? I think we use one in the intro.} \amanat{added}%\co{we don't actually collect data on hand jitter, right? Wonder if we should remove the justification.} \amanat{in the discussion after the results i have cited previous papers who have found accuracy to reduce because of hand jitter. There we explain it. and no we do not have data on hand jitter. we have data on endpiont deviation. but both lateral offset and interaction distance have significant effects on endpoint devitation. So we can not justify that way as lateral offset has no effect on accuracy.}
%xxcomebackxx
%\amanat{DATA}
\end{itemize}
%\subsection{ Amanat}
%\update{
%Ullah et al. \cite{Ullah2023Pointing} performed a a comparative study between a virtual and physical curved display. In their study, they considered two factors: \textit{Interaction Distance} and \textit{Lateral Offset} for different positions. However, they did not find any significant effect of \textit{Lateral Offset} on pointing performance.
%We believe this was due to their task design where they investigated a reciprocal 1D pointing task where the participants had to perform repetitive motor movement without conscious effort as the target positions for the two targets with a certain amplitude is fixed. Therefore, in our study, we follow the same study design for a different pointing task: pointing at 2D targets where targets can appear at any position on the display while maintaining the particular amplitude condition. In our task design, the participants cannot take advantage of muscle memory as the target positions are not fixed and can appear at any location on the screen.}
\subsection{Factors}
This study focuses on the effects of two factors: \textit{Interaction Distance} and \textit{Lateral Offset} (see Fig. \ref{fig:teaser}c).
\textit{Interaction Distance} represents how far the user is in front of the display when considering the display's \textit{main axis}, i.e., the axis passing through the curvature center and the middle of the display.
\textit{Lateral Offset} corresponds to the user's lateral displacement (left or right) relative to the main axis.
Furthermore, as we use a 2D Fitts' Law selection task, we also consider the effects of the selection target's \textit{Amplitude} and \textit{Width} on pointing performance.
\textbf{Interaction Distance}: We consider three distances relative to the radius of the display's curvature ($R = 3.27m$): \(\frac{1}{2}\)R (1.64m), R (3.27m), and \(\frac{3}{2}\)R (4.91m).
Using multiples of $R$ allows our results to generalize beyond the specific large curved display used here.
\textbf{Lateral Offset}: Since target position on axis-symmetric directions has no difference in pointing performance \cite{zhang2012extending,radwin1990method}, we consider two offsets: No-offset where a participant is positioned along the \textit{main axis}, and Left-offset where the participant is positioned \(\frac{1}{2}\)R (1.64m) to the left of the \textit{main axis}.
We further limit recruitment to strictly right-handed participants to control for potential differences due to handedness and studying only left offsets.
User position is thus examined through six ($2\times3$) distinct positions in front of the display (see Fig. \ref{fig:teaser}c). %\kh{Amanat: check the ISMAR journal and add more information here}
\textbf{Target Amplitudes $\And$ Width}: We evaluate pointing performance through three target amplitudes (i.e. distance between the targets): 250cm, 500cm, and 750cm.
Additionally, we consider two target widths (i.e., the diameter of the spherical targets): 20cm, and 70cm.
This results in six distinct Indexes of Difficulty (IDs) ranging from 2.85 to 6.25 bits. % \kh{Amanat: check the ISMAR journal and add more information here}.
The specific levels were selected based on the arc length of the large curved display along its surface, and fine-tuned through a pilot study.
These conditions mimic real-world situations where interface elements might appear anywhere on the curved screen, independently of the user's current position.
\subsection{Apparatus}
We conducted our study on a large physical curved display produced by Mechdyne \cite{mechdyne-new1}.
The display is semi-circular ($180^\circ$ viewing angle), 3m tall, and has a radius of 3.27m.
%\update{(which we refer as 3270R).} %
\update{Here 3270R refers to a screen having a radius of 3270 millimeters or 3.27 meters.}
The display consists of forty 46-inch LED 3D displays set in a 4x10 grid (see Fig. \ref{fig:teaser}a-c).
%\co{@Amanat, is this paragraph correct? Or was the implementation something different? Wondering which comments here are still relevant.} \amanat{yes the appraturs is correct}
We implemented the pointing task using the Unity game engine \cite{unity}. %\kh{this is confusing - where does the Oculus Quest 2 come all of a sudden - even why do we need it} \amanat{removed the Quest 2 controller}
%Participants used a controller for pointing during the experiment.
%\kh{why Quest 2 controller? what was the purpose of that? Also, how it was tracked?}
A ray, corresponding to the user's current pointing direction, was rendered from a Quest 2 controller held by the user to point at the targets (see Fig. \ref{fig:teaser}a-b). %\kh{Figure missing}. \amanat{added}
The ray extended from the controller in the orthogonal direction and a cursor was placed at the endpoint of the ray on the large curved display (Figure \ref{fig:teaser}a-b).
Target selection was done by pressing the controller's trigger button.
Markers were attached to the controller to track its position via an 8-camera Optitrack system \cite{optitrack}.
%For real-time data logging during the study, we used Firebase \cite{google}.
%\remove{Figure 1a visually represent the display.} \amanat{this sentence looks out of place}
%\subsection{Design}
%\co{I'm not convinced we need this section as-is. We are repeating a lot of information that is already in 3.1. Could we instead move this to the start of the section?}
%We used a within-subject design with four factors: \textit{Interaction Distance} (\(\frac{1}{2}\)R, R and \(\frac{3}{2}\)R), \textit{Lateral Offset} (No-offset and Left-offset), \textit{Amplitude} (250cm, 500cm, and 750cm) and \textit{Width} (20cm, 70cm).
%This results in six distinct user positions (two \textit{Interaction Distance} $\times$ 3 \textit{lateral offsets}). %, as well as six IDs (3 \textit{Amplitudes} $\times$ 2 \textit{Widths}). %The target Amplitude and Width combination led to six IDs.
%The \textit{Interaction Distances} and \textit{Offset} led to six different user positions.
%Participants each completed 360 trials (3 \textit{Interaction Distance} $\times$ 2 \textit{Offset} $\times$ 2 \textit{Widths} $\times$ 3 \textit{Amplitude} $\times$ 10 repetitions). % for a total of 4320 trials.
\subsection{Participants and Procedure}
\label{sec:us1:proc}
% \kh{Amanat: this section is similar to what we did for ISMAR paper, please check the paper and update information here}
The experiment was conducted at a North American post-secondary institution, where the large curved display is located.
Ethics approval was granted by the institution.
We invited 12 right-handed participants (6 males and 6 females) from a local university to participate.
They were, on average, 23.25 years old ($SD = 6.24$).
Only two participants had prior experience interacting with large wall-sized displays.
Each participant was compensated with a \$15 honorarium for their time.
Participants first completed a demographics questionnaire focusing on their prior experience using wall-sized displays.
The pointing task was then described.
The study was separated into six blocks, one per user position tested.
Each block began with 15 practice trials to help participants familiarize themselves with the study conditions.
Participants then completed 60 successive trials (six task difficulties - i.e., IDs - repeated 10 times) before taking a two-minute break and moving to the next position.
Each block ended with participants completing a NASA-TLX questionnaire for the current position.
%This process was repeated for each user positions.
User positions were counterbalanced to reduce ordering effects.
Participants each completed 360 trials (3 \textit{Interaction Distances} $\times$ 2 \textit{Lateral Offsets} $\times$ 2 \textit{Widths} $\times$ 3 \textit{Amplitude} $\times$ 10 repetitions).
The study lasted approximately 60 minutes on average.
The \textbf{task} consisted of a 2D Fitts' Law selection task using a hand-held controller.
%The start target's location was randomized.
%For a specific amplitude condition, a random direction was chosen to place the target.
%\co{Were the target positions random? We don't seem to mention it here.} \amanat{these last two sentences answers your questions.}
Participants used the controller to select targets by positioning the cursor (2.5 cm diameter) within the boundary of each target.
Selections were invoked by pressing the controller's trigger button.
Each trial began with participants selecting a white start circle (20 cm radius) rendered at a random location on the display.
Upon selection, the start circle would disappear, and a red circular target would appear on the display with a specific \textit{Width} and \textit{Amplitude} based on the trial's ID.
After selecting the red circle, the trial would end, and the system would present a new start circle.
Participants were instructed to select the targets as quickly and as accurately as possible.
Selections made outside the target boundary were flagged as errors.
\subsection{Measurements}
We recorded movement time and success rate for each trial, i.e., how long it took a participant to select the red target after selecting the white start circle and if they successfully selected the target.
Any selection outside the target's boundary was considered an error.
Additionally, we recorded participants' subjective ratings about the task for every tested position through the NASA-TLX \cite{nasatlx} questionnaire.
%Demographic information was also collected via a questionnaire distributed at the start of the session.
%\co{State what information was collected, or let's just remove this sentence since it's repeating 3.3.} \amanat{removed}
\subsection{Results}
%\co{General comments for this section: we report different p-values throughout the section, e.g. $p<.01$, $p<.001$, $p<.0001$. Can we maybe use the same notation throughout, or does it actually matter to keep them separate? My understanding is the effect is not stronger just because p-val is smaller.
%} \amanat{So the format that we do not report effect size for pairwise comparisions. That what me khalad and william did for our previous papers.}
We analyze movement time with repeated measures ANOVA on log-transformed data and post-hoc pairwise comparisons with Bonferroni corrections. %\co{include reference for use of log-transform, if possible?} \amanat{if the reviewers ask for it let us do then. Willaim said it is fairly standard and does not need a reference}
We use Wilcoxon Signed-Rank test for the error rate data.
For all tests, we report Greenhouse-Geisser corrected p-values and degrees of freedom.
All reported confidence intervals (CIs) are 95\%.
\begin{figure*}[h]
\centering
\includegraphics[width=1\columnwidth]{figures/updated_selection_study1_final.png}
\caption{
(a) Movement Time by \textit{Interaction Distance} for each target \textit{Amplitude};
(b) Movement time by target \textit{Width} for each \textit{Lateral Offset}; (c) Fitts' Law regression lines for Movement Time by \textbf{Fitts ID} for each \textit{Interaction Distance} and (d) Accuracy by \textit{Interaction Distance} for each target \textit{Width}.
%\co{Figure d has incorrect x-axis title (should be interaction distance?).} \amanat{correct. fixed it.}
%\co{Could include symbols to indicate which pairs are significantly different.}
} %\amanat{as there are main effects and also pairwise comparisions it will be difficult to illustrate in the figure. lets keep it as it is.}
\label{fig:exp1-res} % make sure you reference this figure in the task text
\end{figure*}
\subsubsection{Movement Time}
%main effect
We found significant main effects of \textit{Amplitude} ($F_{2,22}=153.58$, $p < .0001$, $\eta^2$ = 0.93), \textit{Width} ($F_{1, 11}=250.92$, $p < .0001$, $\eta^2$ = 0.96), \textit{Interaction Distance} ($F_{2,22}=34.95$, $p< .0001$, $\eta^2$ = 0.76) and \textit{Lateral Offset} ($F_{1,11}=6.92$, $p< .05$, $\eta^2$ = 0.39) on movement time (see Fig. \ref{fig:exp1-res}b-c).
The mean movement time is 1.52s (CI: [1.49, 1.55]) for \textit{Interaction Distance} \(\frac{1}{2}\)R, 1.26s (CI: [1.23, 1.28]) for R and 1.16s (CI: [1.14, 1.18]) for \(\frac{3}{2}\)R. %\co{Mean values are outside the CIs.} \amanat{fixed}
Post-hoc pairwise comparisons reveal that pointing at \(\frac{3}{2}\)R is significantly faster than \(\frac{1}{2}\)R and R (all $p<.0001$).
In addition, pointing at R is significantly faster than \(\frac{1}{2}\)R ($p<.0001$).
%For \textit{Lateral Offset} ($F_{1,11}=0.69$, $p=0.42$), we did not observe any significant effect on movement time.
Furthermore, pointing from the No-offset (mean: 1.29, CI: [1.24, 1.34]) position is significantly faster than pointing from the Left-offset (mean: 1.36, CI: [1.30, 1.42]) position.
%With 4 factors, we found several interaction effects.
Finally, we found a significant interaction effect of \textit{Amplitude} $\times$ \textit{Interaction Distance} ($F_{4,44}=11.45$, $p<.0001$, $\eta^2 = 0.51$) (see Fig. \ref{fig:exp1-res}c) and \textit{Width} $\times$ \textit{Interaction Distance} ($F_{2,22}=10.35$, $p<.001$, $\eta^2 = 0.51$).
%For each \textit{Amplitude}, pairwise comparisons revealed that pointing at 10cm targets had a higher movement time compared to 70cm targets (all $p<.001$). For each \textit{Amplitude}, movement time significantly increased with the increase of user \textit{Interaction Distance} for their pairwise comparisons (all $p<.001$)). %\co{This does not sound like an interaction effect. Is significance maybe due to $R/2$ movement time growing faster than the other distances? Are the pairwise differences more interesting if we group bars by amplitudes instead of distance?} \amanat{It should be good now}
%For each \textit{Amplitude}, pairwise comparisons of No-offset and Left-offset positions did not reveal any significant effect on movement time. %\co{Do you mean there's no significant pairwise differences? Because we said there was a significant interaction effect.} \amanat{yes rephrased.}
%While offset did not have an effect on selection times, for Width = 10cm, the selection was faster in the center (mean: 1.76s, CI:[1.72,1.8]) compared to the left position (mean: 1.82s, CI:[1.77,1.87]) (p<0.05).
%\(\frac{1}{2}\)R, R, and \(\frac{3}{2}\)R
\subsubsection{Accuracy}
% 2 template for the width with only two levels, you should not have X^2, but W=..., Z=...., p<..., r=...
% 3 template There is a significant effect of Width on accuracy [X^2(...)=..., p < ....], with 0.1m targets (M=..., [..., ...]) leading to less accurate results than 0.7m (M=..., [..., ...])
We found a significant main effect of \textit{Interaction Distance} on pointing accuracy ($\chi^2 (2)=20.17, p<.001$). %\co{missing effect size?} \amanat{for factors having three values the effect size is stated in the pair wise comparisons.}
Post-hoc pairwise comparisons revealed that pointing from \(\frac{1}{2}\)R (M = 0.95, [0.93, 0.96]) is significantly more accurate than R (M = 0.90, CI: [0.87, 0.93]) and \(\frac{3}{2}\)R (M = 0.84, CI: [0.8, 0.88]) (see Fig. \ref{fig:exp1-res}d).
Furthermore, pointing from R is significantly more accurate than \(\frac{3}{2}\)R.
We also found a significant main effect of \textit{Width} ($W=0, Z = -3.06, p<.001,r=-0.62$), with 0.1m-sized targets (M = 1.6, CI: [1.56, 1.64]) leading to less accurate results than 0.7m-sized ones (M = 1.04, CI: [1.01, 1.07]) (see Fig. \ref{fig:exp1-res}d).
Finally, we did not find a significant effect of \textit{Lateral Offset} ($p=0.62$) and \textit{Amplitude} ($p=0.56$) on accuracy.
% \begin{figure}[h]
% \centering
% \includegraphics[width=.66\columnwidth]{figures/s1-fitts-draft.png}
% \caption{Fitts' Law regression lines for Movement Time by \textbf{Fitts ID} for each (a) \textit{Lateral Offset} and (b) \textit{Interaction Distance}.
% \co{Should there be error bars for each sample?} \amanat{no in Fitts law we do not do that}}
% \label{fig:exp1-fitts} % make sure you reference this figure in the task text
% \end{figure}
%combined this image with previous image
\subsubsection{Fitts' Law Regression Lines}
We found that Fitts' Law can accurately model movement time for both \textit{Lateral Offsets} ($R^2\ge0.99$ for all).
Interestingly, movement time predictions for \textit{Lateral Offsets} have approximately the same slope, indicating that both no-offset ($y=0.21x+0.41$) and left-offset ($y=0.22x+0.42$) positions exhibit the same tolerance to an increase in task difficulty.
%; See Fig. \ref{fig:exp1-res}a).
%\co{where did the figure go for this?} \amanat{that results did not have significant so i removed it. also prof said to keep less focus on Fitts law. So we have now have one instead of 2.}
%We found that the Fitts law could effectively be applied to both the No-offset ($R^2 = 0.99$) and the Left ($R^2=0.99$) offsets (Figure \ref{fig:exp2-movementTime}c).
Additionally, we observed that Fitts' Law can accurately model movement time with varying \textit{Interaction Distances} ($all R^2\ge0.98$; See Fig. \ref{fig:exp1-res}a). % (regression equations were \(\frac{3}{2}\)R: $y=0.19x+0.39$; R: $y=0.21x+0.40$; \(\frac{1}{2}\)R: $y=0.23x+0.49$). \co{equations are already given in figure}
%From the Fitts law regression lines (Figure \ref{fig:exp2-movementTime}c) we saw similar movement time while pointing from the left offset compared to the center location.
Movement time is found to have a linear relationship with \textit{Interaction Distance} (see Fig. \ref{fig:exp1-res}a and \ref{fig:exp1-res}b), where larger distances result in faster selections when compared to shorter ones, e.g., \(\frac{3}{2}\)R leading to faster selections than R. %\co{this is not the correct figure?} \amanat{Fixed it}
%\william{It seems odd to not speak about VR versus physical anymore. Perhaps we can have 4 lines on the graphs, to show that any context (vr/physical + center/offset) is actually the same.}
%\amanat{Dr. Khalad also agreed. Figure 5(c) and the text is updated}
\begin{figure}[htbp]
\centering
\includegraphics[width=1\columnwidth]{figures/s1-nasatlx-new.png}
\caption{Subjective feedback score: (a) for each \textit{Interaction Distance}, and (b) for each \textit{Lateral Offset} (error bars represent are 95\% confidence intervals).
%\co{Incorrect legend for a? Caption mention interaction distance, legend mentions amplitude.} \amanat{you have a nice eye. Corrected}
}
\label{fig:exp1-nasatlx} % make sure you reference this figure in the task text
\end{figure}
\subsubsection{Preference Scores}
We collected users' feedback on the seven NASA-TLX \cite{nasatlx} criteria along with their preferences and perceived fatigue for each \textit{Interaction Distance} and \textit{Lateral Offset} condition. \update{We calculated the overall workload using raw NATA-TLX values. For enhanced readability, we inverted the performance scores such that a higher score indicates better performance.}
%\co{aren't there six standard criteria and not seven?} \amanat{six + 1 (overall) makes seven. we can include the figures of only the size. let me try to fit all the images in one row. then we can have two rows. one for interaction distance and one for offset.} \co{Is this standard? Otherwise we need to mention it. I havent run into it before personally.} \amanat{it is standard in some papers, I have published two papers following this.}
For \textit{Interaction Distances} \update{(see Figure \ref{fig:exp1-nasatlx}a)}, we found only significant differences for Preference ($\chi^2 (2,N=12)=8.21, p<.05$), Fatigue ($\chi^2 (2,N=12)=10.71, p<.01$), Mental Demand ($\chi^2 (2,N=12)=16.33, p<.001$), Physical Demand ($\chi^2 (2,N=12)=9.91, p<.01$), Performance ($\chi^2 (2,N=12)=6.44, p<.05$), and Frustration ($\chi^2 (1,N=12)=7, p<.05$).
%We did not find significant effects on Temporal Demand ($\chi^2 (2,N=12)=4.07, p=.13$) , Effort ($\chi^2 (2,N=12)=5.16, p=.08$), and Overall task load ($\chi^2 (2,N=12)=5.20, p=.07$).
Pairwise comparisons revealed that \(\frac{3}{2}\)R has higher Preference scores than \(\frac{1}{2}\)R.
Additionally, \(\frac{1}{2}\)R has higher Fatigue, Physical Demand, and Mental Demand, as well as higher than both \(\frac{3}{2}\)R and R.
For \textit{Lateral Offsets} \update{(see Figure \ref{fig:exp1-nasatlx}b)}, we found participants significantly prefer being positioned in No-offset positions compared to Left-offset for three criteria: Preference ($\chi^2 (1,N=12)=5.33, p<.05$), Fatigue ($\chi^2 (1,N=12)=6.40, p<.05$), and Mental Demand ($\chi^2 (1,N=12)=4.50, p<.05$)]. %\co{missing effect sizes.} \amanat{i followed the prof's way same as my previous papers. as we have deadline on 15 Dec lets keep it as it is.}
We found no significant differences across all other dimensions.
%The difference between the other subjective feedback were not statistically significant [Physical Demand ($\chi^2 (1,N=20)=1, p=0.32$), Temporal Demand ($\chi^2 (1,N=20)=2.67, p=0.10$), Performance ($\chi^2 (1,N=20)=2.27, p=0.13$), Effort ($\chi^2 (1,N=20)=2.77, p=0.09$), Frustration ($\chi^2 (1,N=20)=2, p=0.16$), and Overall task load ($\chi^2 (1,N=20)=1.29, p=0.26$)].
\subsection{Discussion}
\subsubsection{Effects of Interaction Distance}
We found movement time to decrease linearly as the user moves farther away from the display (see Fig. \ref{fig:exp1-res}c), resulting in faster pointing speeds.
\textbf{We thus accept H1.}
%Our H1 is valid as movement time lowers as the user is placed farther away.
%We observed a linear relationship between interaction distance and movement time: users perform faster selection with the increase of \textit{Interaction Distance} (see Figure \ref{fig:exp1-res}b).
We additionally found pointing accuracy to decrease as the user moved away from the display's centre (i.e., larger \textit{Interaction Distances}): 5\% error rate at \(\frac{1}{2}\)R, 10\% at R and 16\% at \(\frac{3}{2}\)R (see Figure \ref{fig:exp1-res}d). %\amanat{have to remove current figure and change figure for Accuracy vs Interaction Distance.}.
This result is expected as pointing at a distance introduces jitter from hand tremors \cite{kopper2011understanding}. %\co{reference?} \amanat{added}
The jitter is amplified with increases in \textit{Interaction Distances}, as the same angular jitter results in larger cursor displacements the further away the user, leading to larger error rates.
%Increased jitter leads to significantly higher error rate as the \textit{Interaction Distance} increases.%\co{missing ending to sentence} \amanat{added sentence ending}
\textbf{We therefore accept H3.}
The error rates based on \textit{Interaction Distance} further varied based on the target widths: smaller targets (20cm) at \(\frac{3}{2}\)R, R and \(\frac{1}{2}\)R have error rates of 28\%, 16\% and 10\%, while large targets (75cm) have error rates of 4\%, 3\% and 1\% respectively.
In summary, increased distances from the display correlates with quicker selection times but also higher error rates.
We envision cursor enhancement techniques could reduce cursor speed at large \textit{Interaction Distances}, and increase it at short distances to provide a more uniform experience to users.
%\update{However, this relationship has to be tested on displays with different curvatures, to confirm its validity.}
\subsubsection{Effects of Lateral Offset}
Contrary to prior work \cite{Ullah2023Pointing}, we observed participants were significantly faster at selecting targets from the no-offset positions compared to the left-offset positions. %\co{No interaction effects between distance and offset, right?} \amanat{right. p=0.46}
Furthermore, participants significantly preferred being positioned in No-offset positions compared to Left-offset based on Preference, Fatigue, and Mental Demand.
%\co{This argument is unclear to me. Why would we think jitter would vary based on lateral offset in the first place? And why does the absence of jitter mean that lateral offset does not affect performance?} \amanat{NEED-to-talk}
\textbf{We thus accept H2.}
%xxcomebackxx
\section{User Study 2 - Pointing Augmentations for Large Curved Displays}
We confirmed in our first study the effects of \textit{Interaction Distance} on pointing performance on large curved displays, i.e., both movement time ($p < .0001$, with a large effect size) and accuracy ($p < .001$, with a large effect size).
Furthermore, we observed \textit{Lateral Offset} has a medium-sized effect on pointing speed ($p < .05$).
%\co{@AMANAT: Could you please update this with the latest study 1 results?} \amanat{added.}
%\co{I dont understand how you took the average effect size. Could you please clarify? Also, do the numbers for each effect size indicate that these are "strong"/"large" effect sizes, and not "medium" or "small"?}
Ideally, users should be highly accurate and quick while pointing regardless of their position.
We thus see a need to design cursor-enhancement techniques for large curved displays to improve pointing performance from more difficult positions.
We thus conducted a within-subject user study to evaluate which of six motor- and visual-space-based enhancements best afford users with optimal pointing performance across positions around a large curved display. %\co{we're once again discussing consistency. Modify argument is not relevant anymore.} \amanat{now its is about speed and accuracy instead of consistency}
%\kh{the next paragraph should discuss how we can do motor space enhancement and display space enhancement}
%\co{Missing definitions of motor- and visual-space}
%\amanat{I have defined it below.}
%Although dynamic CD gains have been explored for flat displays \cite{?} \amanat{@DHRUV add citation here}, cursor enhancement techniques have not been explored yet for curved displays. \co{Add sentence to make paragraph flow better if needed.}
%We explore six pointing enhancement techniques along two dimensions: motor-space enhancement, and visual-space enhancement. \co{The rest of this paragraph could be moved to 4.1 if we remove the technique names from the hypotheses.}
We define motor-space enhancements as techniques that modify cursor speed (i.e., Control Display (CD) gain) to facilitate moving towards a target, e.g. based on the controller's movement or user distance from the large display.
On the other hand, visual-space enhancements modify the visual aspects of the cursor (e.g., its size) to facilitate target selection.
In this study, we consider three motor-space enhancement techniques:
1) "Pointer Acceleration" (PA) cursor, an acceleration-based technique \cite{nancel2013high};
%\co{"?" ref}
2) "Position Based Acceleration" (PBA) cursor, where CD gain varies based on \textit{Interaction Distance}; %\co{how does this abbreviation work? What's the "B" and "A" for?} \amanat{now its position based acceleration, we changed it as PDIST and PADIST was very hard to distinguish. I now changed it throughout}
and 3) "Pointer Acceleration and Distance-based" (PADIST), combining PA and PBA.
%(PADIST) cursor, a combination of PA and PBA.
We further consider their visual-space-augmented variants, where the cursor size varies based on motor-space movement: 4) PASIZE; 5) PBASIZE; and 6) PADISTSIZE. %\co{can we modify "controller movement" to "motor-space movement"? We defined CD gain as such, so we could re-use the name here and keep our methodology independent of the method used to move the cursor (e.g., hand movement, controller movement, etc.)} \amanat{I agree it has now been made consistent across.}
Formally, this study consists of a 2D Fitts' Law selection task and involves three factors from the first study: \textit{Interaction Distances} (\(\frac{1}{2}\)R, R, and \(\frac{1}{2}\)R), \textit{Lateral Offsets} (No-offset and Left-offset), as well as \textit{Target Amplitude} (250cm, and 750cm).
We further examine each of the six pointing enhancement techniques with respect to these factors.
%\kh{List the research questions here}
We formulate the following Hypotheses regarding the pointing enhancement techniques:
\begin{itemize}
\item \textbf{H1}) Pointing performance (movement time and accuracy) will be superior for visual-space-based enhancement techniques (i.e., PBASIZE, PASIZE, PADISTSIZE) compared to techniques without visual enhancements (i.e., PBA, PA, PADIST).
\item \textbf{H2}) Techniques combining acceleration and \textit{Interaction Distance} will lead to superior pointing performance (movement time and accuracy) across user positions, compared to techniques that use only \textit{Interaction Distance} or acceleration.
\end{itemize}
%\subsection{Dhruv}
%\update{
%Unlike flat displays, where cursor movement is based solely on two-dimensional transformations \cite{VogelUIST2005,McCallumUIST2009,ForlinesUIST2006,GalloIEEE2010} , curved displays require a more complex approach to accurately track user interactions.
%
%
%our solution involves the use of a virtual cube placed at the center of the interaction space.
%This cube's orientation and position are directly controlled by the user's movements, enabling a more intuitive cursor control.
%As the user rotates their controller, the cube mirrors this rotation, effectively translating these movements into cursor positioning on the curved display.
%Additionally, as users change their physical position, the cube also move to this new location.
%
%To enhance the precision and speed of cursor movement on curved displays, our approach dynamically adjusts the rotation speeds of a virtual cube in response to the user's controller actions, guided by the principles outlined in equations \ref{eq:CDgain-formula} and \ref{eq:find-v-inf-formula}.
%This adjustment is based on a direct correlation between the degree of rotation applied to the controller and the corresponding movement of the virtual cube within the interaction space.
%Specifically, for more precise cursor positioning, a moderate rotation of the controller, such as 10 degrees, results in a relatively smaller movement of the virtual cube, say 5 degrees.
%This ensures fine control over the cursor for detailed tasks.
%
%Conversely, to support rapid cursor navigation across the curved display, the same 10-degree rotation of the controller can be amplified to produce a more significant 20-degree movement of the virtual cube.
%This scalable response mechanism allows users to intuitively control the cursor's movement, seamlessly transitioning between precise adjustments and broad sweeps across the display's curvature. }
\subsection{Techniques}
Numerous techniques have been proposed to enhance pointer speed and acceleration \cite{casiez2008impact, nancel2013high}.
Such techniques typically involve modifying the Control Display (CD) gain or cursor speed, representing the ratio between cursor movement and the input-device movement.
%\co{Need to add sentence relating techniques (or ways) to the next sentence's equations. Otherwise, readers may get confused.}
%\amanat{cite more} \co{"?" ref}
For our techniques, we choose Nancel et al.'s \cite{nancel2013high} equation to change CD gain and cursor size based on motor-space movements as the formula allows mapping an output value (e.g., cursor speed and cursor size) based on different motor-space changes (e.g., controller movement and \textit{Interaction Distance}).
%\update{As we see a stronger effect of \textit{Interaction Distance} compared to \textit{lateral offset}, hence we leverage \textit{Interaction Distance} to dynamically adjust the cursor speed.}
%\co{We need to motivate why we do not consider lateral offset here.} \amanat{added one sentence}
In our study, we consider users' motor-space movement in the 3D space and the \textit{Interaction Distance} of the user as input to generate the CD gain and the current cursor size as outputs.
%, which leads to six different techniques (see Table \ref{tab:cursors}).
We evaluate pointing performance using six cursor enhancement techniques combining visual-space and motor-space enhancements (see Table \ref{tab:cursors}).
Each technique modifies the cursor speed based on the user's \textit{Interaction Distance}, or motor-space movement.
We focus exclusively on \textit{Interaction Distance} and do not consider \textit{Lateral Offset}, as we found the former to have much larger effects on performance and we wish to control the complexity of our study.
Visual-space-augmented variants further modify the cursor's size from a 2.5cm diameter to a 20cm diameter based on controller movement.
%We introduce the equations defining the set of techniques and summarizing the pros and cons of each technique.
%We discuss below how we define each technique.
%We explore 3 motor-space techniques: Pointer Acceleration (PA), PBA (Pointer Distance-based), PADIST (Pointer Acceleration Distance-based) We add visual-space enhancement to these motor-space techniques making the total to 6 techniques.
%By visual-space enhancement we consider making the cursor larger as it approaches the target \cite{GrossmanCHI2005}.
%The cursor size was fixed to 2.5cm diameter for PA, PBA and PADIST and varied between 2.5cm and 20cm range for the cursor enlargement techniques (PASIZE, PBASIZE and PADISTSIZE).
%Furthermore, the cursor did not have a 1-to-1 mapping with the controller; rather the speed of the cursor (i.e., CD gain) varied based on the technique used.
%\co{These two sentences should fit somewhere in section 4.1 (taken from procedure).} \amanat{I feel this is a good position for the two sentences above.}
We define each motor-space enhancement using Nancel et al.'s \cite{nancel2013high} CD-gain equations, which result in cursor speeds increasing along with motor-space speed (e.g., slow motor-space movements will result in slower cursors affording more precise pointing):
\begin{equation} \label{eq:CDgain-formula}
CD(x)=\frac{CD_{\max }-CD_{\min }}{1+e^{-\lambda\left(x - V_{\mathrm{inf}}\right)}} +CD_{\min },
\end{equation}
where CD(x) is the augmented speed of the cursor based on the controller's speed $x$.
$CD_{\max}$ and $CD_{\min}$ correspond to the maximum and minimum speed the cursor can reach. %, which we set to 1.1 and 0.9 based on a pilot study.
%The value for the $CD_{\max}$ is chosen to be 1.1 and the value for $CD_{\min}$ is chosen to be 0.9 based on pilot studies.
Finally, $\lambda$ is a non-negative value defining the slope of the equation at its inflection point $V_{\text {inf }}$.
In our study, we set the value of $\lambda$ to 20 based on the results of pilot studies. %$\epsilon$ [0; $\infty$] defines the slope of the equation at its inflection point ($\lambda$ = 0 yields a constant function, $\lambda$ = $\infty$ a step function). Changing $\lambda$ smooths the curve.
The value of $V_{\text {inf }}$ can be derived as follows:
\begin{equation} \label{eq:find-v-inf-formula}
V_{\text {inf }}=\text { r }_{\text {inf }} \cdot\left(V_{\max }-V_{\min }\right)+V_{\min },
\end{equation}
where $V_{\max}$ and $V_{\min}$ correspond to the maximum and minimum controller speeds, and $ \text {r}_{\text {inf }}$ scales the value of $V_{\text {inf }}$.
All three variables are constants empirically chosen based on the pointing environment.
In our case we used $V_{\max}$ = 1, $V_{\min}$ = 0.1 and $r_{inf}$ = 0.5 based on pilot studies. %\co{Do you mean $V_{inf}$ or $r_{inf}$ (previously called $ratio_{inf}$)?}
We further adapt Equations \ref{eq:CDgain-formula} and \ref{eq:find-v-inf-formula} to consider \textit{Interaction Distance} by modifying the definitions of $CD_{\max}$ and $CD_{\min}$:
\begin{equation} \label{eq:proposed}
\begin{split}
CD_{max} = \overline{CD}_{max}-(a * d_s), \\
CD_{min} = \overline{CD}_{min}-(a * d_s),
\end{split}
\end{equation}
where $d_s$ corresponds to the scaled \textit{Interaction Distance}, which we set to 0 at 0.5R, 0.5 at 1R, and 1 at 1.5R.
$a$, $\overline{CD}_{max}$, and $\overline{CD}_{min}$ are constants which values are empirically chosen to define an optimal range of CD-Gain.
We set their values to be equal to 0.2, 1.2 and 0.8 respectively based on the results of a pilot study.
%\co{I think it would be good to justify a little bit our choice of equations. Why Nancel's and not someone else's? I vaguely recall a discussion around this with William and Pourang.}
%\amanat{Based on what william said in the dicussion i added why we used Nancel's formula at the top of this section "Techniques"}
%From our Study 1, we found that \textit{Interaction Distance} has a linear relationship with pointing performance (both speed and accuracy).
%In this paper we derived a formula from our study 1 analysis that can be used to choose the maximum speed (CDmax) and the minimum speed (CDmin) for any environment given a the \textit{Interaction Distance}.
%We achieve this by first converting our \textit{Interaction Distance} to \textit{Scaled Interaction Distance} that ranges from 0 to 1 using the Min-Max Slacer so that it does not depend on any unit. We use the min-max scaler to find the Scaled distance for
%0.5R = (0.5 - 0.5) / (1.5-0.5) = 0; and Scaled distance for
%1R = (1 - 0.5) / (1.5-0.5) = 0.5; and Scaled distance for
%1.5R = (1.5 - 0.5) / (1.5-0.5) = 1
%we then use the \textit{Scaled Distance} in our proposed formula to define the CDmax and CDmin.
%Equation \ref{proposed}, requires 3 empirically calculated constants: max change, $HighCD_{max}$, and $HighCD_{min}$ to find the optimal range of CD Gain for a given interaction distance.
%In our case we ran pilots to finalize the following values for the three constants: max change = 0.2,$HighCD_{max}$ = 1.2, and $HighCD_{min}$= = 0.8.
%Based on two prior equations \ref{CDgain-formula} and \ref{find-v-inf-formula} from Nancel et al. \cite{nancel2013high} work and our proposed equation \ref{proposed} we define 6 techniques for pointing on large curved displays.
%,Cursor enlargement and PBA+ (Position Based Acceleration + Cursor enlargement). These techniques can be divided into three main categories- Motor space , Visual space and Motor space + Visual space.
%\kh{this section not correct, we have six techniques}
%\amanat{added detials of all 6 techniques}
%\kh{Add table}
%\amanat{added table}
% Please add the following required packages to your document preamble:
% \usepackage{multirow}
\begin{table}[h]
%\begin{center}
\begin{adjustbox}{width=0.75\textwidth}
\begin{tabular}{c|ccc}
%\hline
{\textbf{Visual-Space} } & \multicolumn{3}{c}{\textbf{Motor-Space} } \\ %\hline
& \multicolumn{1}{c}{Acceleration} & \multicolumn{1}{c}{Distance } & Acceleration $+$ Distance \\ \hline %\cline{2-4}
\multicolumn{1}{c|}{None} & \multicolumn{1}{c}{PA} & \multicolumn{1}{c}{PBA} & PADIST \\ %\cline{2-4}
\multicolumn{1}{c|}{Cursor Enlargement} & \multicolumn{1}{c}{PASIZE} & \multicolumn{1}{c}{PBASIZE} & PADISTSIZE \\ %\hline
\end{tabular}
\end{adjustbox}
\caption{Cursor-enhancement technique design space.}%\amanat{Lets see where we stand with our final paper. If not the full table I will move the full table to appendix and keep the sit,stand and all condiitions.}}
\label{tab:cursors}
%\end{center}
\end{table}
%\co{Integrate Dhruv's paragraph here.}
Finally, we note existing pointer enhancement techniques for flat displays must be adapted to extend to curved displays, due to different geometry.
%3D space must be considered to represent the curved surface and the resulting interactions.
%Curved displpresents unique challenges due to the three-dimensional interaction surface of curved screens, which incorporate an additional z-axis along with the traditional x and y axes.
Traditional methods for flat displays, which rely on two-dimensional transformations to navigate cursor movement, cannot be used as-is on curved displays because they do not consider the display's depth and curvature.
We thus adapt existing approaches to account for these factors.
Specifically, we track the user's input device in space and calculate the cursor movement as a function of the controller's rotations from the user's current position (see Appendix \ref{appendix} for more information).
\subsubsection{PA: Pointer Acceleration-based}
The Pointer Acceleration-based (PA) technique \cite{nancel2013high} scales cursor speed based on controller speed, as defined by Equations \ref{eq:CDgain-formula} and \ref{eq:find-v-inf-formula} with $CD_{max}=1.2$ and $CD_{min}=0.8$.
This technique helps the user to move across large pointing distances faster, and facilitates smaller target selections by slowing down the cursor during slower motor-space movements.
However, this technique does not consider how pointing performance may vary based on user position.
%The disadvantage of this technique is that the CD Gain range is constant irrespective of the position of the user. We implement distance based techniques to overcome this drawback.
%\kh{what are the possible advantages of this technique? what are the disadvantages?} \amanat{added}
%The dynamic CD gain is calculated using Equations \ref{CDgain-formula} and \ref{find-v-inf-formula}.
%For the PA formula we used a fixed minimum CD Gain to 0.8 and the maximum CD gain to 1.2.
\subsubsection{PASIZE: Pointer Acceleration-based Cursor Enlargement}
PASIZE augments the PA technique with cursor enlargement defined using Equation \ref{eq:CDgain-formula} with $CD_{max}=20$ and $CD_{min}=2.5$ corresponding to the maximum and minimum size of the cursor in centimeter.
%In the Pointer Acceleration-based cursor enlargement (PASIZE) technique the movement of the cursor is exactly the same as the PA technique.
%We again used Equation \ref{CDgain-formula} and \ref{find-v-inf-formula} to find the CD Gain at a given movement based on the movement of the hand/ controller of the user. %However, the difference between PASIZE and PA is that in PASIZE we use Equation \ref{eq:CDgain-formula} to vary the size of the cursor from 2.5cm to 20cm.
This results in the cursor's size increasing when performing faster motor-space movements to facilitate target selections.
We expect this technique to result in faster movement times than PA, as users will not have to rely as much on slow movements for selections.
%This dynamic resizing is a visual-space enhancement that offers users to approach the target position with ease compared to the techniques that rely on only motor-space enhancement.
However, like PA, this technique does not account for changes in pointing performance based on user position.
%The disadvantages of PASIZE is that we do not leverage the position of the user relative to the display to improve pointing perfromance.
%\kh{what are the possible advantages of this technique? what are the disadvantages?} \amanat{added}
\subsubsection{PBA: Position based Acceleration}
PBA modifies the cursor's speed purely based on the user's \textit{Interaction Distance}.
This change is calculated using Equations \ref{eq:CDgain-formula}, \ref{eq:find-v-inf-formula}, and \ref{eq:proposed} where we use the value of \textit{Interaction Distance} in meters as $x$ and set $CD_{max}=4.5$ meters and $CD_{min}=0.7$ meters. %combto find the CD Gain at any point in time based on the \textit{Interaction Distance} between the controller and the position of the cursor on the screen.
%We use the \textit{Interaction Distance} in our proposed Equation \ref{proposed} to select the $CD_{max}$ and $CD_{min}$.
Given that shorter \textit{Interaction Distances} have led to more accurate selections in our first study, PBA increases cursor speed at short distances and slows it down at longer distances.
%From our study 1, we found that smaller distances lead to more accurate selection while selection becomes less accurate as the user distance increases.
%Therefore, For PBA, we vary the cursor speed based on the distance from the user to the cursor position which ensures that when cursor distance from the user is low the speed is high and when the cursor distance from the user is high then the speed is low.
In adjusting the speed-accuracy trade-off of pointing, we expect pointing performance using PBA to be more balanced across user positions.
%The advantage of this technique is that it tries to find a trade-off between speed and accuracy by leveraging the user's position relative to the curved display.
However, it may further result in more difficult selections at short distances, especially when targets are small. %The disadvantage of PBA is that the cursor speed is high when the cursor distance is low this makes selection hard for targets closer to the users (especially if the target width is small).
%\kh{what are the possible advantages of this technique? what are the disadvantages?} \amanat{added advantages and disadvantages}
\subsubsection{PBASIZE: Position based Acceleration with Cursor Enlargement}
This technique augments PBA by changing the cursor size based on Equation \ref{eq:CDgain-formula}, using the user's motor-space movement.
Like PASIZE, cursor size is set to vary between 2.5 and 20cm.
%Similar to PBA we used the Equation \ref{CDgain-formula}, \ref{find-v-inf-formula} and \ref{proposed} based on the \textit{Interaction Distance} and the user position. The only difference between PBA and PBASIZE is that we also use the movement of the controller in the Equation \ref{CDgain-formula} to vary the size of the cursor from 2.5cm to 20cm.
We expect PBASIZE to further facilitate pointing compared to PBA, but to also suffer from increased pointing difficulty at shorter distances.
%The advantages of PBASIZE is that the large size of the cursor allows for easier target selection.
%The PBASIZE also has the limitation of PBA which is the speed is higher when the cursor position is closer to the user making selection harder for smaller targets close to the user.
%\kh{what are the possible advantages of this technique? what are the disadvantages?} \amanat{added}
\subsubsection{PADIST: Pointer Acceleration and Distance-based}
PADIST combines both augmentations of the PA and PBA techniques to modify cursor speed.
$CD_{max}$ and $CD_{min}$ are calculated using Equation \ref{eq:proposed} with the user's \textit{Interaction Distance.}
These values are then used in Equations \ref{eq:CDgain-formula} and \ref{eq:find-v-inf-formula} with the user's motor-space movement to obtain the cursor's speed.
We expect this method to leverage both PA and PBA's advantages to provide superior pointing performance while minimizing their disadvantages.
However, its use may be more frustrating if the parameter values of Equation \ref{eq:proposed} do not align with the environment, e.g., a small $CD_{min}$ value resulting in much slower selections.
%we use both the hand-movement acceleration and \textit{Interaction Distance} to create an optimal pointing technique. Here we use Equation \ref{CDgain-formula} and \ref{find-v-inf-formula} based on the movement of the controller of the user to change the cursor speed. In addition, we use the Equation \ref{proposed} to find the suitable values of $CD_{max}$ and $CD_{min}$ based on \textit{Interaction Distance}.
%The advantage of PADIST is that it leverages both the controller acceleration and the \textit{Interaction Distance} to find the optimal cursor speed.
%This ensures that the cursor speed is adjusted to the position of the user to fine tune the speed and accuracy trade-off.
%The potential disadvantage of PADIST is that if the CDmin is too low then it leads to slower selections.
%Therefore, the constants for the proposed equation \ref{proposed}: HighCDmax, HighCDmin and max change have to empirically calculated for this technique to work properly.
%\kh{what are the possible advantages of this technique? what are the disadvantages?} \amanat{added}
\subsubsection{PADISTSIZE: Pointer Acceleration and Distance-based Cursor Enlargement}
PADISTSIZE augments PADIST by altering cursor size based on the user's motor-space movement.
This change is calculated using \ref{eq:CDgain-formula} with $CD_{max}=20$ and $CD_{min}=2.5$ corresponding to the maximum and minimum size of the cursor in centimeter.
%Similar to PADIST we use the Pointer Acceleration and Distance-based cursor enlargement (PADISTSIZE) by considering the hand/controller movement to set the CD Gain at a given moment using Equation \ref{CDgain-formula}, \ref{find-v-inf-formula}. The only difference between PADIST and PADISTSIZE is that in PADISTSIZE we use Equation \ref{CDgain-formula} to vary the size of the cursor from 2.5cm to 20cm.
PADISTSIZE combines controller acceleration and \textit{Interaction Distance} along with visual-space enhancement (i.e., cursor enlargement).
In doing so, we expect PADISTSIZE to outperform other techniques due to its richer set of augmentations.
However, we also acknowledge that its performance is highly dependent on the right values used in Equation \ref{eq:proposed}.
%ghSimilar to PADIST, PADISTSIZE ensure speed and accuracy trade-off of selections based on the user position.
%To ensure that the cursor speed is not too low or too high the parameters of the proposed equation \ref{proposed} need to adjusted properly.
%\kh{what are the possible advantages of this technique? what are the disadvantages?} \amanat{added}
\subsection{Apparatus and Implementation}
We used the same large curved display and controller setup as in Study 1.
The Unity application was modified to accommodate the six techniques for pointing at the targets across different user positions.
Selection targets had a width of 10cm, and were distant by 250cm or 750cm (\textit{Target Amplitudes}).
This resulted in two distinct Index of Difficulty for trials: 4.70 and 6.25 bits.
We chose these values based on pilot studies to make the pointing task difficult enough for participants to make selection errors.
The number of IDs was reduced to keep the average session duration under 90 minutes.
\subsection{Participants and Procedure}
We invited 12 right-handed participants (7 males, 5 females) from a local university to participate.
They were on average 24.16 years old ($SD = 4.65$ years).
Few participants had prior experience (M = 0.25 years, SD = 0.32 years) interacting with large displays, and none had experience interacting specifically with large curved displays.
The procedure follows the same steps and task as Study 1 (see Section \ref{sec:us1:proc}).
The experiment was conducted at a North American post-secondary institution, where the large curved display is located.
Ethics approval was granted by the institution.
Participants completed the pointing task at each position for a specific technique, before moving to a new technique.
NASA-TLX questionnaires were completed for each technique only.
Techniques were counter-balanced to reduce ordering effects.
Participants each completed 720 trials (6 \textit{Techniques} $\times$ 3 \textit{Interaction Distances} $\times$ 2 \textit{Lateral Offsets} $\times$ 2 \textit{IDs} $\times$ 10 repetitions).
The study lasted approximately 90 minutes on average.
Each participant was compensated with a \$15 honorarium for their time.
We recorded demographic information, movement time, success rate, and subjective feedback ratings for each of the techniques.
%\subsubsection{Technique 1}
%\textit{Fixed CD}: \kh{Amanat: please update this section based on new CV values and new formula} In this technique we have a fixed CD gain/ratio. We decided on a few CD values (xx,xx,xx,xx) and had a pilot study to get the best CD value.
%We found out that the best CD value was 1.1:1. So in this technique , the speed of the cursor is 1.1 times the speed of the controller.
%The speed is constant and does not change during the trial.
% \subsubsection{PA: Pointer Acceleration-based} \kh{Amanat: please update this section based on new CV values and new formula}
% In this technique, the CD ratio changes dynamically based on the user's motor (hand/controller) movement.
% For this technique, if the controller is moved fast, the speed of the cursor becomes high. Meanwhile, if the controller is moved slowly, the speed of the Cursor becomes low. We the following formula:
% $$
% CD(x)=\frac{CD_{\max }-CD_{\min }}{1+e^{-\lambda\left(x V_{\mathrm{inf}}\right)}} +CD_{\min }
% $$
% Here the CD(x) is the speed of the cursor that will change dynamically based on the movement of the controller in the motor space. $CD_{\max}$ and $CD_{\min}$ are the maximum and minimum speed respectively that the Cursor can reach. The value for the $CD_{\max}$ is chosen to be 1.1 and the value for $CD_{\min}$ is chosen to be 0.9 based on pilot studies. x represents the speed of the controller. $\lambda$ $\epsilon$ [0; $\infty$] defines the curve’s slope at its inflection point ($\lambda$ = 0 yields a constant function, $\lambda$ = $\infty$ a step function). Changing $\lambda$ smooths the curve.
% Here
% $$
% V_{\text {inf }}=\text { ratio }_{\text {inf }} \cdot\left(V_{\max }-V_{\min }\right)+V_{\min }
% $$
% $V_{\max}$ and $V_{\min}$ are the maximum and the minimum input controller speed.$ \text { ratio }_{\text {inf }}$ sets the function’s inflection point within this range.
% \subsubsection{PASIZE: Pointer Acceleration-based cursor enlargement} \kh{Amanat: please update this section based on new CD values and new formula}In this technique, the Size of the cursor changes dynamically based on the speed of the controller.
% So, if the controller is moved fast, the size of the cursor becomes small. This would help to reduce the area covered by the cursor when the cursor is actually moving. Meanwhile, if the controller is moved slowly, the size of the Cursor becomes large. This helps to make the selection better. We modified the formulae for this technique which is as below.
% $$
% Size(x)=\frac{Size_{\max }-Size_{\min }}{1+e^{-\lambda\left(x size_{\mathrm{inf}}\right)}} +Size_{\min }
% $$
% Here the Size(x) is the size of the cursor that will change dynamically based on the speed of the controller. $Size_{\max}$ and $Size_{\min}$ are the maximum and minimum size respectively that the Cursor can reach. The value for the $Size_{\max}$ is chosen to be 5 and the value for $Size_{\min}$ is chosen to be 1. x represents the speed of the controller. $\lambda$ $\epsilon$ [0; $\infty$] defines the curve’s slope at its inflection point ($\lambda$ = 0 yields a constant function, $\lambda$ = $\infty$ a step function). Changing $\lambda$ smooths the curve.
% Here $$
% size_{\text {inf }}=\text { ratio }_{\text {inf }} \cdot\left(V_{\max }-V_{\min }\right)+V_{\min }
% $$
% $V_{\max}$ and $V_{\min}$ are the maximum and the minimum input controller speed.$ \text { ratio }_{\text {inf }}$ sets the function’s inflection point within this range.
% \subsubsection{PBA: Pointer Distance-based} \kh{Amanat: please update this section based on new CV values and new formula}In this technique, the CD ratio changes dynamically based on the distance of the controller from the Cursor.
% For the distance based pointer, we set the speed based on the user location.
% in our case min distance = 0.5
% max distance = 1.5
% We use the min-max scaler to find the Scaled distance
% 0.5R = (0.5 - 0.5) / (1.5-0.5) = 0
% 1R = (1 - 0.5) / (1.5-0.5) = 0.5
% 1.5R = (1.5 - 0.5) / (1.5-0.5) = 1
% max change = 0.2 (based on pilot, has to set for the context)
% High CDmax = 1.2; (based on pilot, has to set for the context )
% High CDmin = 0.8; (based on pilot, has to set for the context)
% as with distance we want to reduce speed
% CDmax = High CDmax - (max change * Scaled distance)
% CDmin = High CDmin - (max change * Scaled distance)
% If the distance of the controller from the cursor is higher then the CD ratio will be low (0.9:1) which means the speed of the cursor will be xx times slower than the controller movement. Meanwhile if the distance of the controller from the cursor is lower then the CD ratio will be higher (1.1:1) which means speeds of the cursor will be xx times faster than the controller movement. This is in accordance with our findings from study 1, As the distance of the user from the screen increases, the error rates also increases and the movement time decreases. Therefore, by making the speed a little lower, User's accuracy can be improved. We can have a similar argument when the distance of the screen from user is low. We modified the previous formulae to be
% $$
% CD(d)=\frac{CD_{\max }-CD_{\min }}{1+e^{\lambda\left(dD_{\mathrm{inf}}\right)}} +CD_{\min }
% $$
% d is the distance of the user from the screen. Rest of the arguments are exactly the same. And $D_{\mathrm{inf}}$ is
% $$
% D_{\text {inf }}=\text { ratio }_{\text {inf }} \cdot\left(D_{\max }-D_{\min }\right)+D_{\min }
% $$
% $D_{\max}$ and $D_{\min}$ are the maximum and the minimum distance of the user from the screen .$ \text { ratio }_{\text {inf }}$ sets the function’s inflection point within this range.
% \subsubsection{PBASIZE: Pointer Distance-based cursor enlargement}: \kh{Amanat: please update this section based on new CV values and new formula}In this technique, the size of the cursor changes dynamically based on the distance of the controller from the Cursor. If the distance of the controller from the cursor is higher then the size of the cursor will be large (5).This is in accordance with our result from study 1. Since the accuracy is low at far distances , increasing the cursor size will help the user to be more accurate. Meanwhile if the distance of the controller from the cursor is lower then the size of the cursor will be small(1).This is because when the user is at near distances from the screen the accuracy is high(study 1). This will help as the cursor will take less space on the screen. We modified the previous formulae to be
% $$
% Size(d)=\frac{Size_{\max }-Size_{\min }}{1+e^{\lambda\left(d size_{\mathrm{inf}}\right)}} +Size_{\min }
% $$
% d is the distance of the user from the screen. Rest of the arguments are exactly the same. And $size_{\mathrm{inf}}$ is
% $$
% size_{\text {inf }}=\text { ratio }_{\text {inf }} \cdot\left(D_{\max }-D_{\min }\right)+D_{\min }
% $$
% $D_{\max}$ and $D_{\min}$ are the maximum and the minimum distance of the user from the screen .$ \text { ratio }_{\text {inf }}$ sets the function’s inflection point within this range.
% \subsubsection{PADIST: Pointer Acceleration and Distance-based} \kh{Amanat: please update this section based on new CV values and new formula}In this technique, the CD ratio changes dynamically based on the distance of the controller from the Cursor. If the distance of the controller from the cursor is higher then the CD ratio will be low (0.9:1) which means the speed of the cursor will be xx times slower than the controller movement. Meanwhile if the distance of the controller from the cursor is lower then the CD ratio will be higher (1.1:1) which means speeds of the cursor will be xx times faster than the controller movement. This is in accordance with our findings from study 1, As the distance of the user from the screen increases, the error rates also increases and the movement time decreases. Therefore, by making the speed a little lower, User's accuracy can be improved. We can have a similar argument when the distance of the screen from user is low. We modified the previous formulae to be
% $$
% CD(d)=\frac{CD_{\max }-CD_{\min }}{1+e^{\lambda\left(dD_{\mathrm{inf}}\right)}} +CD_{\min }
% $$
% d is the distance of the user from the screen. Rest of the arguments are exactly the same. And $D_{\mathrm{inf}}$ is
% $$
% D_{\text {inf }}=\text { ratio }_{\text {inf }} \cdot\left(D_{\max }-D_{\min }\right)+D_{\min }
% $$
% $D_{\max}$ and $D_{\min}$ are the maximum and the minimum distance of the user from the screen .$ \text { ratio }_{\text {inf }}$ sets the function’s inflection point within this range.
% \subsubsection{PADISTSIZE: Pointer Acceleration and Distance-based cursor enlargement}: \kh{Amanat: please update this section based on new CV values and new formula}In this technique, the size of the cursor changes dynamically based on the distance of the controller from the Cursor. If the distance of the controller from the cursor is higher then the size of the cursor will be large (5).This is in accordance with our result from study 1. Since the accuracy is low at far distances , increasing the cursor size will help the user to be more accurate. Meanwhile if the distance of the controller from the cursor is lower then the size of the cursor will be small(1).This is because when the user is at near distances from the screen the accuracy is high(study 1). This will help as the cursor will take less space on the screen. We modified the previous formulae to be
% $$
% Size(d)=\frac{Size_{\max }-Size_{\min }}{1+e^{\lambda\left(d size_{\mathrm{inf}}\right)}} +Size_{\min }
% $$
% d is the distance of the user from the screen. Rest of the arguments are exactly the same. And $size_{\mathrm{inf}}$ is
% $$
% size_{\text {inf }}=\text { ratio }_{\text {inf }} \cdot\left(D_{\max }-D_{\min }\right)+D_{\min }
% $$
% $D_{\max}$ and $D_{\min}$ are the maximum and the minimum distance of the user from the screen .$ \text { ratio }_{\text {inf }}$ sets the function’s inflection point within this range.
%\subsection{Target Width and Amplitude} \kh{Amanat: please update this section - we only used two amplitude and one width- not three}
%We use two amplitudes: 250cm, and 750cm from study 1, and the smaller width: 10cm.
%We use only the small width as the participants in the study 1 had more errors on the small target compared to the large one.
%We want to investigate which of the techniques has the fastest pointing speed while ensuring accuracy for the small targets.
%This results in leads to two distinct Index of Difficulty (ID): XX and 6.25 bits.
%These values are derived from the arc length of the circular display along its surface. Our choice of setup aims to reflect real-world scenarios, ensuring that interface elements can be viewed from any position in front of the large curved screen, regardless of the user position
\subsection{Results}
%\co{Similar to study 1 results comment: can we standardize how we report p-values? Or is it necessary to differentiate .001 from .0001?} \amanat{we used this in our previous paper so should be okay.}
We analyze movement time with repeated measures ANOVA on log-transformed data and post-hoc pairwise comparisons with Bonferroni corrections. %\co{Cite ref OK'ing use of log-transform.} \amanat{lets not cite as willaim said it is a standard procedure.}
We use the Wilcoxon Signed-Rank test to analyze accuracy.
For all tests, we report Greenhouse-Geisser corrected p-values and degrees of freedom.
All reported confidence intervals (CIs) are 95\%.
\begin{figure}[htbp]
\centering
\includegraphics[width=1.0\columnwidth]{figures/s2-results-new-v3.png}
\caption{ For each \textbf{Technique}, we show average (a) Movement time by \textit{Lateral Offset}, (b) Movement time by \textit{Interaction Distance} and (c) Accuracy by \textit{Interaction Distance}. Error bars represent 95\% confidence intervals.}
%\co{Missing multiples of R in (b) x-axis title.} \amanat{fixed.}
%\kh{why do we have throughput here as the first graph when we discussed movement time first in the results sections? Change the order please. Once you change the image order, thoroughly check the text for the reference. The same comment applies to Figure 2}}
%\co{Charts are missing symbols identifying the significant pairwise differences.} \amanat{want to skip}}
\label{fig:exp2-res} % make sure you reference this figure in the task text
\end{figure}
\subsubsection{Movement Time}
%main effect
We found significant main effects of \textit{Technique} ($F_{5,55}=13.36$, $p < .001$, $\eta^2$ = 0.55), \textit{Interaction Distance} ($F_{2,22}=44.47$, $p < .001$, $\eta^2$ = 0.80), \textit{Amplitude} ($F_{1,11}=365.92$, $p < .001$, $\eta^2$ = 0.97), and \textit{Lateral Offset} ($F_{1,11}=5.97$, $p<.05$, $\eta^2$ = 0.35) (see Fig. \ref{fig:exp2-res}a) on movement time.
%We did not find any we did not observe any significant effect of on movement time .
For the \textit{Techniques} (see Fig. \ref{fig:exp2-res}a), the mean movement time is 1.77s (CI: [1.72, 1.82]) for PA, 1.87s (CI: [1.81, 1.92]) for PBA, 1.75s (CI: [1.70, 1.80]) for PADIST, 1.60s (CI: [1.55, 1.65]) for PASIZE, 1.66s (CI: [1.62, 1.71]) for PBASIZE, and 1.54s (CI: [1.50, 1.58]) for PADISTSIZE.
Post-hoc pairwise comparisons reveal that the \textit{Techniques} with visual-space enhancements are significantly faster than their non-visually-enhanced counterparts, i.e., PADISTSIZE, PASIZE and PBASIZE have faster movement times than PADIST, PA and PBA (all $p<.001$).
We further find the PBA techniques to be the slowest out of the three types of motor-space enhancements (all $p<.001$).
For the visual-space enhancements, PADISTSIZE is significantly faster than PBASIZE ($p<.0001$) and PASIZE ($p<.05$).
%xxcomebackxx
%In addition, pointing at R is significantly faster than \(\frac{1}{2}\)R ($p<0.0001$).
Regarding the effect of \textit{Interaction Distance} (see Fig. \ref{fig:exp2-res}b), we find the mean movement times to be 1.79s (CI: [1.75, 1.83]) for \(\frac{1}{2}\)R, 1.63s (CI: [1.60, 1.67]) for R, and 1.68s (CI: [1.64, 1.71]) for \(\frac{3}{2}\)R.
Post-hoc pairwise comparisons reveal that pointing from R is significantly faster than \(\frac{1}{2}\)R and \(\frac{3}{2}\)R (all $p<.0001$).
In addition, pointing at \(\frac{3}{2}\)R is significantly faster than \(\frac{1}{2}\)R ($p<.0001$).
In terms of the \textit{Lateral Offset}(see Fig. \ref{fig:exp2-res}a), the no-offset position (mean time 1.68s, CI: [1.65, 1.72]) was significantly faster ($p<.0001$) than the Left-offset position (mean time 1.71s, CI: [1.68, 1.74]). %\co{This is an example of what I meant by standardizing p-values. .0001 is great, but we're literally talking about a 30ms difference, or 2\% increase.} \amanat{lets keep it as it is, as we did for my previous papers.}
Finally, we found several interaction effects: \textit{Amplitude} $\times$ \textit{Interaction Distance} ($F_{2,22}=9.34$, $p<.01$, $\eta^2 = 0.46$) and \textit{Amplitude} $\times$ \textit{Lateral Offset} ($F_{1,11}=13.66$, $p<.01$, $\eta^2 = 0.55$).
We observed similar trends for both high and low amplitudes in terms of \textit{Interaction Distance} (all $p<0.05$): \(\frac{3}{2}\)R is the fastest followed by R and then followed by \(\frac{1}{2}\)R. %\co{If the effect was the same, this wouldn't be an interaction effect. Could the differences between interaction distances vary based on amplitude? We dont have a figure for this, so it's hard for me to adjust the message.} \amanat{NEED-to-talk}
For the low amplitudes (250cm), the no-offset position was faster in terms of selection time compared to the Left-offset position( $p<.0001$). %\co{Why "all"? Are there more than two amplitudes?}
However, for the higher amplitude (A=750cm), we did not find a significant effect of \textit{Lateral Offset} on Movement time ($p=.11$).
%xxcomebackxx
%While offset did not have an effect on selection times, for Width = 10cm, the selection was faster in the center (mean: 1.76s, CI:[1.72,1.8]) compared to the left position (mean: 1.82s, CI:[1.77,1.87]) (p<0.05).
%\(\frac{1}{2}\)R, R, and \(\frac{3}{2}\)R
\subsubsection{Accuracy}
% 2 template for the width with only two levels, you should not have X^2, but W=..., Z=...., p<..., r=...
% 3 template There is a significant effect of Width on accuracy [X^2(...)=..., p < ....], with 0.1m targets (M=..., [..., ...]) leading to less accurate results than 0.7m (M=..., [..., ...])
We found a significant effect of \textit{Technique} ($\chi^2 (5)=23.1, p<.001$) on pointing accuracy (see Fig. \ref{fig:exp2-res}c).
The mean accuracy rate is 0.89 (CI: [0.86, 0.92]) for PA, 0.88 (CI: [0.85, 0.91]) for PBA, 0.91 (CI: [0.88, 0.94]) for PADIST, 0.92 (CI: [0.89, 0.96]) for PASIZE, 0.93 (CI: [0.90, 0.96]) for PBASIZE, and 0.95 (CI: [0.94, 0.97]) for PADISTSIZE.
Pairwise comparisons revealed that only PADISTSIZE is significantly more accurate than PA ($p<.05$, $r = .68$)and PBA ($p<.05$, $r = .71$).
%xxcomebackxx
We also found a significant effect of\textit{Amplitude} between 250cm (M=0.93, CI: [0.91, 0.94]) and 750cm (M=0.90, [0.88, 0.93]) on accuracy ($W = 77; Z = 2.98; p<.001, r = .61$). %\dhruv{the line after this , is it repeated ?}
%Moreover, we found a significant effect of amplitude between 250cm (M=0.93, CI: [0.91, 0.94]) and 750cm (M=0.90, CI: [0.87, 0.93]) ($W = 77; Z = 2.98; p < .001, r = .61$). %\co{What do 43.8 and 131.41 refer to?}
However, we did not observe a significant effect of \textit{Interaction Distance} ($p=.08$) or \textit{Lateral Offset} ($p=.20$) on accuracy.
% \amanat{DANGER: WILLIAM said you never report Interaction effects of Non-parametric test (e.g., Friedman and Wilcoxon.)}
% \amanat{Marium also said she never saw any paper reporting interaction effects of non-parametric tests.}
% \amanat{But Willaim still performed it and Lets talk about it in Lab.}
% \amanat{DATA}
% \co{This is way above my skill level in stats, so I'll trust your judgment}.
%xxcomebackxx
\subsubsection{Fitts' Law Throughput Analysis}
We evaluated the pointing performance of each technique based on throughput \cite{fitts1954information, iso20009241, zhai2004speed, mackenzie1992fitts}.
%Within the context of Fitts' Law, both movement time and the index of difficulty are influenced by changes in width and amplitude.
%Notably, throughput considers both speed (as represented by movement time) and accuracy (as indicated by endpoint deviation).
%Consequently, for a specific interaction or input device, throughput remains consistent.
%To mitigate potential biases in subjective speed-accuracy trade-offs \cite{zhai2004speed} during pointing tasks,
We used Crossman’s correction \cite{crossman1957speed} to mitigate potential biases \cite{zhai2004speed}, and used the means-of-means methodology \cite{soukoreff2004towards, iso20009241} to calculate throughput for each technique \cite{wobbrock2011effects}.
%The formula for \(ID_e\) is given by \(ID_e =\log _2(\frac{A_e}{W_e}+1)\).
%In this equation, \(W_e\) represents the effective width, which is derived from the standard deviation of endpoints: \(W_e = 4.133\times SD\).
%Additionally, \(A_e\) signifies the average amplitude between the initial and final positions of the pointer for a specific amplitude condition.
%In line with recommendations by Wobbrock et al. \cite{wobbrock2011effects}, We use the means-of-means methodology .
%For each display curvature, throughput (TP) is computed using the formula \(TP=\frac{1}{N}\sum_{i=1}^N\left(\frac{ID_{e_i}}{M T_i}\right)\), where \(N= 2\) represents the product of the cardinalities of amplitude and width conditions: \(card(A_e) \times card(W_e)\).
We found significant main effects of \textit{Technique} ($F_{5,55}=4.11$, $p < .01$, $\eta^2$ = 0.27) on throughput (see Fig. \ref{fig:exp2-nasatlx}b).
We also found significant main effects of \textit{Interaction Distance} ($F_{2,22}=35.65$, $p < .001$, $\eta^2$ = 0.76), \textit{Amplitude} ($F_{1,11}=8.51$, $p < .05$, $\eta^2$ = 0.44), and \textit{Lateral Offset} ($F_{1,11}=5.97$, $p<.05$, $\eta^2$ = 0.35) on throughput.
We found that the mean throughput ($TP$) is 2.97 bits/s (CI: [2.77, 3.18]) for PA, 2.76 bits/s (CI: [2.71, 3.04]) for PBA, 3.01 bits/s (CI: [2.80, 3.22]) for PADIST, 3.01 bits/s (CI: [2.79, 3.23]) for PASIZE, 2.87 bits/s (CI: [2.71, 3.04]) for PBASIZE, 3.11 bits/s (CI: [2.90, 3.31]) for PADISTSIZE.
Pairwise comparisons reveal that PA ($p<.05$), PADIST, ($p<.01$), PASIZE ($p<.05$) and PADISTSIZE ($p<.001$) have significantly higher throughput than PBA.
Additionally, PADIST ($p<.05$) and PADISTSIZE ($p<.01$) have significantly higher throughput than PBASIZE.
For \textit{Lateral Offset} ($F_{1,11}=2.19$, $p=0.17$), we did not observe any significant effect on movement time.
We also did not find any interaction effects between the studied factors.
%Regression lines for each technique are shown in Fig. \ref{fig:exp2-fitts}.
%\co{Use same approach as for previous paragraphs. List all that are relevant.} \amanat{added all relevant info.}
%The regression equation for each of the techniques are as follows. PA:y = 0.29x - 0.87, PADIST: y = 0.27x - 0.74, PADISTSIZE: y = 0.23x - 0.46, PASIZE: y = 0.27x - 0.62, PBA: y = 0.32x - 0.89 and PBASIZE: y = 0.28x - 0.76.
%Fitts' Law regression lines of each \textit{Technique} for Movement Time. PADISTSIZE has the lowest slope (0.23), indicating it offers the fastest movement times.
\subsubsection{Preference Scores}
We collected participant feedback on the seven NASA-TLX criteria along with their preferences across all \textit{Techniques} (see Fig. \ref{fig:exp2-nasatlx}a). \update{We calculated the overall workload using raw NASA-TLX values, but for clarity, we inverted the performance scores.}
We found significant differences for all metrics: Preference ($\chi^2 (5,N=12)=34.88, p<.001$), Fatigue ($\chi^2 (5,N=12)=22.80, p<.001$), Mental Demand ($\chi^2 (5,N=12)=24.97, p<.001$), Physical Demand ($\chi^2 (5,N=12)=33.22, p<0.001$) , Temporal Demand ($\chi^2 (5,N=12)=20.21, p<.001$), Performance ($\chi^2 (5,N=12)=34.76, p<.001$), Effort ($\chi^2 (5,N=12)=18.73, p<.01$), Frustration ($\chi^2 (5,N=12)=37.38, p<.001$), and Overall task load ($\chi^2 (5,N=12)=34.96, p<0.001$).
Pairwise comparisons revealed PADISTSIZE and PASIZE have higher Preference and Performance scores and lower Frustration scores than PBA.
Furthermore, PASIZE and PADISTSIZE were rated having lower mental demand, physical demand, and overall workload than PBA and PADIST.
Finally, PADISTSIZE had lower fatigue and effort ratings than PBA.
%\co{missing some kind of descriptive stats for each pairwise comparisons.} \amanat{as we have figure it gives the idea of the values.}
\begin{figure}[h]
\centering
\includegraphics[width=1\columnwidth]{figures/S2-Final-R2-NASATLX.png}
\caption{(a)Subjective feedback score (NASA-TLX \cite{nasatlx}) for each \textit{Technique} and (b) For each \textbf{Technique}, we show average Throughput.
Error bars represent 95\% confidence intervals.}
\label{fig:exp2-nasatlx} % make sure you reference this figure in the task text
\end{figure}
\subsection{Discussion}
\subsubsection{Benefits of Visual-space Augmentations}
We found visually-augmented techniques to lead to faster, and more accurate pointing (by $4\%$), compared to their strictly motor-space enhancement counterparts.
The main advantage of visual-space enhancements is of facilitating selections by enlarging the cursor or targets and requiring less precise movements by the user.
This suggests visual-space enhancements, like Bubble Cursor \cite{GrossmanCHI2005} may be promising directions to support pointing on large curved displays.
We can however not suggest a specific enhancement, as we only tested one such technique in this study.
\subsubsection{Benefits of Combining Visual- and Motor-space Augmentations}
Although previous studies have solely focused on either visual-space enhancement \cite{GrossmanCHI2005} or motor-space enhancement \cite{nancel2013high}, our results suggest that for large curved displays, combining both visual and motor-space enhancement (i.e., PADISTSIZE) leads to superior performance in terms of movement time, accuracy and throughput.
For instance, we found that PADISTSIZE ($M = 95\%$) led to the most accurate selections on average, and is significantly more accurate than PBA ($M = 88\%$) and PA ($M= 89\%$).
However, we note such techniques may require more advanced parameter tuning to work well across contexts.
Considering all six techniques, we find PADIST and PADISTSIZE to outperform the others.
There are thus clear benefits to account for users \textit{Interaction Distance} in addition to acceleration-based augmentations.
For example, analyzing strictly motor-space enhancements, we found PADIST and PA to result in $6\%$ and $5\%$ faster selections than PBA ($M = 1.87s$).
Similarly, comparing only the visual-space augmented techniques, selections using PADISTSIZE to be $4\%$ faster than PASIZE, and $7\%$ faster than PBASIZE.
Even in terms of throughput, which penalizes off-center selections more commonly encountered with visual-space augmented techniques, we find PADISTSIZE to achieve the higher throughput.
In conclusion, the use of acceleration and distance-based augmentations combined with visual-space augmentations (i.e., PADISTSIZE and PADIST) show promising potential for enhancing user interaction on large curved displays.
Their effectiveness makes them suitable candidates for future studies or practical implementations in real-world scenarios.
\subsubsection{Effects of User Position}
We find the effect of \textit{Interaction Distance} on movement time to be different than in the first study:
here, movement times were nearly the same at R distance and \(\frac{3}{2}\)R.
Separating the times per technique, pointing had faster selection times at R \textit{Interaction Distance} compared to at \(\frac{3}{2}\)R for techniques which adjusted CD gain values based on \textit{Interaction Distance} (i.e., PBA, PBASIZE, PADIST, PADISTSIZE).
This shows the utility of \textit{Interaction Distance}-based augmentations in affording faster selections closer to the display.
However, we note further development is needed to support interactions at shorter distances (e.g., \(\frac{1}{2}\)R).
In terms of \textit{Lateral Offset}, similar to Study 1, we found pointing from the no-offset position was significantly faster than from the left-offset position.
Faster selections from the centered positions were also observed for all cursor enhancement techniques except PADISTSIZE and PASIZE (see Fig. \ref{fig:exp2-res}a).
This suggests that changes in CD gain can notably modulate the effects of position-based factors on pointing speed and accuracy.
\section{Design Implications, Limitations and Future Work}
\subsection{Design Implications}
\subsubsection{Effects of Interaction Distance}
We observed significant differences in pointing performance on large curved displays based on the user's \textit{Interaction Distance} in our first study: the farther away the user from the display, the faster and less accurate their pointing becomes.
For example, pointing from \textit{Interaction Distance} \(\frac{3}{2}\)R compared to \(\frac{1}{2}\)R leads to 16\% increase in speed but 11\% decrease in accuracy.
However, in the second study, we observed faster selection times for \textit{Interaction Distances} of R compared to \(\frac{3}{2}R\), and no significant differences on accuracy for our \textit{Interaction Distance}-based enhancement techniques.
Our proposed techniques thus afforded accurate and fast pointing across different user positions. %Furthermore,
We note our evaluation focused on three specific distances, and may not generalize to much smaller or larger \textit{Interaction Distances} (e.g., 3R).
For instance, the implementation of our cursor enhancement techniques may require different parameter values.
We argue researchers should test their implemented techniques from diverse \textit{Interaction Distances} and \textit{Lateral Offsets} to provide optimal interaction across different locations.
\subsubsection{Effects of Lateral Offset}
Across both studies, pointing from a centered position was consistently faster than from an off-centered position.
We however did not observe any differences in terms of pointing accuracy.
Our results thus suggest benefits to having users interact with large curved displays when positioned along its main axis, but further research is necessary to assess the effects of \textit{Lateral Offsets} in more realistic settings.
Additionally, future research should investigate if specific conditions, postures, or user demographics might be more sensitive to lateral displacements, to ensure a more representative and efficient design approach.
\subsubsection{Motor-Space and Visual-Space Enhancements}
Our second study informs on the impacts of different enhancement techniques on pointing on large curved displays. %
All techniques led to faster movement times and higher accuracy compared to absolute pointing used in Study 1.
Amongst the three motor enhancement techniques (PA, PBA and PADIST), acceleration-based techniques (i.e., PA and PADIST) affected pointing movement time.
Combining acceleration and \textit{Interaction Distance} (PADIST) yielded notably the best-pointing performance in terms of throughput.
We further note the benefits of considering \textit{Interaction Distance} in enhancement techniques, as such techniques minimized the movement time differences between the two larger distances.
We therefore suggest integrating both acceleration and \textit{Interaction Distance} in control schemes, as this combination notably demonstrates superior performance compared to methods relying on a single feature, such as PA and PBA.
Additionally, we observed that augmenting motor-space enhancements with a visual-space enhancement resulted in improved pointing performance across user positions.
Hence, we recommend combining visual-space and motor-space enhancements for optimal pointing performance on large curved displays.
\subsection{Limitations and Future Work}
Our two studies investigated a limited range of six user positions - three \textit{Interaction Distances} and two \textit{Lateral Offsets}.
\update{While this provided valuable insights, the inherent complexities introduced by the curved nature of the display and the dynamic positioning of the user were not fully captured in this study, potentially limiting the generalizability of our findings. Specifically, as users move closer to or further from the display or shift laterally, the angular width of targets changes due to varying interaction distances. Additionally, the actual distance from the user to different points on the screen varies non-linearly along the curvature (e.g., center vs. extreme sides), complicating interactions further. In this study, we employed linear measures for target width, target amplitude, and interaction distance. Future research should, therefore, focus on exploring these factors in terms of angular dimensions and non-linear distances that more accurately reflect the curved display context.}
We acknowledge the range of user positions in real conditions are much richer than our tested conditions.
Future studies may investigate a broader range of user positions to derive a more comprehensive understanding of how spatial configurations influence user interaction, particularly in terms of pointing performance.
Furthermore, we conducted our studies with strictly right-handed participants to mitigate any potential confounding effects of handedness.
However, this has constrained our understanding of how handedness may influence pointing on largue curved displays.
Future studies could incorporate participants of both right and left-handed orientations and explore both left and right \textit{Lateral Offsets}.
We also note we only considered hand-held controllers as the input modality, so our findings may not generalize as-is to other modalities for ray-casting such as head rotations, and eye gaze.
Moreover, our investigation focused only on pointing at 2D targets.
Consequently, our insights may not generalize to other interactions on large curved displays, such as visual searches.
Investigating other forms of interactions will provide a more comprehensive understanding of user experience on large curved displays.
Furthermore, it will be beneficial to examine how our findings extend to real-world applications, such as gaming, visual analytics, or entertainment, as these settings could present unique challenges and considerations not accounted for in isolated tasks.
Lastly, our exploration was conducted on only one large semi-circular curved display with a specific curvature radius (i.e., 3270R and 3m tall) in a private room.
Our findings may thus not generalize to other curvature radii and types of curved displays.
\section{Conclusion}
In this paper, we focused on investigating how the position of a user affects their experience interacting with large curved displays and explored motor- and visual-space-based enhancement techniques to improve pointing performance across different positions.
Our first study revealed a notable trade-off between pointing speed and accuracy relative to the user's position from the display.
The results showed that the farther away users stand, the faster they point, but with less accuracy.
Faster pointing was also observed from no-offset positions compared to left-offset positions.
%Consequently, in our second study, we designed six motor- and visual-space-based enhancement techniques to improve accuracy while also improving pointing speed.
Additionally, we discovered that adapting pointing interactions to the users' motor-space speed andthe distance from the display, combined with cursor enlargement techniques significantly improve pointing performance on large curved displays.
Finally, we provided recommendations to improve pointing performance on large curved displays under varied user positions. %\co{Just to verify, this does not discuss the use of interaction distance.} \amanat{added interaction distance.}
Taken together, our work contributes to our understanding on the design space of large curved displays for pointing interactions, and offers guidelines to design cursor enhancement techniques in this setting.
%Finally, we offer guidelines that aim to improve pointing performance and understanding of user interactions with large curved displays.
% In this paper, we focused on investigating how the position of a user affects their experience with large curved displays and explored motor- and visual-space-based enhancement techniques to offer users optimal performance across different positions around a curved display.
% We conducted a user study to explore users' pointing performance on large curved displays, and the results showed that the farther away users stand, the faster they point, but with less accuracy, especially for smaller targets.
% Interestingly, moving side-to-side did not significantly affect pointing performance.
% Consequently, we designed six motor- and visual-space-based enhancement techniques to improve accuracy while improving pointing speed for smaller targets.
% In another user study, we found that making adjustments based on users' hand movement (i.e., how fast the user points) and cursor enlargement (i.e., expanding the cursor as the user slows down to select the target) helps users point more effectively.
% We provided recommendations to enhance the user experience by improving pointing speed and accuracy on large curved displays.
% Our research offers guidelines for improving pointing performance on large curved displays.
% Our main study focused on how the position of a user affects their experience with large curved displays. Our first study showed that the farther away users stand, the faster they point but with less accuracy especially for smaller targets.
% Interestingly, moving side-to-side did not significantly affect the pointing performance.
% In our second study, we tested six techniques that will improve the accuracy while improving the pointing speed for smaller targets.
% We found that making adjustments based on users' hand movement (i.e., how fast the user points), as well as cursor enlargement (i.e., expanding the cursor as the user slows down to select the target), helps users point better.
% We provide recommendations to improve user experience by improving pointing speed and accuracy on large curved displays.
% In conclusion, Our research provides the initial guidelines for improving pointing performance on large curved displays.
%%% -*-BibTeX-*-
%%% Do NOT edit. File created by BibTeX with style
%%% ACM-Reference-Format-Journals [18-Jan-2012].
\begin{thebibliography}{71}
%%% ====================================================================
%%% NOTE TO THE USER: you can override these defaults by providing
%%% customized versions of any of these macros before the \bibliography
%%% command. Each of them MUST provide its own final punctuation,
%%% except for \shownote{}, \showDOI{}, and \showURL{}. The latter two
%%% do not use final punctuation, in order to avoid confusing it with
%%% the Web address.
%%%
%%% To suppress output of a particular field, define its macro to expand
%%% to an empty string, or better, \unskip, like this:
%%%
%%% \newcommand{\showDOI}[1]{\unskip} % LaTeX syntax
%%%
%%% \def \showDOI #1{\unskip} % plain TeX syntax
%%%
%%% ====================================================================
\ifx \showCODEN \undefined \def \showCODEN #1{\unskip} \fi
\ifx \showDOI \undefined \def \showDOI #1{#1}\fi
\ifx \showISBNx \undefined \def \showISBNx #1{\unskip} \fi
\ifx \showISBNxiii \undefined \def \showISBNxiii #1{\unskip} \fi
\ifx \showISSN \undefined \def \showISSN #1{\unskip} \fi
\ifx \showLCCN \undefined \def \showLCCN #1{\unskip} \fi
\ifx \shownote \undefined \def \shownote #1{#1} \fi
\ifx \showarticletitle \undefined \def \showarticletitle #1{#1} \fi
\ifx \showURL \undefined \def \showURL {\relax} \fi
% The following commands are used for tagged output and should be
% invisible to TeX
\providecommand\bibfield[2]{#2}
\providecommand\bibinfo[2]{#2}
\providecommand\natexlab[1]{#1}
\providecommand\showeprint[2][]{arXiv:#2}
\bibitem[Ahn et~al\mbox{.}(2014)]%
{ahn2014research}
\bibfield{author}{\bibinfo{person}{Sung~Hee Ahn}, \bibinfo{person}{Byungki Jin}, \bibinfo{person}{Sanghyun Kwon}, {and} \bibinfo{person}{Myung~Hwan Yun}.} \bibinfo{year}{2014}\natexlab{}.
\newblock \showarticletitle{A research on curved display comparing to flat display regarding posture, tilt angle, focusing area and satisfaction}.
\newblock \bibinfo{journal}{\emph{Journal of the Ergonomics Society of Korea}} \bibinfo{volume}{33}, \bibinfo{number}{3} (\bibinfo{year}{2014}), \bibinfo{pages}{191--202}.
\newblock
\bibitem[Ames(2023)]%
{nasatlx}
\bibfield{author}{\bibinfo{person}{TLX @~NASA Ames}.} \bibinfo{year}{2023}\natexlab{}.
\newblock \bibinfo{title}{Home.}
\newblock
\newblock
\newblock
\shownote{Retrieved January 15, 2023 from \url{https://humansystems.arc.nasa.gov/groups/tlx/}}.
\bibitem[Andrews et~al\mbox{.}(2011)]%
{andrews2011information}
\bibfield{author}{\bibinfo{person}{Christopher Andrews}, \bibinfo{person}{Alex Endert}, \bibinfo{person}{Beth Yost}, {and} \bibinfo{person}{Chris North}.} \bibinfo{year}{2011}\natexlab{}.
\newblock \showarticletitle{Information visualization on large, high-resolution displays: Issues, challenges, and opportunities}.
\newblock \bibinfo{journal}{\emph{Information Visualization}} \bibinfo{volume}{10}, \bibinfo{number}{4} (\bibinfo{year}{2011}), \bibinfo{pages}{341--355}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1177/1473871611415997}
\showDOI{\tempurl}
\showeprint{https://doi.org/10.1177/1473871611415997}
\bibitem[Ardito et~al\mbox{.}(2015)]%
{ardito2015interaction}
\bibfield{author}{\bibinfo{person}{Carmelo Ardito}, \bibinfo{person}{Paolo Buono}, \bibinfo{person}{Maria~Francesca Costabile}, {and} \bibinfo{person}{Giuseppe Desolda}.} \bibinfo{year}{2015}\natexlab{}.
\newblock \showarticletitle{Interaction with Large Displays: A Survey}.
\newblock \bibinfo{journal}{\emph{ACM Comput. Surv.}} \bibinfo{volume}{47}, \bibinfo{number}{3}, Article \bibinfo{articleno}{46} (\bibinfo{date}{feb} \bibinfo{year}{2015}), \bibinfo{numpages}{38}~pages.
\newblock
\showISSN{0360-0300}
\urldef\tempurl%
\url{https://doi.org/10.1145/2682623}
\showDOI{\tempurl}
\bibitem[Argelaguet and Andujar(2009)]%
{argelaguet2009efficient}
\bibfield{author}{\bibinfo{person}{Ferran Argelaguet} {and} \bibinfo{person}{Carlos Andujar}.} \bibinfo{year}{2009}\natexlab{}.
\newblock \showarticletitle{Efficient 3D Pointing Selection in Cluttered Virtual Environments}.
\newblock \bibinfo{journal}{\emph{IEEE Computer Graphics and Applications}} \bibinfo{volume}{29}, \bibinfo{number}{6} (\bibinfo{year}{2009}), \bibinfo{pages}{34--43}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1109/MCG.2009.117}
\showDOI{\tempurl}
\bibitem[Babic et~al\mbox{.}(2018)]%
{babic2018pocket6}
\bibfield{author}{\bibinfo{person}{Teo Babic}, \bibinfo{person}{Harald Reiterer}, {and} \bibinfo{person}{Michael Haller}.} \bibinfo{year}{2018}\natexlab{}.
\newblock \showarticletitle{Pocket6: A 6DoF Controller Based On A Simple Smartphone Application}. In \bibinfo{booktitle}{\emph{Proceedings of the 2018 ACM Symposium on Spatial User Interaction}} (Berlin, Germany) \emph{(\bibinfo{series}{SUI '18})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{2–10}.
\newblock
\showISBNx{9781450357081}
\urldef\tempurl%
\url{https://doi.org/10.1145/3267782.3267785}
\showDOI{\tempurl}
\bibitem[Baudisch et~al\mbox{.}(2001)]%
{baudisch2001focus}
\bibfield{author}{\bibinfo{person}{Patrick Baudisch}, \bibinfo{person}{Nathaniel Good}, {and} \bibinfo{person}{Paul Stewart}.} \bibinfo{year}{2001}\natexlab{}.
\newblock \showarticletitle{Focus plus Context Screens: Combining Display Technology with Visualization Techniques}. In \bibinfo{booktitle}{\emph{Proceedings of the 14th Annual ACM Symposium on User Interface Software and Technology}} (Orlando, Florida) \emph{(\bibinfo{series}{UIST '01})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{31–40}.
\newblock
\showISBNx{158113438X}
\urldef\tempurl%
\url{https://doi.org/10.1145/502348.502354}
\showDOI{\tempurl}
\bibitem[Bernardos et~al\mbox{.}(2016)]%
{bernardos2016comparison}
\bibfield{author}{\bibinfo{person}{Ana~M Bernardos}, \bibinfo{person}{David G{\'o}mez}, {and} \bibinfo{person}{Jos{\'e}~R Casar}.} \bibinfo{year}{2016}\natexlab{}.
\newblock \showarticletitle{A comparison of head pose and deictic pointing interaction methods for smart environments}.
\newblock \bibinfo{journal}{\emph{International Journal of Human-Computer Interaction}} \bibinfo{volume}{32}, \bibinfo{number}{4} (\bibinfo{year}{2016}), \bibinfo{pages}{325--351}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1080/10447318.2016.1142054}
\showDOI{\tempurl}
\bibitem[Birnholtz et~al\mbox{.}(2007)]%
{birnholtz2007exploratory}
\bibfield{author}{\bibinfo{person}{Jeremy~P. Birnholtz}, \bibinfo{person}{Tovi Grossman}, \bibinfo{person}{Clarissa Mak}, {and} \bibinfo{person}{Ravin Balakrishnan}.} \bibinfo{year}{2007}\natexlab{}.
\newblock \showarticletitle{An Exploratory Study of Input Configuration and Group Process in a Negotiation Task Using a Large Display}. In \bibinfo{booktitle}{\emph{Proceedings of the SIGCHI Conference on Human Factors in Computing Systems}} (San Jose, California, USA) \emph{(\bibinfo{series}{CHI '07})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{91–100}.
\newblock
\showISBNx{9781595935939}
\urldef\tempurl%
\url{https://doi.org/10.1145/1240624.1240638}
\showDOI{\tempurl}
\bibitem[Bowman et~al\mbox{.}(2004)]%
{bowman20043d}
\bibfield{author}{\bibinfo{person}{Doug Bowman}, \bibinfo{person}{Ernst Kruijff}, \bibinfo{person}{Joseph~J LaViola~Jr}, {and} \bibinfo{person}{Ivan~P Poupyrev}.} \bibinfo{year}{2004}\natexlab{}.
\newblock \bibinfo{booktitle}{\emph{3D User interfaces: theory and practice, CourseSmart eTextbook}}.
\newblock \bibinfo{publisher}{Addison-Wesley}.
\newblock
\bibitem[Card et~al\mbox{.}(1999)]%
{card1999information}
\bibfield{author}{\bibinfo{person}{SK Card}, \bibinfo{person}{JD Mackinlay}, {and} \bibinfo{person}{Ben Shneiderman}.} \bibinfo{year}{1999}\natexlab{}.
\newblock \bibinfo{title}{Information Visualization. Readings in Information Visualization}.
\newblock
\newblock
\bibitem[Casiez and Roussel(2011)]%
{casiez2011no}
\bibfield{author}{\bibinfo{person}{G\'{e}ry Casiez} {and} \bibinfo{person}{Nicolas Roussel}.} \bibinfo{year}{2011}\natexlab{}.
\newblock \showarticletitle{No More Bricolage! Methods and Tools to Characterize, Replicate and Compare Pointing Transfer Functions}. In \bibinfo{booktitle}{\emph{Proceedings of the 24th Annual ACM Symposium on User Interface Software and Technology}} (Santa Barbara, California, USA) \emph{(\bibinfo{series}{UIST '11})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{603–614}.
\newblock
\showISBNx{9781450307161}
\urldef\tempurl%
\url{https://doi.org/10.1145/2047196.2047276}
\showDOI{\tempurl}
\bibitem[Casiez et~al\mbox{.}(2008)]%
{casiez2008impact}
\bibfield{author}{\bibinfo{person}{G{\'e}ry Casiez}, \bibinfo{person}{Daniel Vogel}, \bibinfo{person}{Ravin Balakrishnan}, {and} \bibinfo{person}{Andy Cockburn}.} \bibinfo{year}{2008}\natexlab{}.
\newblock \showarticletitle{The impact of control-display gain on user performance in pointing tasks}.
\newblock \bibinfo{journal}{\emph{Human--computer interaction}} \bibinfo{volume}{23}, \bibinfo{number}{3} (\bibinfo{year}{2008}), \bibinfo{pages}{215--250}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1080/07370020802278163}
\showDOI{\tempurl}
\bibitem[Cavens et~al\mbox{.}(2002)]%
{cavens2002interacting}
\bibfield{author}{\bibinfo{person}{Duncan Cavens}, \bibinfo{person}{Florian Vogt}, \bibinfo{person}{Sidney Fels}, {and} \bibinfo{person}{Michael Meitner}.} \bibinfo{year}{2002}\natexlab{}.
\newblock \showarticletitle{Interacting with the Big Screen: Pointers to Ponder}. In \bibinfo{booktitle}{\emph{CHI '02 Extended Abstracts on Human Factors in Computing Systems}} (Minneapolis, Minnesota, USA) \emph{(\bibinfo{series}{CHI EA '02})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{678–679}.
\newblock
\showISBNx{1581134541}
\urldef\tempurl%
\url{https://doi.org/10.1145/506443.506542}
\showDOI{\tempurl}
\bibitem[Choi et~al\mbox{.}(2015)]%
{choi201553}
\bibfield{author}{\bibinfo{person}{Kyungah Choi}, \bibinfo{person}{Hyunjoo Bae}, \bibinfo{person}{Sung-Wook Ju}, {and} \bibinfo{person}{Hyeon-Jeong Suk}.} \bibinfo{year}{2015}\natexlab{}.
\newblock \showarticletitle{53.2: Visual search and attention: What eye-tracking reveals about visual performance in the curved display}. In \bibinfo{booktitle}{\emph{SID symposium digest of technical papers}}, Vol.~\bibinfo{volume}{46}. Wiley Online Library, \bibinfo{pages}{798--801}.
\newblock
\bibitem[Chowdhury et~al\mbox{.}(2023)]%
{chowdhury2023paws}
\bibfield{author}{\bibinfo{person}{Sohan Chowdhury}, \bibinfo{person}{William Delamare}, \bibinfo{person}{Pourang Irani}, {and} \bibinfo{person}{Khalad Hasan}.} \bibinfo{year}{2023}\natexlab{}.
\newblock \showarticletitle{PAWS: Personalized Arm and Wrist Movements With Sensitivity Mappings for Controller-Free Locomotion in Virtual Reality}.
\newblock \bibinfo{journal}{\emph{Proc. of the ACM on Human-Computer Interaction}} \bibinfo{volume}{7}, \bibinfo{number}{MHCI} (\bibinfo{year}{2023}), \bibinfo{pages}{1--21}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1145/3604264}
\showDOI{\tempurl}
\bibitem[Chowdhury et~al\mbox{.}(2022)]%
{chowdhury2022wriarm}
\bibfield{author}{\bibinfo{person}{Sohan Chowdhury}, \bibinfo{person}{A~K M~Amanat Ullah}, \bibinfo{person}{Nathan~Bruce Pelmore}, \bibinfo{person}{Pourang Irani}, {and} \bibinfo{person}{Khalad Hasan}.} \bibinfo{year}{2022}\natexlab{}.
\newblock \showarticletitle{WriArm: Leveraging Wrist Movement to Design Wrist+Arm Based Teleportation in {VR}}. In \bibinfo{booktitle}{\emph{2022 IEEE Int. Symposium on Mixed and Augmented Reality ({ISM{AR}})}}. \bibinfo{pages}{317--325}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1109/ISMAR55827.2022.00047}
\showDOI{\tempurl}
\bibitem[Cockburn et~al\mbox{.}(2009)]%
{cockburn2009review}
\bibfield{author}{\bibinfo{person}{Andy Cockburn}, \bibinfo{person}{Amy Karlson}, {and} \bibinfo{person}{Benjamin~B. Bederson}.} \bibinfo{year}{2009}\natexlab{}.
\newblock \showarticletitle{A Review of Overview+detail, Zooming, and Focus+context Interfaces}.
\newblock \bibinfo{journal}{\emph{ACM Comput. Surv.}} \bibinfo{volume}{41}, \bibinfo{number}{1}, Article \bibinfo{articleno}{2} (\bibinfo{date}{jan} \bibinfo{year}{2009}), \bibinfo{numpages}{31}~pages.
\newblock
\showISSN{0360-0300}
\urldef\tempurl%
\url{https://doi.org/10.1145/1456650.1456652}
\showDOI{\tempurl}
\bibitem[Crossman(1957)]%
{crossman1957speed}
\bibfield{author}{\bibinfo{person}{ERFW Crossman}.} \bibinfo{year}{1957}\natexlab{}.
\newblock \showarticletitle{The speed and accuracy of simple hand movements}.
\newblock \bibinfo{journal}{\emph{The nature and acquisition of industrial skills}} (\bibinfo{year}{1957}).
\newblock
\bibitem[Davis and Chen(2002)]%
{davis2002lumipoint}
\bibfield{author}{\bibinfo{person}{James Davis} {and} \bibinfo{person}{Xing Chen}.} \bibinfo{year}{2002}\natexlab{}.
\newblock \showarticletitle{Lumipoint: multi-user laser-based interaction on large tiled displays}.
\newblock \bibinfo{journal}{\emph{Displays}} \bibinfo{volume}{23}, \bibinfo{number}{5} (\bibinfo{year}{2002}), \bibinfo{pages}{205--211}.
\newblock
\showISSN{0141-9382}
\urldef\tempurl%
\url{https://doi.org/10.1016/S0141-9382(02)00039-2}
\showDOI{\tempurl}
\bibitem[Ens et~al\mbox{.}(2016)]%
{ens2016moving}
\bibfield{author}{\bibinfo{person}{Barrett Ens}, \bibinfo{person}{David Ahlstr\"{o}m}, {and} \bibinfo{person}{Pourang Irani}.} \bibinfo{year}{2016}\natexlab{}.
\newblock \showarticletitle{Moving Ahead with Peephole Pointing: Modelling Object Selection with Head-Worn Display Field of View Limitations}. In \bibinfo{booktitle}{\emph{Proceedings of the 2016 Symposium on Spatial User Interaction}} (Tokyo, Japan) \emph{(\bibinfo{series}{SUI '16})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{107–110}.
\newblock
\showISBNx{9781450340687}
\urldef\tempurl%
\url{https://doi.org/10.1145/2983310.2985756}
\showDOI{\tempurl}
\bibitem[Fitts(1954)]%
{fitts1954information}
\bibfield{author}{\bibinfo{person}{Paul~M Fitts}.} \bibinfo{year}{1954}\natexlab{}.
\newblock \showarticletitle{The information capacity of the human motor system in controlling the amplitude of movement.}
\newblock \bibinfo{journal}{\emph{Journal of experimental psychology}} \bibinfo{volume}{47}, \bibinfo{number}{6} (\bibinfo{year}{1954}), \bibinfo{pages}{381}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1037/h0055392}
\showDOI{\tempurl}
\bibitem[Forlines et~al\mbox{.}(2006)]%
{ForlinesUIST2006}
\bibfield{author}{\bibinfo{person}{Clifton Forlines}, \bibinfo{person}{Daniel Vogel}, {and} \bibinfo{person}{Ravin Balakrishnan}.} \bibinfo{year}{2006}\natexlab{}.
\newblock \showarticletitle{HybridPointing: Fluid Switching between Absolute and Relative Pointing with a Direct Input Device}. In \bibinfo{booktitle}{\emph{Proceedings of the 19th Annual ACM Symposium on User Interface Software and Technology}} (Montreux, Switzerland) \emph{(\bibinfo{series}{UIST '06})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{211–220}.
\newblock
\showISBNx{1595933131}
\urldef\tempurl%
\url{https://doi.org/10.1145/1166253.1166286}
\showDOI{\tempurl}
\bibitem[Gallo et~al\mbox{.}(2010)]%
{GalloIEEE2010}
\bibfield{author}{\bibinfo{person}{Luigi Gallo}, \bibinfo{person}{Mario Ciampi}, {and} \bibinfo{person}{Aniello Minutolo}.} \bibinfo{year}{2010}\natexlab{}.
\newblock \showarticletitle{Smoothed Pointing: A User-Friendly Technique for Precision Enhanced Remote Pointing}. In \bibinfo{booktitle}{\emph{2010 International Conference on Complex, Intelligent and Software Intensive Systems}}. \bibinfo{pages}{712--717}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1109/CISIS.2010.76}
\showDOI{\tempurl}
\bibitem[Grossman and Balakrishnan(2005)]%
{GrossmanCHI2005}
\bibfield{author}{\bibinfo{person}{Tovi Grossman} {and} \bibinfo{person}{Ravin Balakrishnan}.} \bibinfo{year}{2005}\natexlab{}.
\newblock \showarticletitle{The Bubble Cursor: Enhancing Target Acquisition by Dynamic Resizing of the Cursor's Activation Area}. In \bibinfo{booktitle}{\emph{Proceedings of the SIGCHI Conference on Human Factors in Computing Systems}} (Portland, Oregon, USA) \emph{(\bibinfo{series}{CHI '05})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{281–290}.
\newblock
\showISBNx{1581139985}
\urldef\tempurl%
\url{https://doi.org/10.1145/1054972.1055012}
\showDOI{\tempurl}
\bibitem[Haque et~al\mbox{.}(2015)]%
{haque2015myopoint}
\bibfield{author}{\bibinfo{person}{Faizan Haque}, \bibinfo{person}{Mathieu Nancel}, {and} \bibinfo{person}{Daniel Vogel}.} \bibinfo{year}{2015}\natexlab{}.
\newblock \showarticletitle{Myopoint: Pointing and Clicking Using Forearm Mounted Electromyography and Inertial Motion Sensors}. In \bibinfo{booktitle}{\emph{Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems}} (Seoul, Republic of Korea) \emph{(\bibinfo{series}{CHI '15})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{3653–3656}.
\newblock
\showISBNx{9781450331456}
\urldef\tempurl%
\url{https://doi.org/10.1145/2702123.2702133}
\showDOI{\tempurl}
\bibitem[Hennecke et~al\mbox{.}(2013)]%
{hennecke2013investigating}
\bibfield{author}{\bibinfo{person}{Fabian Hennecke}, \bibinfo{person}{Alexander De~Luca}, \bibinfo{person}{Ngo Dieu~Huong Nguyen}, \bibinfo{person}{Sebastian Boring}, {and} \bibinfo{person}{Andreas Butz}.} \bibinfo{year}{2013}\natexlab{}.
\newblock \showarticletitle{Investigating Pointing Tasks across Angularly Coupled Display Areas}. In \bibinfo{booktitle}{\emph{Human-Computer Interaction -- INTERACT 2013}}, \bibfield{editor}{\bibinfo{person}{Paula Kotz{\'e}}, \bibinfo{person}{Gary Marsden}, \bibinfo{person}{Gitte Lindgaard}, \bibinfo{person}{Janet Wesson}, {and} \bibinfo{person}{Marco Winckler}} (Eds.). \bibinfo{publisher}{Springer Berlin Heidelberg}, \bibinfo{address}{Berlin, Heidelberg}, \bibinfo{pages}{720--727}.
\newblock
\showISBNx{978-3-642-40483-2}
\bibitem[Hourcade and Bullock-Rest(2012)]%
{hourcade2012small}
\bibfield{author}{\bibinfo{person}{Juan~Pablo Hourcade} {and} \bibinfo{person}{Natasha Bullock-Rest}.} \bibinfo{year}{2012}\natexlab{}.
\newblock \showarticletitle{How Small Can You Go? Analyzing the Effect of Visual Angle in Pointing Tasks}. In \bibinfo{booktitle}{\emph{Proceedings of the SIGCHI Conference on Human Factors in Computing Systems}} (Austin, Texas, USA) \emph{(\bibinfo{series}{CHI '12})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{213–216}.
\newblock
\showISBNx{9781450310154}
\urldef\tempurl%
\url{https://doi.org/10.1145/2207676.2207706}
\showDOI{\tempurl}
\bibitem[ISO(2000)]%
{iso20009241}
\bibfield{author}{\bibinfo{person}{ISO ISO}.} \bibinfo{year}{2000}\natexlab{}.
\newblock \showarticletitle{9241-9 Ergonomic requirements for office work with visual display terminals (VDTs)-Part 9: Requirements for non-keyboard input devices (FDIS-Final Draft International Standard), 2000}.
\newblock \bibinfo{journal}{\emph{International Organization for Standardization}} (\bibinfo{year}{2000}).
\newblock
\bibitem[James et~al\mbox{.}(2020)]%
{James_2020_personal}
\bibfield{author}{\bibinfo{person}{Rapha{\"{e}}l James}, \bibinfo{person}{Anastasia Bezerianos}, \bibinfo{person}{Olivier Chapuis}, \bibinfo{person}{Maxime Cordeil}, \bibinfo{person}{Tim Dwyer}, {and} \bibinfo{person}{Arnaud Prouzeau}.} \bibinfo{year}{2020}\natexlab{}.
\newblock \showarticletitle{Personal+Context navigation: combining {AR} and shared displays in network path-following}.
\newblock \bibinfo{journal}{\emph{CoRR}} \bibinfo{volume}{abs/2005.10612} (\bibinfo{year}{2020}).
\newblock
\showeprint[arXiv]{2005.10612}
\urldef\tempurl%
\url{https://arxiv.org/abs/2005.10612}
\showURL{%
\tempurl}
\bibitem[Janzen et~al\mbox{.}(2016)]%
{janzen2016modeling}
\bibfield{author}{\bibinfo{person}{Izabelle Janzen}, \bibinfo{person}{Vasanth~K. Rajendran}, {and} \bibinfo{person}{Kellogg~S. Booth}.} \bibinfo{year}{2016}\natexlab{}.
\newblock \showarticletitle{Modeling the Impact of Depth on Pointing Performance}. In \bibinfo{booktitle}{\emph{Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems}} (San Jose, California, USA) \emph{(\bibinfo{series}{CHI '16})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{188–199}.
\newblock
\showISBNx{9781450333627}
\urldef\tempurl%
\url{https://doi.org/10.1145/2858036.2858244}
\showDOI{\tempurl}
\bibitem[Jota et~al\mbox{.}(2010)]%
{jota2010comparisonGIpaper}
\bibfield{author}{\bibinfo{person}{Ricardo Jota}, \bibinfo{person}{Miguel~A. Nacenta}, \bibinfo{person}{Joaquim~A. Jorge}, \bibinfo{person}{Sheelagh Carpendale}, {and} \bibinfo{person}{Saul Greenberg}.} \bibinfo{year}{2010}\natexlab{}.
\newblock \showarticletitle{A Comparison of Ray Pointing Techniques for Very Large Displays}. In \bibinfo{booktitle}{\emph{Proceedings of Graphics Interface 2010}} (Ottawa, Ontario, Canada) \emph{(\bibinfo{series}{GI '10})}. \bibinfo{publisher}{Canadian Information Processing Society}, \bibinfo{address}{CAN}, \bibinfo{pages}{269–276}.
\newblock
\showISBNx{9781568817125}
\urldef\tempurl%
\url{https://doi.org/10.5555/1839214.1839261}
\showDOI{\tempurl}
\bibitem[Khamis et~al\mbox{.}(2017)]%
{khamis2017eyescout}
\bibfield{author}{\bibinfo{person}{Mohamed Khamis}, \bibinfo{person}{Axel Hoesl}, \bibinfo{person}{Alexander Klimczak}, \bibinfo{person}{Martin Reiss}, \bibinfo{person}{Florian Alt}, {and} \bibinfo{person}{Andreas Bulling}.} \bibinfo{year}{2017}\natexlab{}.
\newblock \showarticletitle{EyeScout: Active Eye Tracking for Position and Movement Independent Gaze Interaction with Large Public Displays}. In \bibinfo{booktitle}{\emph{Proceedings of the 30th Annual ACM Symposium on User Interface Software and Technology}} (Qu\'{e}bec City, QC, Canada) \emph{(\bibinfo{series}{UIST '17})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{155–166}.
\newblock
\showISBNx{9781450349819}
\urldef\tempurl%
\url{https://doi.org/10.1145/3126594.3126630}
\showDOI{\tempurl}
\bibitem[Khoo et~al\mbox{.}(2008)]%
{khoo2009designing}
\bibfield{author}{\bibinfo{person}{Eng~Tat Khoo}, \bibinfo{person}{Tim Merritt}, {and} \bibinfo{person}{Adrian~David Cheok}.} \bibinfo{year}{2008}\natexlab{}.
\newblock \showarticletitle{{Designing physical and social intergenerational family entertainment}}.
\newblock \bibinfo{journal}{\emph{Interacting with Computers}} \bibinfo{volume}{21}, \bibinfo{number}{1-2} (\bibinfo{date}{11} \bibinfo{year}{2008}), \bibinfo{pages}{76--87}.
\newblock
\showISSN{0953-5438}
\urldef\tempurl%
\url{https://doi.org/10.1016/j.intcom.2008.10.009}
\showDOI{\tempurl}
\bibitem[K\"{o}nig et~al\mbox{.}(2009)]%
{konig2009adaptive}
\bibfield{author}{\bibinfo{person}{Werner~A. K\"{o}nig}, \bibinfo{person}{Jens Gerken}, \bibinfo{person}{Stefan Dierdorf}, {and} \bibinfo{person}{Harald Reiterer}.} \bibinfo{year}{2009}\natexlab{}.
\newblock \showarticletitle{Adaptive Pointing: Implicit Gain Adaptation for Absolute Pointing Devices}. In \bibinfo{booktitle}{\emph{CHI '09 Extended Abstracts on Human Factors in Computing Systems}} (Boston, MA, USA) \emph{(\bibinfo{series}{CHI EA '09})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{4171–4176}.
\newblock
\showISBNx{9781605582474}
\urldef\tempurl%
\url{https://doi.org/10.1145/1520340.1520635}
\showDOI{\tempurl}
\bibitem[Kopper et~al\mbox{.}(2010)]%
{kopper2010human}
\bibfield{author}{\bibinfo{person}{Regis Kopper}, \bibinfo{person}{Doug~A. Bowman}, \bibinfo{person}{Mara~G. Silva}, {and} \bibinfo{person}{Ryan~P. McMahan}.} \bibinfo{year}{2010}\natexlab{}.
\newblock \showarticletitle{A human motor behavior model for distal pointing tasks}.
\newblock \bibinfo{journal}{\emph{International Journal of Human-Computer Studies}} \bibinfo{volume}{68}, \bibinfo{number}{10} (\bibinfo{year}{2010}), \bibinfo{pages}{603--615}.
\newblock
\showISSN{1071-5819}
\urldef\tempurl%
\url{https://doi.org/10.1016/j.ijhcs.2010.05.001}
\showDOI{\tempurl}
\bibitem[Kopper et~al\mbox{.}(2008)]%
{kopper2008increasing}
\bibfield{author}{\bibinfo{person}{Regis Kopper}, \bibinfo{person}{Mara~G Silva}, \bibinfo{person}{Ryan~Patrick McMahan}, {and} \bibinfo{person}{Douglas~A Bowman}.} \bibinfo{year}{2008}\natexlab{}.
\newblock \bibinfo{booktitle}{\emph{Increasing the precision of distant pointing for large high-resolution displays}}.
\newblock \bibinfo{type}{{T}echnical {R}eport}. \bibinfo{institution}{Department of Computer Science, Virginia Polytechnic Institute \& State~…}.
\newblock
\bibitem[Kopper(2011)]%
{kopper2011understanding}
\bibfield{author}{\bibinfo{person}{R{\'e}gis Augusto~Poli Kopper}.} \bibinfo{year}{2011}\natexlab{}.
\newblock \emph{\bibinfo{title}{Understanding and improving distal pointing interaction}}.
\newblock \bibinfo{thesistype}{Ph.\,D. Dissertation}. \bibinfo{school}{Virginia Tech}.
\newblock
\urldef\tempurl%
\url{http://hdl.handle.net/10919/28183}
\showURL{%
\tempurl}
\bibitem[Kovacs et~al\mbox{.}(2008)]%
{kovacs2008perceptual}
\bibfield{author}{\bibinfo{person}{AJ Kovacs}, \bibinfo{person}{JJ Buchanan}, {and} \bibinfo{person}{CH Shea}.} \bibinfo{year}{2008}\natexlab{}.
\newblock \showarticletitle{Perceptual influences on Fitts’ law}.
\newblock \bibinfo{journal}{\emph{Experimental Brain Research}} \bibinfo{volume}{190}, \bibinfo{number}{1} (\bibinfo{year}{2008}), \bibinfo{pages}{99--103}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1007/s00221-008-1497-3}
\showDOI{\tempurl}
\bibitem[Kyung and Park(2021)]%
{kyung2021curved}
\bibfield{author}{\bibinfo{person}{Gyouhyung Kyung} {and} \bibinfo{person}{Sungryul Park}.} \bibinfo{year}{2021}\natexlab{}.
\newblock \showarticletitle{Curved versus flat monitors: Interactive effects of display curvature radius and display size on visual search performance and visual fatigue}.
\newblock \bibinfo{journal}{\emph{Human Factors}} \bibinfo{volume}{63}, \bibinfo{number}{7} (\bibinfo{year}{2021}), \bibinfo{pages}{1182--1195}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1177/0018720820922717}
\showDOI{\tempurl}
\bibitem[Lin et~al\mbox{.}(2015)]%
{lin2015investigation}
\bibfield{author}{\bibinfo{person}{Chiuhsiang~Joe Lin}, \bibinfo{person}{Sui-Hua Ho}, {and} \bibinfo{person}{Yan-Jyun Chen}.} \bibinfo{year}{2015}\natexlab{}.
\newblock \showarticletitle{An investigation of pointing postures in a 3D stereoscopic environment}.
\newblock \bibinfo{journal}{\emph{Applied ergonomics}} \bibinfo{volume}{48} (\bibinfo{year}{2015}), \bibinfo{pages}{154--163}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1016/j.apergo.2014.12.001}
\showDOI{\tempurl}
\bibitem[Liu et~al\mbox{.}(2020)]%
{liu2020design}
\bibfield{author}{\bibinfo{person}{Jiazhou Liu}, \bibinfo{person}{Arnaud Prouzeau}, \bibinfo{person}{Barrett Ens}, {and} \bibinfo{person}{Tim Dwyer}.} \bibinfo{year}{2020}\natexlab{}.
\newblock \showarticletitle{Design and Evaluation of Interactive Small Multiples Data Visualisation in Immersive Spaces}. In \bibinfo{booktitle}{\emph{2020 IEEE Conference on Virtual Reality and 3D User Interfaces (VR)}}. \bibinfo{pages}{588--597}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1109/VR46266.2020.00081}
\showDOI{\tempurl}
\bibitem[MacKenzie(1992)]%
{mackenzie1992fitts}
\bibfield{author}{\bibinfo{person}{I.~Scott MacKenzie}.} \bibinfo{year}{1992}\natexlab{}.
\newblock \showarticletitle{Fitts' Law as a Research and Design Tool in Human-Computer Interaction}.
\newblock \bibinfo{journal}{\emph{Hum.-Comput. Interact.}} \bibinfo{volume}{7}, \bibinfo{number}{1} (\bibinfo{date}{mar} \bibinfo{year}{1992}), \bibinfo{pages}{91–139}.
\newblock
\showISSN{0737-0024}
\urldef\tempurl%
\url{https://doi.org/10.1207/s15327051hci0701_3}
\showDOI{\tempurl}
\bibitem[McCallum and Irani(2009)]%
{McCallumUIST2009}
\bibfield{author}{\bibinfo{person}{David~C McCallum} {and} \bibinfo{person}{Pourang Irani}.} \bibinfo{year}{2009}\natexlab{}.
\newblock \showarticletitle{Arc-pad: absolute+ relative cursor positioning for large displays with a mobile touchscreen}. In \bibinfo{booktitle}{\emph{Proceedings of the 22nd annual ACM symposium on User interface software and technology}}. \bibinfo{pages}{153--156}.
\newblock
\bibitem[Mechdyne(2023)]%
{mechdyne-new1}
\bibfield{author}{\bibinfo{person}{Mechdyne}.} \bibinfo{year}{2023}\natexlab{}.
\newblock \bibinfo{title}{Mechdyne Corporation}.
\newblock
\newblock
\newblock
\shownote{Retrieved June 15, 2023 from \url{https://www.mechdyne.com/}}.
\bibitem[Myers et~al\mbox{.}(2002)]%
{myers2002interacting}
\bibfield{author}{\bibinfo{person}{Brad~A. Myers}, \bibinfo{person}{Rishi Bhatnagar}, \bibinfo{person}{Jeffrey Nichols}, \bibinfo{person}{Choon~Hong Peck}, \bibinfo{person}{Dave Kong}, \bibinfo{person}{Robert Miller}, {and} \bibinfo{person}{A.~Chris Long}.} \bibinfo{year}{2002}\natexlab{}.
\newblock \showarticletitle{Interacting at a Distance: Measuring the Performance of Laser Pointers and Other Devices}. In \bibinfo{booktitle}{\emph{Proceedings of the SIGCHI Conference on Human Factors in Computing Systems}} (Minneapolis, Minnesota, USA) \emph{(\bibinfo{series}{CHI '02})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{33–40}.
\newblock
\showISBNx{1581134533}
\urldef\tempurl%
\url{https://doi.org/10.1145/503376.503383}
\showDOI{\tempurl}
\bibitem[Nancel et~al\mbox{.}(2013)]%
{nancel2013high}
\bibfield{author}{\bibinfo{person}{Mathieu Nancel}, \bibinfo{person}{Olivier Chapuis}, \bibinfo{person}{Emmanuel Pietriga}, \bibinfo{person}{Xing-Dong Yang}, \bibinfo{person}{Pourang~P. Irani}, {and} \bibinfo{person}{Michel Beaudouin-Lafon}.} \bibinfo{year}{2013}\natexlab{}.
\newblock \showarticletitle{High-Precision Pointing on Large Wall Displays Using Small Handheld Devices}. In \bibinfo{booktitle}{\emph{Proceedings of the SIGCHI Conference on Human Factors in Computing Systems}} (Paris, France) \emph{(\bibinfo{series}{CHI '13})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{831–840}.
\newblock
\showISBNx{9781450318990}
\urldef\tempurl%
\url{https://doi.org/10.1145/2470654.2470773}
\showDOI{\tempurl}
\bibitem[Novak et~al\mbox{.}(2008)]%
{novak2008designing}
\bibfield{author}{\bibinfo{person}{Jasminko Novak}, \bibinfo{person}{Mattias Aggeler}, {and} \bibinfo{person}{Gerhard Schwabe}.} \bibinfo{year}{2008}\natexlab{}.
\newblock \showarticletitle{Designing Large-Display Workspaces for Cooperative Travel Consultancy}. In \bibinfo{booktitle}{\emph{CHI '08 Extended Abstracts on Human Factors in Computing Systems}} (Florence, Italy) \emph{(\bibinfo{series}{CHI EA '08})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{2877–2882}.
\newblock
\showISBNx{9781605580128}
\urldef\tempurl%
\url{https://doi.org/10.1145/1358628.1358777}
\showDOI{\tempurl}
\bibitem[Oh and Stuerzlinger(2002)]%
{oh2002laser}
\bibfield{author}{\bibinfo{person}{Ji-Young Oh} {and} \bibinfo{person}{Wolfgang Stuerzlinger}.} \bibinfo{year}{2002}\natexlab{}.
\newblock \showarticletitle{Laser pointers as collaborative pointing devices}. In \bibinfo{booktitle}{\emph{Graphics Interface}}, Vol.~\bibinfo{volume}{2002}. Citeseer, \bibinfo{pages}{141--149}.
\newblock
\bibitem[OptiTrack(2023)]%
{optitrack}
\bibfield{author}{\bibinfo{person}{OptiTrack}.} \bibinfo{year}{2023}\natexlab{}.
\newblock \bibinfo{title}{OptiTrack - Motion Capture Systems}.
\newblock
\newblock
\newblock
\shownote{Retrieved June 15, 2023 from \url{https://optitrack.com/}}.
\bibitem[Park et~al\mbox{.}(2019)]%
{park2019effects}
\bibfield{author}{\bibinfo{person}{Sungryul Park}, \bibinfo{person}{Gyouhyung Kyung}, \bibinfo{person}{Donghee Choi}, \bibinfo{person}{Jihhyeon Yi}, \bibinfo{person}{Songil Lee}, \bibinfo{person}{Byeonghwa Choi}, {and} \bibinfo{person}{Seungbae Lee}.} \bibinfo{year}{2019}\natexlab{}.
\newblock \showarticletitle{Effects of display curvature and task duration on proofreading performance, visual discomfort, visual fatigue, mental workload, and user satisfaction}.
\newblock \bibinfo{journal}{\emph{Applied Ergonomics}} \bibinfo{volume}{78} (\bibinfo{year}{2019}), \bibinfo{pages}{26--36}.
\newblock
\showISSN{0003-6870}
\urldef\tempurl%
\url{https://doi.org/10.1016/j.apergo.2019.01.014}
\showDOI{\tempurl}
\bibitem[Radwin et~al\mbox{.}(1990)]%
{radwin1990method}
\bibfield{author}{\bibinfo{person}{Robert~G. Radwin}, \bibinfo{person}{Gregg~C. Vanderheiden}, {and} \bibinfo{person}{Mei-Li Lin}.} \bibinfo{year}{1990}\natexlab{}.
\newblock \showarticletitle{A Method for Evaluating Head-Controlled Computer Input Devices Using Fitts' Law}.
\newblock \bibinfo{journal}{\emph{Human Factors}} \bibinfo{volume}{32}, \bibinfo{number}{4} (\bibinfo{year}{1990}), \bibinfo{pages}{423--438}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1177/001872089003200405}
\showDOI{\tempurl}
\showeprint{https://doi.org/10.1177/001872089003200405}
\bibitem[Satyanarayan et~al\mbox{.}(2013)]%
{satyanarayan2013chi}
\bibfield{author}{\bibinfo{person}{Arvind Satyanarayan}, \bibinfo{person}{Daniel Strazzulla}, \bibinfo{person}{Clemens~N. Klokmose}, \bibinfo{person}{Michel Beaudouin-Lafon}, {and} \bibinfo{person}{Wendy~E. Mackay}.} \bibinfo{year}{2013}\natexlab{}.
\newblock \showarticletitle{The CHI 2013 Interactive Schedule}. In \bibinfo{booktitle}{\emph{CHI '13 Extended Abstracts on Human Factors in Computing Systems}} (Paris, France) \emph{(\bibinfo{series}{CHI EA '13})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{2987–2990}.
\newblock
\showISBNx{9781450319522}
\urldef\tempurl%
\url{https://doi.org/10.1145/2468356.2479591}
\showDOI{\tempurl}
\bibitem[Scheible and Ojala(2005)]%
{scheible2005mobilenin}
\bibfield{author}{\bibinfo{person}{J\"{u}rgen Scheible} {and} \bibinfo{person}{Timo Ojala}.} \bibinfo{year}{2005}\natexlab{}.
\newblock \showarticletitle{MobiLenin Combining a Multi-Track Music Video, Personal Mobile Phones and a Public Display into Multi-User Interactive Entertainment}. In \bibinfo{booktitle}{\emph{Proceedings of the 13th Annual ACM International Conference on Multimedia}} (Hilton, Singapore) \emph{(\bibinfo{series}{MULTIMEDIA '05})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{199–208}.
\newblock
\showISBNx{1595930442}
\urldef\tempurl%
\url{https://doi.org/10.1145/1101149.1101178}
\showDOI{\tempurl}
\bibitem[Shupp et~al\mbox{.}(2009)]%
{shupp2009shaping}
\bibfield{author}{\bibinfo{person}{Lauren Shupp}, \bibinfo{person}{Christopher Andrews}, \bibinfo{person}{Margaret Dickey-Kurdziolek}, \bibinfo{person}{Beth Yost}, {and} \bibinfo{person}{Chris North}.} \bibinfo{year}{2009}\natexlab{}.
\newblock \showarticletitle{Shaping the display of the future: The effects of display size and curvature on user performance and insights}.
\newblock \bibinfo{journal}{\emph{Human--Computer Interaction}} \bibinfo{volume}{24}, \bibinfo{number}{1-2} (\bibinfo{year}{2009}), \bibinfo{pages}{230--272}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1080/07370020902739429}
\showDOI{\tempurl}
\bibitem[Siddhpuria et~al\mbox{.}(2018)]%
{siddhpuria2018pointing}
\bibfield{author}{\bibinfo{person}{Shaishav Siddhpuria}, \bibinfo{person}{Sylvain Malacria}, \bibinfo{person}{Mathieu Nancel}, {and} \bibinfo{person}{Edward Lank}.} \bibinfo{year}{2018}\natexlab{}.
\newblock \showarticletitle{Pointing at a Distance with Everyday Smart Devices}. In \bibinfo{booktitle}{\emph{Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems}} (Montreal QC, Canada) \emph{(\bibinfo{series}{CHI '18})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{1–11}.
\newblock
\showISBNx{9781450356206}
\urldef\tempurl%
\url{https://doi.org/10.1145/3173574.3173747}
\showDOI{\tempurl}
\bibitem[Sindhupathiraja et~al\mbox{.}(2024)]%
{sindhupathiraja2024exploring}
\bibfield{author}{\bibinfo{person}{Siddhanth~Raja Sindhupathiraja}, \bibinfo{person}{AKM~Amanat Ullah}, \bibinfo{person}{William Delamare}, {and} \bibinfo{person}{Khalad Hasan}.} \bibinfo{year}{2024}\natexlab{}.
\newblock \showarticletitle{Exploring Bi-Manual Teleportation in Virtual Reality}. In \bibinfo{booktitle}{\emph{2024 IEEE Conference Virtual Reality and 3D User Interfaces (VR)}}. IEEE, \bibinfo{pages}{754--764}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1109/VR58804.2024.00095}
\showDOI{\tempurl}
\bibitem[Slu{\"y}ters et~al\mbox{.}(2023)]%
{sluyters2023consistent}
\bibfield{author}{\bibinfo{person}{Arthur Slu{\"y}ters}, \bibinfo{person}{Quentin Sellier}, \bibinfo{person}{Jean Vanderdonckt}, \bibinfo{person}{Vik Parthiban}, {and} \bibinfo{person}{Pattie Maes}.} \bibinfo{year}{2023}\natexlab{}.
\newblock \showarticletitle{Consistent, Continuous, and Customizable Mid-Air Gesture Interaction for Browsing Multimedia Objects on Large Displays}.
\newblock \bibinfo{journal}{\emph{International Journal of Human–Computer Interaction}} \bibinfo{volume}{39}, \bibinfo{number}{12} (\bibinfo{year}{2023}), \bibinfo{pages}{2492--2523}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1080/10447318.2022.2078464}
\showDOI{\tempurl}
\showeprint{https://doi.org/10.1080/10447318.2022.2078464}
\bibitem[Soukoreff and MacKenzie(2004)]%
{soukoreff2004towards}
\bibfield{author}{\bibinfo{person}{R.~William Soukoreff} {and} \bibinfo{person}{I.~Scott MacKenzie}.} \bibinfo{year}{2004}\natexlab{}.
\newblock \showarticletitle{Towards a standard for pointing device evaluation, perspectives on 27 years of Fitts’ law research in HCI}.
\newblock \bibinfo{journal}{\emph{International Journal of Human-Computer Studies}} \bibinfo{volume}{61}, \bibinfo{number}{6} (\bibinfo{year}{2004}), \bibinfo{pages}{751--789}.
\newblock
\showISSN{1071-5819}
\urldef\tempurl%
\url{https://doi.org/10.1016/j.ijhcs.2004.09.001}
\showDOI{\tempurl}
\newblock
\shownote{Fitts' law 50 years later: applications and contributions from human-computer interaction}.
\bibitem[Sultana and Alam(2024)]%
{sultana2024exploring}
\bibfield{author}{\bibinfo{person}{Samiha Sultana} {and} \bibinfo{person}{Md~Ashraful Alam}.} \bibinfo{year}{2024}\natexlab{}.
\newblock \showarticletitle{Exploring the Impact of Display Field-of-view and Display Type on Text Readability and Visual Text Search on Large Displays in Virtual Reality}. In \bibinfo{booktitle}{\emph{Graphics Interface 2024}}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1145/3670947.3670984}
\showDOI{\tempurl}
\bibitem[Tao et~al\mbox{.}(2021)]%
{tao2021freehand}
\bibfield{author}{\bibinfo{person}{Da Tao}, \bibinfo{person}{Xiaofeng Diao}, \bibinfo{person}{Tieyan Wang}, \bibinfo{person}{Jingya Guo}, {and} \bibinfo{person}{Xingda Qu}.} \bibinfo{year}{2021}\natexlab{}.
\newblock \showarticletitle{Freehand interaction with large displays: Effects of body posture, interaction distance and target size on task performance, perceived usability and workload}.
\newblock \bibinfo{journal}{\emph{Applied Ergonomics}} \bibinfo{volume}{93} (\bibinfo{year}{2021}), \bibinfo{pages}{103370}.
\newblock
\showISSN{0003-6870}
\urldef\tempurl%
\url{https://doi.org/10.1016/j.apergo.2021.103370}
\showDOI{\tempurl}
\bibitem[technologies(2023)]%
{unity}
\bibfield{author}{\bibinfo{person}{Unity technologies}.} \bibinfo{year}{2023}\natexlab{}.
\newblock \bibinfo{title}{Unity Real-Time Development Platform | 2D, 3D, VR \& AR Engine.}
\newblock
\newblock
\newblock
\shownote{Retrieved January 15, 2023 from \url{https://unity.com/}}.
\bibitem[Ullah et~al\mbox{.}(2023a)]%
{Ullah2023LargeCurved}
\bibfield{author}{\bibinfo{person}{AKM~Amanat Ullah}, \bibinfo{person}{William Delamare}, {and} \bibinfo{person}{Khalad Hasan}.} \bibinfo{year}{2023}\natexlab{a}.
\newblock \showarticletitle{Exploring Users Pointing Performance on Large Displays with Different Curvatures in Virtual Reality}.
\newblock \bibinfo{journal}{\emph{IEEE Transactions on Visualization and Computer Graphics}} (\bibinfo{year}{2023}).
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1109/TVCG.2023.3320242}
\showDOI{\tempurl}
\bibitem[Ullah et~al\mbox{.}(2023b)]%
{Ullah2023Pointing}
\bibfield{author}{\bibinfo{person}{A~K M~Amanat Ullah}, \bibinfo{person}{William Delamare}, {and} \bibinfo{person}{Khalad Hasan}.} \bibinfo{year}{2023}\natexlab{b}.
\newblock \showarticletitle{Exploring Users' Pointing Performance on Virtual and Physical Large Curved Displays}. In \bibinfo{booktitle}{\emph{Proceedings of the 29th ACM Symposium on Virtual Reality Software and Technology}} \emph{(\bibinfo{series}{VRST '23})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, Article \bibinfo{articleno}{7}, \bibinfo{numpages}{11}~pages.
\newblock
\showISBNx{9798400703287}
\urldef\tempurl%
\url{https://doi.org/10.1145/3611659.3615710}
\showDOI{\tempurl}
\bibitem[Urakami et~al\mbox{.}(2021)]%
{urakami2021comparing}
\bibfield{author}{\bibinfo{person}{Jacqueline Urakami}, \bibinfo{person}{Henrique Matulis}, \bibinfo{person}{Shio Miyafuji}, \bibinfo{person}{Zhengqing Li}, \bibinfo{person}{Hideki Koike}, {and} \bibinfo{person}{Mark Chignell}.} \bibinfo{year}{2021}\natexlab{}.
\newblock \showarticletitle{Comparing immersiveness and perceptibility of spherical and curved displays}.
\newblock \bibinfo{journal}{\emph{Applied Ergonomics}} \bibinfo{volume}{90} (\bibinfo{year}{2021}), \bibinfo{pages}{103271}.
\newblock
\bibitem[Vogel and Balakrishnan(2005)]%
{VogelUIST2005}
\bibfield{author}{\bibinfo{person}{Daniel Vogel} {and} \bibinfo{person}{Ravin Balakrishnan}.} \bibinfo{year}{2005}\natexlab{}.
\newblock \showarticletitle{Distant Freehand Pointing and Clicking on Very Large, High Resolution Displays}. In \bibinfo{booktitle}{\emph{Proceedings of the 18th Annual ACM Symposium on User Interface Software and Technology}} (Seattle, WA, USA) \emph{(\bibinfo{series}{UIST '05})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{33–42}.
\newblock
\showISBNx{1595932712}
\urldef\tempurl%
\url{https://doi.org/10.1145/1095034.1095041}
\showDOI{\tempurl}
\bibitem[Wei et~al\mbox{.}(2020)]%
{wei2020reading}
\bibfield{author}{\bibinfo{person}{Chunxue Wei}, \bibinfo{person}{Difeng Yu}, {and} \bibinfo{person}{Tilman Dingler}.} \bibinfo{year}{2020}\natexlab{}.
\newblock \showarticletitle{Reading on 3D Surfaces in Virtual Environments}. In \bibinfo{booktitle}{\emph{2020 IEEE Conference on Virtual Reality and 3D User Interfaces (VR)}}. \bibinfo{pages}{721--728}.
\newblock
\urldef\tempurl%
\url{https://doi.org/10.1109/VR46266.2020.00095}
\showDOI{\tempurl}
\bibitem[Wobbrock et~al\mbox{.}(2011)]%
{wobbrock2011effects}
\bibfield{author}{\bibinfo{person}{Jacob~O. Wobbrock}, \bibinfo{person}{Kristen Shinohara}, {and} \bibinfo{person}{Alex Jansen}.} \bibinfo{year}{2011}\natexlab{}.
\newblock \showarticletitle{The Effects of Task Dimensionality, Endpoint Deviation, Throughput Calculation, and Experiment Design on Pointing Measures and Models}. In \bibinfo{booktitle}{\emph{Proceedings of the SIGCHI Conference on Human Factors in Computing Systems}} (Vancouver, BC, Canada) \emph{(\bibinfo{series}{CHI '11})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{1639–1648}.
\newblock
\showISBNx{9781450302289}
\urldef\tempurl%
\url{https://doi.org/10.1145/1978942.1979181}
\showDOI{\tempurl}
\bibitem[Zannoli and Banks(2017)]%
{zannoli2017perceptual}
\bibfield{author}{\bibinfo{person}{Marina Zannoli} {and} \bibinfo{person}{Martin~S Banks}.} \bibinfo{year}{2017}\natexlab{}.
\newblock \showarticletitle{The perceptual consequences of curved screens}.
\newblock \bibinfo{journal}{\emph{ACM Transactions on Applied Perception (TAP)}} \bibinfo{volume}{15}, \bibinfo{number}{1} (\bibinfo{year}{2017}), \bibinfo{pages}{1--16}.
\newblock
\bibitem[Zhai et~al\mbox{.}(2004)]%
{zhai2004speed}
\bibfield{author}{\bibinfo{person}{Shumin Zhai}, \bibinfo{person}{Jing Kong}, {and} \bibinfo{person}{Xiangshi Ren}.} \bibinfo{year}{2004}\natexlab{}.
\newblock \showarticletitle{Speed–accuracy tradeoff in Fitts’ law tasks—on the equivalency of actual and nominal pointing precision}.
\newblock \bibinfo{journal}{\emph{International Journal of Human-Computer Studies}} \bibinfo{volume}{61}, \bibinfo{number}{6} (\bibinfo{year}{2004}), \bibinfo{pages}{823--856}.
\newblock
\showISSN{1071-5819}
\urldef\tempurl%
\url{https://doi.org/10.1016/j.ijhcs.2004.09.007}
\showDOI{\tempurl}
\newblock
\shownote{Fitts' law 50 years later: applications and contributions from human-computer interaction}.
\bibitem[Zhang et~al\mbox{.}(2012)]%
{zhang2012extending}
\bibfield{author}{\bibinfo{person}{Xinyong Zhang}, \bibinfo{person}{Hongbin Zha}, {and} \bibinfo{person}{Wenxin Feng}.} \bibinfo{year}{2012}\natexlab{}.
\newblock \showarticletitle{Extending Fitts' Law to Account for the Effects of Movement Direction on 2d Pointing}. In \bibinfo{booktitle}{\emph{Proceedings of the SIGCHI Conference on Human Factors in Computing Systems}} (Austin, Texas, USA) \emph{(\bibinfo{series}{CHI '12})}. \bibinfo{publisher}{Association for Computing Machinery}, \bibinfo{address}{New York, NY, USA}, \bibinfo{pages}{3185–3194}.
\newblock
\showISBNx{9781450310154}
\urldef\tempurl%
\url{https://doi.org/10.1145/2207676.2208737}
\showDOI{\tempurl}
\end{thebibliography}